From a916244abc81c000257b5acf08bc2751b274f3e8 Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 6 Nov 2026 09:52:50 +0000 Subject: [PATCH 1/2] Sync the archive with Substack again and follow its new cursor paging Substack's archive endpoint began answering HTTP 400 to offset and to a limit above 25, and returns {posts, nextCursor} instead of a list, so the sync had failed since early September. list_posts now pages by cursor (limit=25). The archive gains 36 posts and three titles and subtitles change as Substack has them; companions are rebuilt and match the archive. Tests cover the cursor paging and fail if offset is ever sent again. pytest: 34 passed. build_companions.py --check: PASS. Co-Authored-By: Claude Sonnet 5.5 Claude-Session: https://claude.ai/code/session_011RjaPcC924ePsEE8X6uH72 --- CLAUDE.md | 6 +- ...24-unlearning-to-lead-a-first-step-into.md | 22 +- ...07-good-read-why-economists-dont-listen.md | 12 +- ...07-measurement-and-survey-methods-brief.md | 3 +- ...9-08-research-rundown-daily-7-september.md | 103 ++++ ...9-09-research-rundown-daily-9-september.md | 141 +++++ ...-10-research-rundown-daily-10-september.md | 143 +++++ ...09-11-friday-findings-11-september-2026.md | 93 ++++ ...9-12-good-reads-2-can-india-sustain-its.md | 33 ++ ...ds-3-what-comes-after-industrialisation.md | 82 +++ ...09-12-good-reads-5-can-ai-make-informal.md | 67 +++ ...-09-12-good-reads-6-indias-middle-power.md | 103 ++++ ...-13-research-rundown-daily-13-september.md | 154 ++++++ ...-09-14-methods-monday-14-september-2026.md | 153 ++++++ ...6-09-16-research-rundown-september-2026.md | 516 ++++++++++++++++++ ...-09-17-good-reads-7-what-migration-data.md | 137 +++++ ...-17-research-rundown-daily-16-september.md | 165 ++++++ ...09-18-friday-findings-18-september-2026.md | 154 ++++++ ...-20-research-rundown-daily-19-september.md | 152 ++++++ ...1-good-reads-8-why-are-more-rural-women.md | 113 ++++ ...-09-21-methods-monday-21-september-2026.md | 127 +++++ ...3-good-reads-9-is-the-health-gap-really.md | 161 ++++++ ...-23-research-rundown-daily-23-september.md | 159 ++++++ ...9-25-good-reads-10-when-using-a-service.md | 199 +++++++ ...-09-25-research-rundown-friday-findings.md | 149 +++++ ...7-good-reads-11-what-happens-when-south.md | 207 +++++++ ...-27-research-rundown-daily-26-september.md | 159 ++++++ ...-10-01-good-reads-12-did-mgnrega-narrow.md | 203 +++++++ ...-10-01-methods-monday-28-september-2026.md | 135 +++++ ...-01-research-rundown-daily-29-september.md | 143 +++++ ...01-research-rundown-friday-findings-632.md | 145 +++++ ...-02-research-rundown-daily-27-september.md | 155 ++++++ ...research-rundown-gandhi-jayanti-special.md | 267 +++++++++ ...-07-good-reads-13-when-women-lead-girls.md | 185 +++++++ ...-10-07-research-rundown-daily-3-october.md | 162 ++++++ ...026-10-16-methods-monday-5-october-2026.md | 117 ++++ ...-10-16-research-rundown-daily-6-october.md | 173 ++++++ ...-10-16-research-rundown-daily-7-october.md | 237 ++++++++ ...-17-good-reads-15-who-owns-intelligence.md | 156 ++++++ ...17-research-rundown-friday-findings-c92.md | 253 +++++++++ archive/README.md | 42 +- archive/SECTIONS.md | 1 + companions/praxis.md | 54 ++ companions/reading.md | 230 ++++++++ scripts/sync_substack.py | 26 +- tests/test_sync_substack.py | 41 +- 46 files changed, 6207 insertions(+), 31 deletions(-) create mode 100644 archive/2026-09-08-research-rundown-daily-7-september.md create mode 100644 archive/2026-09-09-research-rundown-daily-9-september.md create mode 100644 archive/2026-09-10-research-rundown-daily-10-september.md create mode 100644 archive/2026-09-11-friday-findings-11-september-2026.md create mode 100644 archive/2026-09-12-good-reads-2-can-india-sustain-its.md create mode 100644 archive/2026-09-12-good-reads-3-what-comes-after-industrialisation.md create mode 100644 archive/2026-09-12-good-reads-5-can-ai-make-informal.md create mode 100644 archive/2026-09-12-good-reads-6-indias-middle-power.md create mode 100644 archive/2026-09-13-research-rundown-daily-13-september.md create mode 100644 archive/2026-09-14-methods-monday-14-september-2026.md create mode 100644 archive/2026-09-16-research-rundown-september-2026.md create mode 100644 archive/2026-09-17-good-reads-7-what-migration-data.md create mode 100644 archive/2026-09-17-research-rundown-daily-16-september.md create mode 100644 archive/2026-09-18-friday-findings-18-september-2026.md create mode 100644 archive/2026-09-20-research-rundown-daily-19-september.md create mode 100644 archive/2026-09-21-good-reads-8-why-are-more-rural-women.md create mode 100644 archive/2026-09-21-methods-monday-21-september-2026.md create mode 100644 archive/2026-09-23-good-reads-9-is-the-health-gap-really.md create mode 100644 archive/2026-09-23-research-rundown-daily-23-september.md create mode 100644 archive/2026-09-25-good-reads-10-when-using-a-service.md create mode 100644 archive/2026-09-25-research-rundown-friday-findings.md create mode 100644 archive/2026-09-27-good-reads-11-what-happens-when-south.md create mode 100644 archive/2026-09-27-research-rundown-daily-26-september.md create mode 100644 archive/2026-10-01-good-reads-12-did-mgnrega-narrow.md create mode 100644 archive/2026-10-01-methods-monday-28-september-2026.md create mode 100644 archive/2026-10-01-research-rundown-daily-29-september.md create mode 100644 archive/2026-10-01-research-rundown-friday-findings-632.md create mode 100644 archive/2026-10-02-research-rundown-daily-27-september.md create mode 100644 archive/2026-10-03-research-rundown-gandhi-jayanti-special.md create mode 100644 archive/2026-10-07-good-reads-13-when-women-lead-girls.md create mode 100644 archive/2026-10-07-research-rundown-daily-3-october.md create mode 100644 archive/2026-10-16-methods-monday-5-october-2026.md create mode 100644 archive/2026-10-16-research-rundown-daily-6-october.md create mode 100644 archive/2026-10-16-research-rundown-daily-7-october.md create mode 100644 archive/2026-10-17-good-reads-15-who-owns-intelligence.md create mode 100644 archive/2026-10-17-research-rundown-friday-findings-c92.md diff --git a/CLAUDE.md b/CLAUDE.md index ab1a039..48301f6 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -85,7 +85,11 @@ Markdown; `archive/` and the generated companions are excluded from all three because their text is not ours to lint. `sync-substack.yml` runs monthly and opens a PR; it never pushes to main. -The sync pages through Substack's archive until a page comes back empty. +The sync follows Substack's archive by cursor until none is returned. On 2026-11-06 +the endpoint answered HTTP 400 to `offset` and to a `limit` above 25 and began +returning `{posts, nextCursor}` instead of a list, which had stopped the sync +since early September (the archive was six weeks behind); it now uses +`limit=25` and `cursor`. Earlier it paged until a page came back empty. Substack returns fewer posts than `limit` on pages that are not the last (23, 26, then 3 overlapping, on 2026-10-19), so stopping at a short page archived only the first 23. The indexes are built from the fetched posts together with diff --git a/archive/2025-08-24-unlearning-to-lead-a-first-step-into.md b/archive/2025-08-24-unlearning-to-lead-a-first-step-into.md index 807f5d7..07f46a8 100644 --- a/archive/2025-08-24-unlearning-to-lead-a-first-step-into.md +++ b/archive/2025-08-24-unlearning-to-lead-a-first-step-into.md @@ -1,20 +1,30 @@ --- -title: "Unlearning to Lead: A First Step into Politics" +title: "Good Reads 4# Unlearning to Lead: A First Step into Politics" date: 2025-08-24 type: other url: https://varna.substack.com/p/unlearning-to-lead-a-first-step-into -subtitle: "" +subtitle: "I have always believed that development is a science of solutions. My entire career has been built on models, data, and the quiet confidence that complex problems can be solved with the right analysis" source: varna.substack.com, fetched by scripts/sync_substack.py --- -# Unlearning to Lead: A First Step into Politics +# Good Reads 4# Unlearning to Lead: A First Step into Politics *2025-08-24. Published at [https://varna.substack.com/p/unlearning-to-lead-a-first-step-into](https://varna.substack.com/p/unlearning-to-lead-a-first-step-into). This file is a generated copy; the Substack post is the original.* -I have always believed that development is a science of solutions. My entire career has been built on models, data, and the quiet confidence that complex problems can be solved with the right analysis. This week, that confidence was gently, and necessarily, dismantled. I walked into the commencement of the Dr. Manmohan Singh Fellowship as a development economist. I walked out as something else entirely—an apprentice in the profound and humbling art of political change. +[![Research Rundown Good Reads cover for “Unlearning to Lead: A First Step into Politics,” with books labelled Question, Listen, Unlearn, Rethink and Act beside an open notebook, map and writing desk, set against a government-building backdrop in the series’ textured cream, teal, rust and ochre editorial style.](https://substackcdn.com/image/fetch/$s_!2Nnm!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fac466fca-8aa8-401d-820c-18b7f0177c89_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!2Nnm!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fac466fca-8aa8-401d-820c-18b7f0177c89_1536x1024.png) + +*A slightly different Good Read this time. Usually, this series looks outward, at an essay, paper or report by someone else. This one looks back at something I wrote a year ago, at the beginning of the Dr. Manmohan Singh Fellowship, about expertise, politics and what it means to unlearn some of the habits we carry into public life.* + +*I am returning to it now because the questions in it still feel unfinished.* + +****** + +This week, that confidence was gently, and necessarily, dismantled. I walked into the commencement of the Dr. Manmohan Singh Fellowship as a development economist. I walked out as something else entirely—an apprentice in the profound and humbling art of political change. The atmosphere was charged not with political rhetoric, but with a sense of historical purpose. Congress President Mallikarjun Kharge framed our journey not as a career shift, but as an answer to a national call. He spoke of an India grappling with extreme inequality and unemployment, and he drew a direct line from the lawyers, economists, and doctors who built this nation to us, the fifty professionals in that room. He reminded us that the most enduring legacies, like those forged under Dr. Manmohan Singh’s leadership, are built on substance, not spectacle—on working more and talking less. For a moment, I saw my econometric models not as solutions, but as mere tools in a much larger, more human toolkit. +[Subscribe now](https://varna.substack.com/subscribe?) + Then, the philosophers of our new journey took the stage. Dr. Salman Khurshid, with the grace of a man who has navigated the highest corridors of power, offered our first and most difficult lesson: the courage to unlearn. He gave us permission to let go of the arrogance of expertise, to shed the illusion that the clean logic of my spreadsheets could ever fully capture the beautiful chaos of society. His most liberating admission was that in the realm of global foreign policy today, “nobody quite understands what the hell is going on.” It was a stunning declaration that the old rulebooks are obsolete. Our task is not to have all the answers, but to help write new questions. This intellectual humility was perfectly complemented by Sachin Rao’s structural clarity. He asked us to become “more political,” and in doing so, reframed politics itself for me. It is not a game of elections, but the very architecture of society—a force that shapes our common sense, a concept that would make Gramsci nod in recognition. As an economist, I think in systems and structures, but this was a reminder that the most important structures are made of power, relationships, and ideas, and that changing them requires more than a perfect policy prescription. @@ -25,6 +35,4 @@ Sitting there as one of the only two economists in the cohort, I felt a familiar The commencement is over. The real education has just begun. ---- - -Subscribe for more dispatches from the front lines of learning, unlearning, and the messy, magnificent work of building a political future. +[Share](https://varna.substack.com/p/unlearning-to-lead-a-first-step-into?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-07-good-read-why-economists-dont-listen.md b/archive/2026-09-07-good-read-why-economists-dont-listen.md index 6ea74df..908fde5 100644 --- a/archive/2026-09-07-good-read-why-economists-dont-listen.md +++ b/archive/2026-09-07-good-read-why-economists-dont-listen.md @@ -1,5 +1,5 @@ --- -title: "Good Read - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic" +title: "Good Reads #1 - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic" date: 2026-09-07 type: other url: https://varna.substack.com/p/good-read-why-economists-dont-listen @@ -7,18 +7,18 @@ subtitle: "Vijayendra Rao | In Development" source: varna.substack.com, fetched by scripts/sync_substack.py --- -# Good Read - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic +# Good Reads #1 - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic *2026-09-07. Published at [https://varna.substack.com/p/good-read-why-economists-dont-listen](https://varna.substack.com/p/good-read-why-economists-dont-listen). This file is a generated copy; the Substack post is the original.* -[![image](https://substackcdn.com/image/fetch/$s_!47Bx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe446fa09-6683-4505-87a3-e08ac76f065e_1672x941.png)](https://substackcdn.com/image/fetch/$s_!47Bx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe446fa09-6683-4505-87a3-e08ac76f065e_1672x941.png) - -For our new series Good Reads, I’d like to begin with this beautiful article: +For our new series Good Reads, I’d like to begin with this beautiful article:[![Research Rundown Good Reads cover for Vijayendra Rao’s “Why Economists Don’t Listen,” with books labelled People, Experience, Evidence and Better Questions alongside field notes, charts and research materials in a textured cream, teal, rust and ochre editorial style.](https://substackcdn.com/image/fetch/$s_!fNOu!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fddda19d5-d302-4c75-812b-bc59599506bc_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!fNOu!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fddda19d5-d302-4c75-812b-bc59599506bc_1536x1024.png) Link: [indevelopmentmag.com/why-economists-dont-listen/](https://indevelopmentmag.com/why-economists-dont-listen/?utm_source=chatgpt.com) Rao begins with a failure of measurement. In a Karnataka village in 1993, he was administering a conventional household survey when a woman was dragged away by her husband for failing to cook lunch. The questionnaire had no place to record what he had just witnessed. A week of staying in the village and talking with women eventually produced evidence about domestic violence that the original instrument had been incapable of seeing. Rao and his colleagues rewrote the survey, and the episode became the starting point for later mixed-methods work on dowry violence and domestic abuse. ([In Development](https://indevelopmentmag.com/why-economists-dont-listen/)) +[Subscribe now](https://varna.substack.com/subscribe?) + The essay develops that episode into a larger criticism of development economics. Rao argues that the discipline’s pursuit of cleaner identification has increased its distance from the people it studies. Surveys and experiments can estimate effects with considerable precision, while the mechanisms, meanings and experiences behind those effects often remain thinly observed. His alternative is methodological pluralism grounded in field engagement: open-ended interviews, ethnography, narrative data, attention to process, and greater involvement of respondents in deciding what researchers should ask. ([In Development](https://indevelopmentmag.com/why-economists-dont-listen/)) The most persuasive parts are the concrete examples. Rao describes a participatory study in which tribal communities designed their own questions and produced a 32,000-household census. Only 17 per cent of those questions overlapped with the standard National Sample Survey. He also revisits a randomized trial of participatory planning in Karnataka where the headline result was null. Embedded ethnography showed that implementation quality, administrative transfers and caste inequality had shaped the outcome in ways the treatment effect alone could not explain. ([In Development](https://indevelopmentmag.com/why-economists-dont-listen/)) @@ -26,3 +26,5 @@ The most persuasive parts are the concrete examples. Rao describes a participato I am less persuaded by the broad disciplinary diagnosis. Development economics is more varied than the essay sometimes allows, and qualitative work can suffer from its own problems of inference, researcher positionality and selective interpretation. Rao’s stronger claim is methodological: the research question should determine the evidence we collect, and some questions become distorted when the instrument defines the world before researchers have spent enough time understanding it. The final section on AI is particularly worth reading. Rao sees real promise in using language models to code qualitative material at scale, while warning that models trained poorly on Bengali or Rohingya text can generate biases that are harder to inspect than errors in a conventional survey. His rule is sensible: computational tools can extend the researcher’s reach; they cannot replace familiarity with people, transcripts and context. ([In Development](https://indevelopmentmag.com/why-economists-dont-listen/)) + +[Subscribe now](https://varna.substack.com/subscribe?) diff --git a/archive/2026-09-07-measurement-and-survey-methods-brief.md b/archive/2026-09-07-measurement-and-survey-methods-brief.md index 164a24a..b0865be 100644 --- a/archive/2026-09-07-measurement-and-survey-methods-brief.md +++ b/archive/2026-09-07-measurement-and-survey-methods-brief.md @@ -11,7 +11,8 @@ source: varna.substack.com, fetched by scripts/sync_substack.py *2026-09-07. Published at [https://varna.substack.com/p/measurement-and-survey-methods-brief](https://varna.substack.com/p/measurement-and-survey-methods-brief). This file is a generated copy; the Substack post is the original.* -[![image](https://substackcdn.com/image/fetch/$s_!taVv!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F233dfb53-912f-4b22-8bbe-12c9dac37f04_1491x1055.png)](https://substackcdn.com/image/fetch/$s_!taVv!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F233dfb53-912f-4b22-8bbe-12c9dac37f04_1491x1055.png) +[![Research Rundown DAILY cover for 8 September 2026, featuring layered South Asian maps, food-safety records, social-protection data, literacy figures and disaster-reconstruction estimates in a textured cream, teal, rust and ochre editorial collage.](https://substackcdn.com/image/fetch/$s_!8Per!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa37ef200-ad4f-4df5-8a48-a127f8e6fd3c_1491x1055.png)](https://substackcdn.com/image/fetch/$s_!8Per!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa37ef200-ad4f-4df5-8a48-a127f8e6fd3c_1491x1055.png) +*Methods Monday - 7th September 2026* ### 1. Meyer et al. — *The Anatomy and Evolution of Survey Error* diff --git a/archive/2026-09-08-research-rundown-daily-7-september.md b/archive/2026-09-08-research-rundown-daily-7-september.md new file mode 100644 index 0000000..6c8f736 --- /dev/null +++ b/archive/2026-09-08-research-rundown-daily-7-september.md @@ -0,0 +1,103 @@ +--- +title: "Research Rundown DAILY - 8 September 2026" +date: 2026-09-08 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-7-september +subtitle: "Today’s strongest signal is regional: climate risk is becoming harder to separate from public health, infrastructure, labour and fiscal policy across South Asia" +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 8 September 2026 + +*2026-09-08. Published at [https://varna.substack.com/p/research-rundown-daily-7-september](https://varna.substack.com/p/research-rundown-daily-7-september). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 8 September 2026, featuring layered South Asian maps, food-safety records, social-protection data, literacy figures and disaster-reconstruction estimates in a textured cream, teal, rust and ochre editorial collage.](https://substackcdn.com/image/fetch/$s_!4p6X!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcd3f9aac-ddac-4634-b8cd-91b006b1604f_1672x941.png)](https://substackcdn.com/image/fetch/$s_!4p6X!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcd3f9aac-ddac-4634-b8cd-91b006b1604f_1672x941.png) +*The Research Rundown 8th September 2026* + +## FIVE THINGS WORTH KNOWING + +### 1. Nepal’s disaster is also an infrastructure-risk story + +Nepal marked a national day of mourning today as rescue operations continued after the catastrophic 26 August flood and glacier-collapse disaster. Reuters reports at least 1,380 deaths and more than 5,500 people still missing across Nepal and Tibet, with hundreds of hydropower workers among those unaccounted for. + +The immediate humanitarian toll dominates, as it should. There is also a harder development question here. Hydropower, roads, settlements and border infrastructure have expanded into mountain systems where cryosphere and cascading-hazard risks are changing. The event should prompt scrutiny of what assumptions about hazard frequency and magnitude sit inside infrastructure appraisal. + +Why it matters: adaptation cannot remain an environmental add-on to infrastructure policy when the infrastructure itself changes exposure. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. Bangladesh’s dengue outbreak is accelerating + +Bangladesh has recorded 42,590 dengue hospitalisations and 117 deaths so far in 2026, according to figures reported by Reuters. The latest 24-hour reporting period produced 1,558 admissions, the highest daily number this year, while hospitals in Dhaka and elsewhere are facing growing pressure. + +The trajectory matters as much as the cumulative count. Weather remains favourable for *Aedes* transmission, and the health system is simultaneously managing a serious measles outbreak. Fogging and generic clean-up campaigns are poor substitutes for surveillance-led vector control when transmission is already widespread. + +Why it matters: Bangladesh is entering the part of the dengue season where hospital capacity, case management and geographically targeted vector control need to be read together. + +### 3. South Asia may have an unusually strong economic case for clean-air investment + +A new assessment from the UN Environment Programme and Climate and Clean Air Coalition estimates that every US$1 spent globally on integrated climate and clean-air measures could generate around US$15 in economic benefits. The South Asian result is even larger: the assessment estimates the region’s existing air-pollution burden at roughly 10 per cent of GDP and a benefit-cost ratio of about 21:1 for the package it models. + +Those are modelled estimates, rather than observed programme returns, and should be read accordingly. Still, the exercise is useful because it puts mortality, morbidity, productivity and climate effects into the same economic accounting rather than treating air pollution as a narrow environmental externality. + +Why it matters: for South Asian finance ministries, clean-air policy increasingly looks like economic policy with measurable health returns. + +### 4. Heat and wildfire smoke are complicating the air-quality story + +The World Meteorological Organization’s new *Air Quality and Climate Bulletin* argues that heatwaves and wildfires can reverse gains from reductions in conventional industrial and transport pollution. The report focuses on PM2.5, ground-level ozone, black carbon and other aerosols, and calls for much better atmospheric observation. Northern India appears explicitly among the regions where local and regional pollution processes interact with climate. + +This matters for measurement. Annual city averages can conceal changing pollutant composition, episodic exposure and regional transport. A monitoring system designed mainly around compliance with fixed concentration thresholds may tell us too little about where exposure is coming from. + +Why it matters: climate change is altering the production and movement of air pollution, which changes what governments need to measure as well as what they need to regulate. + +### 5. India’s factory slowdown has reached employment + +India’s manufacturing PMI fell to 52.8 in August, its weakest reading in five years, alongside slower output and new orders. More consequentially for the labour story, surveyed manufacturers reported falling employment for the first time in roughly two and a half years. + +A diffusion index is a high-frequency business indicator, not a substitute for PLFS employment estimates, and month-to-month movement deserves restraint. The employment component is still worth watching because it sits beside a wider question about whether industrial expansion is producing enough jobs. + +Why it matters: if subsequent labour and industrial data confirm the signal, the slowdown has moved beyond production momentum into employment. + +## ONE PAPER WORTH READING + +### Hidden Assets: The Economic and Health Case for Climate and Clean Air Action + +The most useful new research product today is UNEP and the Climate and Clean Air Coalition’s global economic assessment of joint climate and air-pollution policy. It asks a practical question: what changes when the economic appraisal counts climate mitigation and clean-air benefits together? + +The assessment models 25 measures across energy, industry, transport, agriculture and food systems, household energy and waste. It incorporates implementation costs alongside market gains and health-related welfare benefits. Its headline estimate is striking: integrated action produces roughly US$15 in benefits for every US$1 invested globally, with market benefits alone exceeding costs within ten years. For South Asia, the estimated return rises to about 21:1. + +The caution is substantial. These are model-derived social returns across a package of interventions. They are sensitive to assumptions about valuation of mortality and morbidity, implementation speed, baseline emissions and future policy. A benefit-cost ratio of 21:1 should therefore be read as an estimate from a defined modelling framework, rather than a forecast of fiscal returns available to any individual government programme. + +The report still does something useful. It makes the opportunity cost of treating air quality, health and climate as separate administrative problems empirically visible. + +**Citation:** United Nations Environment Programme and Climate and Clean Air Coalition. 2026. *Hidden Assets: The Economic and Health Case for Climate and Clean Air Action*. Nairobi: UNEP. + +## ONE NUMBER + +### 10% + +That is the estimated current economic burden of air pollution in South Asia as a share of regional GDP in the new UNEP-CCAC assessment. It is the highest burden among the regions assessed. + +The number includes monetised health and wider economic damages, so it should not be interpreted as ten percentage points of measured GDP disappearing from national accounts. Its value is comparative: once health losses are priced, dirty air is large enough to sit alongside the macroeconomic variables that routinely receive far more policy attention. + +## POLICY WATCH + +### India gives employers another route to settle old provident-fund disputes + +EPFO is urging eligible establishments to use VISHWAS 2026 to settle long-running disputes over damages for delayed provident-fund remittances at reduced rates. The scheme covers categories including pending litigation, unpaid final orders and some pre-adjudication cases. + +The policy question is worth separating from the administrative one. Lower penalties can clear litigation and recover outstanding dues faster. They can also weaken deterrence if repeated amnesties teach employers that delayed compliance will eventually become cheaper. The test is therefore what happens to recovery times, fresh defaults and workers’ credited contributions after the scheme, rather than the number of cases formally closed. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## EDITORIAL NOTE + +### Climate adaptation needs a different denominator + +The Nepal disaster exposes a recurring weakness in adaptation debates. We count projects protected, warnings issued, kilometres of embankment built and money allocated. Those measures say little about whether the underlying distribution of risk has changed. + +Infrastructure creates its own exposure. A hydropower project places workers, tunnels, access roads, transmission assets and settlements inside a hazard system. A new road changes where people build. Urban drainage alters land values and settlement patterns. Adaptation appraisal therefore needs to ask how development itself changes the population and assets exposed to a hazard over the lifetime of an investment. + +This also complicates attribution. The immediate physical cause of an individual disaster can remain uncertain while the policy conclusion is already clear: historical hazard distributions are becoming less reliable guides to future risk in a warming mountain system. + +For South Asia, that is a planning problem before it is a climate-finance slogan. diff --git a/archive/2026-09-09-research-rundown-daily-9-september.md b/archive/2026-09-09-research-rundown-daily-9-september.md new file mode 100644 index 0000000..1d69029 --- /dev/null +++ b/archive/2026-09-09-research-rundown-daily-9-september.md @@ -0,0 +1,141 @@ +--- +title: "Research Rundown DAILY 9 September 2026" +date: 2026-09-09 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-9-september +subtitle: "Today’s strongest signal is fiscal: across South Asia, governments are being forced to decide who pays for vulnerability, from Nepal’s multibillion-dollar reconstruction bill to Sri Lanka’s attempt to" +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY 9 September 2026 + +*2026-09-09. Published at [https://varna.substack.com/p/research-rundown-daily-9-september](https://varna.substack.com/p/research-rundown-daily-9-september). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 9 September 2026, showing layered South Asian research records on food safety, literacy, social protection and disaster reconstruction around a circular sampling frame, in a textured cream, teal and rust editorial collage.](https://substackcdn.com/image/fetch/$s_!A6yZ!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F245324c0-8412-4b94-a8ce-b8f40af0384b_1672x941.png)](https://substackcdn.com/image/fetch/$s_!A6yZ!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F245324c0-8412-4b94-a8ce-b8f40af0384b_1672x941.png) +*The Research Rundown Daily - 9th September 2026* + +## five things worth knowing + +### 1. Nepal’s flood recovery bill is moving into the billions + +Nepal’s National Planning Commission has begun a Post-Disaster Needs Assessment after the 26 August floods, with officials expecting final losses to exceed the initial NPR 400 billion estimate by around 30 per cent. The government is preparing for a donor conference and considering a claim to the Loss and Damage Fund, while acknowledging that concessional borrowing may still be necessary. ([Kathmandu Post](https://kathmandupost.com/national/2026/09/08/nepal-looks-for-grants-for-flood-recovery-but-loans-may-be-unavoidable?utm_source=chatgpt.com)) + +This is a substantive development since yesterday’s edition. The disaster is moving from emergency response into a fiscal question, with an estimated US$4 billion to US$5 billion financing requirement and a government already wary of adding to public debt. + +Why it matters: climate loss is becoming sovereign-debt exposure in real time. + +[Kathmandu Post: Nepal looks for grants for flood recovery](https://kathmandupost.com/national/2026/09/08/nepal-looks-for-grants-for-flood-recovery-but-loans-may-be-unavoidable?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. Sri Lanka is putting livelihoods inside its social-protection reform + +Sri Lanka’s cabinet has approved a US$100 million ADB-financed programme covering six districts from November 2026 to October 2031. Among its provisions, 30,000 selected Aswesuma households will receive LKR 200,000 in seed capital, alongside entrepreneurship support and market and financial linkages; the programme also includes care-worker regulation, elderly day-care centres and a national Social Protection Plan. ([EconomyNext](https://economynext.com/sri-lanka-to-roll-out-100mn-adb-funded-social-protection-program-283016/?utm_source=chatgpt.com)) + +The design is worth watching because it tries to connect income support with livelihoods and care services. Whether the seed-capital component produces durable income gains will depend heavily on beneficiary selection, local markets and the quality of accompanying support. + +Why it matters: Sri Lanka is testing a broader conception of social protection during its post-crisis recovery. + +[EconomyNext: Sri Lanka’s US$100 million social-protection programme](https://economynext.com/sri-lanka-to-roll-out-100mn-adb-funded-social-protection-program-283016/?utm_source=chatgpt.com) + +### 3. India’s food-safety problem is becoming harder to dismiss as sporadic enforcement failure + +Government data reviewed by Reuters show that roughly one in five food samples tested annually since 2020 has been classified as unsafe or substandard. Nearly 8,000 food licences have been revoked over the past three years, while regulators have intensified inspections and closures. ([Reuters](https://www.reuters.com/legal/litigation/raids-shutdowns-warning-labels-indias-unprecedented-food-safety-push-2026-09-08/?utm_source=chatgpt.com)) + +The denominator needs care. Regulatory samples are unlikely to constitute a representative probability sample of everything Indians eat, so the 20 per cent figure cannot be interpreted as the prevalence of unsafe food in the market. It does, however, indicate a substantial enforcement problem within the products and establishments reaching regulators. + +Why it matters: India needs better surveillance data alongside tougher enforcement if it wants to distinguish systemic risk from targeted detection. + +[Reuters: inside India’s food-safety crackdown](https://www.reuters.com/world/india/inside-indias-food-safety-crackdown-2026-09-08/?utm_source=chatgpt.com) + +### 4. India’s nutrition programme is getting more specific about what an Anganwadi meal should contain + +The ninth Rashtriya Poshan Maah begins today with dietary diversity and adequate protein among its priorities. More consequential than the campaign itself are revised Supplementary Nutrition Programme guidelines that formally define hot cooked meals, require menu boards and introduce Anganwadi Management Committees. ([Akashvani News](https://newsonair.gov.in/9th-rashtriya-poshan-maah-to-be-launched-in-varanasi-on-wednesday/?utm_source=chatgpt.com)) + +These are delivery rules rather than evidence of better nutrition. Their value will depend on food procurement, menu adherence, meal quality and consumption by children. + +Why it matters: the policy is shifting some attention from enrolment and activity counts toward the content and accountability of nutrition services. + +[Official PIB release on Rashtriya Poshan Maah 2026](https://pib.gov.in/PressReleaseIframePage.aspx?PRID=2307291&utm_source=chatgpt.com) + +### 5. Bangladesh reports a 77.9 per cent literacy rate, but the measure tells us less than it appears to + +Bangladesh’s Bureau of Non-Formal Education marked International Literacy Day on 8 September with a government estimate that 77.9 per cent of people aged seven and above are literate. ([BSS](https://www.bssnews.net/news/national/422378?utm_source=chatgpt.com)) + +The statistic is useful as an access indicator, with a familiar limitation. Literacy classifications do not establish functional reading proficiency, comprehension or learning quality. South Asia’s education problem increasingly lies in the distance between schooling, nominal literacy and acquired skills. + +Why it matters: education monitoring needs measures of what people can actually do with reading and numeracy, alongside participation indicators. + +[Bangladesh Sangbad Sangstha report](https://www.bssnews.net/news/national/422378?utm_source=chatgpt.com) + +## one paper worth reading + +### A useful new paper is harder to find today + +Rather than promote an old working paper because it has been newly indexed, I would leave this section deliberately sparse. + +One apparently new September listing that surfaced in the search, Sampreet Singh Goraya’s *How does Caste Affect Entrepreneurship? Birth vs Worth*, illustrates the problem. The Barcelona School of Economics page currently labels the working paper August 2026, but the downloadable paper itself is dated July 2019. Its empirical analysis uses the 2005 Economic Census and the 2001 and 2006 MSME censuses. ([BSE](https://bse.eu/research/working-papers/how-does-caste-affect-entrepreneurship-birth-vs-worth)) + +The paper remains interesting. It estimates that Scheduled Caste and Scheduled Tribe entrepreneurs owned 13 per cent of non-agricultural establishments despite representing 24 per cent of the population, and uses a calibrated entrepreneurship model to estimate output losses from caste-specific entry and financing constraints. + +It is not new research. Treating repository dates or refreshed publication pages as publication dates would quietly degrade the purpose of this section. + +I would rather publish no paper today than manufacture novelty. + +## one number + +### 20% + +Approximately one in five food samples tested annually by Indian regulators since 2020 has been found unsafe or substandard, according to government data reviewed by Reuters. ([Reuters](https://www.reuters.com/legal/litigation/raids-shutdowns-warning-labels-indias-unprecedented-food-safety-push-2026-09-08/?utm_source=chatgpt.com)) + +There is an important sampling qualification. Food-safety testing is partly risk-directed, so this is not an estimate that 20 per cent of food sold in India is unsafe. The statistic describes the yield among samples regulators actually tested. + +That distinction is exactly why India would benefit from publishing clearer sampling frames and separating surveillance sampling from complaint-driven and enforcement testing. + +[Reuters investigation](https://www.reuters.com/world/india/inside-indias-food-safety-crackdown-2026-09-08/?utm_source=chatgpt.com) + +## policy watch + +### Sri Lanka approves a US$100 million social-protection and livelihoods programme + +Sri Lanka’s cabinet has approved implementation of the Macro-Recovery Sector Development Program for Promoting Resilience and Inclusive Economic Opportunities, financed through a US$100 million concessional ADB loan. It will operate in Kegalle, Matale, Mullaitivu, Kilinochchi, Batticaloa and Nuwara Eliya between November 2026 and October 2031. ([EconomyNext](https://economynext.com/sri-lanka-to-roll-out-100mn-adb-funded-social-protection-program-283016/?utm_source=chatgpt.com)) + +The most interesting design choice is the proposed LKR 200,000 seed-capital grant for 30,000 selected Aswesuma households. Cash transfers and enterprise support solve different problems. Combining them can make sense when households face capital constraints, although enterprise grants have uneven returns and cannot substitute for adequate consumption protection. + +The programme therefore creates a useful evaluation opportunity. Household income, enterprise survival, consumption stability and graduation from Aswesuma should be tracked separately. A fall in benefit receipt by itself would tell us very little about whether livelihoods improved. + +[Programme details reported after the Sri Lankan cabinet decision](https://economynext.com/sri-lanka-to-roll-out-100mn-adb-funded-social-protection-program-283016/?utm_source=chatgpt.com) + +## editorial note + +### The denominator problem + +Two numbers in today’s edition illustrate a recurring problem in development statistics. + +India’s 20 per cent food-sample failure rate sounds like a prevalence estimate. It is not, unless the tested samples represent the food supply. Bangladesh’s 77.9 per cent literacy rate sounds like a measure of learning. It is not, unless the classification captures meaningful reading ability. + +Neither statistic is useless. Both become misleading when the denominator or construct disappears in retelling. + +This matters because administrative systems are producing vastly more numbers than they once did. Publication has become easier. Interpretation has not. + +A useful discipline for reading development data is therefore remarkably old-fashioned: ask who entered the denominator, how they got there, what exactly was measured, and what population the resulting estimate can describe. Many apparently large findings shrink considerably once those four questions are answered. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## sources + +[Kathmandu Post, 8 September: Nepal reconstruction financing](https://kathmandupost.com/national/2026/09/08/nepal-looks-for-grants-for-flood-recovery-but-loans-may-be-unavoidable?utm_source=chatgpt.com) + +[EconomyNext, 8 September: Sri Lanka social-protection programme](https://economynext.com/sri-lanka-to-roll-out-100mn-adb-funded-social-protection-program-283016/?utm_source=chatgpt.com) + +[Reuters, 8 September: India food-safety investigation](https://www.reuters.com/world/india/inside-indias-food-safety-crackdown-2026-09-08/?utm_source=chatgpt.com) + +[Press Information Bureau: Rashtriya Poshan Maah 2026](https://pib.gov.in/PressReleaseIframePage.aspx?PRID=2307291&utm_source=chatgpt.com) + +[Bangladesh Sangbad Sangstha, 8 September: literacy data](https://www.bssnews.net/news/national/422378?utm_source=chatgpt.com) + +[Barcelona School of Economics: Goraya working-paper listing](https://bse.eu/research/working-papers/how-does-caste-affect-entrepreneurship-birth-vs-worth) + +[Original Goraya working paper PDF](https://bw.bse.eu/wp-content/uploads/2019/07/1104-file.pdf) + +## diff --git a/archive/2026-09-10-research-rundown-daily-10-september.md b/archive/2026-09-10-research-rundown-daily-10-september.md new file mode 100644 index 0000000..e325f7e --- /dev/null +++ b/archive/2026-09-10-research-rundown-daily-10-september.md @@ -0,0 +1,143 @@ +--- +title: "Research Rundown DAILY - 10 September 2026 " +date: 2026-09-10 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-10-september +subtitle: "Nepal’s NPR 723 billion post-flood recovery bill puts the cost of climate-resilient reconstruction in sharp focus." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 10 September 2026 + +*2026-09-10. Published at [https://varna.substack.com/p/research-rundown-daily-10-september](https://varna.substack.com/p/research-rundown-daily-10-september). This file is a generated copy; the Substack post is the original.* + +[![Alt text: Research Rundown DAILY cover for 10 September 2026, showing a reconstruction assessment ledger, river-risk map, bridge sketch and planning tools on textured cream paper, reflecting the issue’s focus on climate-resilient rebuilding after Nepal’s floods.](https://substackcdn.com/image/fetch/$s_!g4bm!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb88b33d-e815-4dae-a905-9488b6663026_1672x941.png)](https://substackcdn.com/image/fetch/$s_!g4bm!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbb88b33d-e815-4dae-a905-9488b6663026_1672x941.png) +*Research Rundown Daily - 10th September 2026* + +## five things worth knowing + +### 1. Nepal’s reconstruction estimate has jumped to NPR 723.31 billion + +Nepal’s Rapid Damage and Needs Assessment estimates that recovery from the 26 August Bhotekoshi floods will require NPR 723.31 billion. Infrastructure accounts for NPR 473.57 billion, including NPR 390.62 billion for energy and electricity systems. Thirteen hydropower projects with 759 MW of capacity were affected, alongside 69 kilometres of roads and 33 motorable bridges that were washed away. ([The Kathmandu Post](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com)) + +The figures matter because they are more detailed than the preliminary estimates available earlier this week. They also separate immediate needs, NPR 8.73 billion over six months, from NPR 714.58 billion in longer-term reconstruction. ([The Kathmandu Post](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com)) + +Why it matters: the question now moves from estimating losses to deciding whether Nepal rebuilds the assets it lost or changes where and how infrastructure is built. + +[Kathmandu Post: Rapid Damage and Needs Assessment](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. Nepal has made an early test case of the new loss-and-damage architecture + +Nepal has sought an initial US$20 million from the Fund for Responding to Loss and Damage following the floods. The fund’s co-chairs have told Kathmandu that they are working with its board to consider Nepal’s request for urgent assistance. ([The Kathmandu Post](https://kathmandupost.com/national/2026/09/10/climate-loss-and-damage-fund-moves-to-consider-urgent-support-for-nepal?utm_source=chatgpt.com)) + +The amount is small beside the NPR 723.31 billion recovery estimate. Its significance lies elsewhere. Loss-and-damage finance was created precisely for climate-vulnerable countries facing harms that adaptation cannot fully prevent, but converting that principle into rapid post-disaster finance remains largely untested. + +Why it matters: Nepal could provide an early real-world test of how quickly the international loss-and-damage system can move from attribution and eligibility arguments to usable finance. + +[Kathmandu Post: Fund for Responding to Loss and Damage considers Nepal request](https://kathmandupost.com/national/2026/09/10/climate-loss-and-damage-fund-moves-to-consider-urgent-support-for-nepal?utm_source=chatgpt.com) + +### 3. South-East Asia has adopted an unusually concrete medical-equipment policy + +WHO South-East Asia Member States today adopted the region’s first dedicated resolution on medical devices, diagnostics and assistive technologies. Governments have committed to designate national lead mechanisms by 2027, complete baseline assessments by 2028 and develop or update costed national implementation plans by 2030. ([World Health Organization](https://www.who.int/southeastasia/news/detail/10-09-2026-who-south-east-asia-region-to-improve-access-to-medical-devices--in-vitro-diagnostics-and-assistive-technologies?utm_source=chatgpt.com)) + +The interesting part is what countries will measure. The proposed monitoring system includes whether priority equipment is functional, preventive maintenance, downtime, consumable stock-outs and geographic inequalities in access. Procurement is also expected to account for installation, maintenance, training and total ownership costs. ([World Health Organization](https://www.who.int/southeastasia/news/detail/10-09-2026-who-south-east-asia-region-to-improve-access-to-medical-devices--in-vitro-diagnostics-and-assistive-technologies?utm_source=chatgpt.com)) + +Why it matters: health systems routinely count machines purchased. This framework asks whether those machines can actually deliver a service. + +[WHO South-East Asia: medical devices resolution](https://www.who.int/southeastasia/news/detail/10-09-2026-who-south-east-asia-region-to-improve-access-to-medical-devices--in-vitro-diagnostics-and-assistive-technologies?utm_source=chatgpt.com) + +### 4. Pakistan received US$3.66 billion in remittances in August + +Workers’ remittances reached US$3.656 billion in August, 16.5 per cent above August 2025, according to State Bank of Pakistan data reported on 10 September. Inflows during July and August totalled US$7.3 billion, up 14.7 per cent year-on-year. ([Dawn](https://www.dawn.com/news/2028756/remittances-rise-165pc-in-august?utm_source=chatgpt.com)) + +The increase needs interpretation. Pakistani businesspeople in Gulf states told *Dawn* that some of the rise may reflect investors liquidating assets and repatriating money amid regional uncertainty. If so, part of the recent increase is different from a sustained rise in migrant labour earnings. ([Dawn](https://www.dawn.com/news/2028756/remittances-rise-165pc-in-august?utm_source=chatgpt.com)) + +Why it matters: remittances support household consumption and Pakistan’s external account, but identical balance-of-payments entries can arise from quite different household and investor behaviour. + +### 5. India’s food-label fight has moved materially + +India’s Food Safety and Standards Authority told the Supreme Court today that it is open to introducing stronger front-of-pack warning labels in a single phase. The proposed system would place a red warning symbol on products exceeding the threshold for any one of added sugar, salt or saturated fat, rather than requiring excess levels in two categories. ([Reuters](https://www.reuters.com/legal/litigation/india-food-regulator-says-it-is-open-stricter-labelling-after-court-scrutinizes-2026-09-10/?utm_source=chatgpt.com)) + +A second measurement dispute concerns the denominator. Food companies are pressing for thresholds based on serving sizes, while FSSAI’s proposal uses a standard 100 gram or 100 ml reference. ([The Economic Times](https://m.economictimes.com/industry/cons-products/food/front-of-pack-labels-food-inc-bats-for-per-serving-benchmark-in-sc-not-fssais-per-100-gm-plan/articleshow/133981453.cms?utm_source=chatgpt.com)) + +Why it matters: “per serving” sounds intuitive, but serving sizes can vary across products and manufacturers. The denominator can determine whether the same product receives a warning. + +[Reuters: FSSAI signals stricter warning labels](https://www.reuters.com/legal/litigation/india-food-regulator-says-it-is-open-stricter-labelling-after-court-scrutinizes-2026-09-10/?utm_source=chatgpt.com) + +## one paper worth reading + +### When inequality becomes political conflict + +A new *World Development* paper asks whether changes in local economic inequality cause social unrest in rural India. The authors combine high-frequency bank-account information that tracks village-level inequality over time with data on unrest, and develop a model in which conflict emerges when local elites capture unequal shares of economic surplus. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +Their identification strategy exploits rainfall-driven water shocks that disproportionately increase elite incomes. The headline estimate is substantial: a 10 per cent increase in their inequality measure raises the monthly probability of unrest by about 6.5 per cent. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +The paper is unusually interesting because it moves below state and district averages and observes changing economic distributions within villages. The causal claim still depends on the exclusion restriction: rainfall-induced water shocks must affect unrest through the inequality mechanism being estimated rather than through other channels. In rural India, rainfall can alter employment, agricultural output, prices and local political relations, so that assumption deserves close attention. + +The broader result matters. Distribution within a locality may affect political stability independently of average income. This is a paper I would flag for fuller treatment in a future flagship Research Rundown. + +**Citation:** “Inequality and social unrest in India.” 2026. *World Development* 205: 107418. DOI: 10.1016/j.worlddev.2026.107418. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +[Open-access article and abstract](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com) + +## one number + +### 65.5% + +That is the share of Nepal’s NPR 723.31 billion recovery requirement accounted for by infrastructure, calculated from the government assessment’s NPR 473.57 billion infrastructure estimate. ([The Kathmandu Post](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com)) + +The composition matters more than the percentage itself. Energy and electricity account for NPR 390.62 billion. Transport recovery requires another NPR 73.34 billion. The disaster therefore hit assets with long operating lives and large sunk costs. + +A reconstruction budget that restores these assets to their previous specifications could reproduce part of the exposure that generated the bill. + +## policy watch + +### India’s food-label rules may become substantially stricter + +The Supreme Court’s scrutiny of India’s proposed front-of-pack nutrition labelling regime produced a significant regulatory shift today. FSSAI indicated that it could abandon a phased approach and introduce stronger warning requirements immediately. ([Reuters](https://www.reuters.com/legal/litigation/india-food-regulator-says-it-is-open-stricter-labelling-after-court-scrutinizes-2026-09-10/?utm_source=chatgpt.com)) + +Two design choices now deserve close attention. The first is whether a product receives a warning after exceeding the threshold for one nutrient or only after exceeding several. The second is whether thresholds use a common per-100-gram or per-100-ml denominator or manufacturer-defined serving sizes. Industry representatives have challenged the former approach. ([The Economic Times](https://m.economictimes.com/industry/cons-products/food/front-of-pack-labels-food-inc-bats-for-per-serving-benchmark-in-sc-not-fssais-per-100-gm-plan/articleshow/133981453.cms?utm_source=chatgpt.com)) + +These look like technical details. They determine regulatory coverage. + +A system based on variable serving sizes can make comparisons harder and create incentives around how a “serving” is defined. A standard denominator sacrifices some correspondence with actual consumption while giving consumers and regulators a common comparison unit. The final court and regulatory documents will therefore be worth reading closely. + +[Reuters: food-label proceedings, 10 September](https://www.reuters.com/legal/litigation/india-food-regulator-says-it-is-open-stricter-labelling-after-court-scrutinizes-2026-09-10/?utm_source=chatgpt.com) + +## editorial note + +### Reconstruction needs a counterfactual + +Nepal’s NPR 723.31 billion estimate gives the disaster a price. It does not yet tell us the price of avoiding the next one. + +Post-disaster assessments are good at counting damaged roads, power systems, houses and public facilities. The harder counterfactual concerns where those assets should have been located, what engineering standards would have reduced losses, and whether some investments should exist in the same form at all. + +That distinction matters for climate finance. If “reconstruction” means replacing a washed-away bridge with a similar bridge in the same exposure zone, expenditure can restore capital without reducing vulnerability. If the replacement costs more because its location, design or redundancy changes, the additional expenditure is part of adaptation. + +Nepal’s assessment gives researchers an unusually concrete case for examining this distinction. I would return to it once the full assessment documents and financing plan are available. + +**Reader question:** What evidence should a post-disaster needs assessment require before classifying additional reconstruction expenditure as climate adaptation rather than ordinary asset replacement? + +*This is also a candidate for a future Methods Monday: how should we construct the counterfactual in post-disaster loss and reconstruction estimates?* + +[Subscribe now](https://varna.substack.com/subscribe?) + +## sources + +[Kathmandu Post: Nepal Rapid Damage and Needs Assessment, 10 September 2026](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com) + +[Kathmandu Post: Loss and Damage Fund response to Nepal](https://kathmandupost.com/national/2026/09/10/climate-loss-and-damage-fund-moves-to-consider-urgent-support-for-nepal?utm_source=chatgpt.com) + +[WHO South-East Asia: medical devices, diagnostics and assistive technologies resolution](https://www.who.int/southeastasia/news/detail/10-09-2026-who-south-east-asia-region-to-improve-access-to-medical-devices--in-vitro-diagnostics-and-assistive-technologies?utm_source=chatgpt.com) + +[Dawn: Pakistan remittances, 10 September 2026](https://www.dawn.com/news/2028756/remittances-rise-165pc-in-august?utm_source=chatgpt.com) + +[Business Recorder: Pakistan remittance data](https://www.brecorder.com/news/40438781?utm_source=chatgpt.com) + +[Reuters: India’s front-of-pack food warning labels](https://www.reuters.com/legal/litigation/india-food-regulator-says-it-is-open-stricter-labelling-after-court-scrutinizes-2026-09-10/?utm_source=chatgpt.com) + +[World Development: Inequality and social unrest in India](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com) + +## diff --git a/archive/2026-09-11-friday-findings-11-september-2026.md b/archive/2026-09-11-friday-findings-11-september-2026.md new file mode 100644 index 0000000..7c68e0c --- /dev/null +++ b/archive/2026-09-11-friday-findings-11-september-2026.md @@ -0,0 +1,93 @@ +--- +title: "Friday Findings | 11 September 2026" +date: 2026-09-11 +type: other +url: https://varna.substack.com/p/friday-findings-11-september-2026 +subtitle: "Public systems under pressure: automation in Bangladesh, TB diagnosis in Bihar, Nepal’s reconstruction bill, fertiliser dependence in India, and mounting food-system risks." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Friday Findings | 11 September 2026 + +*2026-09-11. Published at [https://varna.substack.com/p/friday-findings-11-september-2026](https://varna.substack.com/p/friday-findings-11-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Editorial-style Research Rundown cover with a paper-textured research desk featuring charts and sketches on work, tuberculosis diagnosis, Nepal’s flood reconstruction, fertiliser use and food-system pressures to 2050.](https://substackcdn.com/image/fetch/$s_!7RYv!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc5a812cc-7d78-4dfd-824e-c35a0927f7f4_1672x941.png)](https://substackcdn.com/image/fetch/$s_!7RYv!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc5a812cc-7d78-4dfd-824e-c35a0927f7f4_1672x941.png) +*Friday Findings - 11th September 2026* + +I have kept this edition to six items. Several newly indexed papers failed the recency test, while others offered too little evidence to justify inclusion. + +1. Bhattacharya, Debapriya, Towfiqul Islam Khan, Mamtajul Jannat, and Maliha Rahman. 2026. *Is Bangladesh Ready for the Future of Work: Preparing the Labour Market for Automation, AI, and Structural Transition*. Centre for Policy Dialogue. + +Released online on 6 September, although the report itself carries a July 2026 publication period. CPD estimates that automation could place up to 1.22 million ready-made garment jobs at risk and that as much as 60 per cent of women’s RMG employment could be displaced by 2041. More useful than the headline number is the attempt to ask what different labour-market futures might look like. The authors identify 27 drivers of change and construct four scenarios for 2035. + +The obvious caution is that these are foresight estimates. They depend heavily on assumptions about adoption, sectoral adjustment, skills and institutional response. They should not be read as forecasts of job losses. For WEE research, however, the gender composition of Bangladesh’s garment workforce makes the exercise worth close attention. + +[Read the CPD report](https://cpd.org.bd/publication/is-bangladesh-ready-for-the-future-of-work-preparing-the-labour-market-for-automation-ai-and-structural-transition-2/?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +1. Taluja, Kajal, Shailja Shah, Arup Saha, et al. 2026. “Stool Testing for Tuberculosis Diagnosis in Pregnant Women and Under Five Children: A Community Based Mixed Method Study from Rural India.” medRxiv preprint. doi: 10.64898/2026.09.03.26362132. + +An unusual and potentially useful diagnostic study from two districts in Bihar. The researchers worked within primary-care settings and offered Xpert MTB/RIF Ultra stool testing to pregnant women and children under five who could not produce sputum. Stool testing identified 21 TB cases. The accompanying qualitative work identified predictable operational problems: sample handling, transport, stigma and low awareness among patients and providers. + +The study has direct programme relevance because sputum collection is a real barrier in these populations. The evidence remains preliminary. The number of detected cases is small, the study covers two districts, and there is no basis here for assuming comparable diagnostic yield across NTEP settings. It is also a preprint. + +Still, this is exactly the kind of implementation question public-health research often neglects: whether an available diagnostic method can actually work inside ordinary primary-care systems. + +[Read the medRxiv preprint](https://www.medrxiv.org/content/10.64898/2026.09.03.26362132v1?utm_source=chatgpt.com) + +1. National Planning Commission and National Disaster Risk Reduction and Management Authority, Nepal. 2026. Rapid Damage and Needs Assessment of the August Bhotekoshi floods. + +Nepal’s preliminary assessment now puts recovery and reconstruction needs at NPR 723.31 billion. Infrastructure accounts for NPR 473.57 billion. Thirteen hydropower projects, five solar plants, roads, bridges, homes, schools, health facilities and roughly 1,806 hectares of crops were affected. The assessment combines administrative data, sectoral estimates, comparison of pre- and post-disaster statistics and drone surveys. + +These numbers deserve caution. This is an RDNA, and Nepal is simultaneously preparing the fuller post-disaster needs assessment. Estimates may move substantially as asset damage, livelihood losses and reconstruction standards are revised. + +The larger research question is already clear. A climate disaster of this scale changes the development budget itself. When hydropower, roads and productive assets are repeatedly rebuilt, adaptation becomes a fiscal and macroeconomic problem rather than a separate environmental programme. + +This follows directly from the Nepal flood material covered in the Daily earlier this week. I would return to the reconstruction estimates once the PDNA appears. + +[Read the detailed account of the government assessment](https://kathmandupost.com/national/2026/09/10/rs723-billion-needed-to-rebuild-flood-hit-areas-government-assessment-finds?utm_source=chatgpt.com) + +1. Government of Bangladesh, with technical support from ILO and SANEM. 2026. Just Transition chapter and implementation framework under Bangladesh’s NDC 3.0. + +Bangladesh has now attached an investment framework to its just-transition commitments. The chapter maps 258 commitments across 14 national policy documents into 33 priorities and 69 activities covering energy, transport, construction, agriculture, waste and industrial processes. The stated investment envelope is $1.85 billion. Only $250.6 million represents direct programme costs; about $1.6 billion is expected to come through commercial and development finance. Nearly 200 representatives participated across six rounds of consultation. + +This is a policy architecture rather than an evaluation. Its value lies in making several things measurable: responsible institutions, budgets, timelines and indicators. Whether the framework protects workers during sectoral change will depend on the employment and distributional analysis underneath those activities. + +For South Asian climate policy, this deserves watching because “just transition” often remains a statement of intent. Bangladesh is beginning to specify who does what, when and at what cost. + +[Read the open summary of the framework](https://unb.com.bd/category/Business/bangladeshs-just-transition-annex-maps-69-activities-185-billion-investment-envelope/195059?utm_source=chatgpt.com) + +1. Bhardwaj, Aditya Vardhan. 2026. “The Carbon Credit Case for Nano Urea.” *Ideas for India*, 7 September 2026. + +Bhardwaj proposes using India’s Carbon Credit Trading Scheme to pay farmers for verified reductions in conventional urea use when they switch to nano urea. In the proposed mechanism, agricultural emission reductions generate credits purchased by regulated industrial emitters, with the value passed to farmers as a discount on nano urea. + +It is a clever policy design. The evidentiary base is the weak point. The calculations require credible estimates of actual substitution, nitrogen-use efficiency, emissions avoided, farmer behaviour and crop yields. Those assumptions are doing a great deal of work. Verification at farm level may also be far harder than the article suggests. + +I would therefore read this as a proposal worth testing rather than a demonstrated route to reducing India’s fertiliser subsidy or emissions. It would make a good Methods Monday example of how apparently simple policy arithmetic can conceal several counterfactual assumptions. + +[Read the Ideas for India analysis](https://www.ideasforindia.in/topics/trade/the-carbon-credit-case-for-nano-urea?utm_source=chatgpt.com) + +1. Wiebe, Keith D., Nicola Cenacchi, Shahnila Dunston, Sherwin Gabriel, Faaiqa Hartley, Abhijeet Mishra, Richard D. Robertson, Mark W. Rosegrant, Timothy B. Sulser, Timothy S. Thomas, and Dirk Willenbockel. 2026. *Global Agrifood Systems Outlook to 2050*. Washington, DC: International Food Policy Research Institute. + +Released on 10 September, this is the one global report I would make room for in a South Asia list. IFPRI uses long-run scenario modelling to project changes in production, demand, prices, trade, natural-resource use, food security, nutrition, employment and income through 2050. The report covers seven world regions and seven commodity groups and is explicitly framed as projection rather than prediction. + +That distinction matters. Long-horizon models are useful for examining mechanisms and relative trajectories. Their point estimates become increasingly dependent on assumed population, income, climate, productivity and technological pathways as the horizon lengthens. + +[Subscribe now](https://varna.substack.com/subscribe?) + +South Asia is especially interesting here because shifts in diets, labour allocation and agricultural productivity can pull in different directions. IFPRI’s related modelling indicates that healthier-diet scenarios may move low-skilled labour back towards agriculture in parts of South Asia, raising a difficult structural-transformation question rather than producing an uncomplicated welfare story. + +[Open-access report](https://hdl.handle.net/10568/185223?utm_source=chatgpt.com) + +## what I would read first + +Start with the Bihar TB preprint if you work on implementation or public health. Read the Bangladesh future-of-work report for labour and WEE. Keep the Nepal assessment open beside both: its reconstruction numbers show how quickly climate shocks can alter the development constraints within which every other policy operates. + +The nano-urea piece is the one I would argue with most. That is also why it is worth reading. + +## one question for readers + +I am looking for credible India or South Asia evidence that tests nano urea against conventional urea using farm-level yield, nitrogen-use and emissions outcomes. If you know a field experiment or a good quasi-experimental study, send it across. If it survives a methods read, I will credit the recommendation in a future Friday Finding + +[Share](https://varna.substack.com/p/friday-findings-11-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-12-good-reads-2-can-india-sustain-its.md b/archive/2026-09-12-good-reads-2-can-india-sustain-its.md new file mode 100644 index 0000000..5b59294 --- /dev/null +++ b/archive/2026-09-12-good-reads-2-can-india-sustain-its.md @@ -0,0 +1,33 @@ +--- +title: "Good Reads #2: Can India sustain its rise " +date: 2026-09-12 +type: other +url: https://varna.substack.com/p/good-reads-2-can-india-sustain-its +subtitle: "Duvvuri Subbarao begins with a useful provocation. India’s recent economic story looks considerably less secure once we move beyond the headline growth numbers." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #2: Can India sustain its rise + +*2026-09-12. Published at [https://varna.substack.com/p/good-reads-2-can-india-sustain-its](https://varna.substack.com/p/good-reads-2-can-india-sustain-its). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads cover for Duvvuri Subbarao’s “Can India Sustain Its Rise?”, with an illustrated Indian city under construction beside books marked jobs, productivity and inclusion, rendered in a muted teal and rust print style.](https://substackcdn.com/image/fetch/$s_!mjgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F76687dbf-8d7a-4452-978b-4a03e1929481_1672x941.png)](https://substackcdn.com/image/fetch/$s_!mjgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F76687dbf-8d7a-4452-978b-4a03e1929481_1672x941.png) +*Research Rundown Goodreads Issue 2* + +Writing in the September issue of Finance & Development, the former RBI governor asks whether India can sustain its economic rise when employment, productivity and innovation remain persistent constraints. His timing is good. Much commentary on India continues to oscillate between celebration of its growth rates and predictions that it will become the world’s third-largest economy. Subbarao is interested in the machinery underneath those claims. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/can-india-sustain-its-rise-duvvuri-subbarao?utm_source=chatgpt.com)) + +One reason I found the essay worth reading is that it separates economic size from development performance. A country can climb global GDP rankings because of its population and aggregate growth while continuing to struggle with the quality of employment, productivity and the distribution of economic opportunity. Those questions matter more for understanding what growth is doing to people’s lives. + +[Subscribe now](https://varna.substack.com/subscribe?) + +Subbarao also places India’s present difficulties within a changed external environment. Energy dependence, geopolitical instability and the limits of the China+1 opportunity complicate the assumption that India can simply reproduce the development trajectory followed by earlier Asian economies. The harder domestic problem is productivity. India needs workers to move into activities that produce considerably more value, while creating enough decent employment for a very large working-age population. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/can-india-sustain-its-rise-duvvuri-subbarao?utm_source=chatgpt.com)) + +I would push the argument further on labour. Recent evidence on falling real rural wages makes it difficult to treat employment generation as sufficient evidence of economic progress. The composition of employment, earnings and security matter. So does the large amount of women’s work that remains poorly paid, unpaid or statistically difficult to interpret. + +The essay is useful precisely because it resists the inevitability embedded in much discussion of India’s rise. Becoming a larger economy is highly probable. Becoming a richer society with productive work available across a much broader section of the population is a different proposition. + +***That distinction deserves considerably more attention.