🌏 The Research Rundown | May 2026
2026-05-08. Published at https://varna.substack.com/p/the-research-rundown-may-2026. This file is a generated copy; the Substack post is the original.
Dear Readers,
The May 4 election results have consumed every column inch available, and with good reason - West Bengal and Tamil Nadu, in particular, will be studied for years. But development research moves regardless of election cycles, and this month produced some of the most consequential reporting on climate, labour, and welfare measurement that I have seen in a while.
India’s heat crisis is killing people and eroding incomes at a scale that official data still fail to capture. A Harvard white paper released last week says we are systematically underestimating future heat risk. An MSSRF survey found that 97 per cent of informal women workers surveyed lost income during peak summer months. The OECD confirmed the largest single-year collapse in development aid ever recorded. Gig workers got a social security code that recognises them, but six months in, it has produced no actual scheme. And Census Phase 1 is in the field, in May, in a country with 27 per cent rural functional digital literacy.
The elections are a visible event. These are the structural ones.
The Elections: What Actually Happened
The ECI results from five states were declared on May 4:
State Total Seats Winner Seats Runner-up Seats Tamil Nadu 234 TVK (Vijay) 108 DMK 59 West Bengal 294 BJP 207 TMC 80 Kerala 140 UDF 102 LDF 35 Assam 126 NDA 102 INDIA bloc 21 Puducherry 30 NDA 18 INC+DMK 6
Two results stand out.
Tamil Nadu saw the Dravidian duopoly collapse in a single election. C. Joseph Vijay’s TVK, contesting its first Assembly election, won 108 seats and became the single largest party. The BBC attributes the TVK wave to Vijay’s welfare platform - Rs 2,500/month for women (more than double the existing Kalaignar Magalir Urimai Thittam), five lakh government jobs, and Rs 4,000/month for unemployed graduates - combined with an unusual campaign that relied entirely on public meetings without cash distribution. As of today, government formation remains deadlocked: the Governor has yet to invite TVK despite Congress backing it, and back-channel DMK-AIADMK communication is reportedly underway to block Vijay from the Chief Minister’s chair. Sixty years of two-party dominance collapsed on May 4; what replaces it is still being negotiated.
West Bengal is a more troubling story. The BJP won 207 seats - a majority by any measure. But Al Jazeera’s constituency-level analysis documents that in at least 48 seats, the number of voters flagged, deleted, or under adjudication through the Special Intensive Revision process exceeded the winning margin. The SIR, using AI-assisted transliteration software, disproportionately flagged Muslim names for deletion. The Supreme Court ruled that voters with unresolved appeals could not vote; 3.4 million appeals remained pending on polling day. Mamata Banerjee’s TMC lost. Whether the election reflects the will of Bengal’s voters is a different and harder question. (See Disambiguation for a full breakdown of SIR and how it relates to NPR and NRC.)
My take: For readers of “The Hate Effect” - this is what administrative disenfranchisement looks like at scale. The SIR was piloted in Bihar in June 2025 and then expanded to nine states and three Union Territories. West Bengal, with a 27 per cent Muslim population and a BJP that had repeatedly failed there, was where the revision was most consequential. Communal politics increasingly operates through legal and administrative channels rather than through visible violence, and the Bengal result illustrates that with unusual clarity.
Feature: Heat Is Policy
India’s summer 2026 is running above the seasonal average, with Reuters reporting above-average heatwave days forecast through May. But two pieces published this week reframe the story entirely, from weather event to governance failure.
The first is a Harvard Salata Institute white paper on extreme heat in India, covered by Mongabay India on May 5. Its central argument, from climate fellow Kartikeya Bhatotia: “We are still treating heat as an episodic disaster rather than a structural, economy-wide risk.” Roughly 380 million people - three-fourths of India’s workforce - are engaged in heat-exposed labour. The paper also warns that historical temperature records in northern India underestimate future risk because cooling mechanisms (aerosols from crop burning and irrigation-based evapotranspiration) may weaken over the coming decades. Planning calibrated to historical averages will systematically underestimate the exposures populations will face.