*** + +[Read Duvvuri Subbarao, “Can India Sustain Its Rise?”](https://www.imf.org/en/publications/fandd/issues/2026/09/can-india-sustain-its-rise-duvvuri-subbarao?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-2-can-india-sustain-its?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-12-good-reads-3-what-comes-after-industrialisation.md b/archive/2026-09-12-good-reads-3-what-comes-after-industrialisation.md new file mode 100644 index 0000000..54e183a --- /dev/null +++ b/archive/2026-09-12-good-reads-3-what-comes-after-industrialisation.md @@ -0,0 +1,82 @@ +--- +title: "Good Reads #3: What comes after industrialisation?" +date: 2026-09-12 +type: other +url: https://varna.substack.com/p/good-reads-3-what-comes-after-industrialisation +subtitle: "Why the old industrialisation route is narrowing, and what development policy might do instead." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #3: What comes after industrialisation? + +*2026-09-12. Published at [https://varna.substack.com/p/good-reads-3-what-comes-after-industrialisation](https://varna.substack.com/p/good-reads-3-what-comes-after-industrialisation). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads cover for “Rethinking Development,” with signposts pointing toward manufacturing, services and jobs, set against construction cranes, charts, books and a map of Asia in a textured cream, teal, rust and ochre editorial style.](https://substackcdn.com/image/fetch/$s_!HDUk!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd0b44dba-f963-4820-9035-b52637c251e5_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!HDUk!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd0b44dba-f963-4820-9035-b52637c251e5_1536x1024.png) + +For much of development economics, the route from poverty to prosperity had a familiar shape. Workers moved out of low-productivity agriculture, manufacturing expanded, productivity rose, and economies became richer. East Asia supplied the most persuasive examples. + +Gordon Hanson, Dani Rodrik and Rohan Sandhu think that route is becoming much harder to follow. Their new essay, “Rethinking Development,” in the September 2026 issue of *Finance & Development*, asks what development policy should look like when manufacturing can no longer absorb workers at anything like its historical scale. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com)) + +[Read “Rethinking Development” in Finance & Development](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com) + +### The argument + +The starting point is employment. + +Hanson, Rodrik and Sandhu project that roughly two-thirds of workers in developing economies outside China will, a decade from now, be in jobs that require neither a college education nor exposure to international trade. Think retail, food services, domestic work, personal services, small businesses and other largely local activities. Many are informal and have low productivity. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com)) + +That matters because two standard prescriptions in development policy, education and export-oriented industrialisation, reach only part of this labour force. + +Manufacturing remains economically important. The authors point to productivity gains, learning effects, investment and national security. Their narrower claim is about jobs. Manufacturing has become increasingly capital-intensive and cannot be expected to absorb the growing labour forces of developing countries on the scale that it once did. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com)) + +The development question therefore changes. How do you raise productivity and earnings in the sectors where most people actually work? + +### Taking services seriously + +This is where the essay gets interesting. + +Low-productivity services have often been treated as the residue of incomplete structural transformation: workers sell food, drive taxis, run tiny shops or provide personal services because the modern economy has failed to absorb them. + +Hanson, Rodrik and Sandhu ask us to treat these activities as a central object of development policy instead. + +Their prescription has two sides. Workers need better pathways from secondary education into occupations, including vocational and technical training. Small firms need policies directed specifically at productivity: management support, technology, marketing assistance, links with larger firms and other public inputs suited to their actual constraints. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com)) + +This extends an argument Rodrik and Sandhu have been developing for some time. Their earlier work on labour-absorbing services considers policies ranging from public inputs for smaller firms to technologies that complement lower-skilled labour and vocational programmes combined with employment support. ([National Bureau of Economic Research](https://www.nber.org/papers/w32738?utm_source=chatgpt.com)) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### Where I would push the argument + +I am persuaded by the diagnosis more than I am by the implied policy transition. + +There is a real danger in moving too quickly from “most workers will be in services” to “therefore services must become the new engine of development.” Employment shares tell us where the problem is. They do not tell us whether these sectors possess the productivity dynamics that made manufacturing historically unusual. + +A street vendor becoming 15 per cent more productive is good for the vendor. The harder question is whether millions of such improvements can generate sustained economy-wide productivity growth, rising wages, technological learning and movement into progressively more productive activities. + +The authors recognise this problem. They point to emerging evidence of productivity growth within some nontraded services and make experimentation central to their policy proposal. But the empirical burden remains substantial. We know considerably more about how manufacturing generated structural change than we do about whether fragmented, often informal service sectors can reproduce enough of those dynamics. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com)) + +### The India question + +For India, this argument is difficult to ignore. + +India has long presented an awkward case for the conventional structural-transformation story. Its economy moved unusually quickly toward services without first passing through an East Asian-scale manufacturing employment transition. At the same time, a large share of employment remains informal and productivity differences across workers and firms are enormous. + +That makes the authors’ question particularly relevant. If manufacturing is unlikely to absorb enough Indian workers, improving productivity where people already work becomes unavoidable. + +I would still resist treating this as an alternative to industrial policy. India needs both. Manufacturing can matter enormously for productivity, exports and technological capabilities even if its employment share never approaches earlier East Asian levels. The harder policy problem is how productivity gains in manufacturing and high-productivity services connect to the much larger labour force working elsewhere. + +That distributional connection is where the essay could have gone further. + +### Why read it? + +Because it poses a development question that is becoming increasingly difficult to avoid. + +For decades, economists could point to historical experience when asked how poor countries become rich. The sequence differed across countries, but industrialisation occupied a privileged place in most successful accounts. + +Hanson, Rodrik and Sandhu are asking what happens when the historical ladder is still there, but can carry far fewer workers. + +Their answer is provisional. That is precisely what makes the essay useful. The problem is clearer than the solution. + +[Rethinking Development, Gordon Hanson, Dani Rodrik and Rohan Sandhu, Finance & Development, September 2026](https://www.imf.org/en/publications/fandd/issues/2026/09/rethinking-development-gordon-hanson?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-3-what-comes-after-industrialisation?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-12-good-reads-5-can-ai-make-informal.md b/archive/2026-09-12-good-reads-5-can-ai-make-informal.md new file mode 100644 index 0000000..4020547 --- /dev/null +++ b/archive/2026-09-12-good-reads-5-can-ai-make-informal.md @@ -0,0 +1,67 @@ +--- +title: "Good Reads #5: Can AI make informal work better?" +date: 2026-09-12 +type: other +url: https://varna.substack.com/p/good-reads-5-can-ai-make-informal +subtitle: "Anthony Tan argues that technology can improve earnings, credit and skills without requiring workers to become formal. " +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #5: Can AI make informal work better? + +*2026-09-12. Published at [https://varna.substack.com/p/good-reads-5-can-ai-make-informal](https://varna.substack.com/p/good-reads-5-can-ai-make-informal). This file is a generated copy; the Substack post is the original.* + +[![Alt text: Research Rundown Good Reads cover for “Can AI Make Informal Work Better?”, with a street vendor, delivery rider, smartphone map, helmet and work notes against a South Asian city backdrop, in the series’ textured cream, teal, rust and ochre editorial style.](https://substackcdn.com/image/fetch/$s_!K-Sh!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Faf88d70a-7776-4757-bc09-c9a6fc63aeb8_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!K-Sh!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Faf88d70a-7776-4757-bc09-c9a6fc63aeb8_1536x1024.png) +*Good Reads #5 AI and informal work* + +There is a useful reversal at the beginning of Anthony Tan’s essay in the September issue of Finance & Development. Much of the discussion about artificial intelligence begins with formal employment: which jobs will disappear, which occupations will change, and which workers will become more productive. Tan starts elsewhere. + +In Southeast Asia, he writes, 78 per cent of workers operate in the informal economy. If that is where most people work, then asking what AI will do to salaried employment captures only part of the development question. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/point-of-view-ai-and-the-informal-economy-anthony-tan?utm_source=chatgpt.com)) + +It is a sensible place to begin. + +Tan’s larger argument is more contentious. Informality, he suggests, should not simply be understood as a temporary condition from which workers must eventually graduate. Digital platforms can instead make informal work function better. They can use transaction and earnings data to extend credit to people banks do not serve, match workers more efficiently with demand, provide training, and give small businesses tools that would previously have been available mainly to larger firms. + +The examples are striking. Tan reports that Grab disbursed $3.2 billion in loans to drivers and merchants last year, with 68 per cent of borrowers gaining access to formal credit for the first time. More than 1.5 million drivers completed at least one GrabAcademy course. One in four micro, small and medium enterprises on the platform uses its AI assistant, which Grab associates with a 15 per cent increase in sales. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/point-of-view-ai-and-the-informal-economy-anthony-tan?utm_source=chatgpt.com)) + +These numbers deserve attention. They also deserve scrutiny. + +[Subscribe now](https://varna.substack.com/subscribe?) + +Tan is the chief executive and cofounder of Grab. The essay is therefore not an independent assessment of whether platformisation improves informal work. It is an argument from one of the companies building the system being described. The distinction matters. + +The central question is not whether platforms can provide useful services. Plainly, they can. The question is what kind of labour market is created when access to customers, credit, training, information and eventually social protection increasingly passes through privately governed platforms. + +That changes the development problem. + +A worker may gain access to credit because a platform can observe earnings that a bank cannot. But the same arrangement gives the platform an unusually detailed view of that worker’s economic life. An algorithm may reduce the time a driver spends waiting for the next passenger. It can also influence where the driver goes, which work appears worthwhile and how income opportunities are distributed. + +The platform is no longer merely matching buyer and seller. It begins to perform functions previously divided among markets, employers, financial institutions and, in some cases, the state. + +Tan recognises part of the social-protection problem. He argues for benefits that travel with workers across jobs and platforms rather than depending upon their classification as employees. He points to Malaysia’s 2026 Gig Workers Act as an example of extending social-security coverage without making gig workers employees. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/09/point-of-view-ai-and-the-informal-economy-anthony-tan?utm_source=chatgpt.com)) + +That is an interesting direction. But it leaves the political economy of platform power largely outside the frame. + +There is another claim worth examining carefully. Tan argues that policy should be designed around informality rather than treating formalisation as the principal measure of progress. There is something persuasive here. Development policy has often treated the informal economy as a residual category, even when it contains most of the workforce. + +Yet accepting informality as persistent is not the same thing as accepting its present terms. + +For India, that distinction is especially useful. The relevant question is unlikely to be whether digital technology will somehow eliminate informal work. It is whether technology can raise productivity and incomes while workers also gain bargaining power, social insurance, predictable rules and meaningful control over the data through which increasingly large parts of their working lives are organised. + +That makes this less an essay about AI than it first appears. + +It is really an argument about what development policy should do when informality persists and technology becomes one of the institutions governing it. + +Tan offers one answer: design around the economy people actually inhabit. + +I think that is the right starting point. I am much less certain that privately governed platforms can be treated simply as the means of getting there. + +### why read it? + +Because it shifts the AI discussion away from the formal jobs that dominate much of the debate and towards the economic lives of workers in developing countries. And because its blind spots are almost as useful as its argument. + +The essay makes a strong case for taking informality seriously. It also gives us a good reason to ask who should control the digital infrastructure increasingly being built around it. + +[Read Anthony Tan, “AI and the Informal Economy,” Finance & Development, September 2026](https://www.imf.org/en/publications/fandd/issues/2026/09/point-of-view-ai-and-the-informal-economy-anthony-tan?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-5-can-ai-make-informal?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-12-good-reads-6-indias-middle-power.md b/archive/2026-09-12-good-reads-6-indias-middle-power.md new file mode 100644 index 0000000..91f82db --- /dev/null +++ b/archive/2026-09-12-good-reads-6-indias-middle-power.md @@ -0,0 +1,103 @@ +--- +title: "Good Reads #6: India’s middle-power moment" +date: 2026-09-12 +type: other +url: https://varna.substack.com/p/good-reads-6-indias-middle-power +subtitle: "N. K. Singh sees a fractured global order as an opening for India. The harder question is whether diplomatic room for manoeuvre can be converted into economic and institutional influence." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #6: India’s middle-power moment + +*2026-09-12. Published at [https://varna.substack.com/p/good-reads-6-indias-middle-power](https://varna.substack.com/p/good-reads-6-indias-middle-power). This file is a generated copy; the Substack post is the original.* + +[![N. K. Singh sees a fractured global order as an opening for India. The harder question is whether diplomatic room for manoeuvre can be converted into economic and institutional influence.](https://substackcdn.com/image/fetch/$s_!BHFz!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F61e38875-2c9e-48b5-a552-4e3b3e5aa605_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!BHFz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F61e38875-2c9e-48b5-a552-4e3b3e5aa605_1536x1024.png) +*Good Reads 5# on Multilateralism* + +There is a familiar way of describing the current international order: fragmentation, weakening multilateral institutions, trade conflict, industrial policy, strategic competition and a retreat from the assumptions that governed globalisation for several decades. + +N. K. Singh begins from much the same diagnosis. But “The Middle Power Moment,” published in the June 2026 issue of Finance & Development, is less interested in mourning the old order than in asking who gains room to act as it weakens. His answer is the world’s middle powers, with India occupying a particularly interesting position. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +Singh’s central claim is straightforward. Influence in the emerging international system will depend less exclusively on economic or military weight and more on the capacity to convene countries, bridge competing blocs and build coalitions around particular problems. + +That produces a world of smaller arrangements rather than one universal architecture. Climate finance might require one coalition, critical minerals another, migration another, energy security another. These groupings may be temporary and overlapping. Countries need not agree on everything in order to cooperate on something. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +For India, Singh sees an opportunity. + +[Subscribe now](https://varna.substack.com/subscribe?) + +Its size matters, of course. So does its position across several political and economic groupings, its long diplomatic tradition of nonalignment, its relationships with developing economies and its ability to engage countries that increasingly find themselves on opposing sides of geopolitical disputes. + +This is one of the essay’s strongest observations. The return of geopolitics does not necessarily recreate the clean blocs of the Cold War. Economic relationships now cut across political ones. Countries may compete on technology, cooperate on climate, depend upon one another for energy and disagree over security. + +That makes the space between blocs politically valuable. + +Singh argues that India should occupy it deliberately. + +### economics has become foreign policy + +What makes the essay particularly useful for this series is that Singh does not treat foreign policy as something separate from development economics. + +Energy security, industrial capacity, migration, climate finance, technology and skills all appear in the essay as both domestic economic questions and instruments of international power. + +Consider energy. India remains highly exposed to external oil markets while attempting a large transition towards non-fossil energy. Singh places that transition inside a wider problem of climate finance. Developing economies, he notes, require at least $310 billion each year for adaptation, while current flows are about $26 billion. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +Or consider industrial policy. Excess capacity in steel and solar manufacturing is no longer merely a question of prices and efficiency. It now produces tariffs, subsidies and political responses as governments try to protect domestic production while retaining access to global markets. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +Migration is treated similarly. Ageing societies in Europe and East Asia face shrinking workforces while India and much of Africa remain young. Singh argues that migration should be understood as a response to these demographic imbalances rather than primarily as a crisis. But he adds an obvious qualification: a young population becomes an economic advantage only when workers have the skills and employment opportunities to make use of it. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +Then there is technology. Singh points to the gap between India’s demand for AI professionals and the very small proportion of young people receiving formal skills training. His answer is sustained investment in human capital and research, alongside policies intended to spread productivity gains rather than deepen digital inequality. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +Read together, these are not four separate policy problems. + +They amount to an argument that domestic economic capacity increasingly determines foreign-policy capacity. + +### where I hesitate + +I am persuaded by Singh’s description of the opportunity more readily than by the implied measure of Indian power. + +“Middle power” is an unusually elastic category. India is simultaneously a very large economy, a nuclear-armed state, the world’s most populous country, a major military power, a lower-middle-income country and a country with substantial deficits in employment, education, health and state capacity. + +Those facts coexist. + +That makes aggregate size a poor guide to the kind of influence India can exercise. + +A country may possess enough diplomatic weight to convene international coalitions while still facing severe domestic constraints on its ability to provide technology, finance, public goods or market access to partners. The distinction between being present at the table and being able to alter what happens there matters. + +There is also a tension in the essay between plurilateralism as a practical response to institutional paralysis and plurilateralism as a substitute for universal institutions. + +Smaller coalitions can act faster. They can also leave smaller and poorer countries with less influence. A world organised through shifting groups of powerful states may be more flexible than the old multilateral system without necessarily being fairer. + +This is especially relevant for the Global South. India may gain bargaining space from fragmentation. Many much smaller developing economies may experience the same fragmentation primarily as vulnerability. + +The question, then, is not simply whether India can become an effective middle power. + +It is what India would use that position for. + +### the India question + +This is where Singh’s revival of nonalignment becomes interesting. + +He is not proposing a return to the twentieth-century version of the Non-Aligned Movement. His argument is closer to strategic autonomy adapted to economic interdependence: cooperate across blocs, avoid unnecessary alignment and build issue-based coalitions where interests coincide. ([IMF](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh?utm_source=chatgpt.com)) + +India has already practised versions of this. + +But the next stage will require more than diplomatic flexibility. Leadership among developing economies will depend partly on whether India can offer things those countries need: finance, technology, market access, medicines, digital public infrastructure, climate partnerships, knowledge and credible institutions for cooperation. + +That moves the discussion back from diplomacy to development. + +India’s external influence will ultimately rest not just on the cleverness of its positioning but on the economic and institutional capacity beneath it. + +### why read it? + +Because Singh gives us a compact statement of an argument likely to become increasingly common in Indian policy thinking: fragmentation in the global order is not merely a threat to manage but a source of political room for countries that can operate between competing centres of power. + +The essay is most interesting when read alongside a harder question. + +A middle-power moment may exist. + +What, materially, would India need to do with it? + +[Read N. K. Singh, “The Middle Power Moment,” Finance & Development, June 2026](https://www.imf.org/en/publications/fandd/issues/2026/06/point-of-view-the-middle-power-moment-n-k-singh) + +[Share](https://varna.substack.com/p/good-reads-6-indias-middle-power?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-13-research-rundown-daily-13-september.md b/archive/2026-09-13-research-rundown-daily-13-september.md new file mode 100644 index 0000000..560d628 --- /dev/null +++ b/archive/2026-09-13-research-rundown-daily-13-september.md @@ -0,0 +1,154 @@ +--- +title: "Research Rundown DAILY - 13 September 2026" +date: 2026-09-13 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-13-september +subtitle: "The New Delhi BRICS declaration sets out ambitious plans on development finance, payments, agriculture and climate. The test now is implementation." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 13 September 2026 + +*2026-09-13. Published at [https://varna.substack.com/p/research-rundown-daily-13-september](https://varna.substack.com/p/research-rundown-daily-13-september). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 13 September 2026, showing the New Delhi BRICS declaration connected to working files on development finance, agriculture, public health and infrastructure, representing the shift from summit commitments to implementation.](https://substackcdn.com/image/fetch/$s_!hoUw!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42d6484a-18e0-418b-9aaa-736f4864060f_1672x941.png)](https://substackcdn.com/image/fetch/$s_!hoUw!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42d6484a-18e0-418b-9aaa-736f4864060f_1672x941.png) + +## five things worth knowing + +### 1. BRICS has put development-finance guarantees on the table + +The New Delhi Declaration adopted at the BRICS summit moves the proposed BRICS Multilateral Guarantees initiative a little further towards something operational. Leaders called for technical work and pilot transactions, with the stated aim of using guarantees to improve project creditworthiness, lower financing costs and draw private capital into development projects in BRICS countries and the wider Global South. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) + +This deserves attention because guarantees can address a specific development-finance problem. Infrastructure projects in lower-income countries often face financing costs far above those in richer economies, even when the underlying project risk does not explain the entire spread. + +The declaration does not tell us the guarantee structure, capital base, pricing, eligibility rules or allocation of losses. Those details will determine whether the initiative absorbs risks that private investors will not bear, merely subsidises projects that would have been financed anyway, or remains another BRICS proposal under discussion. + +The first pilot transactions will be more informative than the declaration. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. Pakistan has approved an US$58 billion power plan, despite serious disagreement inside the regulator + +Pakistan’s National Electric Power Regulatory Authority has conditionally approved the Integrated System Plan 2025, covering generation and transmission investment through 2035. The plan implies roughly US$47.08 billion in additional generation capacity and US$10.65 billion in transmission investment. ([Dawn](https://www.dawn.com/news/amp/2029253)) + +The regulatory record is unusually revealing. All three NEPRA members wrote dissenting or separate advisory observations. Among their concerns were changes to project selection and whether the Council of Common Interests had been bypassed. The reference scenario assumes average GDP growth of 3.52 per cent and would add 26,045 MW while retiring 2,577 MW, leaving installed capacity at 62,657 MW. ([Dawn](https://www.dawn.com/news/amp/2029253)) + +The number therefore needs to be read as a planning envelope rather than committed expenditure. + +There is a useful research question here. Electricity plans depend heavily on demand forecasts, fuel assumptions and the treatment of distributed generation. Pakistan is making investment choices lasting decades while each of those variables is changing quickly. + +### 3. India’s food-warning-label dispute now has a deadline + +The Supreme Court has ordered the Union government to provide a clear implementation pathway for front-of-pack warning labels within ten days. The order followed concern that FSSAI’s proposed two-stage system could leave the stricter second stage indefinitely delayed. ([Reuters](https://www.reuters.com/legal/litigation/india-court-orders-regulator-provide-timeline-food-warning-labels-2026-09-11/)) + +This is a substantive development from the labelling debate covered earlier in Research Rundown. FSSAI had proposed initially requiring red warning symbols when products exceeded limits in at least two of three categories: added sugar, salt and saturated fat. A stricter second phase would follow. The Court said uncertainty over the timing of that second phase could not be justified by industry reformulation needs or assessments of consumer acceptance. A government source told Reuters that FSSAI was willing to consider implementing both phases together. ([Reuters](https://www.reuters.com/legal/litigation/india-court-orders-regulator-provide-timeline-food-warning-labels-2026-09-11/)) + +The next document matters more than another round of statements. Thresholds, reference quantities, implementation dates and exemptions will determine which products actually receive warnings. + +### 4. Nepal’s flood damage is now drawing down the electricity system’s winter reserve + +The 26 August Bhotekoshi floods knocked more than 400 MW of hydropower capacity offline. Nepal Electricity Authority has consequently been running all three stages of the reservoir-based Kulekhani system continuously during the monsoon. Kulekhani is normally conserved for the dry winter months, when run-of-river generation falls. ([Kathmandu Post](https://kathmandupost.com/national/2026/09/12/bhotekoshi-floods-push-kulekhani-into-continuous-power-generation?utm_source=chatgpt.com)) + +This gives the flood a second-order cost that is easy to miss in damage assessments. Lost generation is one cost. Using stored water now reduces the buffer available against a later seasonal shortage. + +The episode also illustrates why infrastructure resilience cannot be measured asset by asset. Damage to one group of hydropower plants changes how another part of the electricity system must operate. + +Nepal’s reconstruction assessment should therefore be read alongside winter generation adequacy, reservoir levels and electricity-import requirements. + +### 5. Bangladesh has begun a three-month national dengue campaign as the peak season approaches + +Bangladesh launched its Dengue Prevention and Cleanliness Campaign 2026 on 12 September. The three-month programme is being run jointly by the Ministry of Local Government, Rural Development and Co-operatives and the two Dhaka city corporations. ([The Business Standard](https://www.tbsnews.net/bangladesh/pm-opens-three-month-nationwide-dengue-prevention-cleanliness-campaign-1540281?utm_source=chatgpt.com)) + +The campaign itself establishes little about whether transmission will fall. Dengue control is particularly susceptible to activity measures that look impressive administratively: inspections conducted, sites cleaned, awareness events held. + +The useful indicators are epidemiological and operational. Case incidence, spatial concentration, entomological surveillance and the time between identifying breeding sites and removing them would tell us considerably more about whether the intervention changes exposure. + +Bangladesh’s campaign is therefore worth following through its data rather than its launch. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## one paper worth reading + +### Phone tutoring works during school closures. SMS alone is much less reliable. + +Noam Angrist and colleagues report randomized trials of remote primary-school instruction in India, Nepal, Kenya, Uganda and the Philippines. Children received targeted tutoring by phone plus SMS, SMS alone, or usual provision. The combined phone intervention increased the pooled learning outcome by 0.321 standard deviations across the five countries; SMS alone produced a much smaller 0.078 SD effect and worked inconsistently across settings. India contributed 708 participants and Nepal 2,678. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +The cross-country design is the paper’s main strength. Random assignment supports causal inference within the trials, and Nepal and the Philippines also randomized delivery between government and NGO instructors. Government delivery retained substantial effects. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +There is an important qualification. Effect sizes grew across successive trials as targeting and implementation improved. Context and intervention maturity therefore move together, making some cross-country comparisons difficult to interpret. These were also emergency interventions during COVID-era school disruption, when the counterfactual in some countries involved very little schooling. + +The policy result is still useful: low-tech, targeted human instruction travelled across settings far better than messages alone. For education systems planning for climate, conflict or disease-related closures, that is actionable evidence. + +**Citation:** Angrist, Noam, Micheal Ainomugisha, Sai Pramod Bathena, et al. 2026. “Mobile education builds resilience during shocks in five countries.” *Nature*. Published 9 September 2026. DOI: 10.1038/s41586-026-10990-x. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +[Open-access paper](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com) + +## one number + +### 0.321 SD + +The pooled effect of targeted phone tutoring plus SMS on the study’s average numeracy outcome across randomized trials in India, Nepal, Kenya, Uganda and the Philippines. SMS alone produced an estimated effect of 0.078 SD. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +That difference is more interesting than the headline effect size. + +Both interventions used ordinary mobile phones. The larger effect came from adding a person who could assess a child’s level and adjust instruction accordingly. + +Technology access was therefore only one part of the treatment. + +## policy watch + +### BRICS has produced a long list of institutions. Now watch which ones acquire budgets and rules. + +The New Delhi Declaration creates or advances an unusual number of mechanisms with development-policy relevance. + +BRICS members agreed to establish AGRIN, an agro-inputs, genetic-resources and information network, alongside networks on agroecology, regenerative agriculture and digital agriculture. They also backed further work on a BRICS grain-trading platform. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) + +In health, the declaration establishes a network of mental-health Centres of Excellence coordinated by NIMHANS and backs a BRICS Mission for Healthy Lifestyle for 2026–29. It also proposes greater cooperation between medical-product regulators and an integrated early-warning system for infectious diseases. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) + +Finance ministers and central banks will continue work on cross-border payment interoperability and settlement using national currencies. The declaration carefully describes this as continuing technical work and explicitly recognises different national regulatory arrangements. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) + +On climate, the political position is more explicit. BRICS opposes carbon-border adjustment measures it considers unilateral or protectionist, while acknowledging a continuing role for fossil fuels in emerging economies. It simultaneously backs the Paris Agreement, adaptation finance, carbon-market cooperation and community-based adaptation. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) + +These commitments should be separated into three categories as implementation proceeds: institutions that acquire money and operating rules, technical networks that produce usable public goods, and declarations that remain statements of common position. + +The distinction will make the next year of BRICS activity much easier to assess. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## editorial note + +### Count institutions when they start doing something + +International declarations produce an awkward measurement problem. + +The New Delhi document contains networks, centres, platforms, working groups, initiatives, principles, roadmaps and proposed financing mechanisms. Counting them would produce a substantial indicator of international cooperation. + +It would tell us very little about results. + +A useful monitoring framework would start one step later. Does an initiative acquire a secretariat or accountable lead institution? Is there a budget? Are participation and eligibility rules public? Does it produce transactions, datasets, regulatory agreements or services that did not previously exist? + +The proposed Multilateral Guarantees initiative provides a particularly clean test. The declaration now calls for pilot transactions. ([Prime Minister of India](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?comment=disable)) Once those appear, we can ask which projects receive guarantees, how risk is priced, how much private finance follows, and whether borrowing costs actually fall. + +Until then, the correct unit is a proposal. + +**Reader question:** If you were building a scorecard to distinguish functioning international cooperation from summit announcements, which two or three observable indicators would you require before counting an initiative as operational? + +This would make a good future Methods Monday: how to measure institutional implementation when governments report activities more readily than outputs. + +## sources + +[BRICS New Delhi Declaration, Prime Minister’s Office, 12 September 2026](https://www.pmindia.gov.in/en/news_updates/brics-new-delhi-declaration-building-for-resilience-innovation-cooperation-and-sustainability-september-12-2026/?utm_source=chatgpt.com) + +[Dawn: Pakistan’s Integrated System Plan, 12 September 2026](https://www.dawn.com/news/2029253?utm_source=chatgpt.com) + +[Reuters: Supreme Court order on India’s food warning labels, 11 September 2026](https://www.reuters.com/legal/litigation/india-court-orders-regulator-provide-timeline-food-warning-labels-2026-09-11/?utm_source=chatgpt.com) + +[Kathmandu Post: Bhotekoshi floods and Nepal’s winter power reserve, 12 September 2026](https://kathmandupost.com/national/2026/09/12/bhotekoshi-floods-push-kulekhani-into-continuous-power-generation?utm_source=chatgpt.com) + +[The Business Standard: Bangladesh dengue campaign, 12 September 2026](https://www.tbsnews.net/bangladesh/pm-opens-three-month-nationwide-dengue-prevention-cleanliness-campaign-1540281?utm_source=chatgpt.com) + +[Nature: Angrist et al., mobile education during shocks, 9 September 2026](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-13-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +## diff --git a/archive/2026-09-14-methods-monday-14-september-2026.md b/archive/2026-09-14-methods-monday-14-september-2026.md new file mode 100644 index 0000000..6d8a839 --- /dev/null +++ b/archive/2026-09-14-methods-monday-14-september-2026.md @@ -0,0 +1,153 @@ +--- +title: "Methods Monday | 14 September 2026" +date: 2026-09-14 +type: other +url: https://varna.substack.com/p/methods-monday-14-september-2026 +subtitle: "The strongest methods work this week asks a deceptively simple question: how much of a survey result is created by the machinery of measurement itself? " +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Methods Monday | 14 September 2026 + +*2026-09-14. Published at [https://varna.substack.com/p/methods-monday-14-september-2026](https://varna.substack.com/p/methods-monday-14-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Editorial illustration for Methods Monday showing the same survey question in English, Hindi, Bangla, Urdu and Tamil, alongside a South Asia map, magnifying glass, charts, translation notes and measuring tools.](https://substackcdn.com/image/fetch/$s_!FdXh!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F68881d08-5bff-4477-b354-a68402f2df42_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!FdXh!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F68881d08-5bff-4477-b354-a68402f2df42_1536x1024.png) +*Methods Monday - 14th September 2026* + +*Translation choices move means. AI coding performs differently across languages. Fieldwork stopping rules trade cost against uncertainty. Synthetic respondents reproduce some patterns while distorting others. For researchers working across multilingual and uneven survey environments in South Asia, these are practical problems.* + +## 1. translation can change the statistics, even when every version looks defensible + +Chia-Jung Tsai, Clemens Lechner, Dorothée Behr, Ulrike Efu Nkong and Anke Radinger provide unusually direct evidence on questionnaire translation. They asked 16 professional translators and 16 social scientists to translate an English questionnaire into German, using both translation from scratch and human post-editing of machine translation. The resulting 32 questionnaire versions were fielded separately. The analysis ultimately covered 8,377 completed interviews, with roughly 250 responses for each experimental translation. + +The differences were not trivial. Item means varied across translations, sometimes substantially. Post-editing machine translation reduced variation between versions and generally brought results closer to the official benchmark. It introduced a different risk: a subtle error in the machine version could survive human review and produce systematic bias. Social scientists translating from scratch were slightly more likely than professional translators to generate statistical outliers. + +For South Asian survey research, the application is immediate. Translation is still routinely treated as an operational task after questionnaire design. This paper shows that wording decisions can change the distribution of the variable itself. Back-translation does not solve that problem. The authors make a stronger case for collaborative procedures such as translation, review, adjudication, pretesting and documentation. + +Tsai, Chia-Jung, Clemens Lechner, Dorothée Behr, Ulrike Efu Nkong and Anke Radinger. 2026. “The Influence of Translator Backgrounds and Machine Translation on Statistical Properties of Surveys: Evidence from a Survey Experiment.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smag029. Published 9 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article/doi/10.1093/jssam/smag029/8789599?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +**Worth reading: first on my list this week.** + +## 2. LLM coding works reasonably well across several languages, then drops in a lower-resource one + +Mao Li, Stephanie Morales, Kaidar Nurumov and Sunghee Lee test whether large language models can code open-ended responses to a web probe attached to the familiar self-rated health question. Their data span five countries and four languages. Five open-source LLMs were compared with human coding and conventional machine-learning models. + +The best model reached a micro-F1 score of 0.76 when classifying attributes mentioned in respondents’ answers and 0.77 when classifying whether those attributes were expressed positively, negatively or neutrally. Performance remained fairly stable for English, Spanish and German, then fell for Korean. The authors link the difference to Korean’s lower representation in model training data. + +The South Asian implication is an inference rather than something this paper establishes. Performance on English gives us little basis for assuming similar accuracy in Bangla, Sinhala, Tamil, Nepali, Urdu or lower-resource Indian languages. Any LLM-assisted coding workflow used in multilingual research should therefore validate performance separately by language. A pooled accuracy figure can conceal where the model is failing. + +Li, Mao, Stephanie Morales, Kaidar Nurumov and Sunghee Lee. 2026. “Using Large Language Models to Autocode Probing Responses to the Self-Rated Health Question: Strengths and Weaknesses in a Multilingual Survey Context.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smaf059. Published 9 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smaf059/8789739?utm_source=chatgpt.com) + +**Worth reading: yes, especially for multilingual mixed-methods work.** + +## 3. when should field teams stop calling? + +Xinyu Zhang, James Wagner and Michael R. Elliott take on an ordinary fieldwork decision with large budget consequences. Continued attempts may bring harder-to-reach respondents into the sample, while each additional call costs money and may contribute little to the estimates a survey was designed to produce. + +Their proposed risk-conscious stopping rule explicitly incorporates uncertainty in predicted survey costs and errors. Using Health and Retirement Study data and Monte Carlo simulations, they examine both which unresolved cases to stop pursuing and when that decision should be made. In several scenarios, implementing the rule after eight to ten call attempts produced the lowest number of attempts per completed interview while maintaining the same level of data quality for the statistics under study. + +The method suits repeated surveys with good historical paradata and a small set of priority estimates. Transfer to one-off development surveys is less straightforward. A stopping rule based on poor predictions of response propensity or outcomes can simply formalise the wrong decision. + +Zhang, Xinyu, James Wagner and Michael R. Elliott. 2026. “Optimizing Survey Errors and Survey Costs: A Risk-Conscious Stopping Rule and Timing Analysis.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smaf065. Published 9 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smaf065/8789740?utm_source=chatgpt.com) + +**Worth reading: particularly useful for CATI panels, longitudinal surveys and repeated monitoring rounds.** + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 4. a refusal to receive survey questions by text may not be a permanent refusal + +Jim Vine and colleagues use two waves of the Understanding Society Innovation Panel to examine consent for receiving occasional survey questions by SMS. Overall, 69 per cent of respondents with mobile phones consented when asked. Among people who had refused when asked previously, 55 per cent consented when asked again at a later wave. + +Mode mattered. In one wave, 65 per cent of web respondents consented compared with 78 per cent interviewed face to face. Because respondents could move between modes, the authors also exploit randomized initial mode allocation. The estimated mode effect is less precise, but the pattern still suggests that consent itself can be sensitive to how the request is made. Moving the consent question from early to late in the questionnaire changed consent by only 2.4 percentage points and the difference was statistically uncertain. + +This is useful for development research adopting SMS follow-ups, rapid phone modules and panel recontact. Consent should not automatically be treated as a fixed respondent characteristic. Timing, mode and previous contact history may shape it. + +Vine, Jim, et al. 2026. “Getting Consent to Survey in New Ways: Evidence from Experiments about Questions by Text Message in the Understanding Society Innovation Panel.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smag015. Published 9 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article/doi/10.1093/jssam/smag015/8789737?utm_source=chatgpt.com) + +**Worth reading: useful if your study plans repeated respondent contact across modes.** + +## 5. synthetic respondents need their own error framework + +Indira Sen, Georg Ahnert, Leah von der Heyde, Jana Lasser, Bernd Weiß and Markus Strohmaier have proposed a Total Simulated Survey Error framework for studies in which LLMs are asked to act as survey respondents. + +The paper starts from a problem that is becoming harder to ignore. A synthetic sample can reproduce aggregate human response patterns while getting subgroup distributions, associations between variables or particular populations wrong. Their framework separates representation and measurement problems and adds errors specific to simulation design, including the construction of respondent personas and choices about prompts and model settings. The authors illustrate the framework through simulated responses concerning the 2024 US presidential election. + +The framework is more useful than another demonstration that a model can imitate an average respondent. For development research, synthetic respondents remain a poor substitute for empirical observation where the purpose is population inference. The framework gives researchers a better vocabulary for showing why. + +Sen, Indira, Georg Ahnert, Leah von der Heyde, Jana Lasser, Bernd Weiß and Markus Strohmaier. 2026. “Total Simulated Survey Error: Designing and Diagnosing Survey Responses from Large Language Models.” arXiv:2609.10280. Posted 9 September 2026. + +[Read the preprint](https://arxiv.org/abs/2609.10280?utm_source=chatgpt.com) + +**Worth reading: yes, particularly before using synthetic data as anything beyond exploratory material.** + +## 6. administrative data can show who a survey is systematically missing + +Michael Jackson and colleagues use voter records, consumer databases and small-area contextual information to examine differential nonresponse in a US address-based survey. They could match 86 per cent of sampled addresses to voter or consumer databases. A model of partisan leaning agreed with respondents’ reported political leaning in 65 per cent of matched cases. Adding predicted partisanship to demographic weighting improved recalled-vote estimates, while validated turnout information improved them further. + +The electoral application is specialised. The design idea travels well. Most surveys contain detailed data on respondents and almost nothing on nonrespondents. Administrative or frame data available for everyone sampled can turn nonresponse from an invisible problem into something that can at least be studied. + +For South Asian research, feasibility depends on the source. Programme registers, school records, health-system lists and geographic covariates may sometimes provide the relevant frame information. Privacy, identifier quality and incomplete registers remain real constraints. + +Jackson, Michael, et al. 2026. “Who are We Missing? Using Administrative and Contextual Data to Measure and Correct Partisan Nonresponse in Opinion Surveys.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smag017. Published 9 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag017/8789736?utm_source=chatgpt.com) + +**Worth reading: yes. The substantive case is US-specific; the nonresponse design is useful much more widely.** + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 7. web recruitment can accumulate educational bias at several different stages + +Christine Distler, Mustafa Coban and Mark Trappmann examine recruitment into Germany’s IAB-OPAL online panel using administrative employment histories available for respondents and nonrespondents. Educational bias did not arise at one moment. It increased as people moved through successive stages of recruitment and registration. + +People with lower secondary education made up 23.2 per cent of the gross sample and 11.1 per cent of those starting the first-wave interview, a 12.1 percentage-point difference. The study also finds interactions between education and characteristics such as age, nationality and gender. That matters because simply calibrating weights to the population distribution of education assumes away these interactions unless they are explicitly modelled. + +The paper was published on 2 September, so it just falls outside this week’s release window. I am including it because it complements the new nonresponse work above and has direct relevance to the growth of push-to-web and digital-first surveys in development research. + +Distler, Christine, Mustafa Coban and Mark Trappmann. 2026. “Nonresponse in Web Surveys: A Deeper Look into Education Bias.” *Journal of Survey Statistics and Methodology*. DOI: 10.1093/jssam/smag031. Published 2 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article/doi/10.1093/jssam/smag031/8780203?utm_source=chatgpt.com) + +**Worth reading: high priority for anyone moving probability samples online.** + +## south asia methods watch: Sri Lanka launches a three-route citizen feedback survey + +Sri Lanka launched SL-PULSE on 10 September, a national system for measuring citizens’ experiences of public services across all 25 districts and 341 Divisional Secretariats. The programme expects to collect between 13,000 and 15,000 responses through three routes: open public participation, community-based field surveys and independent university-led household surveys. + +That design makes SL-PULSE interesting as a measurement exercise. The three routes imply very different selection mechanisms. Open digital participation offers speed and scale, while the resulting respondents cannot simply be assumed to represent the population. The household component could provide a reference point, depending on how its sample is drawn and how estimates across streams are combined. + +The public material currently describes seven governance areas, including service delivery, administrative procedures, accountability, corruption, environmental governance and digital governance. It does not yet provide enough technical detail to judge weighting, questionnaire validation or integration of the three samples. + +[Read the UNDP announcement](https://www.undp.org/srilanka/press-releases/sl-pulse-jnridm-janaraitama-first-its-kind-digital-survey-tool-strengthen-citizen-centred-public-service-reforms?utm_source=chatgpt.com) + +**Worth watching: yes. I would reserve judgement on the resulting indicators until the technical documentation appears.** + +## also worth watching: Pakistan combines demographic and health survey systems + +Pakistan’s 2026–27 Demographic and Health Survey is being developed within the Pakistan Bureau of Statistics after a government decision to integrate demographic, health, fertility and mortality modules into the existing Pakistan Demographic Survey architecture. PBS says it has reviewed both questionnaire content and survey methodology through technical consultations and plans national and provincial estimates covering population, fertility, maternal and child health, nutrition, mortality, family planning, gender and early childhood development. + +The institutional change is methodologically interesting because consolidation can reduce duplication and respondent burden while creating comparability problems. Changes in sampling, question wording, field protocols or survey ownership can produce breaks in time series that later look like substantive change. + +The next things to look for are the final sampling design, questionnaires and documentation showing how the 2026–27 series maps onto the 2017–18 PDHS. + +[Pakistan Bureau of Statistics PDHS page](https://www.pbs.gov.pk/elementor-28549/?utm_source=chatgpt.com) + +**Worth watching rather than reading: the technical documents will matter more than the announcement.** + +## what I would read first + +Start with Tsai et al. on translation. For multilingual work in South Asia, it addresses a source of measurement error that is routine, consequential and poorly documented. Li et al. comes next if you work with open text or automated coding. Zhang et al. is the most useful piece for fieldwork management. Sen et al. is the one I would keep nearby when somebody proposes replacing respondents with synthetic ones. + +The thread running through this week is fairly plain. Survey error often enters before estimation begins. A different translation, recruitment route, coding system or decision about when to stop fieldwork can alter the dataset that later appears in a regression table. Statistical sophistication downstream does not recover information that the measurement process distorted upstream. + +[Share](https://varna.substack.com/p/methods-monday-14-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-16-research-rundown-september-2026.md b/archive/2026-09-16-research-rundown-september-2026.md new file mode 100644 index 0000000..ba65186 --- /dev/null +++ b/archive/2026-09-16-research-rundown-september-2026.md @@ -0,0 +1,516 @@ +--- +title: "Research Rundown, September 2026" +date: 2026-09-16 +type: edition +url: https://varna.substack.com/p/research-rundown-september-2026 +subtitle: "India has more workers. Are they earning enough?" +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown, September 2026 + +*2026-09-16. Published at [https://varna.substack.com/p/research-rundown-september-2026](https://varna.substack.com/p/research-rundown-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown September 2026 flagship cover, with multiple statistical lines moving in different directions across a faint data grid, reflecting the issue’s examination of employment, wages, poverty, inflation and what improving indicators can tell us about how people are actually living.](https://substackcdn.com/image/fetch/$s_!VAig!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F12b796b8-7017-4a50-8a58-fe010669dc67_1672x941.png)](https://substackcdn.com/image/fetch/$s_!VAig!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F12b796b8-7017-4a50-8a58-fe010669dc67_1672x941.png) +*The Research Rundown - September 2026 * + +## When the numbers get better + +*Employment is up. Poverty is down. Growth has held up. Some rural wages are falling. Food inflation is rising. The problem is not deciding which number is correct. It is deciding what each number allows us to say about how people are living.* + +### From the editor + +*If you have been reading Research Rundown for a while, this issue may feel a little different.* + +When I started the Rundown in 2024, it was a monthly collection of things I thought were worth reading, watching, questioning or trying. Papers sat beside podcasts, research methods, books, datasets, jobs and the occasional bit of research humour. I still like that version of it. + +But Research Rundown has acquired several younger siblings since then. The Daily follows research and policy as they appear. Friday Findings does the heavier paper-finding. Methods Monday gives me somewhere to obsess over measurement and research design. Good Reads gives me room for work that deserves attention without requiring an entire issue. + +That leaves the original Rundown free to do something I have increasingly wanted it to do: stay with a question for longer. + +From this issue, the flagship Research Rundown will appear every two months. Each edition will begin with one problem and stay with it for a while: what the evidence says, where I think it is persuasive, where I think we are asking too much of it, and what I am still not sure about. + +There will still be papers, methods, research practice, things worth reading and the occasional detour. I am keeping the magazine, but giving it more charecter. Ideas will sometimes travel. Something first spotted in a Daily may return here after more evidence appears. A measurement problem from Methods Monday may change how we read a policy claim. A paper from Friday Findings may eventually become part of a larger argument. + +For the first issue in this form, I want to start with a problem that has been following me through several recent releases. + +**India’s economic indicators have been producing quite a lot of good news. I am increasingly interested in what happens when we put the good news next to one another, and then ask what exactly has improved.** + +--- + +## The question + +On 15 September, India’s latest labour-market numbers arrived. + +They were good. + +The labour-force participation rate among people aged 15 and above reached 55.6 per cent in August, up from 55.4 per cent in July. The worker-population ratio reached 52.8 per cent, its highest level since March. Rural employment continued to rise. Female labour-force participation reached 34.8 per cent, 1.1 percentage points above August 2025. Overall unemployment stood at 5.0 per cent. + +A day earlier, another number had arrived. + +Consumer inflation rose to 4.82 per cent in August. Rural inflation was 5.23 per cent. Rural food inflation was 6.13 per cent. + +A week before that, CMIE wage data reported by the *Financial Express* showed average rural monthly wages falling from ₹21,250 in April to ₹19,922 in July. Combined with rising prices, the reported decline in real rural wages over those three months was 8.9 per cent. Urban real wages fell by 1.9 per cent. + +Meanwhile, the World Bank estimates that extreme poverty in India fell from 27.1 per cent in 2011-12 to 2.6 per cent in 2023-24 using its $3.00-a-day 2021 PPP line. At the $4.20 lower-middle-income-country line, poverty fell from 57.7 to 16.1 per cent. + +India’s economy grew 7.6 per cent in FY2025-26. + +None of these figures needs to be false for the others to be true. + +That is the point. + +I think we have a problem in Indian development discussion with the category of “improvement”. We are often careful about the accuracy of an indicator and much less careful about the distance between the indicator and the thing we actually want to know. + +Employment is a particularly good place to see this. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### More work + +The August PLFS result extends a movement that was already visible in July. Overall labour-force participation rose from 54.4 per cent in June to 55.4 per cent in July and 55.6 per cent in August. Rural LFPR reached 58.2 per cent in August. The rural worker-population ratio reached 55.8 per cent, 1.3 percentage points higher than a year earlier. + +These are meaningful changes. + +They also answer fairly narrow questions. + +The monthly PLFS uses Current Weekly Status. Someone who worked for at least one hour on any day during the reference week can enter the employed population under the survey’s activity-status rules. The unemployment rate then tells us what share of the labour force was unemployed. + +This makes unemployment a poor standalone measure of economic security in a labour market where many people cannot afford prolonged open unemployment. + +If a worker loses reasonably paid work and takes irregular casual work at a lower wage, measured employment survives. + +If a woman joins a family enterprise without regular independent earnings, participation rises. + +If a household responds to an income shock by sending another member into poorly paid work, LFPR may improve. + +Employment statistics are doing their job in each case. The interpretation fails when we quietly turn “employed” into “doing well”. + +Women make the problem particularly visible. + +Female LFPR in India has risen dramatically over the longer PLFS period. Under usual status, it increased from 23.3 per cent in 2017-18 to 41.7 per cent in 2023-24. + +I regard that movement as important. I do not regard 41.7 per cent as a sufficient measure of women’s economic progress. + +Employment status matters. So do hours, earnings, occupation, working conditions, continuity of work, unpaid family labour and control over earnings. + +There is an old temptation in development measurement to become attached to an indicator precisely when it starts moving in the direction we hoped it would. Female LFPR deserves more scrutiny now, not less. + +### More employment, less purchasing power? + +This is where the wage numbers become uncomfortable. + +CMIE’s reported rural wage series shows nominal average monthly wages declining 6.25 per cent between April and July. Rural prices rose 2.9 per cent over the same period. Put together, real rural wages fell 8.9 per cent. The rural wage was about 66 per cent of the average urban wage in April and 62 per cent in July. + +I would take this seriously. I would not yet call it evidence of a general rural wage collapse. + +Three months is short. Agricultural labour demand is seasonal. The timing of the monsoon and kharif sowing matters. Composition matters too. + +There is also a measurement problem hidden inside the apparently straightforward term “rural wage”. + +The Labour Bureau’s Wage Rates in Rural India database reports daily wage rates by state, sex and occupation. It distinguishes ploughing, sowing, harvesting, horticulture, general agricultural labour, construction and other occupations. In January 2026, for example, Andhra Pradesh reported ₹613.31 per day for male sowing, planting and weeding workers and ₹443.87 for women doing the same category of work. + +A national monthly wage average cannot tell us whether a decline comes from lower wages within occupations, movement between occupations, fewer days of paid work, a changing mix of workers, or some combination. + +Those mechanisms have different policy implications. + +The distinction between wage rates and earnings matters as well. A daily wage can rise while monthly earnings fall because there are fewer days of work. + +This sounds elementary. Yet public discussion repeatedly slides between wages, earnings, employment and income as though they were interchangeable. + +They are not. + +### Then prices enter the room + +The August CPI makes the question more pressing. + +Headline inflation was 4.82 per cent. Rural inflation was 5.23 per cent. Food inflation was 5.95 per cent nationally and 6.13 per cent in rural India. + +India is also measuring those prices differently now. + +MoSPI moved the CPI base from 2012 to 2024 in February. The new series expanded price collection to 1,465 rural markets and 1,395 urban markets across 434 towns, added online markets and increased the all-India weighted basket from 299 to 358 items. + +This is good statistical housekeeping. Consumption changes. An inflation basket should change with it. + +But every CPI remains a weighted average of a consumption basket. + +A household spending a large share of its budget on food experiences a 6 per cent increase in food prices differently from a household whose spending is distributed across housing, transport, education, services and discretionary consumption. + +The official CPI is not claiming otherwise. + +Our interpretation often does. + +I find this distinction particularly important because inflation is routinely discussed as though the national headline rate describes a shared experience. It describes a price index. + +Those are different propositions. + +--- + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Poverty fell. What exactly have we established? + +The poverty numbers require a different sort of care. + +The World Bank’s latest long-run estimates use the 2011-12 Consumption Expenditure Survey and the 2023-24 Household Consumption Expenditure Survey. At its revised extreme-poverty line of $3.00 per person per day in 2021 PPP terms, it estimates that poverty fell from 27.1 per cent to 2.6 per cent. At the $4.20 line used for lower-middle-income countries, the decline is from 57.7 to 16.1 per cent. + +Those are very large reductions. + +My reading is that the broad conclusion is difficult to dispute: India experienced substantial consumption-poverty reduction between 2011-12 and 2023-24. + +I am considerably less persuaded by attempts to make the 2.6 per cent number carry the entire argument about deprivation in contemporary India. + +The first reason is obvious. A poverty line creates a binary classification around a continuous distribution. + +A household immediately above $3.00 has crossed the extreme-poverty threshold. Its economic security has not undergone a corresponding discontinuous transformation. + +The World Bank’s own $4.20 estimate makes this visible. About 16.1 per cent of Indians remain below that threshold. That is roughly 232 million people in its calculation. + +The second issue is survey comparability. + +The 2023-24 HCES is a much better basis for contemporary consumption analysis than extrapolating indefinitely from 2011-12. But survey design changed over those twelve years. The World Bank therefore uses a harmonisation methodology to construct its long-term trend and has published a reproducibility package for the estimates. + +This is exactly what serious poverty estimation should do. It also means the resulting time series is constructed, not simply read off two identical surveys. + +That distinction is sometimes lost once the estimate enters political or public discussion. + +The third issue is inequality. + +MoSPI’s HCES reports that the Gini coefficient for monthly per-capita consumption expenditure fell from 0.283 in rural India in 2011-12 to 0.237 in 2023-24, and from 0.363 to 0.284 in urban India. + +Again, this is evidence worth taking seriously. + +It is evidence about the distribution of measured consumption. + +It does not settle questions about wealth inequality, income inequality, unequal access to public services, indebtedness, insecurity, caste disparities, gendered control over resources or exposure to shocks. + +The temptation here works in both directions. People sceptical of India’s growth story sometimes dismiss favourable consumption data too readily. People defending the growth story sometimes ask consumption poverty to prove much more than it measures. + +Neither move improves the analysis. + +--- + +## The South Asian version of the problem + +India is hardly alone in this. + +Bangladesh provides a useful counterexample because some of its headline indicators have moved in the opposite direction. The World Bank estimates that Bangladesh’s national poverty rate increased from 18.7 per cent in 2022 to 21.4 per cent in 2025. Inflation remained 8.5 per cent in FY2025-26, and wages among low-income workers failed to keep pace with prices. + +There the signals line up more neatly: high inflation, weak real wages and rising poverty. + +Sri Lanka is more interesting. + +Its economy returned to growth after the crisis, expanding 5 per cent in 2024. Inflation fell sharply. Yet the World Bank estimated poverty at 24.5 per cent, roughly twice its 2019 level, while real wages remained below their 2019 level. Food prices had more than doubled between 2021 and 2024. + +Macroeconomic recovery and household recovery were occurring on different clocks. + +Nepal gives us another version. Extreme poverty fell from 21.5 per cent in 2011 to 3 per cent in 2023, while migration and remittances remain central to household livelihoods. The World Bank reports that 77.3 per cent of Nepali migrant workers are based in the Middle East. + +What does a successful domestic development indicator mean when part of the mechanism producing it is the export of labour? + +I do not mean that rhetorically. Remittances are income. Migration can improve welfare substantially. Workers make rational choices within the opportunities available to them. + +But a fall in poverty generated partly through overseas earnings tells us something different about the domestic economy from a fall generated through rising local labour productivity and wages. + +The poverty statistic alone cannot distinguish the development paths. + +Across South Asia, the World Bank estimates that only 59 per cent of the working-age population is employed and female labour-force participation is about 32 per cent. Formal employment accounts for only around 10 per cent of the workforce. + +This is why I am wary of treating South Asia’s high growth as though the jobs question will eventually resolve itself. + +The region has been growing quickly for years. Its employment structure is still telling us something uncomfortable. + +--- + +## Heat is an economic indicator too + +There is another number I have been thinking about since it appeared in July. + +The World Bank estimates that extreme heat is already costing South Asia labour equivalent to almost 31 million full-time jobs each year. By 2050, it estimates that heat could reduce the region’s economy by nearly 7 per cent. + +The 2050 figure is a modelled projection and should be read as one. The 31-million-job-equivalent figure is also an estimate of labour-productivity losses, not a count of 31 million identifiable people who have lost jobs. + +Still, the underlying problem is real and badly fitted to our conventional labour indicators. + +Imagine a construction worker who remains employed through an extreme-heat period but works fewer effective hours, takes more breaks, loses days to illness, or earns less because output falls. + +Employment status may not change. + +Income can. + +Health can. + +Household welfare can. + +The same problem appears in agriculture, street vending, delivery work, sanitation, transport and other occupations where workers cannot readily move labour indoors or shift working hours. + +Climate exposure therefore creates another wedge between having work and what that work yields. + +This matters for how we evaluate adaptation policy too. Counting heat action plans, cooling centres or alerts is straightforward. Estimating avoided illness, preserved earnings and distributional effects is much harder. + +I increasingly think climate policy has the same measurement problem as employment policy: we count what administrative systems can see. + +--- + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Measurement matters: the indicator is not the construct + +This issue began with a collection of numbers. It has ended up being about construct validity. + +Suppose the construct we care about is something like **material economic security**. + +We cannot observe it directly. + +So we use proxies: + +employment status +earnings +consumption +poverty status +inflation-adjusted income +assets +food expenditure +programme participation +debt +subjective economic security + +Each captures something. + +None is the construct. + +Researchers know this in principle. In practice, the distinction often weakens somewhere between a survey table, a press release and a policy claim. + +There are four questions I would ask whenever a development indicator appears to improve. + +**What exactly was measured?