The paper also identifies a data gap directly relevant to Census 2027: private providers account for the majority of inpatient care in India but are largely absent from national health registries. “We are trying to plan for a systemic risk with partial visibility, which inevitably leads to underestimation,” Bhatotia says.
The second piece is IndiaSpend’s report on how extreme heat affects India’s informal women workers. The numbers are striking. An MSSRF survey of more than 3,300 women across 15 districts found that 97 per cent reported income losses during peak summer months, averaging over Rs 1,500 per month. A South India study of 3,000 informal workers found that a one-unit increase in Wet Bulb Globe Temperature raised the likelihood of productivity loss by 339 per cent for women. In Delhi, shaded workers lost about 20 per cent of output on hot days; unshaded workers lost about 25 per cent. For piece-rate workers, productivity loss and income loss are identical.
The IndiaSpend piece identifies two structural drivers that are specific to women. Women devote over five hours a day to unpaid domestic work (compared to just over an hour for men, per the Time Use Survey 2024). When a heatwave cuts paid work hours, the total workload does not fall - income does. And women are concentrated in the highest-poverty-risk employment categories: own-account workers, casual labourers, and contributing family workers.
Feminism in India’s May 6 piece adds the policy audit: India’s Heat Action Plans, assessed in a 2024 CPR review of 37 plans across 18 states, found that only two conducted vulnerability assessments, only 11 included funding mechanisms, and almost none addressed income loss for informal workers. HAPs list women, children, and informal workers as vulnerable groups. Targeted interventions for those groups remain rare.
My take: JanVayu’s work is built on the premise that air quality data and governance are inseparable - you cannot respond to what you cannot measure. The Harvard paper makes exactly that argument about heat: the data architecture for tracking heat’s economic and health burden does not exist at the level of granularity needed for policy design. Missing: private hospital data, gender-disaggregated income loss data, and informal sector productivity data. What we have instead is a proliferation of Heat Action Plans that are rarely backed by law, rarely funded, and rarely enforced - and are calibrated to historical baselines that may already be obsolete. This is not a climate story. It is a welfare measurement story.
Must-Reads This Month
1. The largest fall in development aid on record
| OECD Data Explainer, April 2026 | Open access |
ODA from DAC members fell 23.1 per cent in 2025, from roughly $226 billion to $174.3 billion - the largest single-year contraction ever recorded. The OECD projects a further 5.8 per cent decline in 2026. Multilateral ODA has now fallen for two consecutive years. Oxfam’s response notes that the headline figure is partly propped up by climate finance - much of it concessional loans rather than grants. African Business’s April piece names the donors who cut most: the US in absolute terms, Germany redirecting ODA to defence.
My take: For India-based civil society, the collapse of ODA and FCRA tightening are occurring simultaneously. The organisations being defunded by FCRA compliance pressure are frequently those doing welfare delivery monitoring and citizen-facing accountability work. The Census is in the field; the groups best positioned to audit whether enumeration is reaching hard-to-reach settlements are being squeezed from both ends simultaneously. That is a research infrastructure crisis, not a funding inconvenience.
2. India’s gig economy is growing faster than its protections
| East Asia Forum, April 9, 2026 | Open access |
The Code on Social Security came into force in November 2025, recognising gig and platform workers for the first time and requiring platforms to contribute 1-2 per cent of annual turnover to a Social Security Fund. Six months in: no concrete central scheme exists. The Code uses the word “may” rather than “shall” when providing for welfare, rendering the entire framework discretionary. Access is conditioned on 90 days of engagement with a single aggregator, a threshold that systematically excludes workers for whom platform-switching and deactivation are structural features of the job. Fewer than 1 per cent of India’s delivery workers are women. Platform algorithms concentrate peak earning windows in the evening hours, when domestic responsibilities are heaviest; gig workers classified as independent contractors have no institutional recourse against client harassment.
Author Kasim Saiyyad’s central argument: India is “digitising informality and calling it reform.” The EU’s 2024 Platform Work Directive establishes a presumption of employment and mandates algorithmic transparency. Singapore’s Platform Workers Act phases in mandatory social security contributions without requiring reclassification. India’s framework avoids the underlying question of whether platforms exercising employer-like control should carry employer-like obligations.
My take: The 90-day threshold is a specific policy design element that warrants scrutiny in evaluation work. Any programme that creates a binary eligibility cliff at an arbitrary threshold - rather than proportionate accrual - will produce perverse outcomes at the margin. A worker who completes 89 days gets nothing; one who completes 90 gets full coverage. In a sector built on volatility, that is, policy design working against the population it claims to serve. Worth reading alongside the PLFS data showing 40 per cent of gig workers earn under Rs 15,000/month before costs.
3. India’s poverty rate: 5 per cent or 24 per cent?
| Frontline, 2026 | Open access |
The World Bank’s $3.00/day line (2021 PPP) gives India a poverty rate of 5.3 per cent. The LMIC line ($4.20/day) - the threshold the World Bank’s own India Development Update recommends for India’s income classification - gives 24 per cent. Data for India’s explainer traces how the shift from URP to MMRP consumption measurement lowered measured poverty mechanically rather than through welfare improvement. The gap between 5 per cent and 24 per cent is entirely a function of which line you choose. Both numbers circulate in policy debates with equal authority.
My take: This is The Measurement Trap’s central argument made visible in a single statistic. An India with 5 per cent poverty looks like a country that has largely solved its development challenge. An India with 24 per cent poverty looks like a country where one in four people cannot meet basic needs. These represent different claims about what the state is obligated to do. The choice of line is a political decision, and the poverty line debate is most honest when practitioners say so explicitly.
Census Watch: Phase 1 in the Field
Census 2027 Phase 1 - the House Listing and Housing Census - began April 1 and runs through September 30. This is India’s first fully digital census, with self-enumeration available online and door-to-door digital enumeration by field officers. The state-wise schedule is staggered: Bihar and several other states are in active self-enumeration through May 31; door-to-door enumeration follows.
Business Standard’s April 1 coverage notes that this census will also include caste enumeration - a politically significant first since 1931. The census portal has a self-enumeration guide with state-specific dates.
The risks that were underreported in April remain underreported in May: 27 per cent rural functional digital literacy, poor connectivity in many districts where undercounting has historically been most severe, and a compressed verification timeline. The Mongabay heat paper, referenced above, also notes that working with “outdated population baselines due to our delayed census” distorts India’s ability to plan for heat risk at all. The census is infrastructure. Every welfare programme, climate adaptation plan, and electoral roll that comes after it will rest on what it captures - and what it misses.
If you are in an area where self-enumeration is open, the portal is at censusindia.gov.in. Worth verifying your household is correctly enumerated, particularly if you live in a densely populated urban informal settlement or a migration-heavy district.
Research Praxis: What a Poverty Line Actually Is
Every development researcher uses a poverty line. Most of us inherited the one our institution adopted without thinking carefully about what it encodes.
A poverty line is a consumption or income threshold below which a person or household is classified as poor. The most commonly cited global lines are the World Bank’s $2.15/day (extreme poverty), $3.00/day (low-income country standard), and $4.20/day (lower-middle-income country standard). India’s Tendulkar poverty line - the last official national benchmark - sits at roughly Rs 1,622 per person per month in rural areas at 2011-12 prices. Adjusted for inflation, this falls well below any of the international lines. As Utsa Patnaik and Jean Dreze have argued, it was calibrated below caloric adequacy thresholds from the start.
Three things to track whenever you read poverty data:
The reference period. India’s HCES 2022-23 is the first comparable consumption survey in fifteen years. Any poverty estimate prior to its release was extrapolated from 2011-12 data using modelling. The assumptions embedded in that extrapolation - which consumption components grew, at what rate - were rarely stated and are rarely audited.
The consumption basket. The HCES 2022-23 uses MMRP (Modified Mixed Recall Period), which captures food and non-food consumption differently from the URP method used in previous rounds. Switching from URP to MMRP reduces measured poverty, not because welfare improved, but because the survey methodology changed. Cross-period comparisons without explicit adjustment are comparisons of different instruments.
The counterfactual. If the poverty rate fell from 27 per cent in 2011-12 to 5 per cent in 2022-23 at the $3.00 line, what drove it? The World Bank attributes the fall to real consumption growth. Researchers, including Dipa Sinha and Himanshu, argue that a significant portion reflects the methodological shift. Clean causal attribution is unavailable given the data breaks; studies that claim it are overstating their evidence.