** + +PLFS unemployment is a labour-force measure. HCES MPCE is a consumption measure. CPI measures price change for a weighted basket. GDP measures production. Each has a definition, reference period and population. + +**What mechanism could have produced the change?** + +A rising female LFPR can reflect expanding opportunity, economic distress, agricultural seasonality or changing unpaid work. A lower poverty rate can result from higher earnings, transfers, remittances, lower dependency ratios or changes elsewhere in household consumption. + +**Who moved?** + +An all-India mean can improve while particular states, occupations or population groups deteriorate. Distribution is not an optional appendix to the average. + +**What would I need to observe before making the stronger claim?** + +This is the question I think we ask least often. + +If employment rises, I want earnings and employment status. + +If women’s participation rises, I want occupation, hours, earnings, unpaid work and control over income. + +If poverty falls, I want the distribution around the threshold and measures of vulnerability to shocks. + +If inflation falls, I want expenditure-specific incidence. + +If a programme reaches more people, I want adequacy and outcomes. + +The stronger claim may eventually be justified. It needs stronger evidence. + +--- + +## Research praxis: build the claim before choosing the indicator + +One habit I recommend to researchers I work with is embarrassingly simple. + +Write the substantive claim before opening the dataset. + +Suppose the claim is: + +> Women’s economic position in rural India improved between two survey rounds. + +Now list what would have to be true for you to believe it. + +Would a rise in LFPR be enough? + +Probably not. + +Would increased paid employment be enough? + +Better. + +Would rising real earnings strengthen the claim? + +Yes. + +Would fewer women reporting unpaid family work matter? + +Yes. + +Would greater control over earnings matter? + +Very much, although PLFS cannot answer it. + +Would reduced time poverty alter the interpretation? + +Again, yes. + +You have now done something useful before running a regression. You have separated the construct from the indicators available to measure it. + +This also exposes the missing-data problem early. + +Many evaluation frameworks do the reverse. They begin with indicators because indicators are available. The programme’s substantive claim is then quietly rewritten to fit them. + +A monitoring system records training completion, so completion becomes “capacity”. + +An administrative database records registrations, so registration becomes “access”. + +A livelihood programme records enterprises started, so enterprise creation becomes “economic empowerment”. + +The mistake happens before estimation. + +No econometric technique repairs a badly specified construct. + +--- + +## What I think is worth reading + +### 1. India Development Update, April 2026 + +The World Bank’s latest India update is worth reading partly because of a technical issue that received less attention than the growth forecast: India’s February 2026 GDP rebasing. + +The base year moved to 2022-23, and the revised methodology incorporates more regular data on the household and informal sectors. MoSPI says annual sources such as ASUSE and PLFS will now contribute directly to household-sector estimates rather than relying as heavily on older benchmark surveys and proxy growth indicators. + +This matters. The informal economy has always been one of the hardest parts of Indian GDP to estimate. + +A rebasing exercise that changes our estimate of the size of that economy is not statistical housekeeping at the margins. It changes the denominator against which several other economic quantities are understood. + +[India Development Update, April 2026](https://thedocs.worldbank.org/en/doc/4262e1e15b463ecb360cec4ad78cf062-0310012026/?utm_source=chatgpt.com) + +### 2. The World Bank’s India poverty reproducibility package + +I wish reproducibility packages received a fraction of the attention that poverty headlines do. + +The package provides the material behind the Bank’s updated 2011-12 to 2023-24 poverty series and specifies the $3.00 and $4.20 2021-PPP poverty lines used in the estimates. + +If you disagree with a poverty estimate, this is where the interesting argument should begin: welfare aggregate, survey comparability, price adjustments, poverty line, treatment of transfers, sampling and sensitivity. + +[India long-term poverty trend reproducibility package](https://reproducibility.worldbank.org/catalog/541?utm_source=chatgpt.com) + +### 3. South Asia Economic Update: Working with Industrial Policy + +I am less interested in its headline regional growth forecast than in the work on jobs and spatial wage differences. + +The report finds persistent subnational wage premiums in South Asia’s larger economies even after controlling for worker characteristics. Better transport connections, skilled workforces, larger firms and service-sector activity are associated with higher residual wage premiums. + +That is a useful corrective to national labour-market averages. Where a worker lives remains economically consequential after observable individual characteristics are taken into account. + +[South Asia Economic Update, April 2026](https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099919404082611180?utm_source=chatgpt.com) + +### 4. A Livable Future: Protecting Jobs and Growth from Extreme Heat in South Asia’s Cities + +Read this alongside labour-market statistics rather than filing it under climate. + +Its central contribution is to treat heat exposure as a labour and productivity problem. I would read the large aggregate estimates cautiously because they depend on modelling assumptions. The occupational logic is much harder to dismiss. Workers exposed to outdoor and poorly cooled environments experience heat through hours, output, illness and earnings. + +That belongs in any serious discussion of South Asia’s future of work. + +[World Bank material on extreme heat, jobs and growth in South Asia](https://www.worldbank.org/en/news/press-release/2026/07/28/extreme-heat-puts-south-asia-s-jobs-and-growth-at-risk-world-bank?utm_source=chatgpt.com) + +--- + +[Subscribe now](https://varna.substack.com/subscribe?) + +## What I’m thinking about + +I keep coming back to the difference between **crossing a threshold and becoming secure**. + +Development research uses thresholds everywhere. + +Poverty lines. Minimum dietary diversity. Anaemia cut-offs. Learning proficiency. Minimum wages. Heat thresholds. Eligibility scores. Anthropometric cut-offs. + +We need them. Policy cannot operate entirely in continuous distributions. + +But thresholds have an odd effect on thought. Once someone crosses one, the classification changes abruptly while the person’s life may have changed only slightly. + +A household at ₹99 against a ₹100 threshold and a household at ₹101 can be almost indistinguishable economically. In the dataset, one is poor and one is not. + +The same problem appears when a child’s test score crosses a proficiency cut-off or a household becomes administratively eligible for a scheme. + +This is hardly a new methodological discovery. What interests me is how readily the categorical result takes over the substantive discussion. + +Perhaps we need to report **distance from thresholds** much more routinely. + +For poverty, that means poverty gaps and vulnerability alongside headcounts. + +For learning, score distributions alongside proficiency rates. + +For nutrition, distributions around clinical cut-offs. + +For programme eligibility, the density immediately above and below the administrative boundary. + +A threshold is often necessary for a decision. + +It is rarely enough for an explanation. + +--- + +## Elsewhere in the Rundown + +Several things that have appeared in the newer Research Rundown series now look related to this question. + +The recent Daily discussion of falling real rural wages is one obvious route into this issue. The Methods Monday work on measurement has repeatedly returned to construct validity and the problem of treating an available variable as though it were the phenomenon of interest. Friday Findings has been turning up work on labour markets, public systems and shocks that makes much more sense when read together than as separate papers. + +This is the sort of connection I want the flagship to make. + +The faster editions can tell us that something happened. + +Here, I want to ask what we are entitled to conclude from it. + +--- + +## Off the desk + +One dataset rather than a book this time: the Labour Bureau’s **Wage Rates in Rural India** database. + +It is not glamorous. The interface could be kinder. It is also the sort of public dataset that rewards wandering around in it. + +Pick an occupation. Compare men and women. Move across states. Then change occupations. + +The national category “rural labour” becomes difficult to sustain for very long. + +In January 2026, Andhra Pradesh’s reported daily wage for male ploughing and tilling workers was ₹691.65. Male workers in sowing, planting, transplanting and weeding averaged ₹613.31; women in that category averaged ₹443.87. + +Those differences are the labour market too. + +[Labour Bureau: Wage Rates in Rural India](https://labourbureau.gov.in/rural-wages?utm_source=chatgpt.com) + +--- + +## Where this leaves me + +I started this issue with employment because the latest numbers looked good. + +I still think they look good. + +August brought higher labour-force participation, a higher worker-population ratio and continued gains in women’s participation. India has also experienced a very large reduction in consumption poverty over the longer term. Those findings should not be waved away because they complicate a preferred account of the economy. + +I also think the evidence is telling us to ask a harder question. + +**For a household, economic progress is experienced through what work pays, how predictable it is, what necessities cost, whether illness or heat interrupts earnings, whether debt is accumulating, whether public services work, and how far the household sits from the next shock.** + +Our major statistical systems observe pieces of this remarkably well. + +The mistake is ours when we make one piece stand for the whole. + +That matters especially now because some of India’s indicators are moving in different directions. Employment is strengthening while recent rural wage data are weak. Food inflation has risen. Female participation is increasing without giving us a simple answer about the quality of women’s work. Consumption poverty has fallen substantially while a much larger population remains below a poverty line appropriate to lower-middle-income countries. + +I do not read that as a contradiction in the data. + +I read it as a warning about the question. + +“Is India doing better?” is too imprecise to be useful. + +Better on what measure, for whom, compared with when, and how far from insecurity? + +Those are questions the data can begin to answer. + +They require more than one number. + +**For readers working with household or labour data in India or elsewhere in South Asia: what measure have you found changes the story most once you put it beside the headline indicator? I would particularly like to see state-level evidence, panel data or work that follows households around a poverty, employment or earnings threshold. If you send me something I use in a future Rundown, I will credit you.** + +[Share](https://varna.substack.com/p/research-rundown-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-17-good-reads-7-what-migration-data.md b/archive/2026-09-17-good-reads-7-what-migration-data.md new file mode 100644 index 0000000..c8452cb --- /dev/null +++ b/archive/2026-09-17-good-reads-7-what-migration-data.md @@ -0,0 +1,137 @@ +--- +title: "Good Reads #7: What migration data cannot tell us" +date: 2026-09-17 +type: other +url: https://varna.substack.com/p/good-reads-7-what-migration-data +subtitle: "Nehal Ahmed’s account of letters, telephone calls and Gulf migration from a village in Bihar is also a reminder that some of the consequences of migration sit outside the categories we usually measure" +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #7: What migration data cannot tell us + +*2026-09-17. Published at [https://varna.substack.com/p/good-reads-7-what-migration-data](https://varna.substack.com/p/good-reads-7-what-migration-data). This file is a generated copy; the Substack post is the original.* + +[![Alt text: *Research Rundown Good Reads cover for “What migration data cannot tell us,” showing old letters and airmail envelopes, a photograph of a postman on a bicycle, a travel case and a distant city, reflecting Nehal Ahmed’s account of communication, family separation and Gulf migration.*](https://substackcdn.com/image/fetch/$s_!fkT2!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F84b53dc6-3a91-476e-af9e-c3013eefb123_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!fkT2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F84b53dc6-3a91-476e-af9e-c3013eefb123_1536x1024.png) +*Good Reads#7 on 17th September 2026* + +[Read Nehal Ahmed, “Ranglal, hamari koi chitthi aayi hai?” in The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai) + +There is a moment near the end of Nehal Ahmed’s essay when Ranglal, the village postman, appears. + +For years, people in the village would see him cycling past and ask whether a letter had arrived from someone working away in the Gulf. Ranglal delivered letters and money orders. He also read letters for people who could not read them and wrote replies for those who could not write. When there was no letter, people sometimes told him what they had wanted to tell the absent person anyway. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +It is a lovely detail. It is also, from a researcher’s point of view, an unusually useful one. + +Much of what we know about migration comes from questions that are necessarily easier to count. Who migrated? Where did they go? For how long? What work did they do? How much did they earn? Did they send money home? + +Ahmed’s essay asks us, without quite putting it this way, what disappears when those become our principal categories for understanding migration. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Migration is also an infrastructure of absence + +The essay follows several generations of Gulf migration from Ahmed’s village in Bihar through the technologies families used to remain connected: letters carried by other migrants, landlines, audio cassettes, cameras and film rolls, calling systems, early mobile phones, cyber cafés and eventually smartphones, WhatsApp and video calls. His own father worked in Saudi Arabia. By the time later generations migrated, both the work migrants did and the technologies connecting them to home had changed. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +Read simply as a history of communications technology, this is interesting enough. + +But the technology is really a way of observing something else: how migration reorganises family life. + +A migrant can be economically present in a household while being physically absent from it. Money travels home. Decisions are made across enormous distances. Children grow up between visits. Marriages are conducted partly through correspondence. Photographs become records of years somebody did not witness directly. + +The cost of migration, then, cannot be reduced to the cost of moving or the conditions of employment at the destination. + +Distance itself has consequences. + +## Who controls the telephone matters + +One of the strongest parts of the essay concerns gender. + +Landlines eventually made communication faster, but Ahmed recalls that they were commonly situated in the men’s section of the household. For young married women living with their husbands’ families while those husbands worked abroad, access to a telephone did not necessarily mean access to a private conversation. + +Letters could offer something the supposedly superior technology could not: privacy. + +Audio cassettes extended that private space further. A woman could record a message away from the household and send it to her husband. Ahmed places this alongside the better-known cassette culture associated with Gulf migration from Kerala. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +This is a small but important methodological lesson. + +If we measured communication by asking whether a household had access to a telephone, we could easily record an improvement. The household had acquired the technology. Contact had become technically possible. + +But access at household level tells us very little about who can actually use a technology, under what conditions, and with whom. + +Those are different variables. + +It is the sort of distinction that matters well beyond this particular story. Household ownership is often a poor proxy for individual control. A phone, bank account, vehicle or internet connection can exist within a household while access to it remains structured by gender, age, income and authority. + +## Technology changes distance. It does not abolish it. + +Ahmed describes another transition that will be familiar to many migrant families. + +With cheap internet and smartphones, communication becomes frequent and immediate. Migrants can see their children rather than discovering, on returning home after years away, how much they have grown. Families exchange photographs, messages, voice notes and video calls and can remain involved in ordinary daily life from thousands of kilometres away. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +Something has plainly improved. + +But I found myself wondering what exactly we would measure if we wanted to understand that improvement. + +Frequency of contact is possible. Cost of communication is possible. Smartphone ownership is possible. Internet access is possible. + +None quite captures the change Ahmed is describing. + +What has changed is partly the experience of absence. + +That is much harder to put into a survey variable. + +## The things surveys flatten + +This is why I wanted to include this piece in Good Reads. + +It is not a research paper and should not be made to pretend to be one. Ahmed is writing from family memory and personal experience, supported by photographs, letters and the history of his village. It cannot tell us how representative these experiences were across Bihar, much less across India’s many migration corridors. + +That is precisely why it is useful alongside quantitative research rather than in competition with it. + +A migration survey can estimate prevalence and patterns. Administrative records can tell us about passports, borders, transfers or employment. Household surveys can examine remittances and consumption. Longitudinal work can follow changes over time. + +But the categories we choose determine what becomes visible. + +If the principal economic story of migration is that a worker moves from a place with fewer opportunities to one with better-paid work and sends income home, we can miss much of the institution that forms around that movement. + +Someone carries the letter. + +Someone waits for the call. + +Someone controls the household telephone. + +Someone writes on behalf of the person who cannot write. + +Someone raises the children who remain behind. + +Someone learns how to operate the cyber café. + +Someone keeps the photograph. + +These are not decorative details around the economic fact of migration. They are part of how migration works. + +## And then there is Ranglal + +By the time Ahmed reaches Ranglal, communications technology has changed almost beyond recognition. + +The postman who once connected migrant families has disappeared, and almost no visual record of him remains. Years later, when Ahmed’s brother tells his own daughters about Ranglal, one of them asks what he looked like. She reconstructs an image of him on an iPad from her father’s description. It becomes the family’s only visual representation of a man who once carried so much of the village’s communication. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +There is an obvious irony in that ending, and Ahmed lets it stand. + +For me, it also raises a question about evidence. + +We tend to think of the digital era as producing an extraordinary abundance of records. Earlier everyday lives can look sparse by comparison. Yet what survives has never been a neutral sample of what mattered. + +Ranglal mattered enormously and left almost no archive. + +That problem should be familiar to researchers. What is recorded is not necessarily what was important. What is measurable is not necessarily what was consequential. And an absence in the data does not mean an absence in social life. + +Good research needs the survey and the story. + +Sometimes the story tells us what we forgot to ask. + +--- + +Source: Nehal Ahmed, “‘Ranglal, hamari koi chitthi aayi hai?’,” *The Migration Story*, 16 September 2026. The essay is the second of a two-part series and was co-published with Arisi Policy Lab. ([The Migration Story](https://themigrationstory.com/post/ranglal-hamari-koi-chitthi-aayi-hai)) + +[Share](https://varna.substack.com/p/good-reads-7-what-migration-data?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-17-research-rundown-daily-16-september.md b/archive/2026-09-17-research-rundown-daily-16-september.md new file mode 100644 index 0000000..86647a7 --- /dev/null +++ b/archive/2026-09-17-research-rundown-daily-16-september.md @@ -0,0 +1,165 @@ +--- +title: "Research Rundown DAILY - 16 September" +date: 2026-09-17 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-16-september +subtitle: "Extreme heat is costing Bangladesh workers time, income and food security, exposing losses that conventional statistics rarely capture." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 16 September + +*2026-09-17. Published at [https://varna.substack.com/p/research-rundown-daily-16-september](https://varna.substack.com/p/research-rundown-daily-16-september). This file is a generated copy; the Substack post is the original.* + +[![**Alt text:** Research Rundown DAILY cover for 16 September 2026, showing a worker’s time-and-wage register with working hours fading as temperatures rise, alongside a map of Dhaka and household expenses, illustrating how extreme heat can reduce working time and income.](https://substackcdn.com/image/fetch/$s_!_HBW!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F35d89ae7-e869-4e20-a7ec-eba2b471ecd2_1672x941.png)](https://substackcdn.com/image/fetch/$s_!_HBW!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F35d89ae7-e869-4e20-a7ec-eba2b471ecd2_1672x941.png) +*Research Rundown DAILY - 16 September 2026* + +## FIVE THINGS WORTH KNOWING + +### 1. Bangladesh had 25 extra days of risky heat this summer because of climate change + +Bangladesh experienced 31 days between June and August when temperatures exceeded the local 90th-percentile threshold. A new Climate Central attribution analysis estimates that human-caused climate change added **25 of those risky-heat days**. ([United News Bangladesh](https://unb.com.bd/category/Bangladesh/climate-change-adds-25-risky-heat-days-to-bangladeshs-summer-new-climate-central-analysis-finds/195482?utm_source=chatgpt.com)) + +That finding matters because exposure is translating into economic loss through work. Separate research released into the policy discussion this week links extreme heat to reduced working hours, lower earnings and food insecurity, particularly among workers whose income depends directly on daily labour. A Bangladesh study by PRAAN also reports heavier financial burdens among women-headed households and points to the absence of a dedicated institutional framework for extreme heat. ([Dhaka Stream](https://www.dhakastream.net/node/198880?utm_source=chatgpt.com)) + +The attribution estimate and the household findings measure different things and should stay separate. Climate Central estimates how climate change altered heat exposure. The household research examines consequences among selected populations. Neither, on its own, gives a national estimate of income lost to heat. + +**Why it matters:** Heat policy is still often organised around temperature alerts and health emergencies. For low-income workers, the relevant outcome may be income forgone because working safely becomes impossible. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. Bangladesh is preparing a Tk 4,039 crore TB, HIV and malaria programme as external financing declines + +Bangladesh has prepared a three-year **Tk 4,039 crore** programme covering tuberculosis, HIV/AIDS and malaria for July 2026 to June 2029. Government financing would provide Tk 3,168.89 crore, with Tk 870.18 crore coming from the Global Fund. ([The Daily Star](https://www.thedailystar.net/news/bangladesh/news/govt-plans-tk-4039cr-drive-against-tb-hiv-malaria-4273956?utm_source=chatgpt.com)) + +The targets are substantial: preventive TB treatment for about 1.5 million contacts of pulmonary TB patients, nutritional support for 824,450 TB patients, HIV testing across all 64 districts and zero malaria deaths by 2029. Bangladesh recorded 1,891 new HIV cases in 2025, up from 1,438 in 2024, while its malaria-elimination deadline has moved from 2030 to 2034. ([The Daily Star](https://www.thedailystar.net/news/bangladesh/news/govt-plans-tk-4039cr-drive-against-tb-hiv-malaria-4273956?utm_source=chatgpt.com)) + +The financing context deserves attention. The previous Health, Population and Nutrition Sector Programme ended in June 2024, and declining donor support has already affected some medicine and testing services. Globally, the Global Fund says recent aid cuts have hit prevention harder than treatment, including TB case-finding and HIV prevention. ([Reuters](https://www.reuters.com/business/healthcare-pharmaceuticals/aid-cuts-hit-disease-prevention-harder-than-treatment-global-fund-chief-says-2026-09-16/?utm_source=chatgpt.com)) + +This is therefore partly a disease-control programme and partly a test of whether domestic public finance can replace external health funding without losing prevention capacity. + +### 3. Maharashtra now has a state climate-finance taxonomy + +Maharashtra released its **State Climate Finance Framework** today, developed by the state government, MITRA and CEEW. It creates a classification and screening system for mitigation, adaptation and cross-cutting investments and is intended for use by government departments, lenders and investors. ([CEEW](https://www.ceew.in/publications/maharashtra-state-climate-finance-framework?utm_source=chatgpt.com)) + +The interesting figure is **80 per cent**. Adaptation accounts for nearly four-fifths of Maharashtra’s proposed climate budget, according to the framework, while remaining the category least attractive to private capital because returns are difficult to monetise and investment horizons are long. ([CEEW](https://www.ceew.in/publications/maharashtra-state-climate-finance-framework?utm_source=chatgpt.com)) + +The framework therefore gives adaptation its own four-step screening pathway, including an explicit test for maladaptation. It is also provisional. Maharashtra intends to revise it when India’s national Climate Finance Taxonomy is formally notified. ([CEEW](https://www.ceew.in/publications/maharashtra-state-climate-finance-framework?utm_source=chatgpt.com)) + +The document is worth watching because classification affects money. Once an expenditure is labelled climate finance, the harder question is whether it produces a measurable reduction in vulnerability. + +### 4. An eight-year railway delay in Bangladesh has produced a much larger bill + +The proposed direct dual-gauge railway between Bogura and Sirajganj has been under development for eight years. Its revised estimated cost is now reported at about **Tk 10,391 crore**, nearly 89 per cent above the original estimate, with the Asian Infrastructure Investment Bank expected to finance 71 per cent. ([The Daily Star](https://online.thedailystar.net/news/bangladesh?utm_source=chatgpt.com)) + +There is already a data problem worth noticing. Another report today gives the revised cost as **Tk 10,242 crore**, an increase of roughly 84 per cent over the original Tk 5,579.70 crore. ([Daily Observer](https://observerbd.com/news/593592?utm_source=chatgpt.com)) + +That discrepancy is a reason to avoid false precision until the revised Development Project Proposal is formally approved. The broader point survives either figure: long project-preparation periods impose costs through price changes, design revisions and delayed benefits. + +For development evaluation, “cost escalation” also needs decomposition. Inflation, exchange-rate movements, scope changes, financing costs and genuine implementation failure are different explanations and should not be collapsed into one percentage. + +### 5. Pakistan’s first fintech sandbox has produced four successful tests, but SBP says that is not regulatory approval + +The State Bank of Pakistan has completed the first cohort of its regulatory sandbox. Six participants were selected for testing between January and June 2026, and four successfully demonstrated products involving inward remittances and open banking. ([Dawn](https://www.dawn.com/news/amp/2030248?utm_source=chatgpt.com)) + +SBP has issued an unusually useful clarification: passing the sandbox **does not constitute permission to launch commercially**. Participants still need the relevant regulatory authorisations. ([Dawn](https://www.dawn.com/news/amp/2030248?utm_source=chatgpt.com)) + +That distinction is easy to lose when innovation programmes report the number of products that “graduate”. A sandbox establishes whether something can be tested successfully under controlled conditions. It does not establish consumer benefit, market viability, financial inclusion effects or safety at scale. + +The first cohort should therefore be treated as regulatory learning rather than evidence that the products themselves have succeeded. + +--- + +# ONE PAPER WORTH READING + +### *No escaping the heat: health and economic impacts of heat stress on informal workers in urban Bangladesh* + +This IIED study is particularly useful alongside today’s Bangladesh heat-attribution numbers. Ritu Bharadwaj and colleagues surveyed **175 informal-worker households in Dhaka**, examining workers at home and at work using the Comprehensive Climate Impact Quantification toolkit. The study attempts to count losses that standard economic statistics often miss: workdays and wages lost to heat, health expenditure, illness and other household effects. ([IIED](https://www.iied.org/22755iied?utm_source=chatgpt.com)) + +Its contribution is measurement rather than national prevalence. Heat exposure becomes economically important before it produces hospitalisation or mortality. A worker who shortens a shift, works more slowly or stays home loses income even though none of those outcomes necessarily appears in conventional disaster-loss accounts. + +The limitation follows directly from the design. **175 households in Dhaka cannot represent Bangladesh’s informal workforce**, and estimates based on reported household experience require care when translated into aggregate economic losses. Occupational exposure also differs substantially across jobs. + +Still, this is exactly the sort of work needed to improve heat-loss accounting. Conventional productivity measures can miss the welfare cost when workers respond to unsafe temperatures by sacrificing either earnings or health. + +**Citation:** Bharadwaj, R., N. Karthikeyan, S. Chaliha, S. Islam and U. Tanni. 2026. *No escaping the heat: health and economic impacts of heat stress on informal workers in urban Bangladesh*. London: International Institute for Environment and Development. ([IIED](https://www.iied.org/22755iied?utm_source=chatgpt.com)) + +[Read the open-access IIED report](https://www.iied.org/22755iied?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +--- + +# ONE NUMBER + +## 25 days + +The estimated number of additional risky-heat days Bangladesh experienced during June-August 2026 because of human-caused climate change. Bangladesh recorded 31 days above the locally defined 90th-percentile temperature threshold during the period. ([United News Bangladesh](https://unb.com.bd/category/Bangladesh/climate-change-adds-25-risky-heat-days-to-bangladeshs-summer-new-climate-central-analysis-finds/195482?utm_source=chatgpt.com)) + +The denominator matters. This is **not** a claim that climate change caused 25 days that would otherwise have been cool. Attribution analysis compares the probability of temperatures under the present climate with a counterfactual climate without human-caused warming. + +That makes the number useful for a different reason: it begins to separate ordinary seasonal exposure from additional exposure attributable to climate change. + +--- + +# POLICY WATCH + +## Maharashtra is trying to define what counts as climate finance + +Climate-finance debates often start with how much money is available. Maharashtra’s new framework starts one step earlier: **which expenditure is entitled to count?** + +Its taxonomy screens proposed activities as mitigation, adaptation or cross-cutting investments and includes safeguards intended to identify maladaptation. The framework says **34 of Maharashtra’s 36 districts are exposed to extreme climate events**, while a quarter of districts show changing patterns between flood and drought exposure. ([CEEW](https://www.ceew.in/publications/maharashtra-state-climate-finance-framework?utm_source=chatgpt.com)) + +The test will come in budget practice. + +A taxonomy can improve consistency without proving effectiveness. If an irrigation project qualifies as adaptation finance, for example, classification says little about whether it actually reduces vulnerability, shifts water risk elsewhere or disproportionately benefits already better-served farmers. + +The useful next step would be to link climate-budget tagging to ex-post outcome measures. Otherwise a better classification system can produce a more accurate count of spending without telling us whether the spending worked. + +[Subscribe now](https://varna.substack.com/subscribe?) + +--- + +# EDITORIAL NOTE + +## Heat creates a measurement problem before it creates a disaster statistic + +Today’s Bangladesh evidence points to something I think development research still handles badly. + +Our systems are much better at recording events than attrition. + +A cyclone destroys a house. A flood washes away a bridge. Both generate identifiable damage. Heat often works differently. A construction worker stops two hours early. A street vendor sells less. Someone works through unsafe conditions because losing the day’s wage is worse. A household spends more on water or electricity. Food expenditure falls because earnings have fallen. + +Each loss can be small enough to disappear from conventional disaster accounting. Repeated across millions of workers and dozens of hot days, the aggregate may be substantial. + +This is also why “productivity loss” is an incomplete outcome. If someone maintains output by continuing to work in dangerous heat, measured productivity may show little deterioration while the health cost rises. + +For informal labour in South Asia, we need measures that jointly observe **exposure, hours worked, earnings and health**. Measuring only one can make adaptation look more successful than it is. + +**Reader question:** What is the best existing longitudinal dataset or study that links daily heat exposure to hours worked, earnings and health outcomes for informal workers in South Asia? I am particularly interested in designs that can distinguish adaptation from workers simply absorbing greater health costs. + +This belongs on the list for a future Methods Monday: how to measure labour-productivity losses from extreme heat when workers can adjust hours, effort and exposure. + +--- + +## Sources + +[Climate-change attribution analysis reported by UNB, 16 September 2026](https://unb.com.bd/category/Bangladesh/climate-change-adds-25-risky-heat-days-to-bangladeshs-summer-new-climate-central-analysis-finds/195482?utm_source=chatgpt.com) + +[PRAAN heat and food-security study reported by Dhaka Stream, 16 September 2026](https://www.dhakastream.net/node/198880?utm_source=chatgpt.com) + +[Bangladesh TB, HIV and malaria programme, The Daily Star, 16 September 2026](https://www.thedailystar.net/news/healthcare/news/govt-plans-tk-4039cr-drive-against-tb-hiv-malaria-4273956?utm_source=chatgpt.com) + +[Global Fund funding and prevention update, Reuters, 16 September 2026](https://www.reuters.com/business/healthcare-pharmaceuticals/aid-cuts-hit-disease-prevention-harder-than-treatment-global-fund-chief-says-2026-09-16/?utm_source=chatgpt.com) + +[Maharashtra State Climate Finance Framework, CEEW, 16 September 2026](https://www.ceew.in/publications/maharashtra-state-climate-finance-framework?utm_source=chatgpt.com) + +[Bangladesh railway cost revision, The Daily Star, 16 September 2026](https://online.thedailystar.net/news/bangladesh?utm_source=chatgpt.com) + +[Pakistan regulatory sandbox, Dawn, 16 September 2026](https://www.dawn.com/news/2030248?utm_source=chatgpt.com) + +[IIED: No escaping the heat](https://www.iied.org/22755iied?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-16-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +--- diff --git a/archive/2026-09-18-friday-findings-18-september-2026.md b/archive/2026-09-18-friday-findings-18-september-2026.md new file mode 100644 index 0000000..5d9a470 --- /dev/null +++ b/archive/2026-09-18-friday-findings-18-september-2026.md @@ -0,0 +1,154 @@ +--- +title: "Friday Findings | 18 September 2026" +date: 2026-09-18 +type: other +url: https://varna.substack.com/p/friday-findings-18-september-2026 +subtitle: "Financial outcomes, hidden urban pollution, Nepal’s federalism baseline, and the design gaps in Bangladesh’s safety nets and garment transition." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Friday Findings | 18 September 2026 + +*2026-09-18. Published at [https://varna.substack.com/p/friday-findings-18-september-2026](https://varna.substack.com/p/friday-findings-18-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Editorial illustration for Research Rundown Friday Findings, 18 September 2026, showing a research desk with charts, policy notes and books on financial inclusion, clean air, federalism, social protection, garments and implementation science, with Kathmandu visible through the window.](https://substackcdn.com/image/fetch/$s_!gCbH!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe2f1ecbb-59bc-4a6b-89bf-b1dcf1f13e96_1672x941.png)](https://substackcdn.com/image/fetch/$s_!gCbH!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fe2f1ecbb-59bc-4a6b-89bf-b1dcf1f13e96_1672x941.png) + +A recurring question runs through this week’s reading: are we measuring the thing we actually care about? Financial access is a weak substitute for financial well being. Air-quality spending tells us little unless we know whether the targeted sources matter and whether interventions work. Institutional decentralisation can exist on paper while fiscal dependence persists. Programme coverage can look respectable while missing the population whose behaviour we hope to change. + +Seven pieces made the cut. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 1. financial inclusion after the bank account + +Ganesan, Priyadarshini, and Aishwarya Iyer. 2026. “The Winding Road from Financial Access to Financial Outcomes.” *Ideas for India*, 16 September 2026. + +This is the piece I would read first. + +Ganesan and Iyer take aim at a familiar measurement shortcut: treating financial access or even product use as evidence of financial inclusion having worked. Drawing on Dvara Research’s Financial Health Survey, they propose an Input-Output-Outcome-Context framework that separately measures product ownership, use, financial health, and household context. + +The empirical application covers 4,000 households across 18 districts in seven Indian states. Households with both physical and digital financial infrastructure record higher financial-health scores than households with digital-only or physical-only access. Households with predictable incomes also perform better across the framework, while the relationship between informal finance and financial outcomes varies substantially by region. + +The important point is methodological. An account, insurance policy or digital-payment channel is an input. Usage is an intermediate result. Neither establishes whether a household is better able to manage routine expenses, cope with shocks or plan ahead. + +The current evidence is cross-sectional, so the observed associations do not establish causal effects. The framework itself is the more interesting contribution. It forces researchers and programmes to specify where they think change occurs along the chain. + +This belongs in a future Methods Monday on proxy outcomes and theories of change. + +[Read the open-access article](https://www.ideasforindia.in/topics/money-finance/the-winding-road-from-financial-access-to-financial-outcomes?utm_source=chatgpt.com) + +## 2. the urban pollution source we still measure poorly + +Ali, Sanjar, Bhavay Sharma, Sree Kumar Kumaraswamy, Vandana Tyagi, and Sreekanth Vakacherla. 2026. *City-Tier-Wise Open Waste Burning (OWB) Characteristics*. WRI India Working Paper, 17 September 2026. + +Open waste burning sits awkwardly in urban air-pollution policy because it is visible enough to be familiar and poorly measured enough to be missing from many serious inventories. + +WRI India surveys open waste burning across 11 Indian cities, covering winter and summer conditions and measuring incident density, estimated burned mass, waste composition, and associated PM2.5 and PM10 emissions. The authors explicitly identify the lack of comparable ground-level data as one reason the source remains poorly characterised. + +This matters for city clean-air plans. An emissions inventory built from weak assumptions about informal burning can produce equally weak decisions about where scarce municipal money should go. + +The paper provides a multicity descriptive baseline. It does not estimate the health burden attributable to open burning or identify the effect of particular waste-management interventions. The 11-city coverage also means the findings should not be casually converted into a national estimate. + +Still, this fills a real evidence gap. + +[Read the WRI India working paper](https://wri-india.org/research/city-tier-wise-open-waste-burning-owb-characteristics?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 3. does cleaning roads actually reduce road dust? + +Kumar, Ritesh, Sree Kumar Kumaraswamy, Sreekanth Vakacherla, and Vandana Tyagi. 2026. *Silt Load Measurement and Investigation of the Effectiveness of Road Dust Cleaning Mechanisms in Three Indian Cities*. WRI India Working Paper, 17 September 2026. + +The companion WRI paper asks a refreshingly concrete question about an intervention Indian cities already spend money on. + +The researchers measured road silt across different road types in Mumbai, Indore and Ahmedabad, compared manual, mechanical, water-based and other cleaning practices, and examined how quickly silt returned after cleaning in Mumbai and Ahmedabad. WRI describes the study explicitly as exploratory. + +That qualification matters. Three cities cannot settle the question of which cleaning technology works across India. Road design, traffic, construction activity, season and local dust sources all affect re-accumulation. + +The policy relevance is immediate, however. National Clean Air Programme funds are already being used for road-dust control, while evidence on the effectiveness of common cleaning practices remains thin. + +This is the sort of mundane implementation research that deserves more attention. Programmes frequently evaluate whether equipment was purchased. Fewer ask what happened to the pollutant afterwards. + +[Read the WRI India working paper](https://wri-india.org/research/silt-load-measurement-and-investigation-effectiveness-road-dust-cleaning-mechanisms-three?utm_source=chatgpt.com) + +## 4. nearly a decade of federalism in Nepal, measured + +Provincial and Local Governance Strengthening Programme. 2025. *Nepal: The State of Federalism Report 2025*. Released online 12 September 2026, with technical support from UNDP. + +Nepal now has its first system-wide baseline of federal governance since the 2015 Constitution. It is intended to be repeated in 2027 and 2029, which makes the measurement architecture at least as interesting as the current scores. + +The report combines quantitative and qualitative evidence across federal, provincial and local government, administrative indicators, interviews and citizen-perception data. The associated Survey of the Nepali People used multistage random sampling and Kish-grid respondent selection, with 12,500 respondents. + +Two numbers capture the tension in the findings. Women hold around 41% of local elected positions, while more than 90% of subnational budgets remain dependent on federal grants and own-source revenue accounts for only 6–8%. + +The report therefore gives us a useful distinction between political decentralisation, administrative capacity and fiscal autonomy. They need not move together. + +As a baseline study, it can tell us where the system stands and how citizens perceive it. Attribution requires later rounds and a much harder identification problem. Changes between 2025 and 2029 cannot automatically be credited to federalism itself. + +[Read the open-access report](https://www.undp.org/nepal/publications/state-federalism-report-2025?utm_source=chatgpt.com) + +## 5. child labour and the targeting problem in Bangladesh’s safety nets + +Moazzem, Khondaker Golam, Nishat Tasnim Nishu, and ASM Shamim Alam Shibly. 2025. *Strengthening Social Safety Nets for Child Labour Elimination in Bangladesh: Coverage, Gaps, and Policy Directions*. Centre for Policy Dialogue. Released online 15 September 2026. + +A date caveat first. CPD posted this report this week, although its stated publication period is December 2025. I am including it as a newly available publication, rather than new fieldwork. + +The central finding is simple and damaging: Bangladesh’s social-safety-net programmes were not designed around the factors along which child labour actually varies. + +The study brings together national survey data, key-informant interviews and regional comparisons to identify gaps in programme coverage and targeting, then proposes a framework more explicitly tied to child-labour risk. + +Its value lies in the mismatch it documents between programme architecture and policy objective. A transfer programme can reach poor households and still be badly designed as a child-labour intervention if eligibility rules, benefit size, timing and accompanying services ignore the mechanisms driving children into work. + +The evidence supports a targeting and programme-design diagnosis. It does not establish the causal effect of existing safety-net participation on child labour. + +[Read the open-access report](https://cpd.org.bd/publication/strengthening-social-safety-nets-for-child-labour-elimination-in-bangladesh/?utm_source=chatgpt.com) + +## 6. why Bangladesh’s garment factories remain carbon-intensive + +Mohammed, Sami, Khondaker Golam Moazzem, Maleehah Sabah Ali, and Mehadi Hasan Shamim. 2026. *The Carbon Stitch: Assessing Carbon Lock-In in the RMG Sector of Bangladesh*. Centre for Policy Dialogue. Released online 14 September 2026. + +This is the second CPD item with a date qualification. The web release is new; CPD records the publication period as May 2026. + +Using plant-level survey data and mixed methods, the authors examine why Bangladesh’s ready-made garment factories remain dependent on carbon-intensive production despite growing commercial and regulatory pressure to decarbonise. They find much greater uptake of relatively cheap energy-efficiency measures than of deeper capital-intensive changes. Financing costs, regulatory constraints, supply-chain arrangements and firm decision-making all appear in the explanation. Renewable-energy adoption also varies by factory size. + +The interesting development question concerns firm heterogeneity. A decarbonisation requirement imposed on an industry does not imply that factories face comparable adjustment costs. Smaller producers may encounter a very different investment problem from large export factories with easier access to finance. + +The mixed-methods design is suited to identifying constraints and mechanisms. It cannot by itself tell us how much emissions would fall under the financing and regulatory measures the authors recommend. + +This is worth keeping beside last week’s Bangladesh future-of-work paper. Automation and decarbonisation are often discussed separately even though the same firms may be making both sets of capital-investment decisions. + +[Read the open-access study](https://cpd.org.bd/publication/the-carbon-stitch-assessing-carbon-lock-in-in-the-rmg-sector-of-bangladesh/?utm_source=chatgpt.com) + +## 7. whose implementation science? + +Rayamajhee, Binod, Gopiram Syangtan, Bhupendra Lama, Saruna Ghimire, Brian Oldenburg, and Uday Narayan Yadav. 2026. “Building Bridges: A Framework for Redefining and Decolonizing Implementation Science in Low- and Middle-Income Countries.” *PLOS Global Public Health* 6(9): e0006155. Published 15 September 2026. doi: 10.1371/journal.pgph.0006155. + +This is a critical review rather than a South Asia-specific empirical study, but several of its authors work in and on Nepal, and the methodological argument belongs in this list. + +The authors examine the extent to which implementation-science frameworks developed in high-income settings travel to LMIC health systems. They propose a framework centred on locally adapted methods, local research leadership, funding arrangements, review structures and recognition of Indigenous and community knowledge. + +I am less persuaded by new frameworks simply because they are new. Implementation science already has a framework proliferation problem. The test here will be whether this framework changes study design, decision rights, measures, budgets or interpretation when researchers actually apply it. + +For now, treat this as an agenda-setting methodological paper. It proposes a way of thinking about implementation research; it does not provide empirical evidence that the proposed framework produces better implementation or health outcomes. + +That makes it another possible Methods Monday piece, perhaps paired with the harder question of when creating a new framework adds analytical value. + +[Read the open-access PLOS article](https://journals.plos.org/globalpublichealth/article?id=10.1371%2Fjournal.pgph.0006155&utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## what I would keep + +The Ganesan-Iyer piece is my pick this week because the argument extends well beyond financial inclusion. Development research repeatedly substitutes an administratively convenient indicator for the outcome in the theory of change. Their framework gives us a useful way to inspect where that substitution occurs. + +I would also keep the two WRI papers together. They move air-pollution research toward questions cities can act on: which sources are being missed, whether interventions reduce them, and how quickly gains disappear. + +The Nepal report is the substantial reference document of the week. Its real value may emerge over the next four years if the planned repeat measurements are carried out consistently. + +I found no new South Asia-specific nutrition or WEE study within 12–18 September that was strong enough, clearly open access and sufficiently recent to earn a place. I would rather leave those categories empty than pad the edition. + +## a question for readers + +Do you know of a South Asian financial-inclusion study that follows households over time and measures financial wellbeing or resilience alongside access and usage? Send it to me. I am particularly interested in evidence that can test, rather than assume, the pathway from owning a financial product to being financially better off. + +[Share](https://varna.substack.com/p/friday-findings-18-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-20-research-rundown-daily-19-september.md b/archive/2026-09-20-research-rundown-daily-19-september.md new file mode 100644 index 0000000..b9b192f --- /dev/null +++ b/archive/2026-09-20-research-rundown-daily-19-september.md @@ -0,0 +1,152 @@ +--- +title: "Research Rundown DAILY - 19 September 2026" +date: 2026-09-20 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-19-september +subtitle: "New evidence on climate extremes in India points to a harder adaptation problem than national averages suggest, while data on gender parity, wages and revenue show why headline indicators mislead." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 19 September 2026 + +*2026-09-20. Published at [https://varna.substack.com/p/research-rundown-daily-19-september](https://varna.substack.com/p/research-rundown-daily-19-september). This file is a generated copy; the Substack post is the original.* + +[![**Alt text:** Research Rundown DAILY cover for 19 September 2026, showing an India climate-extremes map and regional rainfall records, with longer dry spells and more intense one-day rainfall illustrating how similar seasonal totals can conceal different climate risks.](https://substackcdn.com/image/fetch/$s_!XR0k!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcdf6231d-e0da-4b8f-b6d2-254774d50aec_1672x941.png)](https://substackcdn.com/image/fetch/$s_!XR0k!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcdf6231d-e0da-4b8f-b6d2-254774d50aec_1672x941.png) + +## FIVE THINGS WORTH KNOWING + +### 1. India’s climate extremes are becoming more geographically differentiated + +A 52-year analysis of daily rainfall and temperature data finds a statistically significant increase in the area of India exposed to climate extremes. The revised India Climate Extremes Index rises by about **0.36 per cent of area per decade** between 1971 and 2022. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +The national trend conceals different regional changes. Northwestern India shows the strongest increases in maximum-temperature extremes, heat index and heatwave severity. South Peninsular India has seen increasing drought-related extremes. Northeast India is unusual: intense one-day rainfall is increasing alongside longer dry spells and drought episodes. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +That last result matters for adaptation. A district facing heavier downpours and longer dry periods needs water storage, drainage and agricultural planning designed around greater variability. A simple classification as increasingly “wet” or “dry” misses the problem. + +There is a measurement qualification. The researchers combine seven indicators into a revised composite index. Composite indices are useful for tracking several hazards together, but movements in the headline score depend partly on indicator choice and aggregation. The component measures remain important. + +[Read the study in the International Journal of Climatology](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. South Asia’s gender gap is still concentrated in economic participation + +The World Economic Forum’s 2026 Global Gender Gap Report provides a useful 20-year series. Across 145 economies, the global parity score is 69.2 per cent. For the constant sample of 97 economies observed since 2006, the score has risen by 5.2 percentage points. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +South Asia’s problem is particularly economic. The region has closed only **40.9 per cent of its Economic Participation and Opportunity gap**, the lowest regional score in the report. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +India ranks 131st with an overall parity score of 64.5 per cent. Its economic participation score is 41.2 per cent, slightly higher than last year but still below its 2013 peak. Pakistan ranks 143rd, with 59.5 per cent of its overall gender gap closed. Nepal, by contrast, is among the countries recording the largest long-run gains since 2006. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +The rankings attract attention, but changes in scores are more informative. Country ranks move when other countries enter, leave or change position. The report itself warns about data lags and differences in national data availability. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +[World Economic Forum: Global Gender Gap Report 2026](https://www.weforum.org/publications/global-gender-gap-report-2026/?utm_source=chatgpt.com) + +### 3. Bangladesh has gazetted a large public-sector pay increase + +Bangladesh has issued its ninth national pay scale, effective retrospectively from 1 July 2026. Basic salaries rise by as much as **142 per cent**, with the increase being introduced in phases. The existing special allowance will be withdrawn and adjusted against arrears. ([The Business Standard](https://www.tbsnews.net/bangladesh/gazette-issued-national-pay-scale-2026-1547116?utm_source=chatgpt.com)) + +The phasing matters. Employees in Grade 9 and above initially receive 40 per cent of their additional basic pay, while Grades 10 to 20 receive 50 per cent. Higher proportions apply from January 2027 before full implementation. ([The Business Standard](https://www.tbsnews.net/bangladesh/gazette-issued-national-pay-scale-2026-1547116?utm_source=chatgpt.com)) + +This is a substantial fiscal and labour-market intervention. The questions now concern the aggregate wage bill, financing and spillovers into private-sector wage demands. + +Nominal increases also need to be separated from real gains. Bangladesh has experienced sustained inflation since the previous pay structure was introduced. Comparing the old and new salary tables alone will overstate the improvement in purchasing power. + +### 4. Bangladesh’s garment unions want the minimum wage reviewed under the new three-year rule + +The IndustriALL Bangladesh Council has called for a new garment-sector wage board by 30 September and a revised minimum wage by the end of November. The current minimum wage dates from November 2023. ([The Business Standard](https://www.tbsnews.net/economy/rmg/industriall-demands-wage-board-reconstitution-new-minimum-wage-garment-workers-1547676?utm_source=chatgpt.com)) + +The timing follows an amendment to the Bangladesh Labour Act that shortened the statutory minimum-wage review period from five years to three. The present wage therefore reaches its new review point this November. ([The Business Standard](https://www.tbsnews.net/economy/rmg/industriall-demands-wage-board-reconstitution-new-minimum-wage-garment-workers-1547676?utm_source=chatgpt.com)) + +This deserves attention beyond the wage number eventually announced. Bangladesh’s garment industry employs millions of workers and remains central to export earnings. The relevant comparison is between wage growth, inflation and productivity since 2023. + +Compliance matters too. A statutory wage is an institutional floor. Earnings depend on enforcement, hours, overtime payments, employment continuity and whether factories respond by changing workforce composition or hours. + +### 5. India’s direct-tax receipts are growing, but the composition is more interesting than the total + +India collected ₹12.12 lakh crore in net direct taxes between 1 April and 17 September, **12.96 per cent more than during the corresponding period last year**. Gross collections rose 15.19 per cent to ₹14.32 lakh crore, while refunds increased 29.19 per cent to ₹2.20 lakh crore. ([Reuters](https://www.reuters.com/world/india/indias-net-direct-tax-collections-rise-13-april-1-sept-17-2026-09-18/?utm_source=chatgpt.com)) + +Corporate advance-tax payments rose 18.09 per cent. Non-corporate advance tax increased 9.24 per cent. Securities Transaction Tax receipts rose 52.9 per cent to ₹40,214 crore. ([The New Indian Express](https://www.newindianexpress.com/business/2026/Sep/18/net-direct-tax-collections-rise-1296-per-cent-to-rs-1212-lakh-crore?utm_source=chatgpt.com)) + +The STT figure needs careful interpretation because tax rates on derivatives changed from 1 April. Revenue growth therefore reflects a combination of market activity and a changed tax schedule. It cannot be read directly as a 53 per cent increase in the underlying tax base. + +The broader collections data are provisional. At this point in the fiscal year they provide useful information about receipts, rather than a clean measure of household incomes or corporate profitability. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## ONE PAPER WORTH READING + +### Evolution of Climate Extremes Over the Indian Subcontinent Using a Revised CEI + +Thota Mohan and colleagues use IMD gridded daily rainfall and temperature data from **1971 to 2022** to construct a revised India Climate Extremes Index. Seven components capture maximum and minimum temperature, consecutive dry days, heavy precipitation, drought using SPEI, humid heat and heatwave severity. Trend estimation shows widespread increases in heat extremes, particularly in northwestern India; stronger drought signals in southern India; and the unusual combination of more short-duration heavy rainfall and increasing drought episodes in the Northeast. At the national level, the index increases significantly by roughly 0.36 per cent of area per decade. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +The regional decomposition is more useful than the composite headline. Combining different hazards into one index necessarily embeds decisions about indicators and aggregation, while regional averages can still conceal local exposure. The observational record also identifies trends rather than attributing individual changes to specific causal mechanisms. + +Its contribution is practical. Indian climate adaptation cannot sensibly be organised around one national hazard trajectory. Heat, drought and rainfall variability are changing differently across regions. + +**Citation:** Mohan, Thota, M. Rajeevan, Smrutishree Lenka, Kondapalli Niranjan Kumar, et al. 2026. “Evolution of Climate Extremes Over the Indian Subcontinent Using a Revised CEI.” *International Journal of Climatology*. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +[Read the paper](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com) + +## ONE NUMBER + +### 40.9 per cent + +The share of the gender gap in **Economic Participation and Opportunity** that South Asia has closed, according to the 2026 Global Gender Gap Index. It is the lowest score of any region on this subindex. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +The number is more useful than the country league table. + +Education gaps have narrowed substantially across much of South Asia. Translation into employment, earnings and positions of authority has been much slower. India illustrates the gap clearly: its Educational Attainment parity score is 96.6 per cent, compared with 41.2 per cent for Economic Participation and Opportunity. ([World Economic Forum](https://www.weforum.org/publications/global-gender-gap-report-2026/in-full/benchmarking-gender-gaps-2026/?utm_source=chatgpt.com)) + +## POLICY WATCH + +### Bangladesh is about to run two wage experiments at once + +Bangladesh’s new public-sector pay scale and the impending garment-sector minimum-wage review create an unusual opportunity to observe two different wage-setting institutions during the same inflationary period. + +Public salaries are administratively determined and financed through the budget. Garment wages emerge from a statutory wage-board process inside a highly trade-exposed private industry. + +The outcomes should therefore be compared carefully. + +A useful assessment would track real wages rather than announced nominal increases, and examine employment, hours, overtime and worker turnover alongside monthly earnings. For the public sector, the corresponding questions concern the wage bill, recruitment and fiscal displacement. + +There is a broader distributional question. Large formal-sector wage adjustments can restore purchasing power for covered workers while leaving informal workers facing the same price level without an equivalent adjustment mechanism. + +([The Business Standard](https://www.tbsnews.net/bangladesh/gazette-issued-national-pay-scale-2026-1547116?utm_source=chatgpt.com)) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## EDITORIAL NOTE + +### A national climate average is becoming a poor description of the risk + +The climate-extremes paper is useful because it makes a problem with adaptation planning unusually visible. + +Northeast India can experience longer dry spells while extreme one-day rainfall becomes more frequent. There is no contradiction. Seasonal rainfall totals, number of rainy days, rainfall intensity and the length of dry intervals measure different properties of the monsoon. + +Policy often compresses them. + +Agricultural planning may use seasonal rainfall. Urban drainage depends heavily on rainfall intensity. Reservoir management depends on timing and runoff. Household water security can deteriorate when rain arrives in fewer, heavier episodes even if the annual total changes little. + +The methodological implication is straightforward. “Rainfall” is rarely an adequate exposure variable by itself. + +For development research, the choice of climate measure should follow the outcome being studied. A crop-yield model, diarrhoeal-disease study and urban-flood analysis may require quite different representations of the same monsoon. + +**Reader question:** Which climate-exposure measures have proved most useful in South Asian household or agricultural panel data when seasonal rainfall totals conceal changes in dry-spell duration and extreme rainfall intensity? + +This is worth carrying into a future Methods Monday on matching climate exposure measures to development outcomes. + +## Sources + +[International Journal of Climatology: revised India Climate Extremes Index](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com) + +[World Economic Forum: Global Gender Gap Report 2026](https://www.weforum.org/publications/global-gender-gap-report-2026/?utm_source=chatgpt.com) + +[The Business Standard: Bangladesh’s ninth national pay scale](https://www.tbsnews.net/bangladesh/gazette-issued-national-pay-scale-2026-1547116?utm_source=chatgpt.com) + +[The Business Standard: Bangladesh garment wage-board demand](https://www.tbsnews.net/economy/rmg/industriall-demands-wage-board-reconstitution-new-minimum-wage-garment-workers-1547676?utm_source=chatgpt.com) + +[Reuters: India’s direct-tax collections through 17 September](https://www.reuters.com/world/india/indias-net-direct-tax-collections-rise-13-april-1-sept-17-2026-09-18/?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-19-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +## diff --git a/archive/2026-09-21-good-reads-8-why-are-more-rural-women.md b/archive/2026-09-21-good-reads-8-why-are-more-rural-women.md new file mode 100644 index 0000000..9cc34b2 --- /dev/null +++ b/archive/2026-09-21-good-reads-8-why-are-more-rural-women.md @@ -0,0 +1,113 @@ +--- +title: "Good Reads #8: Why are more rural women working?" +date: 2026-09-21 +type: other +url: https://varna.substack.com/p/good-reads-8-why-are-more-rural-women +subtitle: "Rural women’s employment has risen sharply since 2017. The causes remain disputed." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #8: Why are more rural women working? + +*2026-09-21. Published at [https://varna.substack.com/p/good-reads-8-why-are-more-rural-women](https://varna.substack.com/p/good-reads-8-why-are-more-rural-women). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads #8 cover, dated 20 September 2026. The title asks, “Why are more rural women working?” Beside it is a muted editorial collage of two rural women working in a field, a small line chart showing rising female rural employment, and paper-texture elements in cream, rust, ochre and dark teal.](https://substackcdn.com/image/fetch/$s_!RcTY!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F56fce0aa-ca37-45ca-916f-7d65cb3e8177_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!RcTY!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F56fce0aa-ca37-45ca-916f-7d65cb3e8177_1536x1024.png) +*Good Reads #8 on 20th September, 2026* + +One of the largest changes in India’s labour market over the past decade has happened among rural women. + +The rural female workforce participation rate for women aged 15 and above rose from 23.7 per cent in 2017-18 to 44.8 per cent in 2025. In absolute terms, Bishwanath Goldar and Suresh Chand Aggarwal estimate that rural female employment increased from about 75 million women to 150 million over the same period. Much of this increase occurred in agriculture and allied work, including a substantial increase in women classified as unpaid family workers. + +The numbers are striking. Their interpretation is much harder. + +In a recent Ideas for India essay, Goldar and Aggarwal challenge an increasingly common explanation: that women have entered rural employment primarily because household economic conditions have deteriorated. + +[Read “What explains the post-2017 surge in rural female employment in India?” on Ideas for India](https://www.ideasforindia.in/topics/macroeconomics/what-explains-the-post-2017-surge-in-rural-female-employment-in-india?utm_source=chatgpt.com) + +Their argument is based on a longer working paper published in July 2026. + +## the argument + +The distress explanation has considerable intuitive appeal. + +If household income comes under pressure, women who were previously outside measured employment may enter paid work, agriculture or family enterprises. Rising participation can therefore coexist with worsening household circumstances. Several recent analyses have pointed to precisely this possibility, particularly because much of the increase in women’s work has occurred in agriculture, self-employment and unpaid family work. + +Goldar and Aggarwal ask whether the available evidence supports such a broad conclusion. + +One of their more interesting tests looks at consumption groups. + +If economic distress were the dominant explanation, we might expect the increase in women’s employment to be concentrated among poorer households. Their comparison of PLFS data finds increases across the consumption distribution. Rural female workforce participation increased by an average of 17.8 percentage points among the bottom three consumption deciles and by 13 percentage points among the top three. + +Women from relatively better-off farming households also account for a meaningful share of the increase in unpaid agricultural work. + +This creates a problem for a simple distress story. + +The authors also point towards alternative mechanisms. These include expansion of women’s collectives and credit through DAY-NRLM, employment through MGNREGS, growth in agriculture and allied activities, and reductions in the time required for household tasks as access to water, sanitation and clean cooking fuel has expanded. + +Amarjeet Sinha has made a related argument, attributing part of the rise in rural women’s participation to MGNREGS, DAY-NRLM and greater recognition of women’s agricultural activity. His analysis also concedes something worth keeping in view: increased participation through these routes need not produce large increases in women’s incomes. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## where I am persuaded + +Goldar and Aggarwal are right to challenge the tendency to turn one plausible mechanism into a complete explanation. + +The consumption-decile result deserves attention. A phenomenon occurring across the income distribution requires a wider explanation than one driven exclusively by severe household distress. + +Their paper also reminds us that agricultural work classified as “unpaid” cannot automatically be treated as economically meaningless. Family farming combines labour, household production, land, livestock and market activity in ways that conventional employment categories handle rather badly. + +The broader point is methodological. An observed increase in labour supply does not identify the reason for that increase. + +We need evidence on mechanisms. + +## where I remain unconvinced + +I think the authors push their counterargument further than the evidence allows. + +Finding increased participation among higher consumption deciles weakens a pure distress hypothesis. It does not settle whether deteriorating economic conditions contributed substantially to the overall increase. + +Consumption deciles are themselves outcomes of household income, production, transfers, prices and labour supply. Households higher in the consumption distribution can experience income shocks. Agricultural households can respond to changing labour markets by reallocating women’s work within family enterprises. A comparison across deciles therefore gives us useful descriptive evidence, while causal interpretation remains difficult. + +The composition of the employment increase also matters enormously. + +A rise in women’s work has a different welfare meaning when it comes through regular paid employment, profitable own-account activity, MGNREGS employment, or unpaid work on a household farm. + +Singh and Ankita’s competing analysis points directly to this issue. They find much of the increase occurring among unpaid and subsidiary workers and report stagnant or declining real earnings over the period they study. They interpret these patterns as evidence consistent with distress-driven employment. + +These accounts should therefore be read together. + +The disagreement is analytically useful because both sides are trying to infer household behaviour from labour-market outcomes that were never designed to identify motivation. + +## the measurement problem + +This is the part I found most interesting. + +“Why did women start working?” sounds like a straightforward empirical question. PLFS can tell us whether somebody worked, her employment status, industry, occupation and several other characteristics. + +It cannot directly tell us why she entered work. + +Distress is a latent construct here. So are improved opportunity, changing social norms, reduced domestic work burdens and increased recognition of work women were already doing. + +Researchers have to infer these mechanisms from observed patterns. + +That means evidence which discriminates between explanations becomes especially valuable: longitudinal household income and employment data, changes in women’s earnings, transitions between employment categories, hours worked, time-use data, local labour demand, programme exposure, and women’s own accounts of why their work changed. + +Without those pieces, several explanations can fit the same headline statistic. + +## why read it? + +Because the rise in rural women’s employment is too large to treat as a statistical curiosity, and too consequential to explain with a single convenient story. + +Goldar and Aggarwal provide a useful challenge to the distress interpretation. I am less persuaded by their stronger suggestion that the evidence substantially displaces it. + +The more interesting conclusion is that we still do not understand the shift well enough. + +That is a good reason to read the piece. + +And a good reason to keep asking what exactly a rising female workforce participation rate is measuring. + +[Read the Ideas for India article](https://www.ideasforindia.in/topics/macroeconomics/what-explains-the-post-2017-surge-in-rural-female-employment-in-india?utm_source=chatgpt.com) + +[Read the underlying Goldar and Aggarwal working paper](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7024578&utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-8-why-are-more-rural-women?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-21-methods-monday-21-september-2026.md b/archive/2026-09-21-methods-monday-21-september-2026.md new file mode 100644 index 0000000..1830e2a --- /dev/null +++ b/archive/2026-09-21-methods-monday-21-september-2026.md @@ -0,0 +1,127 @@ +--- +title: "Methods Monday | 21 September 2026" +date: 2026-09-21 +type: other +url: https://varna.substack.com/p/methods-monday-21-september-2026 +subtitle: "Voice responses, SMS contact, rooftop sampling and small-area estimation, and where each method shifts survey error." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Methods Monday | 21 September 2026 + +*2026-09-21. Published at [https://varna.substack.com/p/methods-monday-21-september-2026](https://varna.substack.com/p/methods-monday-21-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Editorial desk illustration for Methods Monday showing a household questionnaire, SMS survey invitation, rooftop sampling map, district-level estimates, field notes and measuring tools.](https://substackcdn.com/image/fetch/$s_!JFA5!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb7344e2d-ef6f-42d0-98d9-ffe837f526d4_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!JFA5!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb7344e2d-ef6f-42d0-98d9-ffe837f526d4_1536x1024.png) +*Methods Monday - 21 September 2026* + +This is a short issue. The week’s useful methods work concerns voice responses, SMS contact and small-area estimation. I have also gone back for two recent papers we missed, including an India study on replacing household listing with satellite-derived rooftop data. + +## 1. speaking produces richer open-ended answers. It also produces far more missing ones + +Camille Landesvatter and Paul C. Bauer randomly assigned smartphone respondents to answer nine open-ended survey questions either by typing or recording their voices. The final analytic groups contained 800 respondents in the text condition and 661 in the voice condition. Spoken answers were transcribed with Whisper. + +Among people who answered, voice performed well. Spoken responses were significantly longer for five of the nine questions, with differences ranging from 2.2 to 4.4 words. The authors also use response entropy, a measure of lexical information, and find a smaller advantage for voice there. The presence of other people during the survey reduced response entropy for voice answers on three items. + +The missing-data result changes how I read the paper. Across the nine open-ended items, an average of 47 per cent of respondents in the voice condition supplied an answer, compared with 99 per cent in the text condition. Break-off was also concentrated in the voice arm: 204 cases against 42 in the text arm. The information analysis is therefore conditional on somebody agreeing to record an answer. + +That distinction matters for development surveys. Voice entry could reduce the burden of typing for respondents using phones, particularly where scripts, literacy or device familiarity make text entry cumbersome. Yet a response mode that produces richer material from a much smaller subset can worsen total data quality. Privacy is likely to matter too. Recording an answer about household finances, violence, health or political attitudes while other people are nearby is a different measurement situation from typing it. + +Landesvatter, Camille, and Paul C. Bauer. 2026. “Information Content in Text and Voice Response Formats for Open-Ended Survey Questions.” *Journal of Survey Statistics and Methodology*, smag030. DOI: 10.1093/jssam/smag030. Published 17 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article/doi/10.1093/jssam/smag030/8812674?utm_source=chatgpt.com) + +Worth reading. The response-rate result is at least as interesting as the result on answer length. + +## 2. SMS reminders bring respondents in quickly, while consent to receive them is selective + +Eric Stone, Renee Stepler and Casey Eggleston use the US Census Household Panel to examine SMS invitations and reminders. Their September working paper uses survey responses and paradata from the May 2024 topical, which had approximately 10,000 respondents. + +Responses rose in the hours immediately after text messages were sent. Around half of panel members consented to receive survey texts. Consent varied by race and ethnicity, age, family income, education, marital status, region and respondent type. The route respondents used to enter the survey also differed across demographic groups. + +The lesson for CATI and longitudinal studies is fairly practical. Adding SMS may increase response while changing the composition of respondents reached through that channel. If SMS reminders become part of a mixed-mode protocol, the relevant paradata include consent, delivery, click-through and eventual completion by contact mode. + +The paper examines an established US probability panel whose members had already supplied mobile numbers. Consent and response behaviour could look quite different in a newly recruited panel, or where phones are shared within households. I would want that tested before carrying the findings into South Asian surveys. + +Stone, Eric, Renee Stepler, and Casey Eggleston. 2026. “Evaluating Text Messaging as a Contact Strategy and Predictors of Text Message Consent: Evidence from the Census Household Panel.” US Census Bureau, Research and Methodology Directorate Working Paper RSM2026-09. September 2026. + +[Read the working paper](https://www.census.gov/library/working-papers/2026/adrm/rsm2026-09.html?utm_source=chatgpt.com) + +Worth reading if you run panels or repeated phone surveys. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 3. small-area estimation gets a useful mixed-effects gradient boosting estimator + +Paul Messer and Timo Schmid propose Mixed Effect Gradient Boosting, or MEGB, for unit-level small-area estimation. The method combines gradient boosting with random effects, allowing nonlinearities and interactions while accounting for dependence among observations within areas. They pair the estimator with a nonparametric bootstrap for estimating mean squared error. + +The paper has both model-based simulations and a design-based application using household income data from Chihuahua, Mexico. The population dataset contains 58,578 households in 67 municipalities. Repeated survey samples contain 1,280 households from 22 municipalities, leaving 45 municipalities completely unsampled. The design-based exercise uses 500 Monte Carlo repetitions. + +MEGB does particularly well when the underlying relationship is nonlinear and when within-area dependence matters. Its advantage becomes small under some low signal-to-noise settings. That qualification is useful. A machine-learning estimator does not automatically buy much when the available predictors contain little information. + +There is another practical constraint. Unit-level auxiliary covariates are required for the entire population. Aggregated area-level covariates cannot simply replace them because the fitted function is nonlinear. That requirement may rule the method out in precisely the settings where researchers most want small-area estimates. + +The present application estimates a continuous mean. The authors leave binary and count outcomes, poverty headcounts and inequality measures for later work. For development researchers, those extensions will determine much of the method’s practical value. + +Messer, Paul, and Timo Schmid. 2026. “Gradient Boosting for Hierarchical Data in Small Area Estimation.” *Statistics and Computing* 36: 218. DOI: 10.1007/s11222-026-10970-1. Published 17 September 2026. + +[Read the open-access paper](https://link.springer.com/article/10.1007/s11222-026-10970-1?utm_source=chatgpt.com) + +Worth reading for researchers producing district or subdistrict estimates. I would wait for evidence on poverty and other nonlinear indicators before treating it as a general replacement for established SAE estimators. + +## 4. one we missed: can satellite rooftops replace household listing in India? + +Douglas Johnson and colleagues at IDinsight address an expensive part of conventional area-probability sampling. Once primary sampling units have been selected, survey teams often list every household before drawing the household sample. Where no current frame exists, listing can consume a substantial share of fieldwork time and money. + +Their alternative is rooftop sampling. Researchers sample rooftops from open-source building datasets created by Google and Microsoft and use them to reach households directly. The paper develops a framework for assessing bias and tests the approach using Demographic and Health Survey data, an Indian ground-truth dataset and field tests. + +The authors estimate that the approach introduces bias of around one percentage point for a binary outcome in their tests. Their cost analysis suggests that rooftop sampling can make sense for surveys with fewer than about 1,000 respondents and where conventional listing is infeasible. + +This is one of the more directly useful survey-design papers for India this year. The attraction is obvious in rapidly changing peri-urban settlements and places where existing frames are poor. + +The method moves several errors into new places. A roof need not correspond to one household. Buildings may contain several households, no household, or structures the building dataset failed to detect. Coverage quality may vary geographically. Those errors need local assessment before a rooftop frame can be treated as an adequate substitute for listing. + +Johnson, Douglas, Anjani Balu, Poornima Ramesh, Linh Vo, Rediet Mulu, and Amir Emami. 2026. “Assessing the Reliability of Household Sampling Using Rooftop Data in India.” *Journal of Survey Statistics and Methodology*, smag019. DOI: 10.1093/jssam/smag019. Published 20 August 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag019/8767409?utm_source=chatgpt.com) + +Worth reading. For Indian household-survey teams, I would put this ahead of most of this week’s papers. + +Research Rundown should return to this one. I would like to see evidence on how rooftop sampling performs across dense informal settlements, multi-storey urban housing, tribal settlements and dispersed rural habitations. A national average bias estimate cannot answer that question. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 5. one more from August: predicting question reliability and validity before fieldwork + +Barbara Felderer, Lydia Repke, Wiebke Weber, Jonas Schweisthal and Ludwig Bothmann update the Survey Quality Predictor, an open-access system that predicts the reliability and validity of survey questions from their formal characteristics. + +SQP is based on a database of multitrait-multimethod experiments in which question characteristics vary. For SQP 3.0, the authors compare LASSO, elastic net, boosting and random forest models. Random forest produced the best predictions and became the new prediction engine. The paper also examines which item characteristics contribute most to predictions. + +There is a useful role for this during questionnaire development. Researchers can inspect question form, response categories, labels and other design decisions before an instrument reaches cognitive testing or a pilot. + +I would keep the claim narrow. SQP predicts quality from relationships learned in its existing MTMM database. It cannot establish that a translated item measures the intended construct among a particular population in Bihar, Bangladesh or Nepal. Local cognitive testing, measurement-invariance work and criterion validation answer different questions. + +Felderer, Barbara, Lydia Repke, Wiebke Weber, Jonas Schweisthal, and Ludwig Bothmann. 2026. “Predicting the Validity and Reliability of Survey Questions.” *Survey Research Methods* 20(2): 133–145. DOI: 10.18148/srm/2026.v20i2.8453. Published 10 August 2026. + +[Read the open-access paper and replication files](https://ojs.ub.uni-konstanz.de/srm/article/view/8453?utm_source=chatgpt.com) + +Worth using as a diagnostic during questionnaire design. I would resist presenting its predicted score as validation. + +## what I would read + +The voice experiment is the best new measurement paper this week because its awkward result is the useful one. Voice yields richer answers among respondents who answer, while many more people decline to answer at all. + +For field research in South Asia, Johnson et al. is more consequential. Household listing is expensive enough that a credible alternative deserves testing. The unresolved issue is heterogeneity in coverage error across settlement types. + +Messer and Schmid is the technically strongest new quantitative paper. Its immediate use is narrower because full-population unit-level auxiliary data are a demanding input. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## a question for readers + +If you have tested satellite or map-based household sampling in South Asia, what happened when a sampled building contained several households, was vacant, or did not appear correctly in the building layer? I am particularly interested in field evidence comparing those errors with errors found during conventional household listing. + +That is a question I would like Research Rundown to return to with actual field protocols and validation data rather than treating rooftop sampling as a solved replacement for listing. + +[Share](https://varna.substack.com/p/methods-monday-21-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +## diff --git a/archive/2026-09-23-good-reads-9-is-the-health-gap-really.md b/archive/2026-09-23-good-reads-9-is-the-health-gap-really.md new file mode 100644 index 0000000..03c3c93 --- /dev/null +++ b/archive/2026-09-23-good-reads-9-is-the-health-gap-really.md @@ -0,0 +1,161 @@ +--- +title: "Good Reads #9: Is the health gap really a wealth gap?" +date: 2026-09-23 +type: other +url: https://varna.substack.com/p/good-reads-9-is-the-health-gap-really +subtitle: "Public Health Initiative argues that socioeconomic status is the dominant determinant of health. The social gradient is clear. Explaining what produces it is harder." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #9: Is the health gap really a wealth gap? + +*2026-09-23. Published at [https://varna.substack.com/p/good-reads-9-is-the-health-gap-really](https://varna.substack.com/p/good-reads-9-is-the-health-gap-really). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads #9 cover for “Is the health gap really a wealth gap?”, featuring Michael Marmot’s *The Health Gap*, books on health equity and social determinants of health, a stethoscope, tiffin and notebook listing income, housing, work, gender, environment and care, against an urban backdrop.](https://substackcdn.com/image/fetch/$s_!XejA!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F27f09bab-1837-4f03-b547-ff3ec1a4a5b4_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!XejA!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F27f09bab-1837-4f03-b547-ff3ec1a4a5b4_1536x1024.png) +*Good Reads #9 September 22, 2026* + +Public Health Initiative’s *The Health Gap* begins with two people of the same age. + +One reaches Friday expecting a weekend, some rest, and a return to work on Monday. Somewhere else, another person works through Saturday and Sunday, continues working when sick, and eventually dies earlier. + +The contrast establishes the article’s argument quickly. Health is distributed unequally because the conditions in which people live are distributed unequally. + +The piece draws on Michael Marmot’s work on health inequalities and the social determinants of health. It describes the health gap as the relationship between social and economic position, health and life expectancy, rather than treating health principally as a consequence of biology or access to medical care. + +Then it makes a much stronger claim: + +> “The health gap in reality is a wealth gap.” + +That is where the piece becomes particularly interesting. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## The gradient matters + +The basic insight is difficult to dispute. + +Health is not distributed randomly across populations. Neither is illness. Income, occupation, education, housing, food, working conditions, pollution, sanitation, social protection and the ability to obtain care are all socially patterned. + +The article makes this concrete through everyday differences between richer and poorer households. Someone with greater economic security may have better food, more time, opportunities for physical activity, access to mental-health care and the ability to take time away from work when ill. Someone without that security may have far fewer choices. + +The important word here is not simply poverty. + +It is gradient. + +The Marmot tradition is important partly because it moved discussion beyond a binary division between “the poor” and everybody else. Health outcomes tend to vary across the social hierarchy. The question is therefore not only why the poorest experience worse health. It is why health changes systematically as social position changes. + +That distinction matters for policy. A programme aimed only at people below a poverty threshold addresses something different from a social structure in which health varies across the distribution. + +## But is wealth the determinant? + +Here I would be more cautious than the article. + +Saying that wealth strongly predicts health is not quite the same as saying that wealth determines health. + +Income and wealth buy things that matter for health. They affect housing, food, neighbourhood, transport, education and access to care. They can provide protection against shocks. They influence whether a person can stop working when sick rather than choosing between illness and lost income. + +But socioeconomic position is also entangled with other institutions and relationships. + +Consider two workers earning similar amounts. One has a permanent job, paid sick leave, predictable hours and health insurance. The other works casually, loses income whenever she misses a day, has little control over her schedule and faces regular occupational hazards. + +Their incomes may look similar. Their exposure to health risks does not. + +Or consider two households with comparable incomes living in places with very different air quality, water systems, public health services or exposure to extreme heat. + +Again, income alone does not describe the relevant conditions. + +Gender, caste, race, migration status, disability, geography and political power can affect health independently and through their interaction with economic position. + +Wealth matters enormously. But it is partly a resource and partly a marker for access to a much wider set of resources, protections and forms of control. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Measurement makes this harder + +There is also a measurement problem hiding inside the argument. + +What exactly do we mean by socioeconomic status? + +Income? + +Consumption? + +Assets? + +Occupation? + +Education? + +Housing? + +Wealth? + +A composite index? + +These measures are related, but they are not interchangeable. + +This becomes particularly important in development research. Household consumption may tell us something that reported monthly income does not. An asset index can rank households without telling us whether they have enough liquid resources to survive an emergency. Household wealth tells us little about who within the household controls it. + +Even apparently straightforward measures can therefore produce different versions of the health gradient. + +And then causality runs in both directions. + +Low socioeconomic status can damage health. Poor health can also reduce earnings, interrupt education, exhaust savings, alter employment and push households into poverty. + +A correlation between wealth and health can contain both processes at once. + +## The India question + +This becomes especially interesting in India because socioeconomic position cannot be separated neatly from the institutions through which people encounter health risks. + +A woman’s ability to seek treatment may depend on household income, but also on whether she controls money, can travel alone, has time away from unpaid work and is taken seriously when she reaches a health facility. + +A construction worker’s health is related to earnings, but also to occupational exposure, employment arrangements, housing and access to social protection. + +A child’s health reflects household resources, but also sanitation, air pollution, nutrition, maternal health, schooling and the quality of public services available where the family lives. + +The analytical question is therefore not simply whether poorer people have worse health. + +We know that inequality matters. + +The more useful question is how economic inequality becomes biological inequality. + +What are the pathways? + +Which can public policy interrupt? + +And at what point in the life course? + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Where I disagree with the piece + +*The Health Gap* ends in deliberately stark terms. It describes health inequality as a world in which those with money occupy the position of the strong and those without it the weak, ultimately determining “who lives and dies.” + +I understand the force of that formulation. I am less persuaded by its reduction of the problem to wealth. + +If wealth alone were the principal mechanism, redistribution would do most of the work. + +Redistribution matters. But so do labour law, universal health care, clean air, safe water, housing, nutrition, education, social protection, gender relations, working conditions and the quality of public institutions. + +The distinction is important because diagnosis shapes prescription. + +Calling the health gap a wealth gap directs our attention towards inequality in resources. + +Calling it a social gradient in health asks a broader question: what is it about the organisation of society that allows social position to enter the body? + +That question is harder. + +I think it is also closer to the reason Marmot’s work remains so useful. + +The most interesting thing about the health gap is not simply that money can buy better health. + +It is that inequalities which appear social, economic and political eventually become measurable as differences in illness, disability and years of life. + +Public health begins much earlier than the clinic. + +--- + +**Source:** Public Health Initiative, “The Health Gap: What is the #1 Determinant of Health?”, 28 August 2026. + +[Share](https://varna.substack.com/p/good-reads-9-is-the-health-gap-really?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-23-research-rundown-daily-23-september.md b/archive/2026-09-23-research-rundown-daily-23-september.md new file mode 100644 index 0000000..20da209 --- /dev/null +++ b/archive/2026-09-23-research-rundown-daily-23-september.md @@ -0,0 +1,159 @@ +--- +title: "Research Rundown DAILY - 23 September 2026" +date: 2026-09-23 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-23-september +subtitle: "Digital access is expanding across South Asia, but new evidence shows how much can disappear between formal coverage and effective use." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 23 September 2026 + +*2026-09-23. Published at [https://varna.substack.com/p/research-rundown-daily-23-september](https://varna.substack.com/p/research-rundown-daily-23-september). This file is a generated copy; the Substack post is the original.* + +[![**Alt text:** Research Rundown DAILY cover for 23 September 2026, showing entries declining from availability and eligibility through enrolment, use and benefits received, alongside records on MSME finance, social protection and digital connectivity across South Asia.](https://substackcdn.com/image/fetch/$s_!UJHD!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Febb5715f-1a60-4c47-b5d8-b9e0ea007804_1672x941.png)](https://substackcdn.com/image/fetch/$s_!UJHD!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Febb5715f-1a60-4c47-b5d8-b9e0ea007804_1672x941.png) +*Research Rundown Daily 23rd September 2026* + +## FIVE THINGS WORTH KNOWING + +### 1. ADB now sees El Niño joining the energy shock as South Asia’s next macroeconomic risk + +The Asian Development Bank’s September outlook, released today, raises its 2026 growth forecast for developing Asia and the Pacific slightly to **5.0 per cent**, from 4.9 per cent in July. That small upgrade should not obscure the change in the risk assessment. ADB now expects a strengthening El Niño to threaten harvests and hydropower alongside the continuing energy shock. ([Asian Development Bank](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com)) + +This matters particularly for South Asia because the two shocks can interact. Higher energy costs raise irrigation, fertiliser, transport and electricity costs. Drier conditions can reduce agricultural output and hydropower generation, increasing demand for expensive imported energy. + +The point is not the 0.1 percentage-point forecast revision. Forecast changes of that size are well inside the uncertainty surrounding regional projections. The important development is the **composition of risk**: food, electricity and household purchasing power can now be affected through several channels at once. + +[Asian Development Outlook, September 2026](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com) + +### 2. Bangladesh’s growth forecast has been cut again + +ADB now expects Bangladesh’s economy to grow by **4.0 per cent in FY2027**, down from the 4.5 per cent forecast in July. The earlier forecast had already been reduced because of weak exports, subdued private investment, energy constraints and banking-sector problems. ([Jagonews24](https://www.jagonews24.com/en/biz-econ/news/96351?utm_source=chatgpt.com)) + +The revision matters because Bangladesh is being hit through several channels simultaneously. The energy disruption raises production costs and constrains factories. Inflation remains high. Problems in the banking system restrict investment, while external demand remains uncertain. + +This is also a useful reminder about forecasts. A downward revision is new information about an institution’s assessment of the economy. It is not an observed fall in GDP. The 4.0 per cent figure is conditional on assumptions about energy prices, external demand, domestic policy and the duration of current disruptions. + +For Bangladesh, the next few months of industrial production, exports and private credit will tell us more than the forecast itself. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 3. Women-led MSMEs in Bangladesh are digitally active and still struggle to obtain formal credit + +A study released in Dhaka surveyed **329 MSMEs across all eight administrative divisions**, supplemented by 19 key-informant interviews, 12 in-depth interviews and four focus groups. It found that 87 per cent of surveyed businesses had some digital footprint and 97 per cent used mobile financial services, yet only **11.55 per cent had accessed digitally enabled credit**. ([bdnews24.com](https://bdnews24.com/business/htbuhps4ue?utm_source=chatgpt.com)) + +Women-led businesses were especially interesting. Women entrepreneurs used an average of 2.75 digital platforms, compared with 1.88 among male-led enterprises, and 83 per cent expressed interest in digital credit. Yet women-led businesses receive less than 10 per cent of formal SME credit, according to the study. ([The Daily Star](https://www.thedailystar.net/business/news/women-led-msmes-lead-digital-finance-still-face-heavy-loan-barriers-study-4280846?utm_source=chatgpt.com)) + +The survey is too small to support precise national prevalence estimates, and the sampling design matters before treating its percentages as representative of all Bangladeshi MSMEs. The mechanism it identifies is plausible and important: lenders still rely heavily on collateral and documentation even when digital transactions provide other information about a firm. + +Digital inclusion and credit inclusion are therefore different outcomes. + +### 4. India’s states face a fiscal problem that national public-finance aggregates can hide + +A background paper prepared by two RBI researchers for the Union Finance Ministry’s meeting with state finance ministers argues that demographic ageing, urbanisation, climate change and technological change will place increasing pressure on state budgets. States account for roughly **two-thirds of India’s general-government expenditure**. ([The Financial Express](https://www.financialexpress.com/policy/economy-centre-flags-new-fiscal-strains-for-states-from-ageing-ai-climate-risks-4344915/?utm_source=chatgpt.com)) + +Ageing raises pension and health expenditure. Climate adaptation requires investment before losses occur. Urbanisation creates infrastructure demands. At the same time, states retain substantial debt burdens and limited room to expand some own-source revenues. + +This is worth reading alongside debates about India’s aggregate fiscal position. Many of the services through which citizens experience the state, including health, education, water, transport and substantial parts of social protection, depend heavily on state finances. + +The paper estimates that India would need annual real GDP growth of about 8 per cent to reach the government’s 2047 development objective. That is a scenario requirement, rather than a forecast. ([The Financial Express](https://www.financialexpress.com/policy/economy-centre-flags-new-fiscal-strains-for-states-from-ageing-ai-climate-risks-4344915/?utm_source=chatgpt.com)) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 5. More than 23,000 informal workers across India, Nepal and Pakistan gained access to formalisation and social-protection pathways through two ILO programmes + +The International Labour Organization has published results from two recently completed South Asian programmes. It reports that more than **23,000 informal workers** in India, Nepal and Pakistan obtained improved access to formal employment, social protection, representation or related services. A parallel programme worked on child-labour institutions in India and Nepal. ([International Labour Organization](https://www.ilo.org/resource/news/japan-supported-ilo-initiatives-strengthen-protections-workers-and-children?utm_source=chatgpt.com)) + +The wording matters. “Improved access” combines several possible outcomes and should not be read as 23,000 workers entering permanent formal employment. Organisation, registration, social-protection enrolment and actual receipt of benefits are distinct stages. + +That distinction is particularly important in South Asia, where formalisation is often measured administratively. Registration is relatively easy to count. Whether registration produces insurance coverage, income security, bargaining power or more stable employment requires follow-up data. + +[ILO: worker formalisation and child-labour programmes in South Asia](https://www.ilo.org/resource/news/japan-supported-ilo-initiatives-strengthen-protections-workers-and-children?utm_source=chatgpt.com) + +# ONE PAPER WORTH READING + +### The health production function in South Asia + +Mohammad Ridwan and colleagues assemble annual data for Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka from **2000 to 2025** and examine associations between life expectancy and education, health expenditure, GDP per capita, government effectiveness and urbanisation. Their principal estimates use panel ARDL models, with Driscoll-Kraay errors, PCSE, FGLS, System GMM and bootstrap causality tests as additional specifications. Education and GDP per capita are positively associated with longevity; the health-expenditure coefficient is positive but sensitive to model choice. ([Frontiers](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2026.1866937/full?utm_source=chatgpt.com)) + +The paper is useful partly because its difficulties are instructive. Seven countries over 26 years provide a small macro panel, and national averages hide large within-country differences. More importantly, cointegration, Granger causality and System GMM do not by themselves establish causal effects. Government effectiveness and urbanisation even acquire negative coefficients in some specifications, a warning against attaching simple structural interpretations to the estimates. + +I would read this as evidence about **macro-level associations**, especially the consistent education relationship, rather than as estimates of what an additional dollar of health spending will do to life expectancy. + +**Citation:** Ridwan, M., Z. A. Antor, J. Ko, M. E. Hossain, C. K. Leung, W-K. Ming, and M. Z. Rehman. 2026. “The Health Production Function in South Asia: Panel Evidence on Education, Health Expenditure, Governance, and Longevity.” *Frontiers in Public Health* 14:1866937. Published 16 September 2026. DOI 10.3389/fpubh.2026.1866937. ([Frontiers](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2026.1866937/full?utm_source=chatgpt.com)) + +[Read the open-access paper](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2026.1866937/full?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +# ONE NUMBER + +## 11.55 per cent + +The share of the 329 Bangladeshi MSMEs surveyed that had accessed digitally enabled credit, despite **87 per cent having a digital footprint and 97 per cent using mobile financial services**. ([bdnews24.com](https://bdnews24.com/business/htbuhps4ue?utm_source=chatgpt.com)) + +This is a useful distinction between infrastructure and outcome. + +A firm can have a mobile wallet, digital transaction history and online presence while remaining outside formal credit markets. Digital adoption therefore cannot serve as a proxy for financial inclusion without specifying which financial service is actually being measured. + +The sample is small enough that I would resist treating 11.55 per cent as a national estimate. The gap between digital use and digital borrowing is the more defensible finding. + +# POLICY WATCH + +## Pakistan is putting Rs33 billion into connecting places the market has left behind + +Pakistan’s Universal Service Fund policy committee has approved a **Rs33 billion budget** for expanding digital connectivity in unserved and underserved areas. The programme is explicitly intended to address locations where commercial provision remains weak. ([Dawn](https://www.dawn.com/latest-news/2026-09-23?utm_source=chatgpt.com)) + +This sits neatly beside today’s Bangladesh evidence because it raises the next question in digital-inclusion policy. + +Coverage is the first constraint. Usage is another. Product suitability, affordability and access to credit or other economically useful services come later. + +The evaluation sequence should therefore resist treating network expansion as equivalent to economic inclusion. Useful indicators would move from population covered, to actual subscriptions and regular use, to service quality and price, and eventually to outcomes such as access to markets, finance or employment. + +The distinction sounds obvious. It is frequently lost once programmes begin reporting kilometres of fibre, towers installed or people nominally covered. + +[Subscribe now](https://varna.substack.com/subscribe?) + +# EDITORIAL NOTE + +## Access is a chain, and we keep measuring its first link + +Three items today share a measurement problem. + +Pakistan can extend a telecommunications network. Bangladesh can have almost universal mobile-finance use among a sample of MSMEs. An ILO programme can help an informal worker register for social protection. + +All are meaningful changes. None tells us whether the final economic outcome occurred. + +Development programmes frequently have a chain running from **availability → eligibility → enrolment → use → effective service → outcome**. Administrative systems become progressively weaker as we move along it. + +That creates a systematic bias in what governments and programmes know about themselves. Inputs and registrations are observed quickly and cheaply. Failed transactions, abandoned applications, poor service quality and people who remain eligible but never enrol are harder to see. + +The Bangladesh MSME result is a particularly clean example. A digital footprint exists. Mobile financial services are being used. Credit still does not follow. + +For evaluation, I would increasingly ask programmes to specify **where in the access chain their headline indicator sits** before accepting it as evidence of inclusion. + +**Reader question:** What is the strongest study you know that follows the same people or firms through several stages of an access chain, from eligibility and enrolment through actual use to an economic or welfare outcome? + +This one belongs in a future Methods Monday. There is a useful piece to write on why coverage, take-up, utilisation and effective access should be treated as separate estimands. + +## Sources + +[Asian Development Bank: Asian Development Outlook September 2026](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com) + +[ADB: September outlook release](https://www.adb.org/news/growth-developing-asia-and-pacific-slow-remain-resilient?utm_source=chatgpt.com) + +[ADB Data Library: September 2026 economic forecasts](https://data.adb.org/?utm_source=chatgpt.com) + +[The Daily Star: Women-led MSMEs and digital finance in Bangladesh](https://www.thedailystar.net/business/news/women-led-msmes-lead-digital-finance-still-face-heavy-loan-barriers-study-4280846?utm_source=chatgpt.com) + +[bdnews24: MSME digital-credit study](https://bdnews24.com/business/htbuhps4ue?utm_source=chatgpt.com) + +[Financial Express: fiscal pressures facing Indian states](https://www.financialexpress.com/policy/economy-centre-flags-new-fiscal-strains-for-states-from-ageing-ai-climate-risks-4344915/?utm_source=chatgpt.com) + +[International Labour Organization: informal-worker formalisation programmes](https://www.ilo.org/resource/news/japan-supported-ilo-initiatives-strengthen-protections-workers-and-children?utm_source=chatgpt.com) + +[Dawn: Pakistan Universal Service Fund budget](https://www.dawn.com/latest-news/2026-09-23?utm_source=chatgpt.com) + +[Frontiers in Public Health: South Asian health production function](https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2026.1866937/full?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-23-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-25-good-reads-10-when-using-a-service.md b/archive/2026-09-25-good-reads-10-when-using-a-service.md new file mode 100644 index 0000000..c517de8 --- /dev/null +++ b/archive/2026-09-25-good-reads-10-when-using-a-service.md @@ -0,0 +1,199 @@ +--- +title: "Good Reads #10: When using a service is not the same as benefiting from it" +date: 2026-09-25 +type: other +url: https://varna.substack.com/p/good-reads-10-when-using-a-service +subtitle: "Tamil Nadu’s maternity benefit programme changed how women used health services. A new study asks why those changes did not show up in children’s nutritional outcomes." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #10: When using a service is not the same as benefiting from it + +*2026-09-25. Published at [https://varna.substack.com/p/good-reads-10-when-using-a-service](https://varna.substack.com/p/good-reads-10-when-using-a-service). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads #10 cover for “When using a service is not the same as benefiting from it,” showing maternity and child-health records alongside an evaluation ledger in which entries diminish from eligibility and receipt through use to benefit, illustrating the gap between programme take-up and final health outcomes.](https://substackcdn.com/image/fetch/$s_!joqc!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F10152eed-8192-4c21-9269-f7e310e1c0e9_1672x941.png)](https://substackcdn.com/image/fetch/$s_!joqc!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F10152eed-8192-4c21-9269-f7e310e1c0e9_1672x941.png) +*Good Reads #10 on 24th September, 2026* + +*Tamil Nadu’s maternity benefit programme appears to have changed how women use health services. A new study finds something more difficult: those changes did not translate into measurable improvements in children’s nutritional outcomes.* + +There is a particularly useful result in a new paper by Shreya Biswas, Upasak Das and Narbadeshwar Mishra. + +Tamil Nadu’s Dr. Muthulakshmi Reddy Maternity Benefit Scheme gives pregnant women cash payments tied to behaviours such as registering a pregnancy, delivering in a health facility and completing specified childhood immunisations. The scheme dates to 1987, but the transfers increased substantially from Rs 500 to Rs 6,000 in 2006 and then to Rs 12,000 in 2011. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +The authors ask what happened after those increases. + +Women did more of many of the things the programme wanted them to do. + +Pregnancy registration increased. Institutional deliveries increased. Immunisation increased. Women were also more likely to attend antenatal care, take iron tablets and tetanus injections, and begin breastfeeding immediately after birth. + +And yet the researchers find no corresponding improvement in stunting, wasting or underweight among children. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +That apparent contradiction is what makes this paper worth reading. + +## The difference between doing something and achieving something + +Conditional cash transfers have an appealing programme logic. + +Identify behaviours associated with better health. Attach a financial incentive to them. Reduce some of the financial constraints facing poorer households. More people use the relevant services. Better health follows. + +The first part of that chain appears to have worked rather well in Tamil Nadu. + +Using NFHS-3 from 2005-06 and NFHS-4 from 2015-16, the authors estimate that the expansion of MRMBS increased pregnancy registration by 11.9 percentage points and institutional delivery by 10.1 percentage points. Polio and DPT vaccination also increased. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +The effects extend beyond the behaviours directly tied to payment. + +The estimated probability of receiving at least four antenatal check-ups rose by 10.7 percentage points. Iron-tablet consumption increased by 10 percentage points, tetanus vaccination by 7.2 points and immediate initiation of breastfeeding by 9.9 points. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +Those are meaningful behavioural changes. + +But the children of eligible women were no less likely to be stunted, wasted or underweight. The authors also find no improvement in the corresponding anthropometric z-scores, anaemia or short-term morbidity. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +The programme appears to have moved several intermediate indicators without moving the final indicators the intervention was ultimately intended to improve. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## This is not evidence that the programme failed + +That distinction matters. + +It would be easy to read the null nutritional results and conclude that the cash transfer did not work. I don’t think the paper supports such a simple conclusion. + +Pregnancy registration matters. Antenatal care matters. Immunisation matters. Institutional delivery matters. A programme that increases the use of these services has produced something of value even if anthropometric measures do not subsequently move. + +Nor should we expect every maternal-health intervention to produce readily detectable changes in child height and weight. + +Stunting is cumulative. Wasting reflects a different nutritional process. Both are produced by many exposures occurring before and after birth. A cash transfer encouraging particular maternal and child health behaviours occupies only one part of that causal chain. + +The authors themselves are careful about this. NFHS is a repeated cross-section, not a longitudinal panel. They cannot follow individual children through time. Their estimates therefore capture cohort-level differences associated with programme exposure rather than individual long-term treatment effects. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +That makes the absence of an anthropometric effect important, but not a referendum on the entire programme. + +## What happened between the clinic and the child? + +The authors go looking for mechanisms. + +One possibility would be that households received additional cash but spent it in ways unrelated to child health. They find no evidence of increased spending on what the literature rather unfortunately calls “temptation goods,” nor increased possession of durable assets. + +There is some evidence of greater access to toilets. + +But something else does not change: children’s diets. + +The researchers examine consumption across seven food groups and construct a measure of minimum dietary diversity. They find no evidence that children in beneficiary households received better-quality or more diverse diets. Meal frequency did not improve either. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +Nor did the programme appear to increase engagement with health workers. + +This offers one explanation for the disconnect. + +A woman may attend antenatal care because a transfer encourages her to do so. She may deliver in a health facility. Her child may receive the required vaccinations. + +None of those changes necessarily means that the child’s diet subsequently becomes more diverse, that the household receives sustained nutritional support, or that interactions with the health system are of sufficient quality to alter longer-term nutritional outcomes. + +The transfer can move one part of the chain without moving everything downstream. + +## There is another explanation worth considering: Tamil Nadu + +The setting makes the paper particularly interesting. + +Tamil Nadu was already performing comparatively well on many maternal and child health indicators before the larger transfers were introduced. Among children of eligible mothers in the authors’ pre-expansion data, 28 per cent were stunted and 19 per cent wasted. The state was also improving primary health infrastructure and expanding round-the-clock services at PHCs during this period. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +That creates an important evaluation problem. + +The effect of introducing an intervention into a poorly functioning health system is not necessarily the same as the effect of adding money to a system in which utilisation is already relatively high. + +Marginal gains may become harder to obtain. + +The authors recognise this, describing the study partly as an examination of the returns to a cash transfer in a state that had already made substantial progress in maternal and child health. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +This also makes me cautious about transporting the result elsewhere. + +The finding is not that conditional cash transfers cannot improve child nutrition. It is that this expansion, in this institutional setting, over this period, produced detectable changes in several health behaviours without detectable changes in the nutritional outcomes examined. + +That is both narrower and more useful. + +## And there are identification questions + +The empirical strategy deserves attention too. + +The researchers use a triple-difference design, exploiting changes over time, differences between Tamil Nadu and Karnataka, and eligibility according to birth order. NFHS-3 provides the pre-expansion observations and NFHS-4 the post-expansion observations. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +This is considerably more informative than simply comparing beneficiaries and non-beneficiaries after the programme. + +But the design still asks a lot of observational data. + +The authors test pre-existing trends using NFHS-2 and NFHS-3. For most outcomes, the relevant coefficients are statistically insignificant. But the assumption does not hold for antenatal-care visits or wasting, and some variables cannot be tested because they were not collected in NFHS-2. The authors explicitly caution against a straightforward causal interpretation of those particular estimates. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +There is another complication. Women in the pre-period were not literally untreated. MRMBS already existed. What changed was the size of the payment, dramatically. + +So this is better understood as an evaluation of an expansion in programme generosity than a clean comparison between a world with the programme and one without it. + +None of this makes the results uninformative. It tells us how carefully they should be read. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## The indicator ladder + +What stayed with me after reading the paper was not really a question about cash transfers. + +It was a question about evaluation. + +A programme like this contains a sequence: + +**eligibility → receipt of cash → use of services → quality of care → changes in household behaviour → improved health** + +Each step is an outcome. + +But they are not equivalent outcomes. + +If institutional delivery increases, we have evidence about institutional delivery. We do not yet have evidence about maternal mortality. + +If immunisation increases, we have evidence about immunisation. We do not yet have evidence about nutritional status. + +If antenatal visits increase, we know more women reached a health service. We still need to know what happened during those visits. + +This sounds obvious when written out. It becomes surprisingly easy to forget once programmes acquire dashboards. + +Yesterday’s Research Rundown looked at essentially the same problem in completely different sectors. + +***A Bangladeshi MSME can have a digital footprint and use mobile financial services without obtaining formal credit.*** + +***A worker can register for social protection without necessarily receiving effective protection.*** + +***A village can acquire network coverage without residents obtaining an affordable, reliable service.*** + +***And a pregnant woman can reach a health facility without that contact necessarily producing every health outcome farther down the chain.*** + +***The common mistake is to allow an observable intermediate indicator to stand in for the outcome we actually care about.*** + +That does not make intermediate indicators unimportant. + +It means we should name them correctly. + +## What I take from the paper + +The policy response to these findings should not be to abandon conditional maternity benefits. + +The more interesting response is to ask what additional conditions are required for increased health-service use to become better health. + +The authors point towards diet quality and engagement with health workers. The wider literature they discuss suggests others: quality of care, sanitation, public-health infrastructure and the size of the transfer itself. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +I would add one more. + +Evaluation needs to preserve the distance between stages in the causal chain. + +A programme can succeed at one stage and stall at another. Knowing where that happens is more useful than forcing the entire intervention into a binary judgment of success or failure. + +This Tamil Nadu study gives us a particularly clean example. + +Women changed their behaviour. + +The children’s measured nutritional outcomes did not. + +The research question begins in the space between those two findings. + +--- + +**Source:** Shreya Biswas, Upasak Das and Narbadeshwar Mishra, “Does Better Health Seeking Behavior Ensure Improved Health Outcomes? Evidence from the Muthulakshmi Reddy Maternity Benefit Scheme in India,” *The Journal of Development Studies*, published online 14 September 2026. DOI: 10.1080/00220388.2026.2727161. [Taylor & Francis Online](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +[Read the open-access paper](https://www.tandfonline.com/doi/full/10.1080/00220388.2026.2727161?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-10-when-using-a-service?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-25-research-rundown-friday-findings.md b/archive/2026-09-25-research-rundown-friday-findings.md new file mode 100644 index 0000000..8b3c6b9 --- /dev/null +++ b/archive/2026-09-25-research-rundown-friday-findings.md @@ -0,0 +1,149 @@ +--- +title: "Research Rundown: Friday Findings - 25 September 2026" +date: 2026-09-25 +type: edition +url: https://varna.substack.com/p/research-rundown-friday-findings +subtitle: "Forecast heat deaths, uneven digital access, missing MSME credit, and the fiscal consequences of climate shock in South Asia." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown: Friday Findings - 25 September 2026 + +*2026-09-25. Published at [https://varna.substack.com/p/research-rundown-friday-findings](https://varna.substack.com/p/research-rundown-friday-findings). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Friday Findings cover for 25 September 2026, showing a research desk with a heat-mortality forecast chart, digital-access indicators on a phone, an MSME credit form, an Asian growth chart, and a map of Nepal.](https://substackcdn.com/image/fetch/$s_!YEkS!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5b75363d-0b2d-4ef9-af32-0624e980e1a9_1672x941.png)](https://substackcdn.com/image/fetch/$s_!YEkS!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5b75363d-0b2d-4ef9-af32-0624e980e1a9_1672x941.png) +*Friday Findings - 25th September 2026* + +Heat mortality before the heat arrives, the limits of digital inclusion, Bangladesh’s missing MSME credit, and an Asian economy caught between energy prices and El Niño. + +This was a thinner research week than the previous one. I found six pieces worth carrying. The most interesting material concerns measurement and anticipation: whether we can estimate mortality before a climate shock occurs, whether digital access tells us anything useful about digital independence, and whether formal financial inclusion statistics capture firms that abandon the credit process before applying. + +### 1. India could face 15,800 additional heat deaths from the current El Niño + +Greenstone, Michael, Tamma Carleton, Genevieve Maricle, and Emily Grover-Kopec. 2026. *451,000 at Risk: Where Action Can Save Lives During the Super El Niño*. Climate Impact Lab, September 2026. + +This is the paper I would read first, with one large qualification. + +Climate Impact Lab estimates that the current El Niño could produce 451,000 additional heat-related deaths globally between June 2026 and February 2027 relative to temperatures in an average year. For the six months from September through February, the estimate is 239,000 ± 11,000. India accounts for an estimated 15,800 additional deaths over that six-month period, including about 7,400 between December and February. [Climate & Sustainable Growth Institute](https://climate.uchicago.edu/news/more-than-451000-heat-deaths-projected-from-super-el-nino/?utm_source=chatgpt.com) + +That distinction between periods matters because several reports of the study have incorrectly described 451,000 as a six-month estimate. It covers June through February. The six-month forward estimate is 239,000. [Al Jazeera](https://www.aljazeera.com/features/2026/9/23/scientists-warn-super-el-nino-could-cause-450000-deaths-should-we-worry?utm_source=chatgpt.com) + +The modelling combines estimated temperature-mortality relationships across 24,378 regions with seasonal temperature forecasts, allowing local climate and vulnerability to alter the mortality response. The counterfactual is the corresponding months under average 1996–2025 conditions. [Al Jazeera](https://www.aljazeera.com/features/2026/9/23/scientists-warn-super-el-nino-could-cause-450000-deaths-should-we-worry?utm_source=chatgpt.com) + +There is real analytical value here. Heat policy usually responds to forecasts of temperature. This work attempts to translate those forecasts into expected mortality at geographically useful scales. + +There is also substantial uncertainty. These are modelled deaths, rather than observed mortality. Seasonal temperature forecasts carry error, the temperature-mortality functions are estimated, and the method has yet to appear in the peer-reviewed paper the authors say is forthcoming. The India figure should therefore be treated as a forecast with uncertainty rather than an expected death count waiting to happen. + +The timing is unusual enough to matter for Indian heat planning. Nearly half of the projected September-February mortality occurs during December-February, outside the months in which Indian heat preparedness is usually concentrated. [Climate & Sustainable Growth Institute](https://climate.uchicago.edu/news/more-than-451000-heat-deaths-projected-from-super-el-nino/?utm_source=chatgpt.com) + +This deserves a longer Research Rundown treatment if the underlying methodological paper appears. + +[Read the Climate Impact Lab analysis](https://impactlab.org/research/451000-at-risk/?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. ADB puts numbers on the joint energy and El Niño shock + +Asian Development Bank. 2026. *Asian Development Outlook September 2026: Navigating Prolonged Energy Shocks and El Niño*. Manila: Asian Development Bank. DOI: 10.22617/FLS260387-3. + +ADB now expects developing Asia and the Pacific to grow by 5.0% in 2026 and 5.1% in 2027. South Asia is forecast at 6.4% and 6.5%, respectively. India is projected to grow by 7.0% in FY2026, with consumption, investment and services exports partly offsetting higher energy costs and a weaker monsoon. [Asian Development Bank](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com) + +The useful part is the scenario analysis. ADB models the macroeconomic consequences of a very strong El Niño alongside prolonged disruption in energy markets. Across the region, energy costs and food-supply effects create the familiar combination of slower output growth and higher inflation. [Asian Development Bank](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com) + +Nepal shows why regional averages conceal a great deal. Following the August floods, ADB cut its FY2027 growth forecast from 4.5% to 4.1% and raised its inflation forecast to 5.3%. Damage to hydropower, roads and other productive assets is expected to depress activity while reconstruction raises expenditure and demand for materials. [Kathmandu Post](https://kathmandupost.com/money/2026/09/23/bhotekoshi-floods-prompt-adb-to-cut-nepal-s-growth-forecast-to-4-1-percent?utm_source=chatgpt.com) + +The numbers are forecasts, and they should be read as such. Their value lies in the explicit assumptions and scenario comparisons rather than the second decimal place of any growth estimate. + +There is a direct thread here from the recent Daily on South Asia’s simultaneous energy and El Niño exposure. The full ADO supplies the modelling behind that story. + +[Read the Asian Development Outlook September 2026](https://www.adb.org/publications/asian-development-outlook-september-2026?utm_source=chatgpt.com) + +### 3. India’s digital divide becomes much steeper with age + +Paul, Sohini, and Sudeshna Dey. 2026. “Making India’s Digital Public Infrastructure Work for Older Citizens.” *Ideas for India*, 25 September 2026. Based on Dey and Paul, “Bridging India’s Generational Digital Divide: Rethinking Digital Public Infrastructure for Elderly Inclusion,” *CSI Transactions on ICT*. + +Paul and Dey use the nationally representative NSS Comprehensive Modular Survey on Telecom to separate digital access from capability and safety among older Indians. + +The age gradient is large. Smartphone access falls from 88.6% among adults aged 18–49 to 44.7% among those aged 60–69 and 42% among those aged 70 and above. Among people with smartphone access, recent use remains above 97% across age groups. Access to an appropriate device therefore appears to be part of the constraint. [Ideas for India](https://www.ideasforindia.in/topics/governance/making-indias-digital-public-infrastructure-work-for-older-citizens) + +Capability falls further down the chain. About 53.1% of adults aged 60 and above report being able to undertake online banking, against 71.4% among 18–49-year-olds. Only 8% of people aged 60 and above report knowing how to complain about cybercrime through the official portal. The figure is 6% among older women and 4% among older rural residents. [Ideas for India](https://www.ideasforindia.in/topics/governance/making-indias-digital-public-infrastructure-work-for-older-citizens) + +The authors also estimate regressions controlling for gender, social group, religion, rural or urban residence, household internet access, consumption expenditure and state. Adults aged 70 and above have about 65% lower adjusted odds of being able to use the internet than adults aged 50–59, and 52% lower adjusted odds of conducting online banking. [Ideas for India](https://www.ideasforindia.in/topics/governance/making-indias-digital-public-infrastructure-work-for-older-citizens) + +These remain observational associations. Age captures accumulated differences in schooling, employment, exposure to technology and other characteristics that controls will only partly absorb. Self-reported ability also differs from demonstrated ability. + +The measurement point is stronger. Counting enrolments, IDs, accounts or transactions can substantially overstate effective access to a digital public service. Independence, successful task completion, assistance, failed transactions and grievance resolution belong in the outcome set. + +That connects directly to last week’s financial-health piece. Both papers expose the same problem: counting an input as though it were the outcome. + +[Read the open-access Ideas for India article](https://www.ideasforindia.in/topics/governance/making-indias-digital-public-infrastructure-work-for-older-citizens?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 4. Bangladesh’s MSMEs have digital footprints. Very few receive digital credit + +LightCastle Partners. 2026. *Ecosystem Mapping and Diagnostic Study on MSME Financing in Bangladesh*. Research conducted under The Asia Foundation’s Bridging the Gap programme; findings disseminated 22 September 2026. + +This is a useful study with a modest sample. + +LightCastle surveyed 329 MSMEs across Bangladesh’s eight administrative divisions and supplemented the survey with 19 key-informant interviews, 12 in-depth interviews and four focus groups. It estimates a national MSME financing gap of US$2.8 billion. [Daily Sun](https://www.daily-sun.com/business/899232?utm_source=chatgpt.com) + +The descriptive results reveal a striking break in the credit pathway. Some 87% of surveyed firms had a digital footprint and 97% used mobile financial services, yet only 11.55% had accessed digitally enabled credit. Although 86% reported having borrowed, only 58% had formally applied for loans. [Daily Sun](https://www.daily-sun.com/business/899232?utm_source=chatgpt.com) + +The gender result deserves attention. Women-led firms used more digital platforms and showed strong interest in digital credit, while women-led businesses continue to receive a small share of formal SME lending. The report argues that collateral requirements and documentation remain important barriers. [The Daily Star](https://online.thedailystar.net/business/news/women-led-msmes-lead-digital-finance-still-face-heavy-loan-barriers-study-4280846?utm_source=chatgpt.com) + +I would be careful with the national financing-gap estimate. A survey of 329 enterprises can describe the sampled firms well if the sampling design is sound. Scaling from that sample to a dollar estimate for a sector containing roughly 7.16 million firms requires assumptions that need inspection in the full technical report. The dissemination material available online gives much more information about findings than about weighting and extrapolation. + +The more interesting result may be the application gap. Credit exclusion begins before a lender rejects an application. Firms that expect rejection, cannot satisfy documentation requirements or regard available products as unsuitable disappear from approval-rate statistics. + +[Read the detailed report coverage and methods summary](https://bdnews24.com/business/htbuhps4ue?utm_source=chatgpt.com) + +### 5. Indian cities have more money coming. The institutional problem remains + +Ara, Shamim, and Aditi Pathak. 