For evaluation practitioners: if your programme targets households below a particular poverty line and tracks outcomes against it, your results are only as good as your line. A programme that “lifted” 40 per cent of beneficiaries above the $2.15 line may have moved them to Rs 65 per day, which in Delhi today buys approximately half a meal.
Disambiguation: SIR, NPR, NRC
The West Bengal election cemented SIR (Special Intensive Revision) into the general political vocabulary. Here is a brief guide to how it relates to other roll-related acronyms that recur.
SIR: A process under Section 21 of the Representation of the People Act, 1950, through which the Election Commission can comprehensively review and update electoral rolls. It involves house-to-house verification by Booth Level Officers, followed by a claims-and-objections window. The 2025-26 SIR was notable for its scale (nine states and three UTs simultaneously), its speed, and the use of AI-assisted transliteration software that flagged names for deletion based on script inconsistencies.
NPR (National Population Register): A register of residents maintained under the Citizenship Act. The NPR forms the base for a potential national NRC. The last NPR update was scheduled alongside the 2021 Census. With Census 2027 Phase 1 underway, an updated NPR process may be relaunched.
NRC (National Register of Citizens): A register of citizens determining who legally belongs to India. Currently complete only in Assam, where the 2013-2019 update excluded 1.9 million people. A national NRC has been announced but remains unimplemented.
The critical distinction for practitioners: SIR governs electoral eligibility (who can vote), while NPR/NRC governs citizenship status (who belongs). They are legally separate. They share the same infrastructure: documentary proof, house-to-house verification, and track records of systematically excluding poorer, more mobile, and minority populations, who have the least consistent documentation.
The Bengal SIR operated on the electoral roll rather than the citizenship register, but its effect on Muslim voters’ practical ability to exercise citizenship rights was structurally similar to that of an NRC exercise. This distinction will matter for advocacy and litigation work going forward.
From My Desk
The Measurement Trap - my book is in progress on the politics of welfare measurement. This issue is almost its chapter list: electoral rolls as measurement infrastructure, poverty lines as political choices, census delay as governance strategy, ODA contraction as a data vacuum, Heat Action Plans calibrated to metrics that do not capture what matters. Progress updates to follow.
The Count That Was Never Taken: Census 2027 and the Politics of Not Knowing - Written before Phase 1 launched; the risks flagged there are now live. Digital enumeration gaps and hard-to-enumerate settlements remain the undercovered story.
The Unaccountable State - On FCRA tightening and the defunding of civil society data infrastructure. More urgent now that ODA has also contracted by 23 per cent.
ImpactMojo is running a free workshop on welfare measurement methodology in June - dates to be confirmed. The poverty line module and the consumption survey module are the most directly relevant to this month’s debates. Free, open-access, practitioner-focused.
Visual Treat
The Guardian’s constituency-level maps showing West Bengal electoral roll deletion rates against BJP winning margins are the most analytically useful election visualisations from this cycle. The argument for what the SIR did to the Bengal result is made visually in a way that individual statistics cannot fully convey.
Funsies
A dispatch from Indian constitutional theatre, May 2026 edition: the DMK and the AIADMK - bitter rivals for sixty years, parties born from a single rupture in 1972, whose competition defined Tamil Nadu through four generations - are reportedly in backchannel communication to jointly block a film actor from becoming Chief Minister. MK Stalin, who spent thirty years fighting for the CM’s chair, sits in it as caretaker, reportedly willing to reach an understanding with the man he spent his career opposing, just to keep Vijay out. The TVK MLAs are reportedly camped at Puducherry resorts to avoid pressure from poachers. The people of Tamil Nadu voted for none of this. The whistle - TVK’s party symbol - is being blown outside a Governor’s office that has so far declined to let the winning party win.
Indian politics always finds new levels.
You are reading the Research Rundown, a bi-monthly newsletter on development research and practice from Varna Sri Raman. Subscribe at varna.substack.com. If this was useful, the best thing you can do is send it to someone who would find it equally useful.