2026. “Promoting Ease of Living in Indian Cities.” *Ideas for India*, 21 September 2026. + +This is substantive policy analysis rather than a new empirical paper, and I am including it because it follows neatly from last week’s WRI work on transitional urban areas. + +Ara and Pathak examine municipal finance and service delivery as India’s urban population becomes increasingly concentrated in large urban agglomerations. They draw attention to the Sixteenth Finance Commission’s recommended ₹3.56 lakh crore allocation to urban local bodies for 2026–31, compared with ₹1.21 lakh crore under the Fifteenth Finance Commission. Eighty per cent is structured as basic grants and 20% as performance grants. A ₹56,100 crore special infrastructure component targets wastewater management, while a ₹10,000 crore urbanisation premium addresses peri-urban integration. [Ideas for India](https://www.ideasforindia.in/topics/urbanisation/promoting-ease-of-living-in-indian-cities?utm_source=chatgpt.com) + +The piece is useful beside WRI’s transitional-urban-areas paper because money and administrative geography meet here. If settlements spill across municipal boundaries, assigning funds to individual urban local bodies cannot by itself solve metropolitan transport, wastewater, housing or land-use problems. + +There is little causal evidence in the article. Claims about which cities provide a better quality of life rely partly on composite Ease of Living measures, which combine heterogeneous indicators and citizen perceptions. Rankings of this kind deserve the same scrutiny as any composite index. + +The institutional question is still worth retaining: whether larger transfers can produce better services when metropolitan functions cross jurisdictions and municipalities have limited own-source revenue and planning authority. + +[Read the open-access Ideas for India analysis](https://www.ideasforindia.in/topics/urbanisation/promoting-ease-of-living-in-indian-cities?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 6. Nepal’s flood is now showing up in the public accounts + +Asian Development Bank. 2026. Nepal country assessment in *Asian Development Outlook September 2026*, supplemented by the Government of Nepal’s post-flood assessments. + +Nepal is worth separating from the regional ADO because the August disaster has created an unusually clear link between climate shock, infrastructure loss and fiscal pressure. + +ADB now projects FY2027 growth of 4.1%, down 0.4 percentage points from its July forecast. The fiscal deficit is projected at 4.7% of GDP, compared with a pre-disaster projection of 3.4%. Public debt had already reached about 44.9% of GDP in FY2026. [Kathmandu Post](https://kathmandupost.com/money/2026/09/23/bhotekoshi-floods-prompt-adb-to-cut-nepal-s-growth-forecast-to-4-1-percent?utm_source=chatgpt.com) + +The government’s preliminary reconstruction estimate is NPR 723.31 billion. More than 430 MW of electricity generation, roughly one-tenth of Nepal’s installed capacity, was forced off the grid after the floods. A separate Nepal Tourism Board assessment estimates NPR 52.67 billion in direct and induced tourism losses. [Kathmandu Post](https://kathmandupost.com/money/2026/09/23/bhotekoshi-floods-prompt-adb-to-cut-nepal-s-growth-forecast-to-4-1-percent?utm_source=chatgpt.com) + +The tourism estimate needs particular care. Its induced-loss component uses a six-month projection based partly on tourism income over the previous two years. It is therefore an estimate of foregone activity under stated assumptions, rather than an observed loss. [The Diplomat Nepal](https://diplomatnepal.com/news/preliminary-assessment-shows-tourism-sector-suffered-rs-52-67-billion-losses-from-bhotekoshi-trishuli-floods/?utm_source=chatgpt.com) + +This is becoming an unusually useful case for studying post-disaster public finance. Reconstruction spending can raise measured GDP while households, infrastructure and public balance sheets remain worse off. GDP growth during reconstruction is a poor measure of welfare recovery. + +I would keep Nepal in view for the flagship rather than treat each new damage estimate as a separate Daily. + +[Read ADB’s September 2026 outlook and South Asia assessment](https://www.adb.org/outlook/editions/september-2026?utm_source=chatgpt.com) + +## what I would keep + +The Climate Impact Lab analysis is the week’s most consequential piece and the one requiring the most caution. A mortality forecast is much harder to communicate than an observed mortality estimate. The distinction becomes especially important when the headline number is already being reported with the wrong time period. [Al Jazeera](https://www.aljazeera.com/features/2026/9/23/scientists-warn-super-el-nino-could-cause-450000-deaths-should-we-worry?utm_source=chatgpt.com) + +The Paul-Dey analysis is the cleanest empirical contribution. Its larger lesson is familiar from financial inclusion, social protection and health systems: administrative reach is a poor proxy for effective use. The Bangladesh MSME study reaches a related conclusion from the firm side. Digital traces are widespread. Credit remains scarce. + +I found no new open-access South Asian nutrition study from 19–25 September with enough empirical substance to include. The same applies to WEE research beyond the gender results embedded in the Bangladesh MSME study. Six pieces are enough. + +## for readers + +I am looking for South Asian studies that measure digital public-service access through observed task completion, failed transactions or grievance resolution rather than enrolment or self-reported use. If you know one, send it to me. I will credit reader recommendations that make a future Friday Findings. + +[Share](https://varna.substack.com/p/research-rundown-friday-findings?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-27-good-reads-11-what-happens-when-south.md b/archive/2026-09-27-good-reads-11-what-happens-when-south.md new file mode 100644 index 0000000..173ad0a --- /dev/null +++ b/archive/2026-09-27-good-reads-11-what-happens-when-south.md @@ -0,0 +1,207 @@ +--- +title: "Good Reads #11: What happens when South Asia has fewer children to educate?" +date: 2026-09-27 +type: other +url: https://varna.substack.com/p/good-reads-11-what-happens-when-south +subtitle: "South Asia’s child population is beginning to shrink. A World Bank analysis says that this could release substantial resources for education. Whether it produces better schools is another question..." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #11: What happens when South Asia has fewer children to educate? + +*2026-09-27. Published at [https://varna.substack.com/p/good-reads-11-what-happens-when-south](https://varna.substack.com/p/good-reads-11-what-happens-when-south). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads #11 cover for “What happens when South Asia has fewer children to educate?”, showing an old classroom desk, books and pencils beside a map of South Asia and a chalkboard chart depicting the projected decline in the region’s population of children under 15 from 420 million in 2024 to 352 million in 2050.](https://substackcdn.com/image/fetch/$s_!sgnW!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc7b89fd2-8657-41c8-83cc-ee1cdf831fdc_1672x941.png)](https://substackcdn.com/image/fetch/$s_!sgnW!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc7b89fd2-8657-41c8-83cc-ee1cdf831fdc_1672x941.png) +*Good Reads 11# on 27th September 2026 * + +For most of the past several decades, the education problem in South Asia has been framed around expansion. + +There were more children. Governments needed more schools, more teachers and more classrooms. The policy task was to get children into school and keep them there. + +That demographic assumption is beginning to change. + +Mariana Marchionni, Emmanuel Jose Vazquez and Monica Yanez Pagans point to a striking projection in a recent World Bank essay: South Asia’s population of children under 15 is expected to fall from about 420 million in 2024 to 352 million in 2050. That is a decline of roughly 16 per cent. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +Their argument is deceptively simple. If education systems have fewer children to educate, governments may be able to spend more on each child without spending a larger share of national income on education. + +The underlying World Bank research goes further. Its expenditure modelling suggests that demographic change could make resources equivalent to about 0.6 percentage points of GDP available for reinvestment in basic education by 2050, on average, even after allowing for universal basic education. [DOI](https://doi.org/10.1596/1813-9450-11303?utm_source=chatgpt.com) + +That sounds like unusually good news for a region with a persistent learning problem. + +I think the more interesting question is what has to happen for the arithmetic to become an educational gain. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Fewer children is not the same as cheaper education + +The first problem is that education systems do not shrink neatly. + +Suppose a school goes from 200 pupils to 160. Its roof still needs repairing. It may still need a head teacher, electricity, toilets and teachers across several grades. Transport costs do not disappear. Administrative systems remain. + +A 20 per cent decline in pupils does not therefore produce a 20 per cent decline in costs. + +The World Bank authors make precisely this point. Education contains substantial fixed costs. If governments leave the organisation of schooling unchanged while enrolment falls, expenditure per pupil can rise without releasing much money for improving learning. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +There is useful international experience here. The authors point to Europe and Central Asia, where declining student populations did not automatically generate efficiency gains. School structures often remained in place as enrolment fell. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +This distinction matters because “fiscal space” can sound rather more tangible than it is. + +Demography creates a possibility. Institutions determine whether governments can use it. + +## The regional average is particularly unhelpful + +There is a second problem with the 16 per cent figure. + +There isn’t really a single South Asian demographic transition. + +The underlying World Bank working paper examines eight countries and finds very different trajectories. Afghanistan and Pakistan are expected to continue experiencing growth in their school-age populations, while Bangladesh, Bhutan, India, Maldives, Nepal and Sri Lanka face declines. [World Bank](https://documents1.worldbank.org/curated/en/099120002052611106/pdf/IDU-4fde8dd8-6c02-4225-b4b7-b7a685aa192a.pdf?utm_source=chatgpt.com) + +Even within countries, national averages will conceal large differences. + +India makes this particularly obvious. A falling national child population does not imply that every district will have fewer school-age children at the same rate. Migration, fertility differences and urbanisation will redistribute demand as well as reduce it. + +That creates a planning problem. + +One district may need to decide whether several small schools remain viable. Another may still need additional classrooms because families are moving into its towns and cities. + +The interesting unit of analysis may therefore be much smaller than the country. + +That is something I would have liked the World Bank discussion to push further. National projections tell finance ministries something important about long-run expenditure. They tell education departments much less about where the next teacher, classroom or bus will actually be needed. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Then there is the uncomfortable question of school consolidation + +Once enrolment falls, consolidation enters the conversation. + +Economically, the case can be quite persuasive. Maintaining many very small schools can make it difficult to provide specialist teachers, adequate facilities and sensible pupil-teacher ratios. Combining schools could release resources for better-equipped institutions. + +But a school is not simply a cost centre. + +Closing a small village school can increase the distance a child must travel. That burden will not necessarily fall equally. Age, gender, disability, transport availability, household income and local geography can all affect whether a theoretically accessible school is accessible in practice. + +The authors recognise this problem. They argue that consolidation requires communication with communities, safe transport and better facilities in the schools receiving additional pupils. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +I would put the point more strongly. + +The relevant efficiency measure cannot simply be cost per student. + +It has to include what happens to participation and learning after the system is reorganised. + +A school closure that saves money but increases absence or dropout is not an efficiency gain in any meaningful educational sense. + +## What would we do with the money? + +There is an even larger question behind the paper. + +Suppose the demographic transition does release resources. What should education systems buy with them? + +South Asia has already made substantial progress on enrolment. The World Bank notes that primary-school entry is now close to universal across much of the region and secondary participation has risen substantially. Yet learning remains poor. In Bangladesh and India, the authors report that more than half of children completing primary school are in learning poverty, meaning they cannot read and understand a simple age-appropriate text. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +That changes the policy problem. + +The task is no longer simply: + +**How do we provide schooling to a rapidly growing number of children?** + +Increasingly, it becomes: + +**What should we do differently when the number of children stops growing?** + +The authors favour protecting education budgets and directing demographic savings towards interventions with stronger evidence of effects on learning, including better pedagogy, structured teaching support and teaching at the appropriate learning level. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +That is a sensible starting point. + +But the demographic transition could also permit something less visible: more attention per child. + +Better support for children who are behind. More resources for disability inclusion. Stronger early-childhood education. Better teacher development. Libraries and laboratories that function. Remedial teaching that does not disappear when a project ends. + +A smaller cohort could make some of these choices more affordable. + +It does not make governments choose them. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## The denominator problem + +There is also a measurement issue here that I find particularly interesting. + +Imagine that an education budget remains constant while the number of students declines. + +Expenditure per pupil rises. + +That indicator will look better. + +But nothing whatsoever may have changed inside a classroom. + +No teacher received better training. No textbook arrived. No child learned another word. The denominator simply became smaller. + +The same problem appears when education spending is expressed as a share of GDP. The indicator is useful, but it tells us little by itself about allocation, efficiency or learning. + +This is why the underlying working paper is worth reading alongside the shorter essay. The authors explicitly model education expenditure through demographic change, enrolment and spending per student under different assumptions through 2050. [DOI](https://doi.org/10.1596/1813-9450-11303?utm_source=chatgpt.com) + +The projections are scenarios, not forecasts of what governments will actually do. + +That distinction is important. + +They tell us what fiscal room demographic change could create under particular assumptions. They cannot tell us whether governments will retain education budgets, reorganise schools sensibly or spend additional resources on things that improve learning. + +## A demographic dividend needs politics + +This is where I depart slightly from the optimistic reading of the argument. + +The World Bank essay describes the demographic transition as a potential opportunity to “spend better, not just more.” [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +Perhaps. + +But there is another perfectly plausible response to falling enrolment: governments reduce education budgets. + +If fewer children are entering schools, finance ministries could treat declining enrolment as an opportunity to spend less rather than an opportunity to spend more per child. + +Nothing in the demographic transition itself determines which response wins. + +That is a political and institutional choice. + +Protecting education expenditure while the number of children declines would effectively redistribute demographic savings towards the children who remain in the system. It would represent a decision to convert quantity into quality. + +The demographic dividend in education is therefore rather different from the way the phrase sometimes sounds. + +It is not money that automatically appears when fertility falls. + +It is a choice about what to do when the denominator changes. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## What I take from this + +I like this piece because it takes a demographic fact that could easily remain in the population-projection tables and asks what it means for an actual public system. + +But I would read its numbers as an opening rather than a conclusion. + +A smaller school-age population could create room for higher expenditure per child. + +Higher expenditure per child could permit better teaching and better schools. + +Better schools could improve learning. + +Those are three separate propositions. + +The first is largely demographic arithmetic. The second depends on budgets and institutions. The third depends on what happens inside schools. + +Collapsing them into one story would make the demographic transition look far easier than it is. + +For South Asian education systems, the coming decline in child populations presents an unusual opportunity. After decades spent asking how to accommodate more children, governments may finally have some room to ask what each child actually gets from school. + +Whether they use that room is the question worth watching. + +--- + +**Source:** Mariana Marchionni, Emmanuel Jose Vazquez and Monica Yanez Pagans, “Demographic change is coming to South Asia’s classrooms. Will education systems be ready?”, *World Bank Blogs*, 15 September 2026. [World Bank Blogs](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +[Read the World Bank essay](https://blogs.worldbank.org/en/developmenttalk/demographic-change-is-coming-to-south-asia-s-classrooms--will-ed?utm_source=chatgpt.com) + +The essay draws on Mariana Marchionni, Namrata Raman Tognatta, Emmanuel Vazquez and Monica Yanez-Pagans, *Demographic Transition and Education Expenditure in South Asia: Opportunities and Challenges*, World Bank Policy Research Working Paper 11303, 2026. [DOI](https://doi.org/10.1596/1813-9450-11303?utm_source=chatgpt.com) + +[Read the underlying working paper](https://doi.org/10.1596/1813-9450-11303?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-11-what-happens-when-south?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-09-27-research-rundown-daily-26-september.md b/archive/2026-09-27-research-rundown-daily-26-september.md new file mode 100644 index 0000000..193c8e2 --- /dev/null +++ b/archive/2026-09-27-research-rundown-daily-26-september.md @@ -0,0 +1,159 @@ +--- +title: "Research Rundown DAILY - 26 September 2026" +date: 2026-09-27 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-26-september +subtitle: "The sharp fall in new Gulf jobs for Nepali workers shows how an external shock can reach South Asian households months before it appears in remittance statistics" +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 26 September 2026 + +*2026-09-27. Published at [https://varna.substack.com/p/research-rundown-daily-26-september](https://varna.substack.com/p/research-rundown-daily-26-september). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 26 September 2026, showing falling Gulf labour approvals for Nepali workers alongside still-high remittance inflows, illustrating the lag between weaker overseas recruitment and its eventual effect on household income.](https://substackcdn.com/image/fetch/$s_!vdPg!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa1ae2013-55bc-40cd-a4ef-5bce505281bf_1672x941.png)](https://substackcdn.com/image/fetch/$s_!vdPg!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa1ae2013-55bc-40cd-a4ef-5bce505281bf_1672x941.png) +*Research Rundown DAILY for Setember 26th 2026* + +## FIVE THINGS WORTH KNOWING + +### 1. New Gulf jobs for Nepalis have fallen by more than half + +New labour approvals for Nepalis taking jobs in Gulf countries fell **52.12 per cent** between the start of the Iran conflict and mid-September compared with the same period last year, from 228,165 to 109,249. The decline was steepest for the UAE, at 59.05 per cent, followed by Saudi Arabia at 57.02 per cent. [The Kathmandu Post](https://kathmandupost.com/national/2026/09/26/gulf-job-opportunities-are-narrowing-for-nepalis?utm_source=chatgpt.com) + +This is a labour-market shock that Nepal’s headline remittance numbers do not yet show. Remittance inflows rose 37.1 per cent in FY2025-26, and were still 21.2 per cent higher year-on-year by mid-August. Around 40 to 45 per cent of Nepal’s remittances come from the Gulf. [The Kathmandu Post](https://kathmandupost.com/national/2026/09/26/gulf-job-opportunities-are-narrowing-for-nepalis?utm_source=chatgpt.com) + +The apparent contradiction is mostly timing. Today’s remittances are sent by people already abroad, while new labour approvals measure workers entering the employment pipeline. Renewals also make departures look steadier than new recruitment actually is. + +This makes labour approvals a potentially useful **leading indicator**. If Gulf hiring remains depressed, the effects on remittances and household consumption may appear with a lag. + +### 2. Maharashtra has declared 265 of its 358 talukas drought-affected + +The Maharashtra government has declared **265 talukas across 32 districts** drought-affected after prolonged dry spells and a rainfall deficit. The affected talukas represent roughly 74 per cent of the state. Rainfall between 1 June and 18 September was 18.8 per cent below normal. [Akashvani News](https://newsonair.gov.in/maharashtra-declares-265-talukas-drought-affected/?utm_source=chatgpt.com) + +The declaration activates measures including crop-loan restructuring, land-revenue concessions, delayed recovery of agricultural loans, electricity-bill relief for agricultural pumps, relaxed employment-guarantee rules and provision for drinking water and fodder. District collectors have also been instructed to assess crop losses using photographs and GPS coordinates. [Akashvani News](https://newsonair.gov.in/maharashtra-declares-265-talukas-drought-affected/?utm_source=chatgpt.com) + +The interesting measurement issue is the relationship between rainfall and agricultural loss. An 18.8 per cent seasonal deficit does not translate mechanically into an 18.8 per cent production loss. Timing, crop stage, irrigation access, soil moisture and the length of dry spells all matter. + +That makes the forthcoming crop-loss assessments considerably more important than the statewide rainfall number. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 3. Bangladesh and Nepal are both asking the UN to postpone LDC graduation + +The UN General Assembly’s Second Committee is scheduled to consider Bangladesh’s request for a **three-year postponement of graduation from least-developed-country status on 7 October**. Nepal has also requested a three-year deferral. [The Business Standard](https://www.tbsnews.net/economy/bangladeshs-ldc-graduation-when-and-how-will-un-decide-1554346?utm_source=chatgpt.com) + +Bangladesh is due to graduate in November 2026. The government argues that additional time is needed because of external shocks and the economic adjustment required before preferential trade treatment is withdrawn. Bangladeshi officials say the LDC group, China and the G77 have expressed support in informal consultations, although the General Assembly will make the decision. [The Daily Star](https://www.thedailystar.net/business/economy/news/ldc-deferment-proposal-gaining-major-support-titumir-says-4283586?utm_source=chatgpt.com) + +Graduation is often presented as a development milestone. Economically, it is also a change in policy treatment. Bangladesh in particular must prepare for changes in tariff preferences, pharmaceutical rules and other concessions. + +The useful evaluation question is therefore counterfactual: what adjustment would three additional years permit that cannot reasonably be completed under the existing timetable? + +### 4. Sri Lanka’s services exports are growing much faster than merchandise exports + +Sri Lanka exported an estimated **US$12.01 billion** in goods and services during January to August, 4.26 per cent more than in the corresponding period of 2025. Merchandise exports increased 3.63 per cent to US$9.41 billion. Services exports rose 6.59 per cent to US$2.60 billion. [Sri Lanka Business Portal](https://www.srilankabusiness.com/news/sri-lankas-exports-surpass-us-12-bn-in-first-eight-months-of-2026.html?utm_source=chatgpt.com) + +The August figures make the divergence clearer. Services exports increased 13.97 per cent year-on-year to US$331.87 million, while merchandise exports declined 1 per cent to US$1.27 billion. [Sri Lanka Business Portal](https://www.srilankabusiness.com/news/sri-lankas-exports-surpass-us-12-bn-in-first-eight-months-of-2026.html?utm_source=chatgpt.com) + +There is a qualification. The Export Development Board describes some components as estimates, including services and parts of merchandise trade. These are therefore useful early indicators rather than final balance-of-payments data. + +Still, the composition deserves attention. Sri Lanka’s post-crisis external adjustment is usually discussed through tourism and merchandise trade. ICT/BPM, transport, construction and other traded services increasingly belong in that account. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 5. India has started rating buildings for digital connectivity + +The Telecom Regulatory Authority of India launched its **Digital Connectivity Rating Platform** on 25 September. Buildings can be assessed on fibre readiness, mobile-network availability, in-building systems and Wi-Fi infrastructure, with ratings publicly searchable by consumers. [Press Information Bureau](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2314867&lang=2®=48&utm_source=chatgpt.com) + +This is a different approach to digital infrastructure measurement. Conventional telecom statistics tell us whether a network reaches an area. They say much less about whether the signal works adequately inside the apartment, office, hospital or school where somebody tries to use it. + +The platform therefore moves the unit of measurement from network coverage towards the property. + +Its usefulness will depend on adoption and on whether ratings predict users’ actual experience. A technically compliant building with poor real-world service would expose the same gap between infrastructure and effective access that appeared in the 23 September Daily. + +[TRAI Digital Connectivity Rating announcement](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2314867&lang=2®=48&utm_source=chatgpt.com) + +# ONE PAPER WORTH READING + +### Government support and agroforestry adoption in Madhya Pradesh + +Chandra Singha and colleagues study whether government support changes smallholders’ decisions to grow commercial trees. They combine survey data from **870 farm households in 59 villages in Madhya Pradesh** with geospatial information on climate, ecology and other conditions affecting tree cultivation. Programme participation is associated with a 19 per cent increase in commercial tree-growing participation, 89 per cent more commercial trees overall and 71 per cent more trees per hectare. Farmers who knew about programmes but did not participate showed no comparable increase. [Nature](https://www.nature.com/articles/s43247-026-04098-3?utm_source=chatgpt.com) + +That last distinction is particularly useful. Programme awareness and programme exposure are different treatments. + +The causal language deserves some care. This is observational programme evaluation rather than a randomised allocation of support. Households entering government programmes may differ systematically from non-participants in motivation, land characteristics, market access or administrative connections. Statistical adjustment cannot guarantee that those differences disappear. + +Still, the paper provides useful evidence that implementation matters beyond information provision. Telling farmers that agroforestry programmes exist appears insufficient if seedlings, technical assistance, incentives or other programme components remain inaccessible. + +**Citation:** Singha, C., H. Cardenas-Diaz, Z. Luo, et al. 2026. “Government support drives agroforestry adoption and tree cultivation in smallholder farming systems.” *Communications Earth & Environment*. Published 26 September 2026. DOI: 10.1038/s43247-026-04098-3. [Nature](https://www.nature.com/articles/s43247-026-04098-3?utm_source=chatgpt.com) + +[Read the open-access paper](https://www.nature.com/articles/s43247-026-04098-3?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +# ONE NUMBER + +## 52.12 per cent + +The year-on-year decline in **new labour approvals for Nepalis going to Gulf countries** since the Iran conflict began. Approvals fell from 228,165 to 109,249. [The Kathmandu Post](https://kathmandupost.com/national/2026/09/26/gulf-job-opportunities-are-narrowing-for-nepalis?utm_source=chatgpt.com) + +This is more informative today than Nepal’s still-strong remittance growth. + +Remittances are a lagging stock-flow outcome generated largely by workers already overseas. New labour approvals tell us something about the future population of migrant workers. If the fall persists, the effect on household transfers should appear later. + +It is a good example of why the indicator that moves first is not necessarily the indicator receiving the most attention. + +# POLICY WATCH + +## Maharashtra’s drought declaration turns a climate indicator into an entitlement decision + +Declaring 265 talukas drought-affected does more than describe rainfall. It determines where households and farmers become eligible for specific forms of public assistance. + +The government has announced crop-loan restructuring, land-revenue concessions, agricultural electricity relief, drinking-water and fodder provision, and relaxed rules under the employment-guarantee programme. Financial compensation for crop losses will follow further assessment. [Akashvani News](https://newsonair.gov.in/maharashtra-declares-265-talukas-drought-affected/?utm_source=chatgpt.com) + +That makes the classification system important. + +A rainfall threshold is observable relatively quickly. Household loss is not. Irrigated and rainfed farms within the same taluka can experience very different outcomes, while two locations with similar rainfall deficits may have different soil moisture, reservoir access and crop calendars. + +The next evidence to watch is therefore the **crop-loss assessment**, including how field observations, remote sensing and administrative records are combined and how disputes over eligibility are handled. + +[Subscribe now](https://varna.substack.com/subscribe?) + +# EDITORIAL NOTE + +## Remittances can look healthy after the labour market has turned + +Nepal offers an unusually clear example of a lag problem. + +If recruitment falls today, the stock of workers already employed overseas continues sending money home. Some workers renew permits. Others draw on savings. Exchange-rate changes can alter the domestic value of transfers. Formalisation can move money from informal channels into recorded remittances. + +The headline series can therefore continue rising after the labour market generating future remittances has weakened. + +For research, this means treating migration as a sequence rather than a single variable: **job demand → recruitment → labour approval → departure → employment → earnings → remittance**. + +Different datasets observe different points in that sequence. + +A model using current remittance inflows as its only measure of migration exposure could consequently identify the shock late. Recruitment and labour-approval data may provide earlier information about what households will face six or twelve months later. + +**Reader question:** Are there good South Asian studies that combine destination-country vacancies or recruitment data with labour permits and subsequent household remittances, allowing us to estimate the lag between a migrant-labour demand shock and its effect on household income? + +That is worth returning to in Methods Monday when we look at leading, contemporaneous and lagging indicators in development data. + +## Sources + +[Kathmandu Post: Gulf job opportunities are narrowing for Nepalis, 26 September 2026](https://kathmandupost.com/national/2026/09/26/gulf-job-opportunities-are-narrowing-for-nepalis?utm_source=chatgpt.com) + +[Akashvani: Maharashtra declares 265 talukas drought-affected, 26 September 2026](https://newsonair.gov.in/maharashtra-declares-265-talukas-drought-affected/?utm_source=chatgpt.com) + +[The Business Standard: Bangladesh and Nepal seek LDC-graduation deferrals](https://www.tbsnews.net/economy/bangladeshs-ldc-graduation-when-and-how-will-un-decide-1554346?utm_source=chatgpt.com) + +[The Daily Star: Bangladesh says its LDC deferral request is gaining support](https://www.thedailystar.net/business/economy/news/ldc-deferment-proposal-gaining-major-support-titumir-says-4283586?utm_source=chatgpt.com) + +[Sri Lanka Export Development Board: January-August 2026 exports](https://www.srilankabusiness.com/news/sri-lankas-exports-surpass-us-12-bn-in-first-eight-months-of-2026.html?utm_source=chatgpt.com) + +[TRAI: Digital Connectivity Rating Platform](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2314867&lang=2®=48&utm_source=chatgpt.com) + +[Communications Earth & Environment: Madhya Pradesh agroforestry study](https://www.nature.com/articles/s43247-026-04098-3?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-26-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +### diff --git a/archive/2026-10-01-good-reads-12-did-mgnrega-narrow.md b/archive/2026-10-01-good-reads-12-did-mgnrega-narrow.md new file mode 100644 index 0000000..ed9d26e --- /dev/null +++ b/archive/2026-10-01-good-reads-12-did-mgnrega-narrow.md @@ -0,0 +1,203 @@ +--- +title: "Good Reads #12: Did MGNREGA narrow the gender wage gap?" +date: 2026-10-01 +type: other +url: https://varna.substack.com/p/good-reads-12-did-mgnrega-narrow +subtitle: "Agricultural wages rose in the post-MGNREGA period. A new study asks whether the distance between women’s and men’s wages changed with them." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #12: Did MGNREGA narrow the gender wage gap? + +*2026-10-01. Published at [https://varna.substack.com/p/good-reads-12-did-mgnrega-narrow](https://varna.substack.com/p/good-reads-12-did-mgnrega-narrow). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Good Reads #12 cover, dated 30 September 2026, showing an agricultural wage table comparing male and female wages alongside a cultivated field, farm tool and grain, illustrating the gender wage gap in Indian agriculture.](https://substackcdn.com/image/fetch/$s_!VG_a!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F669d6f20-4c25-4407-bb52-6316776df8fb_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!VG_a!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F669d6f20-4c25-4407-bb52-6316776df8fb_1536x1024.png) +*GoodReads #12 on 30th September 2026* + +There is a tempting story about MGNREGA and women’s wages. + +The programme offers rural households a statutory employment guarantee. Men and women are entitled to the same wage for the same work. In a labour market where women frequently have fewer employment options and weaker bargaining power, an alternative source of paid work should matter beyond the programme itself. + +If an agricultural employer knows that a woman can instead take MGNREGA work at a specified wage, her reservation wage may rise. Employers may have to pay more to retain workers. Public employment can therefore influence private agricultural wages even for workers who are not currently on an MGNREGA worksite. + +This is the possibility examined in a new open-access paper by Padma Lakshmi Govindarajan and K. Thomas Felix, published in *Frontiers in Sustainable Food Systems* on 29 September 2026. + +The paper is interesting because the answer it produces is less tidy than the theory. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Wages rose. The gender gap largely remained. + +The authors use Labour Bureau data from *Wage Rates in Rural India*, covering 20 states between 1995–96 and 2023–24. They examine wages across 11 agricultural occupations and compare patterns before and after the introduction of MGNREGA. + +There is a gender wage gap. + +At the aggregate level, the authors estimate that male agricultural workers earned about ₹5.01 more per day than women in the wage series they analyse. But averages hide considerable variation by task. + +Statistically significant gaps appear in ploughing and tilling, harvesting, threshing, irrigation and general agricultural labour, cane crushing and animal husbandry. The estimated gaps range from ₹3.73 in threshing to ₹7.78 in ploughing and tilling. + +In sowing, weeding, transplantation, picking and winnowing, by contrast, the estimated gender differences are not statistically significant. + +Already, this complicates the idea of a single agricultural gender wage gap. + +There are several. + +They are associated with the work being performed, how that work is gendered, and the labour markets surrounding particular agricultural tasks. + +## Then comes MGNREGA + +The authors find higher agricultural wages in the post-MGNREGA period across the activities they study. + +Their aggregate estimate is an increase of ₹15.43 in agricultural wages. Activity-specific estimates range from ₹7.50 per day for winnowing to ₹15.87 for sowing. + +But here is the result that interested me. + +The gender-by-post-MGNREGA interaction is statistically significant for only three activities: ploughing and tilling, harvesting and winnowing. For most agricultural tasks, the authors do not find statistically significant evidence that women’s wage trajectory relative to men’s changed after MGNREGA. + +At the aggregate level, the interaction is also statistically insignificant. + +In other words, the data are consistent with a period in which agricultural wages increased while gender wage inequality persisted. + +Those are two different outcomes. + +## A rising floor is not the same thing as a closing gap + +This distinction is worth dwelling on. + +Suppose men earn 100 and women earn 80. + +A policy changes labour-market conditions and men subsequently earn 120 while women earn 100. + +Women are earning more. The absolute gender difference is unchanged. + +That is not a trivial improvement for the woman earning 20 more. But it is not wage equality either. + +We often ask a single policy to carry both claims. + +Did it improve women’s economic position? + +Did it reduce gender inequality? + +The answers need not be identical. + +MGNREGA may give women access to paid work, establish an equal-wage rule within the programme and improve their bargaining position outside it. Yet private agricultural labour markets remain organised through occupational segregation, land relations, local labour demand, caste, gender norms, mobility and bargaining power. + +A public employment programme can alter the outside option without rewriting all of those institutions. + +The paper itself points towards these wider structural factors when discussing why wage differences persist across states and activities. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## But we need to be careful with the MGNREGA claim + +This is also where the methods become important. + +The paper uses an interaction-based panel regression comparing female and male wage patterns before and after MGNREGA. It resembles a difference-in-differences setup, but the authors explicitly say that it is not a conventional causal DiD design. + +There is no untreated comparison group. MGNREGA was a national programme applying to rural workers of both sexes. The interaction therefore tells us whether gender wage patterns differed in the post-MGNREGA period, not what wages would have been in a counterfactual India without MGNREGA. + +That distinction becomes even more important when the authors run a placebo test. + +They assign 2000, several years before MGNREGA, as a false intervention date. The gender interaction is statistically significant even there. + +That tells us gender wage trends were already changing before 2005. The pre-policy trajectories were not completely stable. The authors therefore caution against interpreting their post-2005 coefficients as causal effects of MGNREGA. + +I think this is one of the paper’s most useful features. + +A weaker reading would be: MGNREGA increased agricultural wages but failed to eliminate the gender wage gap. + +The evidence does not quite permit that sentence. + +What the study can establish more securely is that agricultural wages were higher in the post-MGNREGA period, gender wage differences persisted, and there is little consistent evidence in these data that the gender gap changed differently after MGNREGA. + +That is less dramatic. It is also more defensible. + +## There is another measurement problem here + +The wage data themselves deserve attention. + +The analysis constructs a long series from official Labour Bureau data covering nearly three decades. Occupational classifications changed during that period. Some activities had to be excluded because comparable female wage data were unavailable or because the series began only in 2014–15. + +The resulting dataset is also effectively a pseudo-panel. It does not follow the same agricultural workers through time. Changes can therefore reflect both changes in wages and changes in the composition of workers represented in successive observations. The authors acknowledge this explicitly. + +There are modelling difficulties too. Standard Hausman tests used to choose between fixed- and random-effects specifications did not produce reliable results for several activities. The authors ultimately use random-effects models with clustered standard errors, while setting out the assumptions this requires. + +None of these issues makes the exercise pointless. + +They change the question we should ask of it. + +This is better read as a study of long-run wage patterns than as an estimate of the treatment effect of MGNREGA. + +## The occupation matters + +For me, the most interesting result may actually be the variation between agricultural activities. + +We routinely discuss “women agricultural workers” as though this describes a homogeneous labour market. + +It does not. + +Women and men are differently distributed across tasks. Some agricultural work is culturally coded as male or female. Tasks differ in mechanisation, physical demands, skill classifications, seasonality and bargaining arrangements. + +The study finds statistically significant gender wage differences in some activities and no statistically significant difference in others. + +That should make us suspicious of a single average gender wage-gap number. + +An aggregate tells us that inequality exists. + +Disaggregation begins to tell us how it is produced. + +And that matters for policy. If wage differences are partly generated through occupational segregation, an equal-wage provision alone cannot solve the problem. The question becomes who can enter which jobs, whose work is classified as skilled, who controls land and productive assets, who can travel for employment, who has competing employment options and how wages are actually negotiated. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## There is a useful lesson here about women’s economic empowerment + +Women’s economic empowerment is sometimes treated as a sequence: + +**employment → income → bargaining power → autonomy** + +But each arrow contains an empirical question. + +Paid employment does not guarantee equal wages. + +Higher wages do not necessarily mean the gender wage gap has narrowed. + +Equal statutory wages do not guarantee equal wages throughout the surrounding labour market. + +And greater earnings do not automatically tell us who controls those earnings inside a household. + +This paper only examines one part of that chain. That is precisely why I find it useful. + +It does not need to establish that MGNREGA caused every change in rural women’s economic position. It shows us something more specific: a major public employment programme can coexist with persistent gender differentiation in the private agricultural labour market. + +## What I take from the paper + +The finding I would carry forward is not that MGNREGA “worked” or “didn’t work” for gender equality. + +The evidence here cannot sustain either verdict. + +Instead, the paper gives us a reason to separate three questions that are too often collapsed: + +Did rural wages rise? + +Did women’s wages rise? + +Did the distance between women’s and men’s wages narrow? + +Those are different tests. + +A programme may do well on the first two while producing much less movement on the third. + +That distinction matters far beyond MGNREGA. It applies whenever we use improvements in women’s absolute economic position as evidence that gender inequality itself has diminished. + +Sometimes everybody moves forward while the distance between them remains. + +That distance is worth measuring too. + +--- + +**Source:** Padma Lakshmi Govindarajan and K. Thomas Felix, “Exploring the gender wage gap in Indian agriculture: an interaction-based panel regression approach,” *Frontiers in Sustainable Food Systems*, 29 September 2026, 10:1872202. DOI: 10.3389/fsufs.2026.1872202. + +[Read the open-access paper](https://www.frontiersin.org/journals/sustainable-food-systems/articles/10.3389/fsufs.2026.1872202/full?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-12-did-mgnrega-narrow?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-10-01-methods-monday-28-september-2026.md b/archive/2026-10-01-methods-monday-28-september-2026.md new file mode 100644 index 0000000..add8c6a --- /dev/null +++ b/archive/2026-10-01-methods-monday-28-september-2026.md @@ -0,0 +1,135 @@ +--- +title: "Methods Monday | 28 September 2026" +date: 2026-10-01 +type: other +url: https://varna.substack.com/p/methods-monday-28-september-2026 +subtitle: "Interviewers, recall, probability scales and sampling designs, and the small choices that change what surveys measure." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Methods Monday | 28 September 2026 + +*2026-10-01. Published at [https://varna.substack.com/p/methods-monday-28-september-2026](https://varna.substack.com/p/methods-monday-28-september-2026). This file is a generated copy; the Substack post is the original.* + +[![Editorial illustration for Methods Monday showing a household questionnaire with Hindi translation notes, interviewer probes, a probability scale, sampling-frame sheet, field notebook, map and magnifying glass.](https://substackcdn.com/image/fetch/$s_!scDM!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F78a82638-8080-4c78-95d5-84db8abafc51_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!scDM!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F78a82638-8080-4c78-95d5-84db8abafc51_1536x1024.png) +*Methods Monday on 28th September, 2026* + +### What interviewers change, what questionnaires create, and when prediction should enter sampling + +Last Monday’s voice-response paper asked what changes when respondents speak rather than type. Two papers published on 25 September take the next step: once speech becomes an interviewer-administered exchange, the interviewer and the question both enter the measurement process. + +## 1. Some questions make interviewers depart from the script + +Rodney Terry, Erica Yu and Holly Fee use wave 1 of the 2014 US Survey of Income and Program Participation to examine when interviewers rephrase questions or probe respondents. Their data combine computer audio-recorded interviews, interviewer observations and responses collected during the household interview. + +Question characteristics were the strongest predictors of interviewer behaviour. Information interviewers acquired during the visit, including observations made at the doorstep and answers given earlier in the interview, contributed somewhat to later deviations, although the evidence for those effects was limited. + +This shifts part of the quality-control problem back to instrument design. If the same item repeatedly produces clarification, probing or rewording, another round of interviewer instruction may leave the source of the problem untouched. The item itself may be difficult to administer as written. + +That matters even more in multilingual South Asian fieldwork. An interviewer who has to translate an awkward phrase, explain an unfamiliar response category or decide what a concept means in a local language is making measurement decisions in real time. The US evidence does not tell us how large those effects are in multilingual CAPI surveys. It gives a good reason to measure them. + +Terry, Rodney L., Erica C. Yu, and Holly Fee. 2026. “What Informs Interviewer Question Asking and Probing? Evidence from the Survey of Income and Program Participation.” *Journal of Survey Statistics and Methodology*, smag012. DOI: 10.1093/jssam/smag012. Published 25 September 2026. + +[Read the paper and supplementary material](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag012/8836145?utm_source=chatgpt.com) + +I would read this first if you run interviewer-administered surveys. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 2. Recorded interviews can be turned into usable survey paradata at scale + +Hanyu Sun, Brad Edwards and Ting Yan use computer-assisted recorded interviewing and machine-learning tools to compare interviewer-respondent interactions across telephone, face-to-face and video interviews in a large US household survey. The system transcribes question-level recordings and derives measures of interviewer behaviour and respondent difficulty. + +They examine whether interviewers read questions verbatim, whether they preserve the intended meaning, how many speaking turns respondents need, and how long respondent speech lasts. Differences across modes exist, but they are generally small. The larger source of variation is the question being asked. + +That is the useful finding. Mode gets enormous attention in mixed-mode survey design. A badly functioning question can create more variation in interview interaction than whether the interview happens by video, telephone or in person. + +The paper also demonstrates a practical quality-control method. Audio monitoring is usually expensive because somebody has to listen and code. Automated transcription and classification can make question-level review possible across far more interviews. I would still want human validation before using automated coding for interviewer assessment, especially across accents and languages that speech-recognition systems handle unevenly. + +Sun, Hanyu, Brad Edwards, and Ting Yan. 2026. “Using Recordings of Survey Interviews and Machine Learning to Evaluate Differences in Interviewer–Respondent Interactions Between Telephone, In-Person, and Video Interviews.” *Journal of Survey Statistics and Methodology*, smag026. DOI: 10.1093/jssam/smag026. Published 25 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag026/8836143?utm_source=chatgpt.com) + +Worth reading alongside Terry et al. The two papers make a fairly persuasive case for treating interview recordings as instrument-diagnostic data rather than material used only to police interviewers. + +## 3. Auxiliary predictions can be used before sampling begins + +P. García-Segador and P. Miranda ask whether predicted values of the outcome can be used to construct a better fixed-size probability sample. + +The ideal sampling design would depend on the true values of the study variable, which are unavailable before data collection. Their method substitutes predictions generated from auxiliary information, including predictions from statistical or machine-learning models, and builds the sampling design around those values. + +In simulations, designs based on imperfect predictions remained close to designs constructed using the true outcome values. They also produced lower estimation error than comparison approaches including cube sampling and conditional Poisson sampling. + +The development-research application is clearest where sampling frames are already rich. Administrative registers, census variables, previous survey waves, satellite measures or programme data may predict the target outcome well enough to influence inclusion probabilities. + +The risk sits in the predictive model. Relationships learned from old data may change. Administrative variables can contain their own selection and measurement errors. A complex prediction-based sample also requires unusually careful documentation if analysts are expected to reconstruct inclusion probabilities later. + +García-Segador, P., and P. Miranda. 2026. “Conditional Optimal Sampling: Enhancing Unequal-Probability Designs with Auxiliary Information.” *Journal of Survey Statistics and Methodology*, smag027. DOI: 10.1093/jssam/smag027. Published 23 September 2026. + +[Read the paper and supplementary material](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag027/8830419?utm_source=chatgpt.com) + +Worth reading for survey statisticians and teams working with administrative or geospatial sampling frames. I would want field applications before using it as a routine design choice. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 4. In urban India, measured harassment prevalence depends on how the questionnaire asks respondents to remember + +Lori Beaman, Seema Jayachandran and Anisha Sharma surveyed more than 4,000 women recruited in public spaces in New Delhi and Bengaluru. Their substantive study examines harassment and women’s use of public space. The measurement result deserves separate attention here. + +Reported past-year harassment prevalence was 65 per cent in New Delhi and 29 per cent in Bengaluru. For groping, stalking or sexual assault, the corresponding figures were 41 and 14 per cent. The authors then show that measured prevalence is more sensitive to questionnaire design where harassment is less common and where respondents have fewer or less recent incidents available for recall. + +That complicates cross-place comparisons. Retrieval from memory is part of the measurement process. Where an experience is frequent and recent, alternative prompts may produce similar answers. Where it is uncommon or distant, wording and recall structure can move the estimate more. + +The result applies well beyond harassment surveys. Researchers measuring discrimination, illness episodes, shocks, violence, informal payments or service failures face the same recall problem. A prevalence estimate may partly reflect how hard the questionnaire makes respondents search their memories. + +The paper is a substantive development-economics study, so the measurement experiment is narrower than a dedicated survey-methods study. That is also what makes it useful. These problems turn up inside ordinary empirical research. + +Beaman, Lori A., Seema Jayachandran, and Anisha Sharma. 2026. “The City Through Her Eyes: Gender-Based Harassment in Public Spaces in Urban India.” NBER Working Paper 35738. DOI: 10.3386/w35738. September 2026. The NBER version was posted on 15 September and revised on 21 September. + +[Read the NBER paper](https://www.nber.org/papers/w35738?utm_source=chatgpt.com) + +This is the regional paper I would keep. Its measurement result has uses far beyond the subject of the study. + +## 5. A respondent choosing 0 or 100 per cent may be telling us about the questionnaire + +Carola Binder, Laura Crespo, Carlos Gento, Luis M. Guirola and Ernesto Villanueva examine probability questions used to measure subjective expectations. Respondents frequently put all probability on one outcome, which researchers often interpret as certainty. + +Using the online ECB Consumer Expectations Survey and the interviewer-administered Spanish Survey of Household Finances, the authors find that this bunching is partly produced by the survey process. It occurs more frequently among respondents with lower financial literacy and increases with panel tenure. + +Two changes provide stronger evidence. Standardising interviewer behaviour reduced bunching by about 15 percentage points. A later questionnaire redesign reduced it by another 18 percentage points. Variation across interviewers also fell following standardisation. + +The development-research relevance is easy to see. Probability elicitation is increasingly used for expectations about rainfall, income, migration, employment, crop yields and shocks. Responses of 0 and 100 are tempting to interpret substantively. Some may instead reflect numeracy, task burden or the way an interviewer explains the scale. + +The exact effect sizes should stay in their original setting. These are European surveys. The measurement issue could be larger where percentage probabilities are unfamiliar. + +Binder, Carola, Laura Crespo, Carlos Gento, Luis M. Guirola, and Ernesto Villanueva. 2026. “(Mis)measuring Uncertainty: Bunching in Probabilistic Expectations.” NBER Working Paper 35733. DOI: 10.3386/w35733. September 2026. + +[Read the NBER paper](https://www.nber.org/papers/w35733?utm_source=chatgpt.com) + +Worth reading if your questionnaire contains expectations or probability modules. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 6. A large lottery did little to recover disengaged panel respondents + +Rasmus Patton, Jannes Jacobsen, Jörg Dollmann and Almuth Lietz test whether a high-value lottery can bring respondents back into a quarterly probability-based online panel. + +Participants were randomly assigned to the standard €10 incentive, the €10 payment plus a lottery offering ten €500 Amazon vouchers, or a conditional version in which continued participation increased the probability of winning. Neither additional treatment produced a statistically significant increase in participation. Subgroup analyses by earlier participation, education, income and country of origin produced no systematic effects either. + +This is a useful null result. Attrition strategies often assume that respondents who have stopped participating require a stronger financial inducement. The experiment suggests that disengagement may have causes that money alone does not address. + +Its relevance to South Asian longitudinal surveys is limited by context. A German online panel already paying €10 is far removed from household panels where attrition comes from migration, phone-number turnover, interviewer continuity, seasonal work or difficulty relocating households. + +Patton, Rasmus, Jannes Jacobsen, Jörg Dollmann, and Almuth Lietz. 2026. “Motivating The Unmotivated? Testing Conditional Lottery Incentives in an Ongoing Panel Survey.” *Journal of Survey Statistics and Methodology*, smag035. DOI: 10.1093/jssam/smag035. Published 25 September 2026. + +[Read the paper](https://academic.oup.com/jssam/advance-article-abstract/doi/10.1093/jssam/smag035/8836144?utm_source=chatgpt.com) + +Worth reading for panel managers. The null finding is more useful than another small positive incentive effect would have been. + +## What I would read + +Terry, Yu and Fee first. Their result gives field teams something concrete to test: whether interviewer deviations concentrate in particular questions. + +The India paper by Beaman, Jayachandran and Sharma is the measurement result I would carry into substantive development research. Recall is part of the estimator long before anyone runs + +[Share](https://varna.substack.com/p/methods-monday-28-september-2026?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-10-01-research-rundown-daily-29-september.md b/archive/2026-10-01-research-rundown-daily-29-september.md new file mode 100644 index 0000000..d5593dd --- /dev/null +++ b/archive/2026-10-01-research-rundown-daily-29-september.md @@ -0,0 +1,143 @@ +--- +title: "Research Rundown DAILY 29 September 2026" +date: 2026-10-01 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-29-september +subtitle: "Floods test recovery systems, Pakistan offers new evidence on life after cash transfers, and food-loss data expose what national averages conceal." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY 29 September 2026 + +*2026-10-01. Published at [https://varna.substack.com/p/research-rundown-daily-29-september](https://varna.substack.com/p/research-rundown-daily-29-september). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 29 September 2026, showing a rainfall chart comparing 66.5 mm with the 7.6 mm norm, a map of northern India and Nepal marked by floodwater, and a household survey sheet tracking work loss, schooling, migration, debt and recovery support.](https://substackcdn.com/image/fetch/$s_!mVZC!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6e268ae0-c18a-4d01-8570-6718040a9f6b_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!mVZC!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6e268ae0-c18a-4d01-8570-6718040a9f6b_1536x1024.png) +*Research Rundown Daily for 29th September, 2026* + +Floods expose the distance between climate risk and preparedness, a Pakistan study asks what happens when women lose cash transfers, and new work puts numbers on food lost before it reaches the market. + +## FIVE THINGS WORTH KNOWING + +### 1. Northern India and Nepal are dealing with another severe flood shock + +Uttar Pradesh received 66.5 mm of rain between 25 and 27 September, against a normal 7.6 mm for those dates. By 29 September, 56 deaths had been reported in India. Nepal reported 14 deaths and 11 people missing, while authorities warned of possible flash floods in 49 of its 77 districts. + +For Nepal, the timing matters. These floods arrived only a month after the late-August disaster that killed at least 1,450 people. Disaster risk is therefore becoming a question of repeated shocks and recovery capacity, rather than isolated emergencies. + +There is a measurement problem here too. Rainfall, mortality and physical damage are reported quickly. Lost work, debt, disrupted schooling, migration and the distribution of recovery spending emerge much later, if they are measured at all. Those are the outcomes that will tell us whether repeated climate shocks are changing household trajectories. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 2. A new Pakistan paper examines what happens when cash transfers suddenly stop + +A new World Bank Policy Research Working Paper by Nasir Iqbal, Amen Jalal, Mahreen Mahmud and Kate Vyborny studies an unusually neglected part of social protection: exit. + +Female-targeted cash transfers are commonly assessed by comparing recipients with non-recipients. This paper instead asks what happens when women unexpectedly cease to qualify. The authors use a regression discontinuity-in-differences design around a change in programme eligibility in Pakistan. + +That distinction matters. A transfer can alter bargaining arrangements inside a household while it is being paid. Whether those changes survive after the money disappears is a separate empirical question. Programme evaluations that stop at the final payment can therefore miss part of the effect. + +More on the paper below. + +### 3. India may be measuring post-harvest loss at the wrong level of aggregation + +A new WRI India working paper examines soyabean, custard apple and onion supply chains in Madhya Pradesh and Maharashtra. Its useful contribution is fairly practical: losses occur at different points for different crops. + +For soyabean, harvesting, threshing and collection emerge as critical points. For onion, farm-level storage dominates. Custard apple losses vary according to whether production comes from forests and common land, farm boundaries or orchards. + +This makes a single category called “post-harvest loss” analytically weak for programme design. An intervention aimed at storage may make sense for onions and miss much of the problem elsewhere. The study uses mixed methods and maps actors and loss points across the selected supply chains. Its geographic and crop coverage is deliberately narrow, so the estimates should not be read as national loss rates. + +### 4. South Asia’s innovation rankings improved, though rankings need careful reading + +WIPO’s 2026 Global Innovation Index places India 38th globally. Pakistan rises four places to 95th, Nepal reaches 104th and Bangladesh 105th. Bhutan enters the index at 74th. India remains the highest-ranked economy in Central and Southern Asia. + +The more interesting finding sits below the league table. India ranks 49th on innovation inputs and 25th on outputs. WIPO classifies India as an innovation “overperformer” relative to its income level for the sixteenth consecutive year. Pakistan also returns to that group, with particular strengths in mobile-app creation and ICT-service exports. + +Composite indices remain sensitive to indicator choice, missing data, normalisation and weighting. Movement of a few ranks should therefore carry less weight than changes in the underlying indicators. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 5. A sovereign rating may tell us less about fiscal fundamentals than its precision suggests + +A paper by Olivier Blanchard, Daniel Leigh and Prachi Mishra examines how sovereign credit ratings relate to debt and projected primary balances. + +Their model-based comparison finds three discrepancies. Ratings place considerably more weight on debt than on forecast primary balances, give too little weight to the difference between interest and growth rates, and contain large country effects. Economies with similar debt and fiscal trajectories can consequently receive very different ratings. + +For developing countries, this is more than an argument about rating methodology. Sovereign ratings affect borrowing costs and perceptions of fiscal space. If persistent country effects partly reflect factors outside the basic debt arithmetic, the question becomes what those effects are measuring. + +The paper is an IMF working paper. It should be read as research in progress rather than an institutional IMF position. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## ONE PAPER WORTH READING + +### What Happens When Women Suddenly Stop Receiving Cash Transfers? + +Nasir Iqbal, Amen Jalal, Mahreen Mahmud and Kate Vyborny, World Bank Policy Research Working Paper 11474, 29 September 2026. + +Most cash-transfer research asks what happens when money arrives. This paper asks the reverse question. + +The setting is Pakistan. The authors exploit a revised eligibility rule that caused some women to lose transfers unexpectedly and use a regression discontinuity-in-differences design around the eligibility threshold. That gives the paper a stronger identification strategy than a simple comparison of former recipients with continuing recipients, since those groups could differ systematically for many other reasons. + +The substantive question is unusually useful: does the position women acquire while bringing programme income into the household persist when that income disappears, and is sudden programme exit associated with backlash? + +The design still identifies effects around an eligibility threshold. External validity therefore needs care. Women farther from that threshold, households leaving programmes for other reasons and planned graduation may behave differently. + +The larger contribution is conceptual. Social-protection evaluations routinely devote considerable attention to targeting and programme entry. Exit receives much less scrutiny. If transfers alter bargaining power, labour decisions or household expectations, the end of a programme is itself an intervention. It should be measured as one. + +## ONE NUMBER + +### 66.5 mm + +Rain recorded in Uttar Pradesh between 25 and 27 September. + +The normal rainfall for the same period is 7.6 mm. In other words, the state received almost nine times the normal amount for those dates. + +A rainfall anomaly does not by itself establish the contribution of anthropogenic climate change to a particular flood. Attribution requires a separate analysis. It does tell us the scale of the immediate hydrological shock confronting infrastructure, agriculture and local disaster-response systems. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## POLICY WATCH + +### Food loss policy needs crop-specific evidence + +India has spent years discussing cold chains, storage and food loss as broad infrastructure problems. The new WRI India study offers a useful corrective. + +Different crops are being lost at different stages. For onion, storage is central. For soyabean, losses arise earlier in harvesting, threshing and collection. Custard apple presents another structure again. + +That changes the policy question. Aggregate estimates can establish the size of a problem. They are a poor guide to where public money should be spent unless they identify where loss occurs, who bears it and why. + +The next useful evidence would connect physical quantities lost at each stage to household income, prices, nutrition and emissions. Without that link, tonnes lost remain an incomplete welfare measure. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## EDITORIAL NOTE + +There is a common measurement problem running through today’s material. + +Flood losses are easiest to count when they involve rainfall, roads and deaths. Food loss becomes one national percentage even when crops disappear at different points in the supply chain. Innovation becomes a rank. Social protection success becomes the effect observed while payments continue. + +Aggregation is necessary. It also removes information. + +For research intended to guide policy, the relevant question is often one level below the headline indicator: where did the loss occur, who experienced it, when did the effect appear, and did it persist? + +That is where several of today’s stories become more interesting. + +### A question for readers + +If a cash transfer changes women’s bargaining position while payments continue, how long should an evaluation follow households after programme exit before we can reasonably say whether that change persisted? I would be particularly interested in comparable evidence from South Asia on programme graduation, discontinuation or payment shocks. + +This is also a measurement question worth returning to in a future Methods Monday: how should evaluations define the treatment period when an intervention may continue to affect behaviour after the intervention itself ends? + +## SOURCES + +World Bank, Iqbal, Jalal, Mahmud and Vyborny, *What Happens When Women Suddenly Stop Receiving Cash Transfers?*, Policy Research Working Paper 11474, 29 September 2026. [Paper and reproducibility information](https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099315009292634527?utm_source=chatgpt.com) + +Reuters, *Fatal floods hit India, Nepal and Thailand*, 29 September 2026. [Reuters report](https://www.reuters.com/sustainability/fatal-floods-hit-india-nepal-thailand-2026-09-29/?utm_source=chatgpt.com) + +WRI India, Sharma, Lamba, Bhatt and Singh, *Counting the Uncounted for People and Climate*, 29 September 2026. [Working paper](https://wri-india.org/research/counting-uncounted-people-and-climate-post-harvest-losses-soyabean-custard-apple-and-onion?utm_source=chatgpt.com) + +WIPO, *Global Innovation Index 2026*. [2026 results](https://www.wipo.int/web-publications-preview/global-innovation-index-2026/en/gii-2026-results.html?utm_source=chatgpt.com) + +Blanchard, Leigh and Mishra, *Ratings, Debt, and Deficits: An Exploration*, IMF Working Paper 2026/195. [IMF working paper](https://www.imf.org/en/publications/wp/issues/2026/09/18/ratings-debt-and-deficits-an-exploration-579729?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-daily-29-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-10-01-research-rundown-friday-findings-632.md b/archive/2026-10-01-research-rundown-friday-findings-632.md new file mode 100644 index 0000000..d41d79f --- /dev/null +++ b/archive/2026-10-01-research-rundown-friday-findings-632.md @@ -0,0 +1,145 @@ +--- +title: "Research Rundown: Friday Findings -1 October 2026" +date: 2026-10-01 +type: edition +url: https://varna.substack.com/p/research-rundown-friday-findings-632 +subtitle: "Coal, drought and pregnancy, tribal representation, programme data, and the thin evidence behind urban climate-health action." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown: Friday Findings -1 October 2026 + +*2026-10-01. Published at [https://varna.substack.com/p/research-rundown-friday-findings-632](https://varna.substack.com/p/research-rundown-friday-findings-632). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Friday Findings cover for 1 October 2026, showing a research desk with charts on changing sources of urban air pollution, drought and pregnancy outcomes, tribal representation and forest conservation, urban climate-health evidence, and a programme data-to-decision flow.](https://substackcdn.com/image/fetch/$s_!_1X-!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2a630746-318c-489e-b45c-209c6fb9d864_1672x941.png)](https://substackcdn.com/image/fetch/$s_!_1X-!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2a630746-318c-489e-b45c-209c6fb9d864_1672x941.png) +*Friday Findings for 1st October 2026* + +### programming note + +Friday Findings is arriving a day early this week. Friday, 2 October, is reserved for a Research Rundown Gandhi Jayanti special. The research window here remains the usual seven days, 25 September to 1 October. + +### coal, drought and pregnancy, tribal representation, programme data, and what we actually know about climate action in cities + +This is a compact issue. Two new peer-reviewed papers on India deserve close attention. One links electricity demand and coal generation to the changing composition of urban air pollution. The other links district-level drought exposure to preterm birth and low birth weight, although a serious survey-design problem tempers the result. A Chhattisgarh study asks whether the level at which Scheduled Tribe political representation occurs affects forest conservation. Two further pieces are worth keeping for methods and implementation. + +My bias this week is toward work where the design teaches us something beyond the headline result. + +## 1. India’s economic boom changed the composition of its air-pollution problem + +Swain, Basudev, Aishwarya Singh, Debashis Tripathy, Rui Song, Adrien Deroubaix, Marco Vountas, Pritanjali Shende, Chenchen Huang, Luca Lelli, and Ankit Tandon. 2026. “Economic-boom-driven electricity demand linked to urban air pollution and health in India.” *Nature Cities*. Published 25 September 2026. DOI: 10.1038/s44284-026-00518-9. + +This is my pick of the week. + +Between 2000 and 2020, India’s GDP per capita roughly tripled while electricity demand rose almost fourfold. Coal supplied about 76% of generation. Over the same period, the authors estimate that urban aerosol loading doubled and sulfate came to account for about 46% of that burden. Estimated urban PM2.5-attributable deaths rose from roughly 0.7 million to 0.9 million. Household air-pollution mortality, meanwhile, fell as clean-cooking access expanded. + +The paper earns attention because it separates sources rather than treating PM2.5 as a single exposure. The authors combine MERRA-2 reanalysis, MODIS MAIAC satellite observations, GEOS-Chem atmospheric modelling, TROPOMI SO2 observations and Central Pollution Control Board measurements. GEOS-Chem attributes a growing share of urban aerosol loading to the power sector, whose estimated contribution to total urban aerosol optical depth rises from around 16% in 2000 to 36% in 2020. + +The causal language needs discipline. Economic growth, coal generation, urbanisation and pollution changed together over two decades. Atmospheric source-attribution models can identify plausible sectoral contributions with considerable sophistication. They do not turn the historical association between GDP growth and mortality into a clean causal estimate of the health cost of growth. + +The mortality numbers also come from comparative risk assessment using Global Burden of Disease estimates. They are modelled population burdens, with uncertainty in exposure, risk functions and baseline mortality. + +The result I would keep is narrower and more useful. India’s move away from household solid-fuel exposure has coincided with an increasingly electricity-linked urban pollution burden. Public-health policy therefore has to follow the energy system as exposure shifts outdoors. + +[Read the open-access paper](https://doi.org/10.1038/s44284-026-00518-9?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 2. Drought exposure is associated with preterm birth and low birth weight. The design deserves scrutiny. + +Bhattacharjee, Sujayita, Arup Jana, and Aparajita Chattopadhyay. 2026. “Drought assessment and linkages with selected pregnancy outcomes in India using satellite-based aridity analysis and large-scale data, 2019–2021.” *Discover Public Health* 23: 1524. Published 25 September 2026. DOI: 10.1186/s12982-026-02694-4. + +This paper links MODIS-derived aridity measures with NFHS-5 data. The analyses cover 148,746 institutional births for preterm birth, 232,920 children with recorded birth weight, and pregnancy histories from 724,115 women. Around 43.7% of the study population was classified as drought-exposed. + +After adjustment, drought exposure was associated with 12% higher odds of preterm birth and 8% higher odds of low birth weight. The adjusted miscarriage association was effectively null. + +The sample size is impressive. The inference is weaker than the N suggests. + +The authors explicitly state that their analyses do not incorporate NFHS sampling weights, primary-sampling-unit clustering or stratification. For a complex national household survey, that is a substantial limitation. + +Exposure measurement adds another problem. District-level drought conditions are attached to pregnancy outcomes without precise gestational timing. A woman classified as drought-exposed may therefore have experienced a different exposure window from the one biologically relevant to the outcome. + +I would read the paper as evidence of a spatial association worth pursuing. A stronger design would use more precise temporal exposure assignment and respect the NFHS survey design. + +This is an obvious Methods Monday candidate. Large N does very little to rescue a mismatch between sampling design, exposure construction and the inferential claim. + +[Read the open-access paper](https://doi.org/10.1186/s12982-026-02694-4?utm_source=chatgpt.com) + +## 3. Tribal political representation appears to conserve forests when the office carries enough authority + +Agarwal, Bina, Shamindra Nath Roy, and Shiva Chakravarti Sharma. 2026. “Forest conservation and indigenous political reservation: What level of representation is effective?” *Ideas for India*, 29 September 2026. The article reports findings from the authors’ 2026 *World Development* study of Chhattisgarh. + +This is a new public-facing analysis of research published earlier this year, so I am treating it as substantive new analysis rather than a newly published underlying paper. + +The study uses satellite-derived forest-cover data and GIS mapping across roughly 20,000 Chhattisgarh villages from 2001 to 2019. It separates villages into seven categories according to whether Scheduled Tribe representation applied at assembly-constituency level, through PESA village councils, through both, or through neither. The authors then estimate panel, OLS and logistic regressions with controls for population change, cultivation dependence, roads, proximity to towns, nearby forests, poverty and baseline forest cover. + +Villages that experienced some form of ST reservation gained an average 22.71 hectares of forest cover over 2001–2019, compared with 5.38 hectares among never-reserved villages. The disaggregated result matters more: villages with sustained assembly-level reservation performed better, while villages covered solely by PESA did poorly. + +The authors interpret this through institutional authority. MLAs control discretionary funds and have greater access to the bureaucracy. ST sarpanches operate with fewer resources, mixed local electorates and, during much of the study period, weak implementation of PESA’s forest powers. + +This is much more interesting than a simple “representation improves conservation” claim. The study suggests that the administrative level at which representation occurs may determine whether political voice can alter an environmental outcome. + +Selection into reserved areas and the policy changes surrounding the 2008 delimitation still complicate causal interpretation. The authors deal with these institutional changes explicitly and test several specifications, but the design remains observational. + +For work on decentralisation, tribal governance or community natural-resource management, this one deserves a full read. + +[Read the open-access Ideas for India analysis](https://www.ideasforindia.in/topics/environment/forest-conservation-and-indigenous-political-reservation-what-level-of-representation-is-effective?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 4. Large livelihood programmes collect enormous amounts of data. The harder question is whether anybody uses it. + +Das, Kumar, and Neha Dhingra. 2026. “Unlocking the power of administrative data to strengthen programme delivery.” *Ideas for India*, 28 September 2026. + +This is an implementation piece rather than an evaluation, but I am including it because the problem will be familiar to anyone who has worked with a large government MEL system. + +Das and Dhingra use Samaveshi Aajeevika Yojana, the Ministry of Rural Development’s economic-inclusion programme for ultra-poor and marginalised rural households, to examine what happens to administrative data as a programme scales. SAY generates data through participant identification, household profiling, livelihood plans, coaching visits, financial transactions and convergence with government schemes. + +Their argument is that the binding constraint increasingly lies in governance around data use: who interprets an emerging pattern, who has authority to act on it, and whether information travels back to frontline staff. The authors also warn that linking incentives too tightly to reported indicators can shift staff behaviour towards target achievement. + +There is no impact estimate here. The article does not show that a particular administrative-data system improves programme outcomes. It gives a management diagnosis and a set of design propositions. + +I would still keep it. Development programmes spend heavily on MIS platforms and dashboards while giving much less attention to the decision rule attached to each indicator. A dashboard that generates no specified action is mostly a display. + +This has a direct connection to our recurring Methods Monday discussion about measurement systems. The next useful step would be empirical: identify cases where changing the feedback architecture of programme data altered implementation quality or participant outcomes. + +[Read the open-access analysis](https://www.ideasforindia.in/topics/governance/unlocking-the-power-of-administrative-data-to-strengthen-programme-delivery) + +## 5. The evidence for city climate-health action is much thinner than the rhetoric + +Hutchinson, Emma, Tobias Magnus Schuster, Melanie Crane, James Milner, Kristine Belesova, Roberto Picetti, Simon Lloyd, Catalina Turcu, Jane Falconer, Blanca Anton, et al. 2026. “Systematic overview of climate actions for health in cities.” *Nature Cities*. Published 29 September 2026. DOI: 10.1038/s44284-026-00515-y. + +This review is global. I am making room for it because Indian cities are rapidly adopting climate, clean-air, mobility and heat interventions, while evidence of health effect often comes from a much narrower set of settings. + +The authors screened 5,550 records, critically appraised 60 systematic reviews and retained 49 whose findings met their confidence threshold. Most evidence concerned transport, urban form and buildings. Fifty reviews drew on studies from high- or middle-income countries; only ten contained information from low-income settings. + +The quality assessment is sobering. Only six of the 60 reviews received medium or high confidence ratings. Seven reviews, about 12%, meta-analysed effects. Evidence on food and waste interventions was particularly sparse, and only seven of the 49 included reviews examined implementation barriers or enablers. + +The literature is strongest around active transport and housing improvements. Evidence on city-wide, cross-sector programmes capable of producing large climate and health gains is absent from the published review base the authors examined. + +That does not establish that integrated urban climate programmes fail. It tells us that the evaluation literature has not caught up with the claims being made for them. + +This connects directly with the WRI India road-dust and open-waste-burning papers in the 18 September Friday Findings. Those studies were useful precisely because they measured specific sources and interventions. This review explains why that apparently mundane work matters. + +[Read the open-access paper](https://doi.org/10.1038/s44284-026-00515-y?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## What I would keep + +Swain et al. is the week’s strongest substantive paper. The source-attribution work gives a more useful account of India’s changing pollution burden than another national PM2.5 trend. + +Bhattacharjee et al. is the paper I would use in a methods classroom. The very large NFHS sample makes the unmodelled survey design easier to miss, which is precisely why it is instructive. + +The Agarwal, Roy and Sharma study raises the most interesting political-economy question: whether representation changes outcomes only when the represented office has enough fiscal and administrative authority to act. + +The urban climate-health review supplies the corrective. We have many plausible mechanisms and policy proposals. Evidence from implemented, cross-sector city programmes remains sparse. + +I found several new nutrition and public-health papers during the search that I have left out. One Indian child-obesity study, for example, drew its sample from 163 children in a single affluent private school. It is too narrow for Friday Findings. I also found current Bangladesh and Pakistan commentary on macroeconomics and climate finance, but little new empirical work within the seven-day window that justified adding items simply for geographic balance. + +## For readers + +I am looking for South Asian evaluations in which routine programme or administrative data were actually used to trigger a management change, with evidence on whether that change altered delivery or participant outcomes. If you know one, send it across. Reader recommendations that make a future Friday Findings will be credited. + +[Share](https://varna.substack.com/p/research-rundown-friday-findings-632?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +## diff --git a/archive/2026-10-02-research-rundown-daily-27-september.md b/archive/2026-10-02-research-rundown-daily-27-september.md new file mode 100644 index 0000000..4fbc0a0 --- /dev/null +++ b/archive/2026-10-02-research-rundown-daily-27-september.md @@ -0,0 +1,155 @@ +--- +title: "Research Rundown DAILY | 27 September 2026" +date: 2026-10-02 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-27-september +subtitle: "Housing finance in Bangladesh, child development in Pakistan, clean cooking in India and the growing weight of debt servicing. " +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY | 27 September 2026 + +*2026-10-02. Published at [https://varna.substack.com/p/research-rundown-daily-27-september](https://varna.substack.com/p/research-rundown-daily-27-september). This file is a generated copy; the Substack post is the original.* + +[![Housing finance in Bangladesh, child development in Pakistan, clean cooking in India and the growing weight of debt servicing. ](https://substackcdn.com/image/fetch/$s_!jWcK!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F0ed2ed34-fe04-4275-b3bf-bdb0f3503194_1672x941.png)](https://substackcdn.com/image/fetch/$s_!jWcK!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F0ed2ed34-fe04-4275-b3bf-bdb0f3503194_1672x941.png) +*Research Rundown 27th September 2026* + +*Today’s evidence asks a familiar question: what changes when policy reaches households that conventional systems have left out?* + +## Five things worth knowing + +### 1. Bangladesh is trying housing finance without the mortgage + +The Asian Development Bank approved a **$100 million concessional loan** for affordable and climate-resilient housing in Bangladesh. The interesting part is the financing model. + +The money will flow through Palli Karma-Sahayak Foundation and eligible microfinance institutions, providing long-term, collateral-free housing loans to low- and middle-income households. Conventional mortgage markets often exclude households with informal or irregular incomes, even when those households can service smaller loans. [Asian Development Bank](https://www.adb.org/news/adb-approves-100-million-loan-expand-access-affordable-resilient-housing-finance-bangladesh) + +ADB expects the programme to reach at least **40,000 households by 2031**, including at least 32,000 women and 24,000 borrowers living in areas exposed to natural hazards. Forty million dollars of the financing is classified as climate adaptation finance. [Asian Development Bank](https://www.adb.org/news/adb-approves-100-million-loan-expand-access-affordable-resilient-housing-finance-bangladesh) + +There are at least three interventions bundled together here: access to credit, women’s housing security and climate-resilient construction. That makes the eventual evaluation interesting. Loan disbursement will be the easiest outcome to count. Whether households gain secure tenure, safer housing and greater resilience after a shock will take longer to establish. + +[ADB: Inclusive Affordable Housing Finance Project](https://www.adb.org/news/adb-approves-100-million-loan-expand-access-affordable-resilient-housing-finance-bangladesh?utm_source=chatgpt.com) + +### 2. Bangladesh is also paying out considerably more foreign debt than it is receiving + +There is another financing story in Bangladesh today. + +Economic Relations Division data reported by *The Business Standard* show that Bangladesh received **$294.56 million** in foreign loans and grants during July and August of FY2026–27, while paying **$698.92 million** in foreign debt service. + +Of that repayment, $516.04 million was principal and $182.88 million interest. Foreign assistance disbursements during the same two months of the previous financial year were $750 million. [The Business Standard](https://www.tbsnews.net/economy/bangladesh-pays-699m-foreign-debt-receives-295m-july-august-1554971) + +Two months do not establish an annual trend, and disbursement schedules can be lumpy. But the comparison is still useful. Gross development-finance commitments tell us little about the net resource flow unless repayments are placed beside them. + +[The Business Standard: Bangladesh foreign debt flows](https://www.tbsnews.net/economy/bangladesh-pays-699m-foreign-debt-receives-295m-july-august-1554971?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 3. Child development in Pakistan does not follow a single provincial pattern + +A new study uses data on **46,067 children aged 36–59 months** across four Pakistani provinces to examine developmental attainment. + +Its most useful finding may be the absence of a simple provincial hierarchy. Performance varies across developmental domains, while socioeconomic gradients appear in every province. Early-childhood education attendance is associated with higher developmental attainment; stunting and functional difficulties are associated with poorer attainment. The relationship with adult stimulation differs across settings. [MDPI](https://www.mdpi.com/2227-9067/13/10/1314?utm_source=chatgpt.com) + +This matters for how we use composite early-childhood indicators. A single national or provincial score can conceal different deficits in literacy-numeracy, physical development, learning and socio-emotional development. + +It also argues against treating “early childhood development” as one intervention problem. Nutrition, disability, preschool participation, household resources and caregiving environments do not operate through identical pathways. + +[Read the study in Children](https://www.mdpi.com/2227-9067/13/10/1314?utm_source=chatgpt.com) + +### 4. Bangladesh’s manufacturing problem is not simply that it imports too much + +A new open-access *Scientific Reports* paper examines Bangladesh’s electrical-equipment manufacturing sector through trade dependence, domestic industrial capacity and competitiveness. + +The sector is an interesting case for industrial policy. Expanding electricity systems and urbanisation create domestic demand for electrical equipment, but high demand does not by itself create domestic manufacturing capability. The paper examines the relationship between import dependence and the constraints facing local production rather than treating import substitution as a slogan. [Nature](https://www.nature.com/articles/s41598-026-72639-z?utm_source=chatgpt.com) + +The distinction matters. Reducing imports can be an objective on paper. Building firms capable of producing reliable goods at competitive prices requires technology, capital, standards, skills, supply chains and sufficient market scale. + +For development policy, the better question is therefore not simply *what does Bangladesh import?* It is *which capabilities prevent domestic firms from producing it competitively?* + +[Read the Scientific Reports paper](https://www.nature.com/articles/s41598-026-72639-z?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 5. Climate migration policy in Bangladesh is moving beyond the place people leave + +A workshop organised by the Refugee and Migratory Movements Research Unit in Chattogram has proposed a five-part coordination model involving government, business, civil society, media and local communities to support migrants arriving in the city because of climate and other pressures. [The Business Standard](https://www.tbsnews.net/bangladesh/migration/pentagonal-model-proposed-coordinate-jobs-services-ctg-migrants-1555286?utm_source=chatgpt.com) + +The proposal itself is preliminary. The more interesting shift is conceptual. + +Climate-migration policy is often organised around vulnerable places of origin: eroding coastlines, flood-prone settlements or areas exposed to salinity and extreme weather. But adaptation through migration also creates policy obligations at the destination. + +Where do people live? How do they enter urban labour markets? Can children enter school? Can families obtain health services and social protection? What happens when documentation remains tied to another place? + +If migration is an adaptation strategy, receiving cities are part of climate-adaptation policy too. + +[Read the report from Chattogram](https://www.tbsnews.net/bangladesh/migration/pentagonal-model-proposed-coordinate-jobs-services-ctg-migrants-1555286?utm_source=chatgpt.com) + +## Paper of the day + +### What happened when India subsidised LPG connections for women? + +Bharti Nandwani and Manisha Jain have a new paper in the *Journal of Environmental Management* examining the Pradhan Mantri Ujjwala Yojana, which provided subsidised LPG connections to women from economically and socially disadvantaged households. + +The paper was published online on 26 September. Its question is larger than fuel adoption: what effect did a gender-targeted clean-energy policy have on women’s well-being? [PubMed](https://pubmed.ncbi.nlm.nih.gov/42800272/?utm_source=chatgpt.com) + +PMUY is especially useful for evaluation because it sits at the intersection of energy poverty, household air pollution and gender. Women disproportionately undertake cooking, so the time and health costs associated with solid fuels are also unequally distributed within households. + +But an LPG **connection** is not the same thing as sustained LPG **use**. Refill affordability, fuel prices, household preferences and access to distributors can all affect what happens after a connection is issued. + +That distinction is central to evaluating clean-cooking programmes. Administrative data can tell us how many connections exist. The harder outcomes are fuel stacking, frequency of LPG use, exposure to smoke, time use and changes in women’s welfare. + +[Nandwani and Jain, Journal of Environmental Management](https://pubmed.ncbi.nlm.nih.gov/42800272/?utm_source=chatgpt.com) + +[Subscribe now](https://varna.substack.com/subscribe?) + +## One number + +### $698.92 million + +Bangladesh’s foreign debt repayments during July and August 2026. + +Foreign loans and grants received during the same period: **$294.56 million**. [The Business Standard](https://www.tbsnews.net/economy/bangladesh-pays-699m-foreign-debt-receives-295m-july-august-1554971) + +The ratio is more informative than either number on its own. + +## Policy watch + +### Housing is becoming part of climate finance + +The Bangladesh housing project is worth watching for a reason beyond its $100 million size. + +ADB has designated **$40 million** of its financing as climate adaptation finance, while the housing loans are intended to support construction capable of better withstanding floods, cyclones and extreme heat. [Asian Development Bank](https://www.adb.org/news/adb-approves-100-million-loan-expand-access-affordable-resilient-housing-finance-bangladesh) + +This broadens what adaptation spending looks like. + +Climate resilience is often discussed through embankments, drainage, early-warning systems or large infrastructure. But for a household exposed to repeated flooding or extreme heat, the design and location of the house itself may be adaptation infrastructure. + +The evaluation question will be whether financing produces homes that are measurably safer when hazards occur, rather than simply financing more construction. + +## A note from me + +There is an interesting thread running through today’s material. + +A housing loan is not secure housing. + +An LPG connection is not clean cooking. + +Foreign assistance received is not the same as the net flow after debt repayment. + +Migration away from a climate-exposed place is not successful adaptation if the destination offers insecure housing, poor work and weak access to services. + +And an early-childhood-development score can hide very different developmental problems. + +These are all reminders to ask what sits between an intervention and the outcome attached to it. + +We often measure the first observable change because it arrives quickly and sits neatly in administrative data. Connections. Loans. Enrolments. Disbursements. Migration counts. + +The harder work starts one step later. + +**What happened after access was created?** + +That is usually where the development question becomes more interesting. + +[Share](https://varna.substack.com/p/research-rundown-daily-27-september?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +--- diff --git a/archive/2026-10-03-research-rundown-gandhi-jayanti-special.md b/archive/2026-10-03-research-rundown-gandhi-jayanti-special.md new file mode 100644 index 0000000..f1b1da1 --- /dev/null +++ b/archive/2026-10-03-research-rundown-gandhi-jayanti-special.md @@ -0,0 +1,267 @@ +--- +title: "Research Rundown | Gandhi Jayanti Special" +date: 2026-10-03 +type: edition +url: https://varna.substack.com/p/research-rundown-gandhi-jayanti-special +subtitle: "Gandhi imagined power moving towards self-governing villages. Ambedkar asked what happens when a village itself is unequal. Nearly eighty years later, development policy is still wrestling with both." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown | Gandhi Jayanti Special + +*2026-10-03. Published at [https://varna.substack.com/p/research-rundown-gandhi-jayanti-special](https://varna.substack.com/p/research-rundown-gandhi-jayanti-special). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown Gandhi Jayanti Special cover for 2 October 2026, with Gandhi and B. R. Ambedkar above a layered rural Indian scene featuring a village meeting, farmland, a charkha, the Constitution, water infrastructure and solar power, reflecting the essay’s examination of village swaraj, decentralisation and unequal local power.](https://substackcdn.com/image/fetch/$s_!Mrmd!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4f212b63-7a01-4107-b8fc-e986cea28e9c_1672x941.png)](https://substackcdn.com/image/fetch/$s_!Mrmd!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4f212b63-7a01-4107-b8fc-e986cea28e9c_1672x941.png) +*A Research Rundown Special on Gandhi Jayanti 2026* + +## The village was never the real question + +*Gandhi imagined economic and political power moving closer to ordinary people. Ambedkar asked what happens when the local community itself is unequal. Rural India still lives with the problem they identified from opposite directions.* + +Gandhi Jayanti has not been particularly kind to Gandhi’s economic thought. What survives in public commemoration is usually a small repertoire of symbols: the charkha, khadi, sanitation, simplicity and, above all, the village. Detached from the arguments in which Gandhi placed them, these can make his economics seem either quaint or easily admirable. Neither reading is especially useful. + +Gandhi’s interest in the village belonged to a much larger argument about the organisation of economic and political power. He distrusted an industrial order in which ownership, production and decision-making became increasingly concentrated. He worried about forms of technological change that could increase output while leaving large numbers of people without adequate work. His preference for local production was therefore connected to employment, control over productive assets and the distribution of economic power. In the writings collected as *Village Swaraj*, he argued that the means of producing basic necessities should remain under broad social control and that people should be able to obtain sufficient work to support themselves. [Mahatma Gandhi Website](https://www.mkgandhi.org/village_swaraj/preface.php?utm_source=chatgpt.com) + +Village Swaraj was not simply a plea to preserve rural India as Gandhi found it. His description of an ideal village included food production, clean water, schools, public spaces, cooperative economic activity, elected local government, and the abolition of caste hierarchy and untouchability. He imagined a village that was substantially capable of meeting its basic needs while remaining interdependent with the world beyond it. [Mani Bhavan](https://www.gandhi-manibhavan.org/gandhian-philosophy/philosophy-village-idealvillage.html?utm_source=chatgpt.com) + +That distinction matters because the Indian village as it existed was precisely where Ambedkar’s objection began. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Ambedkar’s problem with the village + +In the Constituent Assembly on 4 November 1948, B. R. Ambedkar responded sharply to members who wanted the Constitution to give greater importance to traditional village communities. His description of the village as a “sink of localism” is now frequently quoted. The more important sentence for our purposes comes immediately afterwards. Ambedkar welcomed the Draft Constitution because it had discarded the village and “adopted the individual as its unit.” [Constitution of India](https://www.constitutionofindia.net/debates/04-nov-1948/?utm_source=chatgpt.com) + +The disagreement was therefore deeper than a dispute over whether village life was desirable. + +Gandhi began with the possibility of transforming the community. His ideal village would have neither caste hierarchy nor untouchability and would govern itself through democratic cooperation. Ambedkar began from the social relations of the village that actually existed. Caste was not an unfortunate feature that could simply be removed from the village model. It structured property, labour, residence, social status and political power. A political system that devolved authority to the village without changing those relations could give greater authority to the people who already dominated it. + +This is one reason the Gandhi-Ambedkar disagreement remains useful for development policy. It identifies a problem that has never been solved simply by choosing centralisation or decentralisation. + +Moving power closer to people can make institutions more responsive. It can also move power closer to existing systems of local domination. + +The important question is not therefore whether decentralisation is good. It is how power is distributed within the unit to which authority is being decentralised. + +## India eventually tried to do both + +Independent India’s system of rural local government can be read, cautiously, as an attempt to live with both propositions. + +The 73rd Constitutional Amendment gave Panchayati Raj institutions constitutional status and established elected local government across rural India. It also required reservation of seats for Scheduled Castes and Scheduled Tribes and reserved at least one-third of seats and chairperson positions for women. Many states subsequently increased women’s reservation to 50 per cent. + +That second part is important. Decentralisation says that decisions should move downwards. Political reservation asks who will be present when those decisions are made. + +There is now substantial evidence that this institutional design has had consequences. + +One of the best-known studies remains Raghabendra Chattopadhyay and Esther Duflo’s analysis of randomly reserved village council leadership positions in West Bengal and Rajasthan. Using data from 265 village councils, they found that women leaders changed the composition of local public spending, investing more in public goods that were particularly relevant to women’s expressed priorities. Representation affected policy. [DOI](https://doi.org/10.1111/j.1468-0262.2004.00539.x?utm_source=chatgpt.com) + +Later research has found effects extending beyond the period in which seats are reserved. Evidence from Jharkhand, for example, suggests that councils previously exposed to gender quotas continue to elect more women after the quota ceases to bind. Other research has found effects on women’s political participation beyond local government, although these effects vary by level of government and institutional setting. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X23001109?utm_source=chatgpt.com) + +None of this means that a reserved seat automatically redistributes political power. Research on Panchayati Raj has also documented constraints created by caste, gender relations, institutional design and differences in representatives’ ability to exercise authority. [Korea Journal Central](https://journal.kci.go.kr/jrsd_ipaid/archive/articleView?artiId=ART001615735&utm_source=chatgpt.com) + +That combination of findings is more useful than either a celebratory or dismissive account of decentralisation. Institutions can change who participates and what governments do. The social relations surrounding those institutions continue to matter. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## MGNREGA contains the same tension + +Consider MGNREGA. + +It is difficult to imagine a less purely Gandhian programme. The employment guarantee is created by national legislation, financed overwhelmingly beyond the village and supported by administrative and digital systems operating at enormous scale. + +Yet its implementation also gives local institutions a significant role. Gram panchayats are involved in planning and implementing works, and the programme’s social-audit architecture attempts to create a form of public accountability in which expenditure, works and worker entitlements can be examined locally. The government’s own social-audit system records works, expenditure, wage payments and entitlements at gram-panchayat level. [MGNREGA](https://mnregaweb4.nic.in/netnrega/SocialAuditFindings/SA-GPReport.aspx?utm_source=chatgpt.com) + +The design is interesting because it does not assume that proximity itself creates accountability. It attempts to build an institution through which local decisions can be inspected. + +And even that is insufficient by itself. Research from Karnataka, for example, found that formal evidence that social audits had occurred did not necessarily mean workers had participated effectively. In the gram panchayats studied, the researchers found that local elites could influence the process and exclude poorer labourers. [ResearchGate](https://www.researchgate.net/publication/246546990_How_Effective_are_Social_Audits_under_MGNREGS_Lessons_from_Karnataka?utm_source=chatgpt.com) + +This is almost a textbook example of the problem Ambedkar identified. + +A gram sabha is closer to a worker than a ministry in Delhi. That does not guarantee that the worker has equal power inside the gram sabha. + +The distinction is important for development practice because we often use words such as “community”, “participation” and “local ownership” without specifying who inside the community participates, whose preferences determine the outcome, or who has enough social power to disagree publicly. + +A community is not a unitary actor. + +## The economic question is different, but connected + +Gandhi’s economics raises a second problem that deserves more attention than the charkha usually receives. + +He was concerned with employment as well as production. His argument for dispersed production rested partly on the proposition that an economic system in a labour-abundant country should provide people with enough work to live. His opposition to concentrated control over the production of basic necessities followed from the same concern. [Mahatma Gandhi Website](https://www.mkgandhi.org/village_swaraj/preface.php?utm_source=chatgpt.com) + +Taken literally as an industrial strategy for India in 2026, much of this is difficult to sustain. A modern economy cannot be organised as hundreds of thousands of substantially self-sufficient village economies. Modern medicine, universities, railways, electricity grids, pharmaceutical production, telecommunications and advanced manufacturing depend on specialisation, large networks and forms of coordination extending far beyond a village. + +Scale produces real gains. + +Cities also matter for reasons that a romantic account of rural community can obscure. Migration can provide higher earnings, education and new social possibilities. Leaving a village may be particularly important for people whose opportunities are constrained by local caste or gender relations. Rural development should not be judged by whether it succeeds in preventing people from moving. + +But rejecting village self-sufficiency does not settle Gandhi’s economic question. + +India can produce more without creating enough secure and adequately paid employment. Agricultural productivity can increase while labour demand falls. Platform economies can create new work while transferring economic risk to workers. Automation can raise output while reducing the labour required to produce it. + +The contemporary question is therefore not whether India should return to village production. It is whether growth in output is sufficient evidence of economic progress when the distribution of work, income, risk and economic power moves differently. + +That is a question development economics recognises perfectly well, even if we no longer use Gandhi’s vocabulary to ask it. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## The village economy has already changed + +There is another reason not to turn Gandhi’s village into a contemporary policy prescription: the rural economy itself no longer corresponds to the agricultural village that dominates much public imagination. + +Rural households can combine farming, casual labour, salaried employment, public works, self-employment and remittances. Workers move between villages, small towns and large cities. Goods, credit, information and public benefits travel through networks extending far beyond local boundaries. Mobile phones and digital payments have altered the economic geography of rural life. + +The rural and urban economies are not separate systems. + +This matters because self-reliance can easily become a misleading development objective if it means reducing the opportunities available outside a locality. A village from which nobody needs to migrate is not necessarily more developed than one whose residents can move freely and obtain better work elsewhere. + +The better question concerns choice. Can people remain without being trapped by the absence of alternatives? Can they leave without losing access to rights and public services? Can local economies provide viable livelihoods without making mobility a sign of failure? + +That formulation is quite different from either preserving the village or abandoning it. + +## Climate change gives decentralisation another meaning + +There is, however, one part of Gandhi’s concern with local production and restraint that looks different in an era of climate change. + +Development has spent decades asking how economies can produce and consume more. Climate change forces an additional question: what material and ecological systems make that production possible? + +Food, water, energy and housing are unusually local when something goes wrong. A national electricity grid may be centrally organised, but a household experiences a blackout locally. Food markets may be national or global, but crop failure begins in particular fields. Climate models operate at large scales, while heat stress is experienced by particular workers in particular places. + +Local capacity therefore matters for resilience. + +This is not an argument for economic autarky. It is an argument for examining dependence more carefully. + +A community dependent on a single crop, employer, water source or transport connection may be efficient under ordinary conditions and extremely vulnerable when conditions change. Diversification and redundancy can look inefficient until a shock occurs. + +Here the Gandhian suspicion of an economy organised solely around increasing production intersects with a contemporary problem he could not have formulated in our terms. Economic systems have environmental costs, and those costs are unevenly distributed. + +The question is not merely how much an economy produces. It is who receives the benefits, who performs the work, and who absorbs the ecological cost. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## The measurement problem + +There is also a measurement problem here, which is where this argument connects most directly with the concerns of Research Rundown. + +Suppose we wanted to know whether a village had become more capable of governing its own development. + +Income per capita would tell us something. It would not tell us enough. + +We would want to know about employment and wages, but also land ownership and indebtedness. We would need measures of drinking water, sanitation, healthcare and schooling, but also whether access differed by caste or settlement. Women’s labour-force participation would matter, but so would wages, unpaid work, control over earnings and freedom of movement. + +Political representation could be counted. Effective political authority would be harder to measure. + +A village could elect a woman sarpanch while decisions were exercised informally by male relatives. It could have a functioning gram panchayat while particular hamlets remained poorly served. Average household income could increase while land concentration worsened. Infrastructure could improve while groundwater deteriorated. + +None of these contradictions is unusual. They are what development data look like once we stop asking one indicator to represent an entire social condition. + +This is also why the individual matters. + +Household income is not individual income. Household food security does not tell us how food is distributed within the household. Women’s employment does not tell us who controls their wages. Village-level access to water does not establish equal access across neighbourhoods or caste groups. + +Ambedkar’s insistence on the individual as the constitutional unit has an interesting methodological echo here. Development researchers repeatedly need aggregate units. We sample households, villages, districts and states because research and administration require them. But improvements measured at those levels cannot simply be assumed to reach every person within them. + +The average village does not experience development. + +People do. + +## Gandhi’s ideal village also presents an evaluation problem + +There is a further difficulty with Gandhi’s village that becomes apparent when we read his description closely. + +His ideal village already contains many of the outcomes that political and economic institutions are supposed to produce. It has democratic government, cooperation, clean water, education, and no caste hierarchy or untouchability. [Mani Bhavan](https://www.gandhi-manibhavan.org/gandhian-philosophy/philosophy-village-idealvillage.html?utm_source=chatgpt.com) + +That makes it morally attractive, but analytically difficult. + +If the village is already free of the social inequalities that make local government problematic, decentralisation becomes easier to defend. The harder policy problem is how democratic local government functions while those inequalities still exist. + +Ambedkar begins closer to that problem. + +He asks us to start with unequal society rather than assume that equality has already been produced. + +This is why the disagreement cannot be resolved by saying Gandhi imagined an egalitarian village and therefore had already answered Ambedkar’s criticism. The disagreement concerns the route to democratic community as much as its desired form. + +Can local democracy dismantle entrenched hierarchy? + +Or does equality require rights and institutions strong enough to protect individuals against local power before meaningful local democracy becomes possible? + +India’s experience suggests that these are not necessarily alternatives. Constitutional rights, political reservation, nationally financed entitlements and decentralised institutions can operate together. Their relationship is often uncomfortable, and implementation varies considerably, but that may be precisely the point. + +The answer to excessive centralisation need not be unprotected localism. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## What remains useful in Gandhi + +I would not use Gandhi as a guide to the industrial structure India should adopt in 2026. Nor would I use the self-sufficient village as a benchmark for rural development. + +But several of his economic concerns remain worth taking seriously. + +One is concentration. Who owns productive assets and how much economic power follows from that ownership? + +Another is employment. Does technological and economic change produce livelihoods as well as output? + +A third is scale. Which decisions benefit from being made locally, and which require institutions operating at much larger levels? + +A fourth is ecological restraint. How much production and consumption can an economy sustain, and who bears their environmental costs? + +These questions survive even when Gandhi’s answers do not. + +Ambedkar makes the inquiry harder by requiring us to examine power within the community as carefully as power above it. Decentralisation can bring government closer to citizens, but distance from Delhi is not a measure of democracy. A decision made in the village can still exclude the person standing outside its dominant caste, class or gender relations. + +The evidence from Panchayati Raj is useful precisely because it refuses a simple conclusion. Political reservation has changed representation and, in some settings, public investment and subsequent political participation. At the same time, local institutions continue to operate inside unequal social structures. [DOI](https://doi.org/10.1111/j.1468-0262.2004.00539.x?utm_source=chatgpt.com) + +That is not evidence that decentralisation has succeeded or failed. It tells us that institutional design changes incentives and participation ,while social power does not disappear when an institution is redesigned. + +## The village was never the real question + +The interesting argument between Gandhi and Ambedkar is therefore not about whether villages are good places. + +It is about power. + +Gandhi asks what happens when economic and political power becomes too distant and too concentrated. Ambedkar asks what happens when power is moved closer to people without confronting hierarchy among those people. + +Development policy still moves between these two problems. + +Centralised systems can achieve scale, redistribution and enforceable rights that local institutions cannot. They can also become unresponsive to local conditions and difficult for citizens to influence. + +Local institutions can use knowledge that central governments do not possess and provide citizens with more direct routes into decision-making. They can also reproduce the inequalities of the societies in which they operate. + +Neither scale solves the problem of power by itself. + +That seems a more useful way to return to Gandhi on 2 October than asking whether India has remained faithful to his vision of the village. India has changed too much for that to be an especially productive test. + +The harder question is whether development has answered the problems that made Gandhi suspicious of concentrated economic power in the first place: insecure work, unequal ownership, dependence and the distance between ordinary people and the institutions that shape their lives. + +Ambedkar requires us to add another test. When power is redistributed, who actually receives it? + +That is where the argument becomes contemporary. + +And it is why the Indian village remains interesting, not as a model to which India should return, but as a place in which some of the central problems of development are unusually easy to see. + +--- + +### Sources and further reading + +M. K. Gandhi’s writings on village swaraj set out the relationship he saw between local government, cooperative production, employment, access to basic necessities and the abolition of caste hierarchy and untouchability. [Mani Bhavan](https://www.gandhi-manibhavan.org/gandhian-philosophy/philosophy-village-idealvillage.html?utm_source=chatgpt.com) + +[Read Village Swaraj](https://www.mkgandhi.org/village_swaraj/village_swaraj.php?utm_source=chatgpt.com) + +B. R. Ambedkar’s intervention in the Constituent Assembly on 4 November 1948 provides the clearest primary source for his criticism of village republics and his defence of the individual as the constitutional unit. [Constitution of India](https://www.constitutionofindia.net/debates/04-nov-1948/?utm_source=chatgpt.com) + +[Read the Constituent Assembly debate of 4 November 1948](https://www.constitutionofindia.net/debates/04-nov-1948/?utm_source=chatgpt.com) + +Raghabendra Chattopadhyay and Esther Duflo’s 2004 *Econometrica* paper remains an important empirical study of women’s political reservation and local public-goods provision in West Bengal and Rajasthan. [DOI](https://doi.org/10.1111/j.1468-0262.2004.00539.x?utm_source=chatgpt.com) + +[Women as Policy Makers: Evidence from a Randomized Policy Experiment in India](https://doi.org/10.1111/j.1468-0262.2004.00539.x?utm_source=chatgpt.com) + +Research on electoral gender quotas in Jharkhand provides more recent evidence on whether reservation can have effects on women’s representation after reserved terms end. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X23001109?utm_source=chatgpt.com) + +[Contemporaneous and lasting effects of electoral gender quotas](https://www.sciencedirect.com/science/article/pii/S0305750X23001109?utm_source=chatgpt.com) + +The Ministry of Rural Development’s MGNREGA social-audit system shows the formal architecture through which gram-panchayats scrutinize work, expenditure, payments, and worker entitlements. [MGNREGA](https://mnregaweb4.nic.in/netnrega/SocialAuditFindings/SA-GPReport.aspx?utm_source=chatgpt.com) + +[MGNREGA Social Audit portal](https://mnregaweb4.nic.in/netnrega/SocialAudit/sa_status_report.aspx?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/research-rundown-gandhi-jayanti-special?utm_source=substack&utm_medium=email&utm_content=share&action=share) + +### diff --git a/archive/2026-10-07-good-reads-13-when-women-lead-girls.md b/archive/2026-10-07-good-reads-13-when-women-lead-girls.md new file mode 100644 index 0000000..18f18a9 --- /dev/null +++ b/archive/2026-10-07-good-reads-13-when-women-lead-girls.md @@ -0,0 +1,185 @@ +--- +title: "Good Reads #13 | When women lead, girls marry later. But there is a catch." +date: 2026-10-07 +type: other +url: https://varna.substack.com/p/good-reads-13-when-women-lead-girls +subtitle: "Women’s political representation in rural India delays marriage by expanding young women’s economic opportunities. A new study finds that marriage markets respond with a higher dowry." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #13 | When women lead, girls marry later. But there is a catch. + +*2026-10-07. Published at [https://varna.substack.com/p/good-reads-13-when-women-lead-girls](https://varna.substack.com/p/good-reads-13-when-women-lead-girls). This file is a generated copy; the Substack post is the original.* + +[![image](https://substackcdn.com/image/fetch/$s_!04Rz!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F48702b71-f911-4394-8651-0ea7d41e44b4_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!04Rz!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F48702b71-f911-4394-8651-0ea7d41e44b4_1536x1024.png) +*Research Rundown 4 October, 2026* + +A new study of women’s political representation in rural India finds that women leaders can delay marriage by improving young women’s labour-market opportunities. In more patriarchal settings, families appear to pay a price for that delay. + +A result in Priyama Majumdar’s new paper has been on my mind because it refuses to resolve into a straightforward success story. + +Women become village council leaders. + +Young women marry later. + +That sounds like precisely the kind of intergenerational effect we might hope political representation would produce. + +But there is another result. + +Women who marry later also pay higher dowries. And that penalty is largest in more patriarchal states. [DOI](https://doi.org/10.1016/j.jdeveco.2026.103859?utm_source=chatgpt.com) + +So what should we conclude? + +## What the paper asks + +Majumdar studies the consequences of India’s 73rd Constitutional Amendment, which reserved one-third of village council leadership positions for women. + +The institutional feature that makes this useful for research is the rotation of reserved seats. Villages were exposed to women leaders in a manner the paper treats as quasi-random, allowing the author to compare marriage outcomes among young women with different exposure to female political leadership. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0304387826001422?utm_source=chatgpt.com) + +There is already a substantial literature on women’s political reservation in India. It has examined public goods, political participation, aspirations, attitudes towards women leaders, and investments in girls. + +This paper asks something slightly different. + +Can women’s political representation change marriage practices? + +That matters because early marriage and dowry are not simply private household decisions. They are institutions through which gender, labour markets, family status and social norms interact. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## Women leaders delay marriage + +The first result is encouraging. + +Exposure to a woman village head delays marriage among young women. [DOI](https://doi.org/10.1016/j.jdeveco.2026.103859?utm_source=chatgpt.com) + +The proposed mechanism is particularly interesting. + +The paper finds evidence that women leaders increase labour-market opportunities for young women. When remaining unmarried for longer becomes economically more useful, the incentives surrounding marriage timing change. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0304387826001422?utm_source=chatgpt.com) + +That is a different story from one in which a woman leader simply acts as a role model and families consequently adopt more progressive attitudes. + +Representation changes material opportunities. + +Those opportunities change the value of delaying marriage. + +And that changes behaviour. + +I find that mechanism more persuasive than the vague claim that representation automatically produces “empowerment”. It gives us something observable to investigate between political representation and a social outcome. + +## Then comes the uncomfortable result + +The marriage market does not necessarily accept the change without adjustment. + +Women who delay marriage pay higher dowries. [DOI](https://doi.org/10.1016/j.jdeveco.2026.103859?utm_source=chatgpt.com) + +The dowry penalty is strongest in more patriarchal states, which the author interprets as evidence of backlash against delayed marriage. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0304387826001422?utm_source=chatgpt.com) + +This is the part of the paper worth sitting with. + +A policy can improve one dimension of women’s lives while creating a cost somewhere else. + +That does not mean the original improvement was illusory. A woman marrying later may have additional education, employment, earnings or autonomy. + +Nor does the dowry result mean women’s political reservation caused gender relations to deteriorate overall. + +It tells us something more specific. + +Institutions can respond to social change. + +If the expected age of marriage changes while the institution of dowry remains intact, the marriage market can partly absorb that change through price. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## This complicates how we measure empowerment + +Suppose we evaluated women’s political reservation using only age at marriage. + +We might conclude that exposure to women leaders improved women’s outcomes. + +Suppose instead we looked only at dowry. + +We might reach a much darker conclusion. + +Both would miss what is happening. + +The interesting result lies in observing the two together. + +Political representation appears to expand opportunities outside marriage. Women consequently marry later. But an existing institution then imposes a cost on some of that departure from the norm. + +This is why I am increasingly wary of empowerment indices that combine several outcomes into a single score. + +An intervention can increase employment and bargaining power while leaving unpaid work unchanged. + +It can increase political representation without changing control over political decisions. + +It can delay marriage while increasing the financial cost associated with marriage. + +These are not statistical inconveniences to be averaged away. They tell us how gender institutions respond to change. + +## Representation matters, but not in isolation + +There is another reason I like this paper. + +Arguments about political reservation sometimes become arguments about whether women leaders “make a difference”. + +That formulation puts an enormous burden on representation. + +Electing women into political office does not remove the institutions surrounding them. It changes who holds political authority within those institutions. + +The effects can then travel outward through public spending, employment opportunities, aspirations, information and social expectations. + +But caste, class, labour markets, marriage practices and patriarchy remain. + +The paper’s finding that the dowry penalty is stronger in more patriarchal states is therefore important. [ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0304387826001422?utm_source=chatgpt.com) + +The same political intervention can encounter different social institutions and produce different secondary effects. + +Context here is not background information. It is part of the causal story. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## There is also a useful lesson about unintended effects + +Development research often uses the language of “unintended consequences” rather loosely. + +This case is more interesting. + +The higher dowry associated with delayed marriage is not necessarily an unrelated side effect. It may be part of the process through which an existing institution reacts to women acquiring different opportunities. + +That distinction matters. + +If so, the policy question is not whether women should therefore marry earlier. + +It is why delaying marriage carries a penalty in the first place. + +The appropriate response to backlash against greater freedom is not less freedom. It is to understand the institution producing the backlash. + +## What I take from the paper + +I would resist turning this into either a cheerful story about women leaders changing villages or a pessimistic story about patriarchy cancelling political reform. + +The evidence is more interesting than either. + +Women’s political representation appears to change young women’s economic opportunities. + +Those opportunities alter marriage timing. + +Marriage institutions respond. + +And the response differs according to the surrounding gender regime. + +That is a much more plausible account of social change than the idea that one intervention moves an entire society neatly from one equilibrium to another. + +Policies enter systems. + +Systems react. + +Sometimes the reaction itself is the finding. + +--- + +**Source:** Priyama Majumdar, “Political Empowerment and Marital Practices,” *Journal of Development Economics*, vol. 183, September 2026, article 103859. The paper is open access. [DOI](https://doi.org/10.1016/j.jdeveco.2026.103859?utm_source=chatgpt.com) + +[Read the paper](https://doi.org/10.1016/j.jdeveco.2026.103859?utm_source=chatgpt.com) + +[Share](https://varna.substack.com/p/good-reads-13-when-women-lead-girls?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-10-07-research-rundown-daily-3-october.md b/archive/2026-10-07-research-rundown-daily-3-october.md new file mode 100644 index 0000000..5f366a5 --- /dev/null +++ b/archive/2026-10-07-research-rundown-daily-3-october.md @@ -0,0 +1,162 @@ +--- +title: "Research Rundown DAILY - 3 October 2026" +date: 2026-10-07 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-3-october +subtitle: "Energy prices surge, Pakistan moves mental health into primary care, and Nepal’s floods expose the health costs that disaster counts miss." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown DAILY - 3 October 2026 + +*2026-10-07. Published at [https://varna.substack.com/p/research-rundown-daily-3-october](https://varna.substack.com/p/research-rundown-daily-3-october). This file is a generated copy; the Substack post is the original.* + +[![Research Rundown DAILY cover for 3 October 2026, showing a 25.7 per cent September surge in energy prices alongside a South Asia map and research notes on fuel imports, food prices, fiscal effects and social protection.](https://substackcdn.com/image/fetch/$s_!5Vvx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F71736eaa-dc51-45c0-8142-592483e2a796_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!5Vvx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F71736eaa-dc51-45c0-8142-592483e2a796_1536x1024.png) + +Energy prices jump again, Pakistan takes mental health into primary care, Sri Lanka keeps its inflation target at 5 per cent, and Nepal’s floods show how quickly a climate disaster becomes a maternal-health crisis. + +## FIVE THINGS WORTH KNOWING + +### 1. The energy shock is getting considerably larger + +The World Bank’s latest commodity data, released on 2 October, show its energy price index rising **25.7 per cent in September alone**. Crude oil increased 21.2 per cent and natural gas 15.3 per cent. Food prices rose 3.8 per cent. Fertiliser prices were comparatively stable, increasing 0.4 per cent. + +[World Bank commodity price data](https://www.worldbank.org/en/research/commodity-markets?utm_source=chatgpt.com) + +For South Asia, the energy number matters well beyond petrol and electricity. India, Pakistan, Bangladesh, Sri Lanka and Nepal remain exposed to imported fuel costs to varying degrees. Higher energy prices can pass through to transport, fertiliser production, food distribution, industrial costs and government budgets. + +The monthly change is striking, though it should not be mistaken for a forecast. Commodity prices can reverse quickly. The relevant question for South Asian economies is how long the shock persists and how much governments absorb through taxes, subsidies, administered prices or fiscal support. + +### 2. Pakistan has adopted a national mental health policy centred on primary and community care + +Pakistan launched its National Mental Health Policy on 1 October. The policy proposes moving much of mental-health provision closer to households through primary healthcare, schools and community services. + +Health Minister Syed Mustafa Kamal said around **24 million people** in Pakistan require psychiatric assistance, while the country has fewer than one psychiatrist per 500,000 people. + +[Dawn report on Pakistan’s National Mental Health Policy](https://www.dawn.com/news/2034059/govt-launches-policy-to-bring-mental-healthcare-closer-to-families?utm_source=chatgpt.com) + +The direction makes sense in a system with such severe specialist shortages. Implementation will be harder. Moving services into primary care requires trained workers, referral systems, medicines, supervision and financing. A policy that decentralises the location of care without supplying those inputs simply relocates scarcity. + +The useful indicators now are therefore operational: how many primary-care facilities actually provide mental-health services, who delivers them, what happens after referral, and whether poorer and rural households use them. + +### 3. Pakistan is putting Rs87 billion into coastal resilience in Sindh + +The Sindh government has launched a **Rs87 billion coastal resilience project** covering Thatta, Sujawal and Badin, with a stated target of one million beneficiaries. + +The project involves the provincial government, the International Fund for Agricultural Development, Asian Development Bank and Green Climate Fund. + +[Dawn report on the Sindh Coastal Resilience Project](https://www.dawn.com/news/2034097/rs87bn-coastal-resilience-project-launched-for-three-sindh-districts?utm_source=chatgpt.com) + +The geography matters. These coastal districts face overlapping exposure to salinity, cyclones, flooding, declining agricultural productivity and livelihood insecurity. + +The one-million-beneficiary figure deserves scrutiny as implementation begins. Climate programmes often report people “reached” without distinguishing between direct recipients, households living within a project area and people experiencing measurable reductions in risk. Those are different quantities. + +This project will be worth following for its measurement system as much as its expenditure. + +[Subscribe now](https://varna.substack.com/subscribe?) + +### 4. Sri Lanka has kept its inflation target at 5 per cent for another three years + +Sri Lanka’s government and central bank have retained a **5 per cent inflation target** under the country’s Monetary Policy Framework Agreement. + +The Central Bank of Sri Lanka said the decision followed a scheduled review covering the country’s economic structure, empirical evidence, monetary-policy considerations, stakeholder views and international experience. + +[Central Bank of Sri Lanka announcement](https://www.cbsl.gov.lk/en/news/SLs-inflation-target-is-set-at-5-for-the-next-three-years-20261002?utm_source=chatgpt.com) + +This is significant because Sri Lanka’s recent experience with inflation was extreme. The country is now moving from crisis stabilisation into the harder task of establishing a credible medium-term monetary framework. + +A numerical target cannot establish credibility on its own. Fiscal policy, exchange-rate pressures, energy and food prices, and the central bank’s capacity to respond to shocks will determine how binding the target becomes in practice. + +### 5. Nepal’s floods are disrupting maternity care weeks after the water receded + +At least **33 pregnant women from Rasuwa and Nuwakot have travelled to Kathmandu to give birth** because health services and roads remain disrupted after the 26 August floods. Four women in labour had to be airlifted. + +[Kathmandu Post report on maternity services after the floods](https://kathmandupost.com/health/2026/10/02/floods-drive-33-pregnant-women-to-kathmandu-for-childbirth?utm_source=chatgpt.com) + +This is a useful example of why disaster mortality alone is an inadequate health indicator. Flood exposure can continue through interrupted antenatal care, inaccessible facilities, medicine shortages, damaged roads and delayed emergency referrals. + +The denominator also matters. Thirty-three transfers tell us that disruption occurred. They do not tell us the proportion of pregnant women affected, how many missed antenatal visits, whether institutional delivery rates changed or whether maternal and neonatal outcomes worsened. + +Those are the data needed to measure the health burden of the disaster rather than its immediate physical damage. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## ONE PAPER WORTH READING + +### Does ageing matter for public health spending in South Asia? + +Jalandhar Pradhan, Soumyanayani Mahali and Tapan Swain examine the relationship between population ageing and government health expenditure across South Asian countries in a new peer-reviewed paper in *BMC Health Services Research*, published on 1 October. + +[Read the open-access paper](https://link.springer.com/article/10.1186/s12913-026-15545-4?utm_source=chatgpt.com) + +The authors use cross-country panel data and several econometric specifications, including Driscoll-Kraay standard errors, panel-corrected standard errors and three-stage least squares. The paper treats demographic change alongside physician availability, unemployment and other macro-level determinants of public health expenditure. + +Its regional focus is useful. South Asia is ageing while several countries still carry large maternal, child-health and infectious-disease burdens. Public systems therefore face overlapping rather than sequential epidemiological demands. + +The limitation is identification. This is an observational macro-panel analysis. Alternative estimators can address particular statistical problems, including cross-sectional dependence and endogeneity assumptions, but they do not turn the relationship between ageing and expenditure into a clean causal estimate. Country-level expenditure data also conceal large differences in who receives public spending and what governments purchase. + +Read it for the fiscal question it raises rather than a single coefficient: how should health budgets change when population ageing begins before older health burdens have disappeared? + +## ONE NUMBER + +### 25.7% + +The increase in the World Bank’s energy price index in **September 2026**. + +Crude oil rose 21.2 per cent during the month and natural gas 15.3 per cent. + +[World Bank commodity markets data](https://www.worldbank.org/en/research/commodity-markets?utm_source=chatgpt.com) + +For South Asia, persistence matters more than the September spike itself. A short-lived increase produces one set of fiscal and inflation effects. Several months of high energy prices produce another. + +## POLICY WATCH + +### SDG budgeting is moving into the machinery of state finance + +NITI Aayog, the Government of West Bengal and UNDP held a national workshop on **SDG budgeting** in Kolkata on 2 October under the State Support Mission. + +[NITI Aayog](https://www.niti.gov.in/?utm_source=chatgpt.com) + +The idea is straightforward: connect government expenditure with measurable Sustainable Development Goal outcomes rather than treating SDG reporting as a separate planning exercise. + +The difficult part is attribution. A budget line can be tagged to an SDG with relative ease. Establishing whether the expenditure changed an outcome requires programme data, plausible indicators, expenditure tracking and, where the question warrants it, evaluation. + +This distinction matters because budget tagging can create an appearance of measurement without producing evidence about spending effectiveness. The next step should therefore be to link financial classifications to administrative and outcome data that allow states to ask what each rupee purchased. + +[Subscribe now](https://varna.substack.com/subscribe?) + +## EDITORIAL NOTE + +Today’s stories share a problem that public policy repeatedly encounters: **the headline indicator often arrives before the useful one**. + +Pakistan can announce 24 million people needing mental-health support. Sindh can identify one million intended beneficiaries of a climate project. Nepal can count 33 pregnant women transferred after floods. Governments can classify expenditure against the SDGs. + +Each number establishes something. None tells us whether a public system worked. + +The harder measures come later: treatment received, risk reduced, care interrupted, outcomes changed, money converted into services. + +That gap between administrative counting and outcome measurement is where a surprising amount of development policy disappears. + +### A question for readers + +For large climate-adaptation programmes such as Sindh’s new coastal project, what should count as a beneficiary: residence inside the project area, receipt of an intervention, demonstrated reduction in exposure, or a measurable improvement in household resilience? + +If you know of adaptation programmes that have handled this denominator problem well, send them my way. It is a measurement question worth examining in Methods Monday. + +## SOURCES + +World Bank, Commodity Markets data, updated 2 October 2026. + +Pakistan National Mental Health Policy launch, reported by *Dawn*, 2 October 2026. + +Sindh Coastal Resilience Project launch, reported by *Dawn*, 2 October 2026. + +Central Bank of Sri Lanka, *Sri Lanka’s Inflation Target Is Set at 5% for the Next Three Years*, 2 October 2026. + +Arjun Poudel, *Floods Drive 33 Pregnant Women to Kathmandu for Childbirth*, *Kathmandu Post*, 2 October 2026. + +Jalandhar Pradhan, Soumyanayani Mahali and Tapan Swain, *Does Aging Matter? Mapping the Government Health Expenditure in South Asian Countries*, *BMC Health Services Research*, published 1 October 2026. + +NITI Aayog, National Workshop on SDG Budgeting, 2 October 2026. + +[Share](https://varna.substack.com/p/research-rundown-daily-3-october?utm_source=substack&utm_medium=email&utm_content=share&action=share) diff --git a/archive/2026-10-16-methods-monday-5-october-2026.md b/archive/2026-10-16-methods-monday-5-october-2026.md new file mode 100644 index 0000000..a49f053 --- /dev/null +++ b/archive/2026-10-16-methods-monday-5-october-2026.md @@ -0,0 +1,117 @@ +--- +title: "Methods Monday | 5 October 2026" +date: 2026-10-16 +type: other +url: https://varna.substack.com/p/methods-monday-5-october-2026 +subtitle: "Question order, video interviews, local estimates and financial-health indicators: five developments worth examining before the next survey. " +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Methods Monday | 5 October 2026 + +*2026-10-16. Published at [https://varna.substack.com/p/methods-monday-5-october-2026](https://varna.substack.com/p/methods-monday-5-october-2026). This file is a generated copy; the Substack post is the original.* + +[![image](https://substackcdn.com/image/fetch/$s_!Bq5P!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb3fae16e-3e12-4059-b6f2-e4834b9c824a_1536x1024.png)](https://substackcdn.com/image/fetch/$s_!Bq5P!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb3fae16e-3e12-4059-b6f2-e4834b9c824a_1536x1024.png) +*Research Rundown Methods Monday - 5 October, 2026* + +A questionnaire can record an answer accurately and still measure the wrong thing. A change in question order may alter what respondents think they are being asked. A different interview mode may change what they are willing to disclose. A statistical model may produce precise local estimates while making assumptions that are difficult to defend. + +*Three recent studies examine how question order changes factual responses, whether video interviews produce comparable data, and how statistical models can improve estimates for small populations.* + +## Question order, interview modes and the problem of small-area estimates + +Researchers devote considerable attention to sampling, estimation and statistical inference, but some consequential decisions are made much earlier, when questionnaires are designed and data collection procedures are selected. Three papers published in the *Journal of Survey Statistics and Methodology* on 29 September examine different parts of this process. Their findings are particularly relevant to development researchers working with repeated surveys, mixed-mode data collection and small geographical populations. + +The first paper demonstrates how the order of two apparently straightforward questions can influence factual responses. The second tests the measurement consequences of replacing face-to-face interviews with video calls. The third addresses a different problem: obtaining reliable estimates for geographical areas where survey samples are too small to support precise direct estimation. + +Together, they provide useful evidence about decisions researchers often make before considering their implications for measurement quality. + +## 1. Even factual questions are sensitive to question order. + +Daniel Goldstein's paper, *Room for Error: Identifying Question Order Effects in Factual Measures When Meaning or Context Shifts*, examines how respondents interpret questions about the number of rooms and bedrooms in their homes. + +Using data from three surveys conducted in New York City, Goldstein investigates differences in reported housing characteristics when respondents encounter these questions in different sequences. In surveys where respondents were asked about total rooms before bedrooms, interview records indicated greater confusion. When the bedroom question appeared first, or the two questions were presented closer together, respondents tended to report a larger number of total rooms. + +Interviewer feedback suggested a possible explanation. Some respondents interpreted a question about rooms as asking about bedrooms, apparently because bedroom counts were a more familiar way of describing housing size. Asking about bedrooms first helped establish the distinction between the two concepts. + +The finding matters because the questions concern characteristics that might ordinarily be regarded as objective. Housing size does not change when a questionnaire changes. Its reported value can. + +There are implications for longitudinal surveys, particularly where researchers revise instruments between waves. A change in reported housing characteristics may reflect a change in the measurement process rather than an actual change in living conditions. Researchers examining time trends therefore need to consider whether modifications to question order have introduced discontinuities in the resulting series. + +The relevance extends to Indian household surveys. Questions about household composition, employment, dwelling characteristics and access to public services frequently require respondents to distinguish between related concepts. The appropriate sequence may depend on which categories respondents already understand and use in everyday conversation. + +Goldstein's evidence comes from comparisons across survey implementations and associated interview records. The study provides a persuasive account of how question order can influence comprehension, although other differences between survey implementations must be considered when interpreting the magnitude of the effects. + +I would recommend this paper to researchers designing panel questionnaires or revising established instruments. It is a useful reminder that preserving questionnaire wording alone does not guarantee measurement comparability. + +Goldstein, Daniel. 2026. “Room for Error: Identifying Question Order Effects in Factual Measures When Meaning or Context Shifts.” *Journal of Survey Statistics and Methodology*, published 29 September 2026. + +[Read the paper](https://doi.org/10.1093/jssam/smag032). + +## 2. Video interviews produce results similar to face-to-face interviews, with some qualifications. + +Marc Asensio, Matt Brown, Richard Silverwood, Tim Hanson and Gabriele Durrant examine whether video interviewing offers a credible alternative to conventional face-to-face data collection. + +Their study, *Assessing the Impact of Video Interviewing on Measurement Quality: Evidence from an Experimental Study on Mode Effects*, randomly assigned 1,692 respondents aged between 20 and 40 to video, web or face-to-face interviewing. Participants came from a nonprobability convenience sample. + +The authors examine two dimensions of measurement quality: item nonresponse and the distribution of answers across interview modes. + +Video interviewing produced slightly lower item nonresponse than face-to-face interviewing, although the differences were very small. Responses obtained through the two interviewer-administered modes were also broadly similar. + +The more substantial differences appeared between video interviews and self-administered web surveys. These were particularly apparent for questions concerning sensitive subjects, including mental wellbeing and financial difficulties. + +This distinction deserves attention. Video interviewing retains the presence of an interviewer, even when that interaction occurs remotely. Respondents may therefore experience some of the same pressures to provide socially acceptable answers that arise during face-to-face interviews. + +The experiment supports the possibility of using video interviewing as an alternative mode for some types of surveys. However, its design places limits on how widely the findings can be applied. Random allocation identifies mode differences among the recruited participants, while the convenience sample restricts claims about the wider population. The experiment also examines a relatively narrow age group. + +For development research in India, the practical implications depend on the population being surveyed. Video interviewing requires access to suitable devices, stable connectivity and sufficient digital familiarity. These requirements can introduce coverage and participation problems that a measurement experiment among digitally accessible respondents may fail to capture. + +There are also questions about interview privacy. A respondent answering questions about household finances, gender relations or personal experiences may have less control over who can hear the conversation during a video call. + +The paper is particularly useful for research teams planning mixed-mode surveys. It suggests that video and face-to-face interviews can generate comparable responses under some conditions. However, researchers would still need to assess selection into interview modes, item sensitivity and the conditions under which respondents participate. + +The distinction between measurement quality and population coverage is particularly important when considering whether remote interviewing can replace household visits. + +Asensio, Marc, Matt Brown, Richard Silverwood, Tim Hanson, and Gabriele Durrant. 2026. “Assessing the Impact of Video Interviewing on Measurement Quality: Evidence from an Experimental Study on Mode Effects.” *Journal of Survey Statistics and Methodology*, published 29 September 2026. + +[Read the paper](https://doi.org/10.1093/jssam/smag024). + +## 3. Small-area estimation models need to account for differences between areas. + +Many development surveys are designed to produce estimates at national or state level, although policymakers frequently require information for districts, blocks or smaller administrative units. Direct estimates for these populations can be unreliable when the number of observations is small. + +Small-area estimation addresses this problem by combining survey information with auxiliary variables through statistical models. A common approach uses area-level models that assume a shared relationship between the outcome and its predictors across geographical areas. + +Xin Wang and Konnor Payne examine what happens when this assumption is too restrictive. + +Their paper, *Model-Based Estimators Using Area-Level Models With Clustered Effects*, proposes an estimator that allows regression coefficients to differ across groups of areas. Rather than requiring researchers to specify these groups in advance, the method identifies clusters through a penalised estimation procedure. + +The basic idea is that different geographical areas may share similar relationships between an outcome and its predictors without all areas following the same model. + +The authors evaluate the estimator through simulation studies and an application involving forest inventory data from 168 counties in the western United States. Their empirical example estimates mean tree carbon per acre using forest inventory measurements and remotely sensed information about vegetation. + +The proposed method identifies four groups of counties with different regression coefficients. In the application, allowing these differences reduces the coefficient of variation of estimates relative to conventional models with common regression slopes. + +The method has potential applications in development statistics. Relationships between household characteristics and poverty, nutrition, employment or service access may vary across districts. An auxiliary variable that predicts an outcome well in one group of districts may perform differently elsewhere. + +This is particularly relevant when administrative records, census variables or geospatial indicators are used to generate estimates below the geographical level supported by a survey's original design. + +There are limitations. The estimator assumes that areas can be grouped according to shared regression coefficients, and its performance depends on how well that structure represents the data. The authors acknowledge computational difficulties when the number of areas becomes large. Their framework also does not directly accommodate areas without observed survey estimates unless further information is available. + +The practical test for Indian applications would involve comparing the method with existing small-area estimators using survey and auxiliary datasets where direct estimates are sufficiently reliable for validation. Model-based precision alone would provide an incomplete assessment, particularly where auxiliary data quality differs systematically across geographical areas. + +I would recommend this paper to researchers working with district-level indicators, poverty mapping or the integration of survey and administrative data. It offers a useful methodological extension, although applications to development indicators will need empirical validation. + +Wang, Xin, and Konnor Payne. 2026. “Model-Based Estimators Using Area-Level Models With Clustered Effects.” *Journal of Survey Statistics and Methodology*, published 29 September 2026. + +[Read the paper](https://doi.org/10.1093/jssam/smag033). + +## What I would take from these papers + +Goldstein's paper is the most immediately useful for researchers developing household survey instruments. It identifies a measurement problem that can arise even when questions appear unambiguous to the people who designed them. The implications for questionnaire pretesting and longitudinal comparability are direct. + +The video-interviewing experiment provides evidence for decisions about survey administration, particularly when fieldwork costs or access constraints make conventional interviewing difficult. Its findings should be considered alongside evidence on coverage, participation and respondent privacy. + +Wang and Payne's contribution is more technical. Their estimator addresses heterogeneity across small areas, which is relevant to the increasing demand for district-level and subdistrict-level statistics. Its practical value will depend on the availability and quality of auxiliary information and on whether the assumed clusters provide a credible description of the populations being studied. + +I would begin with Goldstein's paper because questionnaire sequencing is a design decision that nearly every survey researcher encounters. It also offers a fairly inexpensive opportunity for empirical testing. A small randomised experiment during piloting could establish whether alternative question sequences materially change responses before the final instrument enters the field. diff --git a/archive/2026-10-16-research-rundown-daily-6-october.md b/archive/2026-10-16-research-rundown-daily-6-october.md new file mode 100644 index 0000000..fa10a5b --- /dev/null +++ b/archive/2026-10-16-research-rundown-daily-6-october.md @@ -0,0 +1,173 @@ +--- +title: "Research Rundown | Daily | 6 October 2026" +date: 2026-10-16 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-6-october +subtitle: "South Asia's growth outlook improves, Bangladesh's poverty and banking indicators deteriorate, Sri Lanka's recovery remains uneven, and India proposes a new way to measure services." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown | Daily | 6 October 2026 + +*2026-10-16. Published at [https://varna.substack.com/p/research-rundown-daily-6-october](https://varna.substack.com/p/research-rundown-daily-6-october). This file is a generated copy; the Substack post is the original.* + +[![image](https://substackcdn.com/image/fetch/$s_!J4Ip!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5163784b-74ec-41d6-92d2-ec61bb1c82d7_1672x941.png)](https://substackcdn.com/image/fetch/$s_!J4Ip!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5163784b-74ec-41d6-92d2-ec61bb1c82d7_1672x941.png) +*Research Rundown Daily 6 October 2026* + +The World Bank released its latest South Asia economic outlook on 6 October, alongside country updates that reveal substantial differences in economic performance across the region. India also announced changes to its statistical system, while new administrative figures raised familiar questions about how public healthcare access should be measured. + +## FIVE THINGS WORTH KNOWING + +### 1. South Asia is expected to grow by 6.9 per cent, although the regional average conceals considerable differences. + +The World Bank's October 2026 *South Asia Economic Update*, titled *Adopting AI for Growth*, projects regional economic growth of 6.9 per cent in 2026, followed by 6.7 per cent in 2027. Strong domestic demand, remittances and recent economic reforms have supported growth despite higher energy prices and uncertainty in international markets. + +The report gives particular attention to artificial intelligence as a possible source of productivity growth. Approximately 23 per cent of Indian firms report using AI, compared with 43 per cent of firms in the United States. The Bank identifies opportunities in global value chains and public services, including agriculture, health and education. + +The adoption figures require careful interpretation. Reported use tells us relatively little about the intensity of use, changes in productivity or consequences for employment. A firm that uses an AI application occasionally is quite different from one that has reorganised its production processes around the technology. + +The immediate development question concerns the distribution of benefits. Smaller firms often lack the digital infrastructure, skilled workers and financing needed to adopt new technologies. Productivity gains concentrated among a relatively small number of larger firms would have different consequences for wages and employment from gains distributed across enterprises. + +The report's emphasis on locally adapted applications is sensible, particularly where connectivity, language and administrative capacity remain constraints. Their effectiveness will need to be assessed through actual service outcomes and labour-market data. + +[Source: World Bank, South Asia Economic Update, 6 October 2026](https://www.worldbank.org/en/news/press-release/2026/10/06/south-asia-s-strong-growth-defies-global-headwinds) + +### 2. Bangladesh's slowdown is now visible in poverty, employment and banking indicators. + +The World Bank's *Bangladesh Development Update*, released on 6 October, projects GDP growth of 3.4 per cent in both FY2026 and FY2027. This is substantially below the regional projection, although the fiscal-year and calendar-year estimates refer to different reporting periods. + +The report identifies weaknesses in the banking sector, limited domestic revenue collection, energy constraints and sluggish private investment as continuing obstacles to recovery. + +Its distributional findings deserve particular attention. The Bank estimates that approximately 2.1 million additional people were living in poverty in FY2026 compared with the preceding year. Employment creation has stalled, and women have experienced job losses. + +Banking-sector distress has also worsened. The reported non-performing loan ratio increased from 30.6 per cent in December 2025 to 33.2 per cent in June 2026. Domestic revenue collection stood at approximately 8.3 per cent of GDP, leaving limited fiscal room for social expenditure and investment. + +The social protection findings are particularly concerning. According to the report, approximately half of the poorest households remain outside social protection programmes. The government is developing a Dynamic Social Registry to improve beneficiary identification and enrolment. + +Better targeting could reduce exclusion, although a registry cannot compensate for insufficient programme financing or weak delivery systems. Its performance should be judged by changes in coverage, exclusion errors, timeliness of transfers and the adequacy of benefits. + +Bangladesh's current difficulties demonstrate how financial-sector problems, employment insecurity and limited fiscal capacity can reinforce one another. The consequences extend well beyond the national growth rate. + +[Source: World Bank, Bangladesh Development Update, 6 October 2026](https://www.worldbank.org/en/news/press-release/2026/10/06/bangladesh-needs-urgent-financial-energy-and-revenue-reforms-to-restore-growth-cre) + +### 3. Sri Lanka's economy has recovered its pre-crisis output, while poverty remains elevated. + +Sri Lanka's economic recovery has reached an important milestone. According to the World Bank's *Sri Lanka Development Update*, released on 6 October, real GDP has returned to its 2018 level following twelve consecutive quarters of expansion. + +The Bank projects growth of 4.4 per cent in 2026 and 4.2 per cent in 2027. Industry and services have contributed to the recovery, while fiscal performance has improved. + +Household welfare has recovered more slowly. The report places poverty at 16.9 per cent, substantially above its pre-crisis level. Employment and household incomes continue to lag behind aggregate economic activity. + +This distinction matters for assessing economic recovery. Returning to an earlier level of national output does not establish that households have regained their purchasing power or that employment opportunities have recovered across occupations and regions. + +The report also identifies agribusiness as a possible source of future employment and investment. Primary agriculture accounts for approximately 8 per cent of GDP, while the wider agrifood system contributes around one-sixth of output and more than 40 per cent of employment. + +Agricultural productivity therefore has consequences for a much larger workforce than the national accounts category for primary agriculture might suggest. + +Sri Lanka's recovery needs to be assessed through household incomes, employment quality, real wages and poverty transitions alongside national output. These measures will establish whether economic stabilisation is translating into improvements in living standards. + +[Source: World Bank, Sri Lanka Development Update, 6 October 2026](https://www.worldbank.org/en/news/press-release/2026/10/02/sri-lanka-s-economy-reaches-pre-crisis-levels-but-recovery-uneven) + +### 4. India is trying to measure a larger share of its services economy every month. + +The Ministry of Statistics and Programme Implementation released an approach paper on 6 October proposing to expand the coverage of India's Index of Services Production. + +The experimental monthly index currently covers 19 broad service subsectors representing approximately 60 per cent of services-sector gross value added. It relies substantially on administrative records and information derived from GST transactions. + +The proposed expansion would incorporate education, human health and residential care, and public administration and defence. Together, these activities represent approximately 21 per cent of services-sector GVA. The ministry estimates that the expanded index would cover approximately 78.4 per cent of services GVA. + +This is an important statistical development because services account for more than half of India's gross value added, while high-frequency indicators have traditionally provided much better coverage of industrial production. + +The methodological problems are substantial. Services delivered outside conventional market transactions present particular difficulties for measuring changes in real output. In education, for example, the number of students enrolled or classes delivered provides information about activity, while educational outcomes require separate measurement. + +Health services present similar difficulties. Consultations, admissions and procedures measure different aspects of activity, and changes in their composition can complicate comparisons over time. + +The usefulness of the proposed index will depend on the coverage of informal and non-market services, the construction of volume indicators and the treatment of changes in service quality. + +MoSPI has invited comments on the approach paper until 16 October 2026. The consultation provides an opportunity for researchers working on national accounts, public services and administrative data to examine the proposed methods before they become part of regular statistical reporting. + +[Source: Ministry of Statistics and Programme Implementation, 6 October 2026](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2319661&lang=1®=3) + +### 5. India's eSanjeevani platform has crossed 500 million consultations. The next question concerns continuity of care. + +The Ministry of Health and Family Welfare announced on 6 October that eSanjeevani, India's national telemedicine service, had recorded more than 50 crore consultations. + +The cumulative figure reached 500,092,730 on 15 September 2026. According to the ministry, women accounted for 57.14 per cent of utilisation, while people aged 46 years and above accounted for 40.15 per cent. + +These figures indicate considerable use of digital consultations within the public healthcare system. They also raise questions about what administrative utilisation statistics can establish. + +The number of consultations is different from the number of patients served. Repeat consultations may reflect appropriate follow-up care, continuing health needs or difficulties obtaining treatment elsewhere. + +A fuller assessment would examine the proportion of consultations resulting in completed treatment, referrals requiring further care, medicine availability, continuity of care and patient outcomes. + +There is also a geographical question. National totals may conceal differences in the availability of telemedicine services across districts and population groups. The distribution of consultations across socioeconomic categories would be especially useful for assessing equity. + +The reported scale of eSanjeevani makes these questions worth examining systematically. Administrative data can provide a starting point, although claims about improvements in health outcomes require additional evidence. + +[Source: Ministry of Health and Family Welfare, 6 October 2026](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2319427&lang=1®=3) + +## ONE PAPER WORTH READING + +### Does migration increase the likelihood of experiencing discrimination in later life? + +Vasim Ahamad, Rahul Rahul, Ram B. Bhagat and T. V. Sekher examine the association between internal migration and perceived discrimination among Indians aged 45 and above in a paper published in *BMC Public Health* on 5 October 2026. + +The authors analyse 64,711 respondents from the first wave of the Longitudinal Ageing Study in India, conducted during 2017–2018. They classify migration using respondents' place of last residence and measure perceived discrimination through an everyday discrimination scale. + +Approximately 57 per cent of respondents were classified as internal migrants. After adjusting for observed socioeconomic characteristics, migrants had 1.38 times the odds of reporting discrimination compared with non-migrants, with a 95 per cent confidence interval of 1.28–1.48. The association was stronger among men, interstate migrants and recent migrants. + +The paper offers useful evidence on an understudied dimension of ageing in India. Its principal limitation concerns causal interpretation. Cross-sectional associations cannot establish whether migration caused discrimination, and the migrant category captures movement over the life course rather than a single type of migration experience. The subjective discrimination measure also requires careful interpretation. + +The findings justify closer examination of migration histories, gender, social identity and access to services among older adults. Longitudinal analysis would help establish how these relationships change with duration of residence. + +[Read the open-access paper](https://doi.org/10.1186/s12889-026-29800-0) + +## ONE NUMBER + +### 33.2 per cent + +This was Bangladesh's reported non-performing loan ratio in June 2026, according to the World Bank's latest development update. The corresponding figure in December 2025 was 30.6 per cent. + +The increase matters because bank balance-sheet problems can restrict credit, discourage investment and impose additional demands on financial supervision and public resources. + +The ratio measures the share of loans classified as non-performing. It does not directly measure losses ultimately incurred by banks, which also depend on recovery, provisioning and collateral values. + +For Bangladesh, changes in this indicator should be read alongside trends in private credit, investment and employment. + +## POLICY WATCH + +### NAVYA 2.0 will expand vocational training to 100 Aspirational Districts. + +The Ministry of Skill Development and Entrepreneurship and the Ministry of Women and Child Development held a preparatory orientation for NAVYA 2.0 on 6 October. + +The programme targets young women and adolescent girls aged 16–25 and is intended to operate across 100 Aspirational Districts in 24 states and Union Territories. Its predecessor was implemented in 27 districts across 19 states and Union Territories. + +The ministries also released standard operating procedures for the expanded programme. + +Its stated objective is to connect vocational training with apprenticeships, entrepreneurship and livelihood opportunities. The important distinction is between participation in a training programme and successful entry into paid work. + +Programme monitoring should therefore record completion, placement, earnings, job retention and changes in employment status. These outcomes should be disaggregated by age, educational attainment, social group and location. + +The earlier pilot also offers an opportunity for learning. Evidence on which training courses produced sustained employment, which participants completed them and why others dropped out would be useful before expansion. + +The 6 October announcement concerns preparations for implementation. Actual coverage and results will have to be assessed as the programme is rolled out. + +[Source: Ministry of Skill Development and Entrepreneurship, 6 October 2026](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2320163&lang=2®=48) + +## EDITORIAL NOTE + +The economic reports released today provide a useful illustration of why growth statistics need to be read alongside distributional indicators. South Asia's projected growth rate is high, while Bangladesh faces worsening poverty and Sri Lanka continues to struggle with the household consequences of its economic crisis. + +India's announcements present a related measurement problem. A larger services index could improve the timeliness of economic statistics, and eSanjeevani's administrative records provide evidence of substantial service use. Their value for policy will depend on what they capture, whom they exclude and how they relate to outcomes. + +Better measurement begins with identifying the question each indicator can answer. + +## SOURCES + +1. World Bank. 2026. [South Asia Economic Update: Adopting AI for Growth](https://www.worldbank.org/en/news/press-release/2026/10/06/south-asia-s-strong-growth-defies-global-headwinds). Released 6 October. +2. World Bank. 2026. [Bangladesh Development Update](https://www.worldbank.org/en/news/press-release/2026/10/06/bangladesh-needs-urgent-financial-energy-and-revenue-reforms-to-restore-growth-cre). Released 6 October. +3. World Bank. 2026. [Sri Lanka Development Update: From Recovery to Transformation](https://www.worldbank.org/en/news/press-release/2026/10/02/sri-lanka-s-economy-reaches-pre-crisis-levels-but-recovery-uneven). Released 6 October. +4. Ministry of Statistics and Programme Implementation. 2026. [Approach Paper on Expanding the Scope and Coverage for the Index of Services Production](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2319661&lang=1®=3). Released 6 October. +5. Ministry of Health and Family Welfare. 2026. [eSanjeevani Teleconsultation Statistics](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2319427&lang=1®=3). Released 6 October. +6. Ahamad, Vasim, Rahul Rahul, Ram B. Bhagat, and T. V. Sekher. 2026. [The Influence of Migration Status on Perceived Discrimination Among Older Adults in India: A Nationwide Cross-Sectional Study](https://doi.org/10.1186/s12889-026-29800-0). *BMC Public Health*. Published 5 October. +7. Ministry of Skill Development and Entrepreneurship. 2026. [Preparations for NAVYA 2.0](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2320163&lang=2®=48). Released 6 October. diff --git a/archive/2026-10-16-research-rundown-daily-7-october.md b/archive/2026-10-16-research-rundown-daily-7-october.md new file mode 100644 index 0000000..f44eaf5 --- /dev/null +++ b/archive/2026-10-16-research-rundown-daily-7-october.md @@ -0,0 +1,237 @@ +--- +title: "Research Rundown | Daily | 7 October 2026" +date: 2026-10-16 +type: daily +url: https://varna.substack.com/p/research-rundown-daily-7-october +subtitle: "India prepares to account for its natural resources, a national travel survey reveals inequalities in mobility, and new reporting requirements may make financial inclusion easier to examine." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown | Daily | 7 October 2026 + +*2026-10-16. Published at [https://varna.substack.com/p/research-rundown-daily-7-october](https://varna.substack.com/p/research-rundown-daily-7-october). This file is a generated copy; the Substack post is the original.* + +[![image](https://substackcdn.com/image/fetch/$s_!Knq7!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F14814adf-3f18-4146-9cc7-5771ab6ba6bd_1672x941.png)](https://substackcdn.com/image/fetch/$s_!Knq7!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F14814adf-3f18-4146-9cc7-5771ab6ba6bd_1672x941.png) + +India’s statistical system is beginning to account for resources and activities that conventional economic indicators have historically captured inadequately. On 7 October, the Ministry of Statistics and Programme Implementation released a new strategy for environmental-economic accounting covering 2026–2030. A national household travel survey published two days earlier provides detailed information about commuting patterns, including substantial differences between men and women. Developments in banking oversight, healthcare regulation and rural water supply raise further questions about the quality of administrative information and its usefulness for public policy. + +## FIVE THINGS WORTH KNOWING + +### 1. India has released a new environmental-economic accounting strategy for 2026–2030. + +The Ministry of Statistics and Programme Implementation (MoSPI) released its *Strategy for Environmental Economic Accounts in India 2026–2030* on 7 October. The document builds on work initiated under the ministry’s previous strategy for 2022–2026 and identifies priorities for expanding India’s environmental accounts. + +The proposed areas include forests, minerals, soil resources, land, water, carbon stocks, biodiversity and environmentally related economic activities. The ministry has already undertaken work on wetlands, ocean accounts, croplands, energy, water quality and pollination services. + +Environmental-economic accounting attempts to establish consistent statistical relationships between economic production and changes in natural resources. The United Nations System of Environmental-Economic Accounting (SEEA) provides the international framework for this work. + +Consider groundwater extraction. Agricultural production that depends on groundwater contributes to national income, while the depletion of the underlying resource presents a separate economic and environmental problem. Conventional measures of production provide limited information about the sustainability of the resource base supporting that activity. + +Environmental accounts can help describe these relationships through physical measures of resource stocks and flows, supplemented by monetary estimates where appropriate. Their construction requires information from several sources, including administrative records, field surveys, remote sensing and scientific assessments. + +The methodological difficulties are considerable. Environmental data differ in geographical coverage, reporting periods, classification systems and measurement procedures. Measuring changes in the condition of an ecosystem also requires decisions about which characteristics should be included and how those characteristics should be interpreted. + +The new strategy identifies data gaps and proposes further work to develop environmental accounts at national and state levels. Its practical value will depend on the publication of comparable statistical series, transparent methods and sufficiently detailed information for policy analysis. + +For India, where groundwater depletion, land degradation and ecosystem loss have direct consequences for livelihoods and economic production, these accounts could become an important part of environmental policy assessment. + +[Source: MoSPI, 7 October 2026](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2320063&lang=2®=48) + +### 2. India’s first National Household Travel Survey reveals substantial gender differences in work-related travel. + +The National Statistical Office released findings from India’s first National Household Travel Survey on 5 October. Fieldwork was conducted between July 2025 and June 2026 and covered 492,023 households and 2,034,371 people across 14,274 villages and 10,377 urban blocks. + +The survey collected information about journeys undertaken for employment, education and other purposes, including travel distance, duration, expenditure and modes of transport. + +Among workers aged fifteen and above, 57.6 per cent reported work-related journeys exceeding one kilometre. The corresponding proportion was 61.9 per cent among male workers and 41.9 per cent among female workers, a difference of twenty percentage points. + +The survey also found that two-wheelers were the primary mode of transport for 42.6 per cent of workers who reported qualifying work-related journeys. Walking accounted for 28.3 per cent. + +These figures have different denominators. The gender-specific percentages describe the proportion of workers undertaking journeys exceeding one kilometre. The transport-mode percentages describe the distribution of primary transport modes among workers who made those journeys. + +This distinction matters for interpretation. + +The lower proportion of women reporting work-related travel may reflect differences in occupations, workplace locations, employment arrangements and household responsibilities. Restrictions on mobility and difficulties accessing transport may also contribute. + +The survey establishes the observed differences. Identifying their causes requires further analysis. + +The definition of travel introduces another consideration. Journeys shorter than one kilometre were excluded. This is relevant for employment undertaken near home and for short trips associated with domestic work and caregiving. + +Commuting expenditure provides additional evidence. Workers travelling to fixed workplaces spent an average of ₹785 per month. The corresponding expenditure was ₹1,044 in urban areas and ₹612 in rural areas. + +These amounts should be examined relative to earnings and household income. The financial burden of commuting can be substantial for low-paid workers even when their expenditure is below the national average. + +The survey provides an opportunity to study whether transport availability and affordability influence access to employment. Its greatest contribution may emerge when the underlying microdata are analysed alongside information about employment, income, household responsibilities and geographical location. + +[Source: MoSPI, National Household Travel Survey, 5 October 2026](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2319180) + +### 3. Banks will have to report Scheduled Caste beneficiaries separately under major financial schemes. + +On 7 October, the National Commission for Scheduled Castes announced that the Department of Financial Services had instructed State Level Bankers’ Committees and their Union Territory equivalents to report assistance provided to Scheduled Caste beneficiaries under major government financial schemes. + +The directions followed an intervention by the Commission on 24 September. It had observed that existing committee reports did not adequately record the proportion of Scheduled Caste families and beneficiaries receiving assistance. + +The schemes identified include Pradhan Mantri Mudra Yojana, Stand-Up India, the Prime Minister’s Employment Generation Programme, PM SVANidhi and other forms of MSME credit assistance. + +Under directions issued on 5 October, committees must include scheme-wise and bank-wise information on the number and percentage of Scheduled Caste beneficiaries in their periodic reviews. The Department of Financial Services has also requested an action-taken report by 9 October. + +This is a useful administrative change. Financial inclusion is frequently assessed through aggregate indicators such as the number of accounts opened, loans sanctioned and amounts disbursed. These measures provide information about programme activity, while their distribution across social groups is necessary for examining inequalities in access. + +The proposed reporting system could establish where assistance is concentrated and identify banks or schemes with relatively low Scheduled Caste participation. + +Its analytical value will depend on the information collected. + +The proportion of Scheduled Caste beneficiaries among approved borrowers, for example, should be interpreted alongside the composition of eligible applicants, application rates and loan approval rates. Loan amounts matter as well. Similar beneficiary shares can coexist with substantial differences in the average value of credit received. + +There is also a need to distinguish applicants, sanctioned borrowers and people who actually receive funds. + +The Commission’s intervention could improve monitoring of financial inclusion if these categories are clearly defined and consistently reported. Publication of comparable data would allow independent researchers to examine differences between banks, schemes and geographical areas. + +[Source: National Commission for Scheduled Castes, 7 October 2026](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2320345&lang=1®=3) + +### 4. India has introduced new registration and training requirements for allied healthcare professionals. + +The National Commission for Allied and Healthcare Professions announced regulatory measures on 7 October covering institutional assessment, professional registration and training standards. + +The measures include autonomous boards under State Allied and Healthcare Councils, a Central Assessment Committee for areas without functioning State Councils, and a national registration framework for allied healthcare professionals. + +The Registration of Allied and Healthcare Professionals Regulations, published in the Gazette on 15 September, establish State and Central Registers with unique identification numbers and publicly verifiable digital certificates. + +Registration will remain valid for five years, subject to renewal and continuing professional development requirements. + +The Commission has also introduced a competency-based curriculum for the Diploma in Medical Radiology and Imaging Technology. The two-year programme includes six months of clinical internship and will become mandatory for relevant institutions from the 2027–28 academic year. + +These developments concern a substantial section of India’s healthcare workforce. Allied healthcare professionals undertake diagnostic, rehabilitation and clinical support activities that affect the availability and quality of services in hospitals, primary healthcare facilities and diagnostic centres. + +A common registration framework could improve information about professional qualifications and workforce availability. Public registers may also help patients and employers verify credentials. + +The immediate implementation problem concerns professionals already working in the sector whose qualifications differ from the prescribed standards. The Commission has indicated that it is developing arrangements for provisional registration and bridge courses for eligible practitioners. + +The design of these arrangements will influence how existing workers enter the new regulatory system. + +For health workforce planning, registration data will be useful if they can be connected with geographical location, professional specialisation and employment status. The number of registered professionals provides only a partial account of the services actually available within a district. + +[Source: Ministry of Health and Family Welfare, 7 October 2026](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2320314&lang=1®=3) + +### 5. Jal Jeevan Mission 2.0 is turning its attention to the reliability of rural water services. + +The Department of Drinking Water and Sanitation held a regional review meeting on Jal Jeevan Mission 2.0 at IIM Bangalore on 7 October. + +The discussions concerned water-source sustainability, service reliability, water quality monitoring, operational expenditure and the management of rural drinking water infrastructure. + +According to the department, more than 15.96 crore rural households now have tap water connections, compared with 3.27 crore households when the mission began in August 2019. + +The expansion represents a substantial increase in reported household access. Its consequences for water security depend on the performance of the systems supplying those connections. + +Officials reported that pipeline leakages and breakages were leading causes of service interruptions identified through *Jal Seva Adhayan* assessments. Power failures and low-voltage conditions also affected water delivery. + +The department has consequently called for greater attention to maintenance, financial sustainability and water quality surveillance. + +These operational problems have direct implications for households. Interruptions in piped supply may require families to store water, purchase water or obtain it from alternative sources. The resulting costs can vary considerably across locations and income groups. + +For programme monitoring, the availability of a tap connection needs to be accompanied by information about service performance. + +Useful indicators include the number and duration of supply interruptions, water quality test results, repair times and the expenditure households incur when piped services fail. + +Household reports can provide an independent source of evidence on the actual availability of water. Comparing these responses with administrative records would help identify differences between infrastructure reported as functional and services experienced by users. + +Jal Jeevan Mission 2.0 offers an opportunity to improve the measurement of rural water supply. The challenge is to produce information that allows service reliability to be assessed consistently across districts and states. + +[Source: Ministry of Jal Shakti, 7 October 2026](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2320327&lang=2®=48) + +## ONE PAPER WORTH READING + +### Does community sanitation deprivation affect depression among older Indians? + +Anoop Jain, Rockli Kim and S. V. Subramanian examine the association between community-level toilet deprivation and depressive symptoms among older adults in India in a paper published in *GeroScience* on 7 October 2026. + +Their study uses data from the first wave of the Longitudinal Ageing Study in India, collected during 2017–2018. The analytical sample comprises 63,564 adults above the age of 45 living in 2,439 villages and urban wards. + +The authors examine whether the proportion of residents living in households without toilets within a community is associated with depressive symptoms and probable major depression. + +They estimate multilevel regression models that account for individual characteristics, household socioeconomic conditions, residential location and household toilet access. + +The findings indicate a substantial association between community sanitation deprivation and mental health. + +For one measure of depressive symptoms, respondents living in communities with the lowest levels of toilet coverage had an estimated odds ratio of 1.38, with a 95 per cent confidence interval of 1.08–1.76, relative to the model’s reference condition. + +The authors report broadly similar patterns using alternative measures of probable major depression. + +The analysis is interesting because community sanitation conditions may affect people through mechanisms extending beyond their own household facilities. + +Living in an area with widespread sanitation deprivation may involve exposure to environmental contamination, difficulties accessing alternative facilities and social stigma associated with the residential environment. + +The study examines whether community conditions modify the relationship between household toilet access and depression. Although descriptive patterns suggest higher depressive symptom prevalence in communities with lower toilet coverage, the estimated interaction terms were statistically insignificant. + +There are important limitations. + +The analysis is cross-sectional, and the data were collected in 2017–2018. Communities with poor sanitation may also experience other forms of deprivation associated with mental health. The regression adjustments cannot eliminate the possibility of unobserved confounding. + +The study also lacks information about whether households without private toilets have access to shared or public facilities, which could influence the interpretation of sanitation deprivation. + +These limitations restrict causal inference. The results nevertheless suggest that household-level sanitation indicators may provide an incomplete account of the environmental conditions associated with health. + +I would recommend this paper to researchers working on sanitation, ageing and social determinants of health. Its principal contribution lies in specifying community-level sanitation deprivation as an exposure and examining its relationship with mental health outcomes using a large national dataset. + +Future work using longitudinal observations could establish whether changes in community sanitation conditions precede changes in depressive symptoms. Qualitative research could also examine the proposed mechanisms involving stigma, environmental conditions and access to facilities. + +Jain, Anoop, Rockli Kim, and S. V. Subramanian. 2026. “Association Between Community Toilet Deprivation and Depression Among Older Adults in India: A Multilevel Analysis of the Longitudinal Aging Study in India from 2017–2018.” *GeroScience*. Published 7 October 2026. + +[Read the open-access paper](https://doi.org/10.1007/s11357-026-02537-9) + +## ONE NUMBER + +### 41.9 per cent of women workers reported journeys exceeding one kilometre for work. + +India’s National Household Travel Survey found that 41.9 per cent of female workers aged fifteen and above reported work-related journeys exceeding one kilometre. The corresponding proportion among male workers was 61.9 per cent. + +The difference was twenty percentage points. + +These estimates concern people classified as workers and measure participation in work-related travel above the survey’s distance threshold. + +They cannot establish how much of the observed difference arises from transport access, employment arrangements, household responsibilities or other factors. + +The findings nevertheless provide a useful basis for studying women’s geographical access to employment. + +A more detailed analysis would examine employment type, commuting distance, earnings and the availability of public transport. It would also need to account for workers whose employment takes place at home or within one kilometre of their residence. + +The national survey offers the possibility of investigating these relationships across rural and urban areas. + +## POLICY WATCH + +### Banks have until 9 October to report progress on Scheduled Caste financial inclusion data. + +The Department of Financial Services has asked State Level Bankers’ Committees to submit an initial action-taken report by 9 October concerning the availability of Scheduled Caste beneficiary information under major government financial schemes. + +The direction requires committees to identify existing data and report difficulties encountered during compilation. + +The National Commission for Scheduled Castes has also sought regular access to the resulting information. + +The immediate administrative task concerns consistency. Banks and implementing agencies will need common definitions of beneficiaries, loan sanctions and disbursements if their figures are to support comparisons. + +Public reporting would make the information more useful for independent assessment. It would allow researchers to examine differences across schemes, institutions and geographical areas, subject to the availability of appropriate denominators. + +The direction is an initial measure. Its value will depend on the quality, completeness and accessibility of the resulting data. + +## EDITORIAL NOTE + +Several developments in today’s edition concern the relationship between official statistics and the conditions people experience. + +India’s environmental-economic accounting strategy seeks to improve information about natural resources and economic activity. The National Household Travel Survey introduces new measures of mobility, including differences between men and women that deserve closer analysis. + +The announcements concerning banking, healthcare professionals and rural water supply show how administrative records can describe different parts of public service delivery. + +Their policy value depends on the questions being asked of the data. + +A register of healthcare professionals can inform workforce planning when qualifications and geographical distribution are recorded accurately. Information about bank beneficiaries can support analysis of social inequalities when loan values and eligible populations are available. Reports of tap connections become more informative when accompanied by reliable measures of water availability. + +The methodological questions concern definitions, coverage, accuracy and interpretation. Addressing them requires independent scrutiny of official statistics and the administrative systems from which they are produced. + +## SOURCES + +1. Ministry of Statistics and Programme Implementation. 2026. [Strategy for Environmental Economic Accounts in India 2026–2030](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2320063&lang=2®=48). Released 7 October 2026. +2. Ministry of Statistics and Programme Implementation. 2026. [National Household Travel Survey, July 2025–June 2026](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2319180). Released 5 October 2026. +3. National Commission for Scheduled Castes. 2026. [Mandatory Reporting of Scheduled Caste Beneficiaries under Key Financial Schemes](https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2320345&lang=1®=3). Released 7 October 2026. +4. Ministry of Health and Family Welfare. 2026. [National Commission for Allied and Healthcare Professions: Regulatory Measures](https://www.pib.gov.in/PressReleasePage.aspx?PRID=2320314&lang=1®=3). Released 7 October 2026. +5. Ministry of Jal Shakti. 2026. [Regional Review of Jal Jeevan Mission 2.0](https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2320327&lang=2®=48). Released 7 October 2026. +6. Jain, Anoop, Rockli Kim, and S. V. Subramanian. 2026. [Association Between Community Toilet Deprivation and Depression Among Older Adults in India](https://doi.org/10.1007/s11357-026-02537-9). *GeroScience*. Published 7 October 2026. diff --git a/archive/2026-10-17-good-reads-15-who-owns-intelligence.md b/archive/2026-10-17-good-reads-15-who-owns-intelligence.md new file mode 100644 index 0000000..fef6e52 --- /dev/null +++ b/archive/2026-10-17-good-reads-15-who-owns-intelligence.md @@ -0,0 +1,156 @@ +--- +title: "Good Reads #15 | Who owns intelligence? Arundhati Roy on AI and extraction" +date: 2026-10-17 +type: other +url: https://varna.substack.com/p/good-reads-15-who-owns-intelligence +subtitle: "Arundhati Roy's new essay asks who controls the knowledge, labour and resources behind artificial intelligence. Her political economy argument deserves serious attention, even where her predictions outrun the evidence." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Good Reads #15 | Who owns intelligence? Arundhati Roy on AI and extraction + +*2026-10-17. Published at [https://varna.substack.com/p/good-reads-15-who-owns-intelligence](https://varna.substack.com/p/good-reads-15-who-owns-intelligence). This file is a generated copy; the Substack post is the original.* + +Arundhati Roy begins her new essay with a small irritation that many writers will recognise. Artificial intelligence has made even the ordinary habits of writing suspect. A familiar turn of phrase, an unusual sentence or a particular punctuation mark may now be interpreted as evidence that a machine produced the text. For a novelist who has spent decades developing a distinctive voice, this is an unsettling prospect. + +By the end of *Apocalypse Now and Then*, published in EQUATOR on 15 October, Roy has moved considerably further. She believes artificial intelligence threatens the relationship between human beings and language, transfers the accumulated knowledge of civilisation into corporate hands, and may eventually make human labour and thought dispensable. + +It is an angry essay, filled with historical comparisons and unsettling predictions. Roy writes as a novelist and political critic, and she makes no effort to conceal her convictions. Some of her claims are speculative. Others concern economic and political arrangements that can already be examined through available evidence. + +I found the latter considerably more persuasive. + +## Who gets to own what we know? + +The most compelling part of Roy's essay concerns her own experience with Anthropic. She describes receiving a notice offering compensation for the use of her books in training a large language model. The books had reportedly been obtained from websites distributing unauthorised copies. + +The settlement is substantial. On 20 July 2026, a federal court in California granted final approval to the class-action settlement in *Bartz v. Anthropic*, concerning books obtained from unauthorised digital libraries. The litigation has also raised wider questions about the legal treatment of copyrighted works in AI training. + +Roy's response is personal, although the problem extends far beyond publishing. She has spent years producing books whose language, arguments and literary form belong to her work as a writer. A technology company has incorporated copies of that work into a commercial process that she neither initiated nor authorised. + +Copyright law provides one framework for addressing the dispute. The economic problem is considerably broader. + +Large AI systems depend on enormous quantities of material produced by writers, artists, software developers, scientists, researchers and public institutions. Some of that material is copyrighted, some is licensed for particular uses, and some belongs to the public domain. The legal conditions differ. So do the circumstances under which the material was originally produced. + +Research universities, for example, generate knowledge through a combination of public financing, academic labour, private funding and institutional resources. Their publications may circulate through commercial journals, open repositories and databases. AI companies can subsequently draw on these accumulated bodies of knowledge to develop systems sold to universities, governments and businesses. + +The result raises questions about the distribution of economic returns across institutions. Public expenditure may contribute to the creation of knowledge while commercial firms control the infrastructure through which it is subsequently processed and sold. + +The distinction between owning information and controlling access to services built around it deserves closer examination. AI firms are acquiring positions within the infrastructure of knowledge production. Their pricing, licensing arrangements and technical restrictions can influence who can use those services and on what terms. + +This gives Roy's argument a wider significance. Compensation for authors addresses one particular form of appropriation. Publicly financed knowledge, academic research and collective intellectual work present additional questions for which existing copyright arrangements provide only partial answers. + +Roy pushes the claim much further, suggesting that corporations may eventually control access to human knowledge itself. That outcome remains speculative. The existing concentration of computing capacity, proprietary models and commercial distribution systems is sufficient reason to examine the problem. + +## The corporation and the state + +Roy compares the expansion of AI corporations with the history of the East India Company. India's colonial experience makes the comparison particularly forceful. A commercial enterprise acquired political authority, exercised control over territory and resources, and became part of the machinery of imperial rule. + +Contemporary technology corporations operate under very different legal and political conditions. Their relationship with governments nevertheless deserves scrutiny. + +AI companies provide services to public institutions and develop systems used in administration, surveillance, defence and other government activities. Governments, in turn, influence the industry's development through procurement contracts, infrastructure investment, regulation and access to public resources. + +Roy is especially suspicious of technology executives who warn about existential risks while asking governments to regulate the industry. She suspects that large corporations may benefit from regulation that establishes their legitimacy while raising the costs faced by smaller competitors. + +There is a familiar political economy problem here. Firms with considerable financial and technical resources can influence the rules governing their industries. Compliance requirements may impose proportionately greater costs on new entrants. Public authorities can also become dependent on the companies whose conduct they are expected to oversee. + +The argument needs greater discrimination than Roy allows. Regulatory proposals differ in their objectives, obligations and enforcement arrangements. Competition law, data protection, procurement rules and independent technical audits can all restrict the discretion of established firms. Their effects depend on institutional design and implementation. + +The more immediate concern is the extent to which states retain the capacity to make independent decisions about systems on which essential public functions depend. + +For India, this includes the terms governing government contracts for cloud services, access to public datasets, the location and ownership of computing infrastructure, and the possibility of transferring services between providers. + +The concentration of technological capacity creates problems for public accountability when governments lack the personnel, information or alternative suppliers necessary to scrutinise procurement decisions. + +Roy's historical analogy makes these relationships visible, although the comparison cannot establish that contemporary AI companies will acquire powers equivalent to those exercised by colonial enterprises. + +The present distribution of commercial and public authority deserves investigation in its own right. + +## The workers and resources behind the machine + +Roy is particularly effective when she turns to the physical economy of artificial intelligence. Computers require electricity, data centres require land and cooling systems, and AI development depends on workers performing tasks distributed across several countries. + +These systems have measurable costs. + +The International Energy Agency estimated that data centres consumed approximately 415 terawatt-hours of electricity worldwide in 2024, representing about 1.5 per cent of global electricity consumption. Its 2025 *Energy and AI* report projected consumption of approximately 945 terawatt-hours by 2030 under its base-case scenario. + +The projection covers data centres generally, with AI accounting for a substantial part of the expected increase. Its uncertainty is considerable because future electricity demand depends on the pace of adoption, improvements in computing efficiency and constraints within energy systems. + +The national figures are useful, although the consequences of new facilities are often local. Data centres may create concentrated demands for electricity and water in places where infrastructure already serves competing users. + +Roy approaches this problem through an account of a rayon factory established along the Chaliyar River in Kerala during the state's early period of industrial development. The factory created industrial employment while, in her account, its operations damaged the river and affected people whose livelihoods depended on the surrounding environment. + +The historical details merit independent examination. The distributional problem she identifies is familiar to development economists. + +The financial accounts of an industrial enterprise may record the value of its output, investment and direct employment. Its activities can also impose environmental costs and affect employment in other sectors. Those costs may be borne by households and communities with limited influence over the original investment decision. + +The comparison is relevant to India's emerging data-centre industry. An adequate assessment would consider the employment created during construction and operation, the value of investment, the public concessions offered, and the demands placed on local infrastructure. + +Environmental assessment is particularly important where electricity and water supplies are constrained. The opportunity cost of allocating these resources to one activity depends on what other activities might have used them. + +Roy also draws attention to workers engaged in data collection, annotation and related activities. Some of this work takes place through subcontracting arrangements that make employment conditions and responsibility for labour protections difficult to establish. + +She anticipates that workers who help develop AI systems may eventually find some of their own tasks automated. The possibility deserves investigation, although her predictions about widespread displacement exceed what can currently be established. + +The International Labour Organization's 2025 assessment estimated that one-quarter of global employment was in occupations with some exposure to generative AI. Approximately 3.3 per cent fell into the highest exposure category. + +Occupational exposure measures the potential for AI systems to perform particular tasks. Actual employment consequences depend on the decisions of employers, the costs of adoption, changes in demand and the capacity of workers to move into other activities. + +Even where technology increases productivity, the allocation of the resulting gains remains a matter of economic institutions and bargaining power. + +Roy's concerns about subcontracted labour are therefore worth investigating independently of her predictions about the eventual disappearance of work. + +## Where Roy's argument becomes less convincing + +The essay makes several claims about the future of human intelligence and independent thought. Roy worries that reliance on AI-generated language will weaken people's capacity to reason, create and recognise authentic expression. + +Her concerns are understandable, particularly in education and academic writing. The widespread availability of generated text creates new problems for assessment, authorship and the evaluation of intellectual work. + +The evidence concerning long-term cognitive effects remains incomplete. Existing research examines different uses of AI under different conditions. It cannot yet establish that dependence on these systems will produce widespread or irreversible cognitive decline. + +The distinction between different forms of use is significant. A student who delegates an entire assignment to a machine may learn very little from the exercise. A researcher who uses software to compare alternative specifications while independently checking the assumptions and results is engaged in a different activity. + +The outcomes depend partly on the organisation of work and the demands placed on users. + +Roy's argument also gives limited attention to the potential benefits of AI applications in settings where professional services are scarce. Translation, accessibility tools and selected administrative applications may reduce the time or cost required to perform certain tasks. + +Claims about these benefits also require empirical testing. Evidence of faster task completion, for example, does not establish improvements in accuracy, equity or final outcomes. The appropriate assessment depends on the particular application. + +Roy's larger claim that artificial intelligence represents capitalism's final destination is difficult to evaluate as an empirical proposition. It expresses a political judgment about the concentration of ownership and the extraction of value from human activity. + +The judgment draws attention to existing arrangements. Its predictive force is less clear. + +## The question that remains + +I came away from Roy's essay thinking about the terms on which societies are allowing AI infrastructure to develop. + +Consider a public university that pays commercial providers for access to AI services trained partly on publicly financed research. Or a government that offers infrastructure concessions to attract data-centre investment without publishing a complete assessment of the environmental costs. Or workers who perform essential tasks in AI supply chains while remaining outside the employment protections available to directly hired staff. + +These examples concern arrangements that can be examined through contracts, expenditure records, labour surveys and environmental assessments. The distribution of costs and benefits is open to empirical investigation. + +Roy is more interested in the larger political question of who should control the technology. She is sceptical of proposals that compensate the public financially while leaving corporate ownership of the underlying systems intact. + +There is a substantial argument here, although the alternatives also require examination. Public ownership can create new problems of accountability. Regulation can fail through weak enforcement or the influence of established firms. National control can leave existing inequalities within a country undisturbed. + +The quality of the institutions governing technology will therefore influence its distributional consequences. + +I would recommend *Apocalypse Now and Then* for the questions Roy raises about knowledge, ownership and political power. The essay is uneven. Its predictions are sometimes far more confident than the evidence permits, and its historical comparisons need greater care. Her account of the material resources and human work behind AI provides a useful corrective to discussions that treat these systems as autonomous technological achievements. + +For researchers in development economics, the most productive response is to examine the arrangements through which public resources, labour and knowledge contribute to privately controlled systems, and to establish who receives the returns. + +Roy believes those arrangements are already producing a dangerous concentration of power. There is enough evidence of concentration to justify serious public scrutiny, even while its eventual consequences remain uncertain. + +--- + +### Original essay + +Arundhati Roy. [“Apocalypse Now and Then.”](https://www.equator.org/articles/arundhati-roy-artificial-intelligence) *EQUATOR*, 15 October 2026. + +### Further reading + +International Energy Agency. 2025. *[Energy and AI](https://www.iea.org/reports/energy-and-ai)*[.](https://www.iea.org/reports/energy-and-ai) + +International Labour Organization. 2025. *[Generative AI and Jobs: A Refined Global Index of Occupational Exposure](https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure)*[.](https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure) + +United States Copyright Office. 2025. *[Copyright and Artificial Intelligence, Part 3: Generative AI Training](https://copyright.gov/ai/)*[.](https://copyright.gov/ai/) + +[Bartz v. Anthropic, settlement information and court documents.](https://www.anthropiccopyrightsettlement.com/) diff --git a/archive/2026-10-17-research-rundown-friday-findings-c92.md b/archive/2026-10-17-research-rundown-friday-findings-c92.md new file mode 100644 index 0000000..e116d71 --- /dev/null +++ b/archive/2026-10-17-research-rundown-friday-findings-c92.md @@ -0,0 +1,253 @@ +--- +title: "Research Rundown | Friday Findings | 9 October 2026" +date: 2026-10-17 +type: edition +url: https://varna.substack.com/p/research-rundown-friday-findings-c92 +subtitle: "New evidence on rural employment, women's enterprises, digital learning, public health and environmental risk, with studies from India, Bangladesh, Nepal and beyond." +source: varna.substack.com, fetched by scripts/sync_substack.py +--- + +# Research Rundown | Friday Findings | 9 October 2026 + +*2026-10-17. Published at [https://varna.substack.com/p/research-rundown-friday-findings-c92](https://varna.substack.com/p/research-rundown-friday-findings-c92). This file is a generated copy; the Substack post is the original.* + +[![image](https://substackcdn.com/image/fetch/$s_!Xgo5!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbafb5b9f-3fb3-45fe-a465-2de4356213f0_1672x941.png)](https://substackcdn.com/image/fetch/$s_!Xgo5!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbafb5b9f-3fb3-45fe-a465-2de4356213f0_1672x941.png) +*Friday Findings October 9 2026* + +This week's papers examine several familiar assumptions in development policy. An increase in rural nonfarm employment may coexist with persistent labour-market segmentation. Training women entrepreneurs can improve their access to corporate procurement, although previous experience influences who secures contracts. Even brief delays on educational platforms appear to affect learning. + +Research from Nepal, Bangladesh and Manipur adds evidence about environmental exposure, food systems and the use of geographical information for public decisions. The selection also includes two studies with implications for public health and land-use policy. + +Eight papers are worth examining this week. + +## 1. Rural nonfarm employment in India is expanding, with unequal returns. + +Shadab Hashmi and Abdul Latheef K examine rural nonfarm employment using nationally representative microdata from India's Periodic Labour Force Survey (PLFS), 2021–22. Their paper, published in *Humanities and Social Sciences Communications* on 8 October, investigates wage differences between agricultural and non-agricultural employment and the characteristics associated with access to better-paying work. + +The authors combine a Mincerian wage regression with quantile regressions and a Heckman two-step selection model. This allows them to examine average wage differences, variation across the conditional wage distribution and potential selection into different forms of employment. + +They find a positive average nonfarm wage premium. The estimated premium declines towards the upper end of the wage distribution, however, and access to better-paying nonfarm work remains associated with education, technical training, formal employment and geographical location. Women, informal workers and socially disadvantaged groups experience persistent disadvantages. + +These findings complicate the interpretation of structural transformation. An increase in employment outside agriculture provides limited information about whether workers have moved into more productive or secure occupations. The distinction is especially relevant where workers enter casual construction, petty trade and other activities offering irregular earnings. + +The paper has methodological limitations. Its estimates come from cross-sectional data, and the Heckman correction depends on assumptions about selection and model specification. A statistically significant selection term does not establish why workers entered particular occupations. The estimated wage premium should therefore be interpreted as a conditional association. + +The authors' larger claim about changes in employment quality over time also requires evidence from comparable survey rounds. + +I would recommend this paper to researchers working on employment diversification, rural livelihoods and structural transformation. Its strongest contribution is the attention it gives to differences within nonfarm employment, particularly the unequal returns associated with education, gender and social position. + +**Paper:** Hashmi, Shadab, and Abdul Latheef K. 2026. “Rural Nonfarm Employment in India: Structural Transformation, Wage Determinants, and Labour Market Segmentation.” *Humanities and Social Sciences Communications*. Published 8 October. + +[Read the open-access paper](https://doi.org/10.1057/s41599-026-09283-5). + +## 2. Procurement training helps women-owned firms bid for contracts. Previous experience influences who wins. + +Rachel Cassidy, David McKenzie and Kalyan Neelamraju examine whether targeted procurement training can help women-owned and women-led enterprises enter corporate supply chains. + +Their World Bank working paper, released on 7 October, reports findings from a phased randomised controlled trial in Colombia. The programme provided procurement training through masterclasses, bootcamps and mentoring, with outcomes measured at baseline and approximately six and twelve months later. + +The intervention increased registration on corporate procurement portals by ten percentage points and the likelihood of submitting a bid by eight percentage points. + +The effects on winning contracts were concentrated among firms with previous procurement knowledge or experience. For these enterprises, the probability of winning a contract increased by approximately eight to nine percentage points, representing around 20 per cent of the control-group mean. + +The distinction between entering a procurement process and winning a contract is important. Registration and bid submission are intermediate outcomes. Contract acquisition depends on additional factors, including the firm's production capacity, commercial experience and ability to meet buyer requirements. + +The study also has an interesting methodological contribution. Participation occurred through several stages, and some enterprises assigned to the programme did not complete all its components. The authors use complier-weighted treatment effects to examine outcomes under these conditions, alongside intention-to-treat estimates and sensitivity analyses. + +This is relevant to enterprise-development evaluations where enrolment, attendance, completion and subsequent business activity form a sequence of decisions. + +The experiment provides credible causal evidence within the Colombian programme. Its results should be applied cautiously to other settings. Procurement practices, buyer behaviour and the characteristics of participating enterprises may differ substantially. + +For India, the findings are relevant to efforts to increase women's participation in government procurement and corporate supply chains. They suggest that training may improve entry into procurement systems, while the benefits of winning contracts could depend heavily on firms' existing capabilities. + +I would place this among the week's strongest papers because the research design permits causal interpretation and the outcomes extend beyond knowledge gained during training. + +**Paper:** Cassidy, Rachel, David McKenzie, and Kalyan N. Neelamraju. 2026. “Breaking into Supply Chains: Experimental Evidence on Procurement Training for Women-Owned SMEs.” World Bank Policy Research Working Paper 11481. Published 7 October. + +[Read the working paper](https://ideas.repec.org/p/wbk/wbrwps/11481.html). + +[Access the reproducibility materials](https://reproducibility.worldbank.org/catalog/653). + +## 3. A two-second delay can affect online learning. + +Asaf Mazar and colleagues examine a problem that educational technology evaluations frequently overlook: the effect of brief loading delays during online learning. + +Their paper, published in *Nature Human Behaviour* on 8 October, combines two longitudinal quasi-experimental field studies. + +The first uses observations from 26,707 American primary and secondary school students completing online mathematics assignments. The second examines 55,184 aspiring programmers, primarily in Latin America, using an online coding platform. + +Across both settings, loading delays of approximately one to two seconds were associated with poorer engagement. + +Students experiencing these delays completed fewer assignments, took longer to finish their work, became distracted more frequently and obtained lower scores. Aspiring programmers spent less time actively engaged, completed fewer lessons and were less likely to reach an important course milestone. + +The authors exploit variation in loading times experienced by the same users and test the findings across alternative specifications. This provides stronger evidence than a simple comparison between users with fast and slow internet connections, since those groups may differ in income, education and other characteristics. + +There are limits to the causal interpretation. Loading times were observed in functioning platforms under real-world conditions, and the studies rely substantially on within-person variation rather than random assignment of delays. Unobserved conditions that vary over time could still influence the estimates. + +The findings are particularly relevant to educational technology programmes in countries where internet access is uneven. + +Evaluations often record whether schools and students have connectivity, how frequently a platform is used and whether users complete assigned activities. The quality of the connection during individual learning tasks receives much less attention. + +This paper suggests that technical performance should be treated as a potential determinant of learning outcomes. Platform design, responsiveness and network quality may influence whether students remain engaged. + +I would recommend it to researchers evaluating digital learning interventions. It offers an empirically grounded reason to measure the actual user experience alongside access to devices and connectivity. + +**Paper:** Mazar, Asaf, Geoff Tomaino, Abubakir Siedahmed, and colleagues. 2026. “Microdelays Disrupt Online Learning.” *Nature Human Behaviour*. Published 8 October. + +[Read the paper](https://doi.org/10.1038/s41562-026-02598-y). + +## 4. Himalayan flood exposure is increasing, and the underlying building data need careful validation. + +C. Scott Watson and colleagues examine changes in the exposure of buildings and populations to potential flooding in Nepal and adjoining Himalayan catchments in India and China. + +Their paper, published in *Scientific Reports* on 8 October, combines satellite-derived building inventories, terrain information and modelled glacial lake outburst flood inundation areas. + +An important part of the study concerns the accuracy of the underlying geographical datasets. The authors compare five building-footprint products against manually identified reference data. Google Open Buildings Temporal 2.5D performed best, with a median F1 score of 0.68, followed by Google Open Buildings V3 at 0.62. + +These values indicate that substantial classification errors remain even in the better-performing datasets. Differences between products can materially affect estimates of the buildings exposed to hazards. + +Using the selected temporal dataset, the researchers estimate that building area within fifty metres of vertical elevation above nearby drainage channels increased by 46 per cent between 2016 and 2023. Building area within modelled glacial lake outburst flood inundation zones increased by 77 per cent. + +The authors estimate that approximately 293,800 people, representing 18 per cent of the population in their study-area catchments, lived within the specified fifty-metre vertical proximity zone in 2026. + +The meaning of exposure requires care. The height-above-nearest-drainage measure identifies geographical proximity in elevation to drainage channels. It does not represent an observed flood event or directly estimate the probability of flooding at each building. + +The population figures also rely on projections and are less certain than the building estimates. + +The methodological contribution is substantial. Disaster-risk assessments increasingly use remotely sensed settlement data, often without adequately examining their accuracy in mountainous terrain. Errors in building identification or positioning can alter exposure estimates considerably. + +This study demonstrates why validation should precede the use of such datasets for investment and disaster planning. + +For Himalayan states and districts, the research offers useful methods for examining settlement expansion in areas potentially exposed to floods. The next step would be to combine these data with infrastructure condition, household vulnerability and the consequences of previous flood events. + +**Paper:** Watson, C. Scott, Lauren D. Rawlins, Mohan B. Chand, and colleagues. 2026. “Increasing Exposure to Potential Flooding in High-Mountain Catchments.” *Scientific Reports*. Published 8 October. + +[Read the open-access paper](https://doi.org/10.1038/s41598-026-70268-0). + +## 5. Personalised risk messages increased influenza vaccination in three randomised trials. + +Gail Rosenbaum and colleagues examine whether personalised information about health risks can influence vaccination decisions. + +Their paper, published in *Nature Human Behaviour* on 5 October, reports three preregistered randomised field trials involving more than 90,000 unique patients in a United States healthcare system. + +The researchers used an existing machine-learning model to identify patients at elevated risk of influenza and related complications. Participants were randomly assigned to receive no message or one of several vaccination reminders. + +Some messages provided personalised information about elevated risk. Others disclosed that an algorithm had identified the patient's risk level or explained the factors contributing to that assessment. + +The personalised high-risk messages increased vaccination by approximately 1.1 to 1.4 percentage points compared with ordinary reminders. Relative to patients who received no message, the increases ranged from 1.7 to 3.5 percentage points. + +The researchers found little evidence that explicitly mentioning the algorithm or explaining its prediction produced additional improvements in vaccination uptake. + +These are modest absolute effects, although their public health significance depends partly on the size and characteristics of the population reached and the costs of delivering the messages. + +The randomised design supports causal interpretation of the messages' effects on recorded vaccination within the study population. The findings are less informative about how patients interpreted the messages or why particular versions worked. + +The trial also took place within a healthcare system capable of identifying high-risk patients through existing electronic records. Its applicability to settings with incomplete patient information requires further examination. + +For public health programmes in India, the study raises practical questions about targeting and communication. The effectiveness of personalised messages would depend on the accuracy of the underlying risk information, the ability to contact patients and the availability of vaccination services. + +The paper is worth reading for its experimental design and its separation of message content from the use of an algorithm to select recipients. + +**Paper:** Rosenbaum, Gail M., Amir Goren, Maheen Shermohammed, and colleagues. 2026. “Algorithm-Assisted Personalized Risk Communication to Encourage Flu Vaccination in the USA: Three Randomized Field Trials.” *Nature Human Behaviour*. Published 5 October. + +[Read the paper](https://doi.org/10.1038/s41562-026-02603-4). + +## 6. Home gardens remain part of urban food systems in Imphal. + +Arati Ningombam and colleagues examine urban and peri-urban agriculture in Imphal, Manipur, in a paper published in *Scientific Reports* on 6 October. + +The study uses a mixed-methods survey of 150 households to investigate household food production, gardening practices and the relationship between these activities and local food traditions. + +The researchers describe how traditional homestead gardens have changed as residential density has increased. Households increasingly use smaller spaces, including terraces, rooftops and containers, to produce vegetables and herbs. + +Women manage most of the gardens examined in the study. The authors also document concerns about pesticide residues in commercially purchased vegetables as one reason households pursue home cultivation. + +These findings provide useful information about the persistence of household food production within an increasingly urban environment. + +The study's conclusions about nutritional security require more caution. Growing vegetables at home may contribute to dietary diversity, but the extent of that contribution depends on production quantities, seasonal availability, household consumption and the foods displaced or supplemented. + +The research design cannot establish that household gardening caused improvements in nutritional status. The sample is also too small to support broad claims about the prevalence of gardening across Imphal without further evidence about how participating households were selected. + +The authors' wider claims concerning social resilience should be read in the same way. + +The paper is nevertheless valuable for documenting a form of urban food production that can be overlooked in conventional agricultural statistics. + +There are also questions worth pursuing through further research. The economic value of home-produced food could be estimated alongside household food purchases, women's time use and the availability of space and water. + +I would recommend the paper to researchers working on urban food systems and household nutrition, particularly those interested in combining survey methods with qualitative research. Its descriptive evidence is useful, while its claims about nutritional and social outcomes need stronger testing. + +**Paper:** Ningombam, Arati, Blessa Sailo, Sonia Chongtham, and colleagues. 2026. “Role of Urban and Peri-Urban Agriculture in Nutritional Diversity, Cultural Resilience and Community-Based Food Security in an Eastern Himalayan City.” *Scientific Reports*. Published 6 October. + +[Read the open-access paper](https://doi.org/10.1038/s41598-026-72784-5). + +## 7. Researchers identify limited suitable habitat for fish sanctuaries in three major river basins of Bangladesh. + +Murad Ahmed Farukh and colleagues examine habitat suitability across the Ganges–Padma, Brahmaputra–Jamuna and Barak–Meghna river systems in Bangladesh. + +Their study, published in *Scientific Reports* on 9 October, combines satellite-derived land-use information, seasonal water observations, hydrochemical measurements and geographical multi-criteria analysis. + +The researchers examine changes in mapped land cover between 2000 and 2020, together with indicators of water quality and habitat conditions. + +They report a 1.5 per cent decline in mapped monsoon water extent and a 12.7 per cent decline in the monsoon vegetation class, alongside a 25 per cent increase in dry-season vegetation. + +Water quality presents additional constraints. Approximately 37.5 per cent of site-season observations were classified as poor, very poor or unsuitable. The measurements included low dissolved oxygen and elevated concentrations of particular pollutants. + +The resulting ecological suitability index classified only 4.9 per cent of the mapped area as highly suitable for fish sanctuaries. + +The analysis is relevant to fisheries management because the suitability of a sanctuary depends on several conditions operating together. Habitat connectivity, seasonal water availability and pollution can influence whether protected areas provide appropriate ecological conditions. + +The study also demonstrates a limitation of spatial decision-support models. Their outputs depend on the indicators selected, the weights assigned to those indicators and the quality of the underlying measurements. + +A classification of highly suitable habitat should therefore be validated against ecological observations, including fish distribution and breeding outcomes where data permit. + +The research could inform sanctuary planning, although management decisions also need to consider fishing livelihoods, seasonal access and local institutional arrangements. + +I would recommend this paper to researchers interested in combining remote sensing, environmental monitoring and spatial planning. Its methods are potentially transferable to other South Asian river systems, subject to local validation. + +**Paper:** Farukh, Murad Ahmed, M. Zulfikar Rahman, Md. Abdus Salam, and colleagues. 2026. “Integrating Land-Use Dynamics, Water Quality, and Multi-Criteria Ecological Suitability for Fish Sanctuary Prioritization in Three Major River Basins of Bangladesh.” *Scientific Reports*. Published 9 October. + +[Read the open-access paper](https://doi.org/10.1038/s41598-026-72239-x). + +## 8. Reclaiming wetlands for agriculture delivers limited production gains while increasing flood exposure. + +Xi Zhang and colleagues examine the agricultural benefits and flood exposure associated with converting wetlands into cropland. + +Their paper, published in *Nature Sustainability* on 9 October, identifies wetland-reclaimed cropland globally between 2000 and 2022 and compares its production performance and flood exposure with existing agricultural land. + +The findings are substantial. Nearly half of reclaimed wetland cropland had lower yields than existing cropland. The reclaimed areas contributed approximately 0.1 per cent of total global crop production. + +More than half of the reclaimed cropland faced severe flood hazards, with an average flood exposure probability estimated to be twenty-three times that of cropland generally. + +The authors classify approximately 60 per cent of reclaimed wetland cropland as experiencing the combined disadvantages of limited agricultural gains and high flood exposure. + +These findings challenge the economic justification for agricultural expansion through wetland conversion in many locations. + +The study is based on geographical analysis and modelled estimates. Its conclusions depend on the accuracy of wetland classification, crop production estimates and flood exposure measures. The results should therefore be interpreted as evidence about patterns across mapped locations. + +They do not establish the financial returns or welfare consequences for every individual farming household cultivating reclaimed land. + +The policy implications extend to South Asia, where wetlands support agriculture, fisheries, flood regulation and other economic activities. + +Assessing conversion proposals requires information about agricultural output and the services lost through changes in land use. Flood exposure is especially relevant where communities already experience frequent inundation. + +The paper provides evidence that many reclamation projects may generate limited agricultural gains while exposing production to substantial environmental risk. + +I would recommend this study to researchers working on land-use policy, agricultural economics and climate adaptation. Its value lies in comparing agricultural gains with environmental exposure within a common geographical framework. + +**Paper:** Zhang, Xi, Jinwei Dong, Yiqun Shang, and colleagues. 2026. “Limited Agricultural Gains but High Flood Risk from Global Wetland Reclamation.” *Nature Sustainability*. Published 9 October. + +[Read the paper](https://doi.org/10.1038/s41893-026-01965-x). + +## What I would read first + +The procurement experiment by Cassidy, McKenzie and Neelamraju is my first recommendation this week. It has a credible experimental design, reports outcomes extending to actual procurement behaviour, and addresses a problem common to enterprise-development programmes. The distinction between programme assignment, participation and eventual commercial outcomes is particularly useful for evaluation design. + +Hashmi and Latheef's paper on rural nonfarm employment deserves attention for a different reason. It questions what can be inferred from changes in employment categories and directs attention towards earnings, labour-market segmentation and selection. The cross-sectional design limits causal claims, but the distributional analysis raises worthwhile questions. + +The Himalayan flood-exposure study is especially relevant for researchers using geospatial data. Its validation exercise demonstrates how differences between building datasets affect the apparent distribution of risk. + +I would also keep the online-learning paper close at hand. Its findings suggest that brief interruptions during digital learning may have measurable consequences for engagement and achievement, an issue that deserves greater attention in educational technology evaluations. + +The remaining papers are useful for more specialised research interests. The Imphal study is primarily descriptive, the Bangladesh paper offers a spatial planning method, and the wetland study provides a comparative assessment of agricultural gains and environmental exposure. The influenza vaccination trials contribute credible experimental evidence about personalised health communication. + +Across the eight papers, the central methodological lesson concerns the relationship between the outcome being measured and the conclusion that can be supported. Registration, employment category, geographical exposure and programme participation each describe a particular condition or activity. Their consequences for income, welfare, learning or health require additional evidence. diff --git a/archive/README.md b/archive/README.md index c1d3ecf..69a51e4 100644 --- a/archive/README.md +++ b/archive/README.md @@ -1,19 +1,55 @@ # Archive -13 posts from [varna.substack.com](https://varna.substack.com), newest first. Generated by `scripts/sync_substack.py`; do not edit by hand. +49 posts from [varna.substack.com](https://varna.substack.com), newest first. Generated by `scripts/sync_substack.py`; do not edit by hand. | Date | Type | Title | |---|---|---| +| 2026-10-17 | edition | [Research Rundown \| Friday Findings \| 9 October 2026](2026-10-17-research-rundown-friday-findings-c92.md) | +| 2026-10-17 | other | [Good Reads #15 \| Who owns intelligence? Arundhati Roy on AI and extraction](2026-10-17-good-reads-15-who-owns-intelligence.md) | +| 2026-10-16 | daily | [Research Rundown \| Daily \| 7 October 2026](2026-10-16-research-rundown-daily-7-october.md) | +| 2026-10-16 | daily | [Research Rundown \| Daily \| 6 October 2026](2026-10-16-research-rundown-daily-6-october.md) | +| 2026-10-16 | other | [Methods Monday \| 5 October 2026](2026-10-16-methods-monday-5-october-2026.md) | +| 2026-10-07 | other | [Good Reads #13 \| When women lead, girls marry later. But there is a catch.](2026-10-07-good-reads-13-when-women-lead-girls.md) | +| 2026-10-07 | daily | [Research Rundown DAILY - 3 October 2026](2026-10-07-research-rundown-daily-3-october.md) | +| 2026-10-03 | edition | [Research Rundown \| Gandhi Jayanti Special](2026-10-03-research-rundown-gandhi-jayanti-special.md) | +| 2026-10-02 | daily | [Research Rundown DAILY \| 27 September 2026](2026-10-02-research-rundown-daily-27-september.md) | +| 2026-10-01 | edition | [Research Rundown: Friday Findings -1 October 2026](2026-10-01-research-rundown-friday-findings-632.md) | +| 2026-10-01 | other | [Good Reads #12: Did MGNREGA narrow the gender wage gap?](2026-10-01-good-reads-12-did-mgnrega-narrow.md) | +| 2026-10-01 | daily | [Research Rundown DAILY 29 September 2026](2026-10-01-research-rundown-daily-29-september.md) | +| 2026-10-01 | other | [Methods Monday \| 28 September 2026](2026-10-01-methods-monday-28-september-2026.md) | +| 2026-09-27 | daily | [Research Rundown DAILY - 26 September 2026](2026-09-27-research-rundown-daily-26-september.md) | +| 2026-09-27 | other | [Good Reads #11: What happens when South Asia has fewer children to educate?](2026-09-27-good-reads-11-what-happens-when-south.md) | +| 2026-09-25 | edition | [Research Rundown: Friday Findings - 25 September 2026](2026-09-25-research-rundown-friday-findings.md) | +| 2026-09-25 | other | [Good Reads #10: When using a service is not the same as benefiting from it](2026-09-25-good-reads-10-when-using-a-service.md) | +| 2026-09-23 | daily | [Research Rundown DAILY - 23 September 2026](2026-09-23-research-rundown-daily-23-september.md) | +| 2026-09-23 | other | [Good Reads #9: Is the health gap really a wealth gap?](2026-09-23-good-reads-9-is-the-health-gap-really.md) | +| 2026-09-21 | other | [Methods Monday \| 21 September 2026](2026-09-21-methods-monday-21-september-2026.md) | +| 2026-09-21 | other | [Good Reads #8: Why are more rural women working?](2026-09-21-good-reads-8-why-are-more-rural-women.md) | +| 2026-09-20 | daily | [Research Rundown DAILY - 19 September 2026](2026-09-20-research-rundown-daily-19-september.md) | +| 2026-09-18 | other | [Friday Findings \| 18 September 2026](2026-09-18-friday-findings-18-september-2026.md) | +| 2026-09-17 | other | [Good Reads #7: What migration data cannot tell us](2026-09-17-good-reads-7-what-migration-data.md) | +| 2026-09-17 | daily | [Research Rundown DAILY - 16 September](2026-09-17-research-rundown-daily-16-september.md) | +| 2026-09-16 | edition | [Research Rundown, September 2026](2026-09-16-research-rundown-september-2026.md) | +| 2026-09-14 | other | [Methods Monday \| 14 September 2026](2026-09-14-methods-monday-14-september-2026.md) | +| 2026-09-13 | daily | [Research Rundown DAILY - 13 September 2026](2026-09-13-research-rundown-daily-13-september.md) | +| 2026-09-12 | other | [Good Reads #6: India’s middle-power moment](2026-09-12-good-reads-6-indias-middle-power.md) | +| 2026-09-12 | other | [Good Reads #5: Can AI make informal work better?](2026-09-12-good-reads-5-can-ai-make-informal.md) | +| 2026-09-12 | other | [Good Reads #3: What comes after industrialisation?](2026-09-12-good-reads-3-what-comes-after-industrialisation.md) | +| 2026-09-12 | other | [Good Reads #2: Can India sustain its rise](2026-09-12-good-reads-2-can-india-sustain-its.md) | +| 2026-09-11 | other | [Friday Findings \| 11 September 2026](2026-09-11-friday-findings-11-september-2026.md) | +| 2026-09-10 | daily | [Research Rundown DAILY - 10 September 2026](2026-09-10-research-rundown-daily-10-september.md) | +| 2026-09-09 | daily | [Research Rundown DAILY 9 September 2026](2026-09-09-research-rundown-daily-9-september.md) | +| 2026-09-08 | daily | [Research Rundown DAILY - 8 September 2026](2026-09-08-research-rundown-daily-7-september.md) | | 2026-09-07 | brief | [Measurement & Survey Methods Brief — 4 September 2026](2026-09-07-measurement-and-survey-methods-brief.md) | | 2026-09-07 | daily | [Research Rundown: The Daily 7 September 2026](2026-09-07-research-rundown-the-daily-7-september.md) | -| 2026-09-07 | other | [Good Read - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic](2026-09-07-good-read-why-economists-dont-listen.md) | +| 2026-09-07 | other | [Good Reads #1 - Why Economists Don’t Listen: Tales From an Anthropology Spy in Economic](2026-09-07-good-read-why-economists-dont-listen.md) | | 2026-09-05 | daily | [5 September 2026 Research Rundown \| DAILY](2026-09-05-5-september-2026-research-rundown.md) | | 2026-09-04 | daily | [Research Rundown DAILY— 4 September 2026](2026-09-04-research-rundown-daily-4-september.md) | | 2026-07-08 | edition | [🌏 The Research Rundown: July 2026](2026-07-08-the-research-rundown-july-2026.md) | | 2026-05-08 | edition | [🌏 The Research Rundown \| May 2026](2026-05-08-the-research-rundown-may-2026.md) | | 2026-03-18 | edition | [🌏 The Research Rundown — March 2026 Edition](2026-03-18-the-research-rundown-march-2026-edition.md) | | 2025-09-01 | edition | [🌏 The Research Rundown - August 2025 Edition (erm... in September)](2025-09-01-the-research-rundown-august-2025.md) | -| 2025-08-24 | other | [Unlearning to Lead: A First Step into Politics](2025-08-24-unlearning-to-lead-a-first-step-into.md) | +| 2025-08-24 | other | [Good Reads 4# Unlearning to Lead: A First Step into Politics](2025-08-24-unlearning-to-lead-a-first-step-into.md) | | 2025-06-26 | edition | [The Research Rundown \| June 2025](2025-06-26-the-research-rundown-june-2025-edition.md) | | 2025-04-28 | edition | [The Research Rundown \| April 2025](2025-04-28-the-research-rundown-april-2025-edition.md) | | 2024-06-05 | edition | [The Research Rundown \| June 2024](2024-06-05-the-research-rundown-june-edition.md) | diff --git a/archive/SECTIONS.md b/archive/SECTIONS.md index ff01372..9e1456d 100644 --- a/archive/SECTIONS.md +++ b/archive/SECTIONS.md @@ -10,6 +10,7 @@ Where each recurring section of the newsletter appears, by edition. Generated; d ## Research praxis +- 2026-09-16: [Research Rundown, September 2026](2026-09-16-research-rundown-september-2026.md) - 2026-07-08: [🌏 The Research Rundown: July 2026](2026-07-08-the-research-rundown-july-2026.md) - 2026-05-08: [🌏 The Research Rundown | May 2026](2026-05-08-the-research-rundown-may-2026.md) - 2026-03-18: [🌏 The Research Rundown — March 2026 Edition](2026-03-18-the-research-rundown-march-2026-edition.md) diff --git a/companions/praxis.md b/companions/praxis.md index beb58e0..641c3c4 100644 --- a/companions/praxis.md +++ b/companions/praxis.md @@ -4,6 +4,60 @@ Every research-practice note the newsletter has run, newest first. Each is a way Generated by `scripts/build_companions.py` from `archive/`; edit the source, not this file. +## Build the claim before choosing the indicator + +*From [Research Rundown, September 2026](../archive/2026-09-16-research-rundown-september-2026.md), 2026-09-16.* + +One habit I recommend to researchers I work with is embarrassingly simple. + +Write the substantive claim before opening the dataset. + +Suppose the claim is: + +> Women’s economic position in rural India improved between two survey rounds. + +Now list what would have to be true for you to believe it. + +Would a rise in LFPR be enough? + +Probably not. + +Would increased paid employment be enough? + +Better. + +Would rising real earnings strengthen the claim? + +Yes. + +Would fewer women reporting unpaid family work matter? + +Yes. + +Would greater control over earnings matter? + +Very much, although PLFS cannot answer it. + +Would reduced time poverty alter the interpretation? + +Again, yes. + +You have now done something useful before running a regression. You have separated the construct from the indicators available to measure it. + +This also exposes the missing-data problem early. + +Many evaluation frameworks do the reverse. They begin with indicators because indicators are available. The programme’s substantive claim is then quietly rewritten to fit them. + +A monitoring system records training completion, so completion becomes “capacity”. + +An administrative database records registrations, so registration becomes “access”. + +A livelihood programme records enterprises started, so enterprise creation becomes “economic empowerment”. + +The mistake happens before estimation. + +No econometric technique repairs a badly specified construct. + ## How to Read a Deletion Rate *From [The Research Rundown: July 2026](../archive/2026-07-08-the-research-rundown-july-2026.md), 2026-07-08.* diff --git a/companions/reading.md b/companions/reading.md index 6fea786..87d10b5 100644 --- a/companions/reading.md +++ b/companions/reading.md @@ -4,6 +4,236 @@ Everything the newsletter recommended reading, edition by edition, newest first, Generated by `scripts/build_companions.py` from `archive/`; edit the source, not this file. +## 16 October 2026 + +*ONE PAPER WORTH READING, from [Research Rundown | Daily | 7 October 2026](../archive/2026-10-16-research-rundown-daily-7-october.md), 2026-10-16.* + +### Does community sanitation deprivation affect depression among older Indians? + +Anoop Jain, Rockli Kim and S. V. Subramanian examine the association between community-level toilet deprivation and depressive symptoms among older adults in India in a paper published in *GeroScience* on 7 October 2026. + +Their study uses data from the first wave of the Longitudinal Ageing Study in India, collected during 2017–2018. The analytical sample comprises 63,564 adults above the age of 45 living in 2,439 villages and urban wards. + +The authors examine whether the proportion of residents living in households without toilets within a community is associated with depressive symptoms and probable major depression. + +They estimate multilevel regression models that account for individual characteristics, household socioeconomic conditions, residential location and household toilet access. + +The findings indicate a substantial association between community sanitation deprivation and mental health. + +For one measure of depressive symptoms, respondents living in communities with the lowest levels of toilet coverage had an estimated odds ratio of 1.38, with a 95 per cent confidence interval of 1.08–1.76, relative to the model’s reference condition. + +The authors report broadly similar patterns using alternative measures of probable major depression. + +The analysis is interesting because community sanitation conditions may affect people through mechanisms extending beyond their own household facilities. + +Living in an area with widespread sanitation deprivation may involve exposure to environmental contamination, difficulties accessing alternative facilities and social stigma associated with the residential environment. + +The study examines whether community conditions modify the relationship between household toilet access and depression. Although descriptive patterns suggest higher depressive symptom prevalence in communities with lower toilet coverage, the estimated interaction terms were statistically insignificant. + +There are important limitations. + +The analysis is cross-sectional, and the data were collected in 2017–2018. Communities with poor sanitation may also experience other forms of deprivation associated with mental health. The regression adjustments cannot eliminate the possibility of unobserved confounding. + +The study also lacks information about whether households without private toilets have access to shared or public facilities, which could influence the interpretation of sanitation deprivation. + +These limitations restrict causal inference. The results nevertheless suggest that household-level sanitation indicators may provide an incomplete account of the environmental conditions associated with health. + +I would recommend this paper to researchers working on sanitation, ageing and social determinants of health. Its principal contribution lies in specifying community-level sanitation deprivation as an exposure and examining its relationship with mental health outcomes using a large national dataset. + +Future work using longitudinal observations could establish whether changes in community sanitation conditions precede changes in depressive symptoms. Qualitative research could also examine the proposed mechanisms involving stigma, environmental conditions and access to facilities. + +Jain, Anoop, Rockli Kim, and S. V. Subramanian. 2026. “Association Between Community Toilet Deprivation and Depression Among Older Adults in India: A Multilevel Analysis of the Longitudinal Aging Study in India from 2017–2018.” *GeroScience*. Published 7 October 2026. + +[Read the open-access paper](https://doi.org/10.1007/s11357-026-02537-9) + +## 16 October 2026 + +*ONE PAPER WORTH READING, from [Research Rundown | Daily | 6 October 2026](../archive/2026-10-16-research-rundown-daily-6-october.md), 2026-10-16.* + +### Does migration increase the likelihood of experiencing discrimination in later life? + +Vasim Ahamad, Rahul Rahul, Ram B. Bhagat and T. V. Sekher examine the association between internal migration and perceived discrimination among Indians aged 45 and above in a paper published in *BMC Public Health* on 5 October 2026. + +The authors analyse 64,711 respondents from the first wave of the Longitudinal Ageing Study in India, conducted during 2017–2018. They classify migration using respondents' place of last residence and measure perceived discrimination through an everyday discrimination scale. + +Approximately 57 per cent of respondents were classified as internal migrants. After adjusting for observed socioeconomic characteristics, migrants had 1.38 times the odds of reporting discrimination compared with non-migrants, with a 95 per cent confidence interval of 1.28–1.48. The association was stronger among men, interstate migrants and recent migrants. + +The paper offers useful evidence on an understudied dimension of ageing in India. Its principal limitation concerns causal interpretation. Cross-sectional associations cannot establish whether migration caused discrimination, and the migrant category captures movement over the life course rather than a single type of migration experience. The subjective discrimination measure also requires careful interpretation. + +The findings justify closer examination of migration histories, gender, social identity and access to services among older adults. Longitudinal analysis would help establish how these relationships change with duration of residence. + +[Read the open-access paper](https://doi.org/10.1186/s12889-026-29800-0) + +## 7 October 2026 + +*ONE PAPER WORTH READING, from [Research Rundown DAILY - 3 October 2026](../archive/2026-10-07-research-rundown-daily-3-october.md), 2026-10-07.* + +### Does ageing matter for public health spending in South Asia? + +Jalandhar Pradhan, Soumyanayani Mahali and Tapan Swain examine the relationship between population ageing and government health expenditure across South Asian countries in a new peer-reviewed paper in *BMC Health Services Research*, published on 1 October. + +[Read the open-access paper](https://link.springer.com/article/10.1186/s12913-026-15545-4?utm_source=chatgpt.com) + +The authors use cross-country panel data and several econometric specifications, including Driscoll-Kraay standard errors, panel-corrected standard errors and three-stage least squares. The paper treats demographic change alongside physician availability, unemployment and other macro-level determinants of public health expenditure. + +Its regional focus is useful. South Asia is ageing while several countries still carry large maternal, child-health and infectious-disease burdens. Public systems therefore face overlapping rather than sequential epidemiological demands. + +The limitation is identification. This is an observational macro-panel analysis. Alternative estimators can address particular statistical problems, including cross-sectional dependence and endogeneity assumptions, but they do not turn the relationship between ageing and expenditure into a clean causal estimate. Country-level expenditure data also conceal large differences in who receives public spending and what governments purchase. + +Read it for the fiscal question it raises rather than a single coefficient: how should health budgets change when population ageing begins before older health burdens have disappeared? + +## 1 October 2026 + +*ONE PAPER WORTH READING, from [Research Rundown DAILY 29 September 2026](../archive/2026-10-01-research-rundown-daily-29-september.md), 2026-10-01.* + +### What Happens When Women Suddenly Stop Receiving Cash Transfers? + +Nasir Iqbal, Amen Jalal, Mahreen Mahmud and Kate Vyborny, World Bank Policy Research Working Paper 11474, 29 September 2026. + +Most cash-transfer research asks what happens when money arrives. This paper asks the reverse question. + +The setting is Pakistan. The authors exploit a revised eligibility rule that caused some women to lose transfers unexpectedly and use a regression discontinuity-in-differences design around the eligibility threshold. That gives the paper a stronger identification strategy than a simple comparison of former recipients with continuing recipients, since those groups could differ systematically for many other reasons. + +The substantive question is unusually useful: does the position women acquire while bringing programme income into the household persist when that income disappears, and is sudden programme exit associated with backlash? + +The design still identifies effects around an eligibility threshold. External validity therefore needs care. Women farther from that threshold, households leaving programmes for other reasons and planned graduation may behave differently. + +The larger contribution is conceptual. Social-protection evaluations routinely devote considerable attention to targeting and programme entry. Exit receives much less scrutiny. If transfers alter bargaining power, labour decisions or household expectations, the end of a programme is itself an intervention. It should be measured as one. + +## 20 September 2026 + +*ONE PAPER WORTH READING, from [Research Rundown DAILY - 19 September 2026](../archive/2026-09-20-research-rundown-daily-19-september.md), 2026-09-20.* + +### Evolution of Climate Extremes Over the Indian Subcontinent Using a Revised CEI + +Thota Mohan and colleagues use IMD gridded daily rainfall and temperature data from **1971 to 2022** to construct a revised India Climate Extremes Index. Seven components capture maximum and minimum temperature, consecutive dry days, heavy precipitation, drought using SPEI, humid heat and heatwave severity. Trend estimation shows widespread increases in heat extremes, particularly in northwestern India; stronger drought signals in southern India; and the unusual combination of more short-duration heavy rainfall and increasing drought episodes in the Northeast. At the national level, the index increases significantly by roughly 0.36 per cent of area per decade. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +The regional decomposition is more useful than the composite headline. Combining different hazards into one index necessarily embeds decisions about indicators and aggregation, while regional averages can still conceal local exposure. The observational record also identifies trends rather than attributing individual changes to specific causal mechanisms. + +Its contribution is practical. Indian climate adaptation cannot sensibly be organised around one national hazard trajectory. Heat, drought and rainfall variability are changing differently across regions. + +**Citation:** Mohan, Thota, M. Rajeevan, Smrutishree Lenka, Kondapalli Niranjan Kumar, et al. 2026. “Evolution of Climate Extremes Over the Indian Subcontinent Using a Revised CEI.” *International Journal of Climatology*. ([Royal Meteorological Society (RMetS)](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com)) + +[Read the paper](https://rmets.onlinelibrary.wiley.com/doi/10.1002/joc.70467?utm_source=chatgpt.com) + +## September 2026 + +*What I think is worth reading, from [Research Rundown, September 2026](../archive/2026-09-16-research-rundown-september-2026.md), 2026-09-16.* + +### 1. India Development Update, April 2026 + +The World Bank’s latest India update is worth reading partly because of a technical issue that received less attention than the growth forecast: India’s February 2026 GDP rebasing. + +The base year moved to 2022-23, and the revised methodology incorporates more regular data on the household and informal sectors. MoSPI says annual sources such as ASUSE and PLFS will now contribute directly to household-sector estimates rather than relying as heavily on older benchmark surveys and proxy growth indicators. + +This matters. The informal economy has always been one of the hardest parts of Indian GDP to estimate. + +A rebasing exercise that changes our estimate of the size of that economy is not statistical housekeeping at the margins. It changes the denominator against which several other economic quantities are understood. + +[India Development Update, April 2026](https://thedocs.worldbank.org/en/doc/4262e1e15b463ecb360cec4ad78cf062-0310012026/?utm_source=chatgpt.com) + +### 2. The World Bank’s India poverty reproducibility package + +I wish reproducibility packages received a fraction of the attention that poverty headlines do. + +The package provides the material behind the Bank’s updated 2011-12 to 2023-24 poverty series and specifies the $3.00 and $4.20 2021-PPP poverty lines used in the estimates. + +If you disagree with a poverty estimate, this is where the interesting argument should begin: welfare aggregate, survey comparability, price adjustments, poverty line, treatment of transfers, sampling and sensitivity. + +[India long-term poverty trend reproducibility package](https://reproducibility.worldbank.org/catalog/541?utm_source=chatgpt.com) + +### 3. South Asia Economic Update: Working with Industrial Policy + +I am less interested in its headline regional growth forecast than in the work on jobs and spatial wage differences. + +The report finds persistent subnational wage premiums in South Asia’s larger economies even after controlling for worker characteristics. Better transport connections, skilled workforces, larger firms and service-sector activity are associated with higher residual wage premiums. + +That is a useful corrective to national labour-market averages. Where a worker lives remains economically consequential after observable individual characteristics are taken into account. + +[South Asia Economic Update, April 2026](https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099919404082611180?utm_source=chatgpt.com) + +### 4. A Livable Future: Protecting Jobs and Growth from Extreme Heat in South Asia’s Cities + +Read this alongside labour-market statistics rather than filing it under climate. + +Its central contribution is to treat heat exposure as a labour and productivity problem. I would read the large aggregate estimates cautiously because they depend on modelling assumptions. The occupational logic is much harder to dismiss. Workers exposed to outdoor and poorly cooled environments experience heat through hours, output, illness and earnings. + +That belongs in any serious discussion of South Asia’s future of work. + +[World Bank material on extreme heat, jobs and growth in South Asia](https://www.worldbank.org/en/news/press-release/2026/07/28/extreme-heat-puts-south-asia-s-jobs-and-growth-at-risk-world-bank?utm_source=chatgpt.com) + +--- + +[Subscribe now](https://varna.substack.com/subscribe?) + +## 13 September 2026 + +*one paper worth reading, from [Research Rundown DAILY - 13 September 2026](../archive/2026-09-13-research-rundown-daily-13-september.md), 2026-09-13.* + +### Phone tutoring works during school closures. SMS alone is much less reliable. + +Noam Angrist and colleagues report randomized trials of remote primary-school instruction in India, Nepal, Kenya, Uganda and the Philippines. Children received targeted tutoring by phone plus SMS, SMS alone, or usual provision. The combined phone intervention increased the pooled learning outcome by 0.321 standard deviations across the five countries; SMS alone produced a much smaller 0.078 SD effect and worked inconsistently across settings. India contributed 708 participants and Nepal 2,678. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +The cross-country design is the paper’s main strength. Random assignment supports causal inference within the trials, and Nepal and the Philippines also randomized delivery between government and NGO instructors. Government delivery retained substantial effects. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +There is an important qualification. Effect sizes grew across successive trials as targeting and implementation improved. Context and intervention maturity therefore move together, making some cross-country comparisons difficult to interpret. These were also emergency interventions during COVID-era school disruption, when the counterfactual in some countries involved very little schooling. + +The policy result is still useful: low-tech, targeted human instruction travelled across settings far better than messages alone. For education systems planning for climate, conflict or disease-related closures, that is actionable evidence. + +**Citation:** Angrist, Noam, Micheal Ainomugisha, Sai Pramod Bathena, et al. 2026. “Mobile education builds resilience during shocks in five countries.” *Nature*. Published 9 September 2026. DOI: 10.1038/s41586-026-10990-x. ([Nature](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com)) + +[Open-access paper](https://www.nature.com/articles/s41586-026-10990-x?utm_source=chatgpt.com) + +## 10 September 2026 + +*one paper worth reading, from [Research Rundown DAILY - 10 September 2026](../archive/2026-09-10-research-rundown-daily-10-september.md), 2026-09-10.* + +### When inequality becomes political conflict + +A new *World Development* paper asks whether changes in local economic inequality cause social unrest in rural India. The authors combine high-frequency bank-account information that tracks village-level inequality over time with data on unrest, and develop a model in which conflict emerges when local elites capture unequal shares of economic surplus. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +Their identification strategy exploits rainfall-driven water shocks that disproportionately increase elite incomes. The headline estimate is substantial: a 10 per cent increase in their inequality measure raises the monthly probability of unrest by about 6.5 per cent. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +The paper is unusually interesting because it moves below state and district averages and observes changing economic distributions within villages. The causal claim still depends on the exclusion restriction: rainfall-induced water shocks must affect unrest through the inequality mechanism being estimated rather than through other channels. In rural India, rainfall can alter employment, agricultural output, prices and local political relations, so that assumption deserves close attention. + +The broader result matters. Distribution within a locality may affect political stability independently of average income. This is a paper I would flag for fuller treatment in a future flagship Research Rundown. + +**Citation:** “Inequality and social unrest in India.” 2026. *World Development* 205: 107418. DOI: 10.1016/j.worlddev.2026.107418. ([ScienceDirect](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com)) + +[Open-access article and abstract](https://www.sciencedirect.com/science/article/pii/S0305750X26001075?utm_source=chatgpt.com) + +## 9 September 2026 + +*one paper worth reading, from [Research Rundown DAILY 9 September 2026](../archive/2026-09-09-research-rundown-daily-9-september.md), 2026-09-09.* + +### A useful new paper is harder to find today + +Rather than promote an old working paper because it has been newly indexed, I would leave this section deliberately sparse. + +One apparently new September listing that surfaced in the search, Sampreet Singh Goraya’s *How does Caste Affect Entrepreneurship? Birth vs Worth*, illustrates the problem. The Barcelona School of Economics page currently labels the working paper August 2026, but the downloadable paper itself is dated July 2019. Its empirical analysis uses the 2005 Economic Census and the 2001 and 2006 MSME censuses. ([BSE](https://bse.eu/research/working-papers/how-does-caste-affect-entrepreneurship-birth-vs-worth)) + +The paper remains interesting. It estimates that Scheduled Caste and Scheduled Tribe entrepreneurs owned 13 per cent of non-agricultural establishments despite representing 24 per cent of the population, and uses a calibrated entrepreneurship model to estimate output losses from caste-specific entry and financing constraints. + +It is not new research. Treating repository dates or refreshed publication pages as publication dates would quietly degrade the purpose of this section. + +I would rather publish no paper today than manufacture novelty. + +## 8 September 2026 + +*ONE PAPER WORTH READING, from [Research Rundown DAILY - 8 September 2026](../archive/2026-09-08-research-rundown-daily-7-september.md), 2026-09-08.* + +### Hidden Assets: The Economic and Health Case for Climate and Clean Air Action + +The most useful new research product today is UNEP and the Climate and Clean Air Coalition’s global economic assessment of joint climate and air-pollution policy. It asks a practical question: what changes when the economic appraisal counts climate mitigation and clean-air benefits together? + +The assessment models 25 measures across energy, industry, transport, agriculture and food systems, household energy and waste. It incorporates implementation costs alongside market gains and health-related welfare benefits. Its headline estimate is striking: integrated action produces roughly US$15 in benefits for every US$1 invested globally, with market benefits alone exceeding costs within ten years. For South Asia, the estimated return rises to about 21:1. + +The caution is substantial. These are model-derived social returns across a package of interventions. They are sensitive to assumptions about valuation of mortality and morbidity, implementation speed, baseline emissions and future policy. A benefit-cost ratio of 21:1 should therefore be read as an estimate from a defined modelling framework, rather than a forecast of fiscal returns available to any individual government programme. + +The report still does something useful. It makes the opportunity cost of treating air quality, health and climate as separate administrative problems empirically visible. + +**Citation:** United Nations Environment Programme and Climate and Clean Air Coalition. 2026. *Hidden Assets: The Economic and Health Case for Climate and Clean Air Action*. Nairobi: UNEP. + ## 7 September 2026 *One paper worth reading, from [Research Rundown: The Daily 7 September 2026](../archive/2026-09-07-research-rundown-the-daily-7-september.md), 2026-09-07.* diff --git a/scripts/sync_substack.py b/scripts/sync_substack.py index e0c8975..b741960 100644 --- a/scripts/sync_substack.py +++ b/scripts/sync_substack.py @@ -26,6 +26,7 @@ import sys import time import urllib.error +import urllib.parse import urllib.request from datetime import date from html.parser import HTMLParser @@ -84,22 +85,29 @@ def fetch_json(url: str): def list_posts() -> list[dict]: - """Every post, newest first. Substack returns fewer than `limit` posts on - a page that is not the last (23, then 26, then 3 overlapping, then none, on - 2026-10-19), so only an empty page ends the listing. Pages can overlap, so - posts are kept once by id, falling back to slug.""" - out, seen, offset = [], set(), 0 + """Every post, newest first. Since November 2026 the archive endpoint answers + HTTP 400 to `offset` and to a `limit` above 25, and returns an object with + `posts` and `nextCursor` where it used to return a bare list. Pages are + followed by cursor until none is given. The earlier endpoint also returned + short pages that were not the last, so a short page does not end the + listing, and pages can overlap, so posts are kept once by id, falling back to + slug. A bare list is still accepted, in case Substack reverts.""" + out, seen, cursor = [], set(), None for _ in range(PAGE_CAP): - page = fetch_json(f"{PUBLICATION}/api/v1/archive?sort=new&offset={offset}&limit=50") - if not page: - break + url = f"{PUBLICATION}/api/v1/archive?limit=25" + if cursor: + url += f"&cursor={urllib.parse.quote(cursor)}" + data = fetch_json(url) + page = data if isinstance(data, list) else data.get("posts", []) + cursor = None if isinstance(data, list) else data.get("nextCursor") for post in page: key = post.get("id") or post.get("slug") if key in seen: continue seen.add(key) out.append(post) - offset += len(page) + if not page or not cursor: + break else: print(f"stopped after {PAGE_CAP} pages; the listing may be incomplete", file=sys.stderr) return out diff --git a/tests/test_sync_substack.py b/tests/test_sync_substack.py index cc188b2..2ab69f4 100644 --- a/tests/test_sync_substack.py +++ b/tests/test_sync_substack.py @@ -85,28 +85,59 @@ def test_a_404_is_not_retried(monkeypatch): # ---------------------------------------------------------------- list_posts def _pages(monkeypatch, pages): + """Serve `pages` in order, each as the cursor-shaped answer Substack gives + since November 2026: {"posts": [...], "nextCursor": "..." or None}. The URLs + asked for are kept in the returned list.""" it = iter(pages) - monkeypatch.setattr(sync, "fetch_json", lambda url: next(it)) + urls = [] + + def fake(url): + urls.append(url) + return next(it) + + monkeypatch.setattr(sync, "fetch_json", fake) + return urls def _posts(start, n): return [{"id": i, "slug": f"p{i}"} for i in range(start, start + n)] +def _page(posts, cursor): + return {"posts": posts, "nextCursor": cursor} + + def test_short_pages_do_not_end_the_listing(monkeypatch): - # Substack returned 23 and then 19 on 6 October 2026, with limit=50. - _pages(monkeypatch, [_posts(0, 23), _posts(23, 19), []]) + # Substack returned 23 and then 19 on 6 October 2026. Only a missing cursor + # or an empty page ends the listing. + _pages(monkeypatch, [_page(_posts(0, 23), "a"), _page(_posts(23, 19), "b"), _page([], None)]) assert len(sync.list_posts()) == 42 def test_overlapping_pages_keep_each_post_once(monkeypatch): - _pages(monkeypatch, [_posts(0, 23), _posts(20, 6), []]) + _pages(monkeypatch, [_page(_posts(0, 23), "a"), _page(_posts(20, 6), "b"), _page([], None)]) got = sync.list_posts() assert [p["id"] for p in got] == list(range(26)) +def test_the_listing_follows_the_cursor_and_never_sends_offset(monkeypatch): + # Substack answers HTTP 400 to `offset` and to a `limit` above 25. + urls = _pages(monkeypatch, [_page(_posts(0, 25), "abc=="), _page(_posts(25, 24), None)]) + assert len(sync.list_posts()) == 49 + assert len(urls) == 2 + assert all("offset" not in u for u in urls) + assert "cursor=" not in urls[0] and "limit=25" in urls[0] + assert "cursor=abc%3D%3D" in urls[1] + + def test_the_page_cap_stops_a_listing_that_never_ends(monkeypatch): - monkeypatch.setattr(sync, "fetch_json", lambda url: [{"id": url, "slug": url}]) + count = iter(range(10**6)) + + def endless(url): + n = next(count) + return _page([{"id": n, "slug": f"p{n}"}], f"c{n}") + + monkeypatch.setattr(sync, "fetch_json", endless) assert len(sync.list_posts()) == sync.PAGE_CAP From 07b7bafebf226492b9e21d1dd5af35cd0b676db3 Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 6 Nov 2026 09:52:59 +0000 Subject: [PATCH 2/2] SignalStack notes: say what is known about the sync failure, and 34 tests Co-Authored-By: Claude Sonnet 5.5 Claude-Session: https://claude.ai/code/session_011RjaPcC924ePsEE8X6uH72 --- CLAUDE.md | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/CLAUDE.md b/CLAUDE.md index 48301f6..e9eebb0 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -78,7 +78,7 @@ label. ## Testing -`.github/workflows/tests.yml` runs `python -m pytest tests/` (33 tests) and +`.github/workflows/tests.yml` runs `python -m pytest tests/` (34 tests) and `build_companions.py --check` on pull requests and pushes to main. `lint-content.yml` runs markdownlint, cspell and lychee over hand-written Markdown; `archive/` and the generated companions are excluded from all three @@ -87,8 +87,9 @@ opens a PR; it never pushes to main. The sync follows Substack's archive by cursor until none is returned. On 2026-11-06 the endpoint answered HTTP 400 to `offset` and to a `limit` above 25 and began -returning `{posts, nextCursor}` instead of a list, which had stopped the sync -since early September (the archive was six weeks behind); it now uses +returning `{posts, nextCursor}` instead of a list, which stops the sync (the +archive was six weeks behind when this was found; when the change began is not +known, and the 403 below is a second cause on the scheduled run); it now uses `limit=25` and `cursor`. Earlier it paged until a page came back empty. Substack returns fewer posts than `limit` on pages that are not the last (23, 26, then 3 overlapping, on 2026-10-19), so stopping at a short page archived