VoxDev Development Economics

This is an episode from VoxDev's new podcast series, Ideas in Development. This series has a separate podcast feed, where you can find every episode of Oliver Hanney’s conversations on evidence.
YouTube: https://www.youtube.com/watch?v=I-e0GAFGaJ0 
Apple Podcasts: https://podcasts.apple.com/us/podcast/moving-billions-towards-evidence/id1866874059?i=1000772910756 
Spotify: https://open.spotify.com/episode/12UqCmkB4oLLWqoHpT5Lyt?si=6be55d2c81884d65 
Audioboom: https://audioboom.com/posts/8917092-moving-billions-towards-evidence 
Substack: https://ideasindevelopment.substack.com/p/moving-billions-towards-evidence

In 2022, Dean Karlan became Chief Economist at USAID, tasked with steering the world's largest bilateral aid agency towards evidence-backed approaches. He left in 2025, as the agency was being dismantled, having moved roughly $1.7 billion of funding in the process.
In this episode of Ideas in Development, Dean joins Oliver Hanney to discuss what evidence-based policy actually looks like inside a government institution; how his team picked their battles; why collaboration beat prescription; and where the limits of taking goals as given lie. They also cover the rise of embedded evidence labs in countries like Rwanda and Peru, the synthesis and implementation gaps between academia and policy, and whether there are questions in development economics, like the impacts of cash transfers, on which we now have enough evidence.
Dean Karlan is Professor of Economics and Finance at Northwestern University, founder of Innovations for Poverty Action, and former Chief Economist of USAID.

A farmer in rural Ghana can use her phone to take a picture of a blighted leaf, upload it, and within seconds she gets a diagnosis that takes account of the soil, the weather, and the other local conditions. Score one for AI. We don't have enough human experts because they are hard to train, and harder still to retain, so apps like this have huge potential. But it doesn't mean they will succeed.
Iqbal Dhaliwal, global executive director at J-PAL, has watched ambitious tech dreams for development fall apart before. Remember one laptop per child, the plan that failed because schools didn't have technical support, the right lesson plans -- or even electricity? He tells Tim Phillips that AI can succeed, as long as whoever is using it has thought through the theory of change, designs for impact rather than downloads, and can scale up successful experiments. J-PAL has launched the AI Evidence Playbook to help policymakers solve these knotty problems.
The research behind this episode

Dhaliwal, Iqbal, Sam Carter, Attaullah Abbasi, and Audrey Lorvo. 2026. AI Evidence Playbook: A Practical Guide. Cambridge, MA: Abdul Latif Jameel Poverty Action Lab (J-PAL).
To cite this episode

Phillips, Tim, and Iqbal Dhaliwal. 2026. "How AI can put 20 years of development evidence to work." VoxDev Talks (podcast).

About the guest

Iqbal Dhaliwal is global executive director of the Abdul Latif Jameel Poverty Action Lab (J-PAL), based at MIT, and co-chairs Project AI Evidence. He co-directs J-PAL's South Asia office with Esther Duflo. Before joining J-PAL in 2009 he served in the Indian Administrative Service, where he ran a statewide welfare department and led a public company, delivering large-scale programmes in the field. His work spans the design, evaluation, and scale-up of anti-poverty programmes, and the question of when a technology genuinely changes lives rather than dashboards.
Research and concepts cited in this episode

Project AI Evidence (PAIE) is J-PAL's initiative to identify, evaluate, and scale applications of AI for social good, and to scale down those that may cause harm. The AI Evidence Playbook is its practical output, a reference for policymakers, practitioners, and donors weighing whether and how to adopt AI-enabled programmes.
The six pathways. The playbook groups two decades of development evidence into six areas where AI could raise impact or cut cost: improving needs prediction and targeting; increasing access to personalised, timely support; maximising the effectiveness of frontline service providers; improving organisational and programmatic efficiency; reducing bias and ensuring fairness; and boosting government resource mobilisation, including more progressive taxation.
Machine learning targeting in Togo. During the Covid-19 pandemic, Togo's government wanted to reach its poorest households but lacked the administrative data to find them. Researchers used satellite imagery to identify likely-poor neighbourhoods from features such as house size and roof quality, then paired it with mobile phone records to narrow down poorer individuals within them; a practical, rapid way to fill a data gap without a social registry.
Theory of change. The sequence of steps connecting an input to a final outcome. Dhaliwal's point is that most people hold an optimistic hypothesis rather than a theory of change: the inputs and the hoped-for outcome are clear, but the intermediary links, adoption, trust, workflows, repair, are missing. Writing it on paper is where the gaps show.
Design for impact, design for scale. Engagement numbers and downloads are a first step, not proof of impact. Designing for scale means designing for a farmer who may lack a device, may lack reliable connectivity, and may not trust the technology without the extension worker they have known for fifteen years.
One laptop per child, and smokeless stoves. Dhaliwal raises both as technologies whose theory of change was sound but whose delivery was not thought through: the school system could not absorb the laptop, the electricity and connectivity were not there, and no one planned for breakage. The relevant cautionary parallel for AI, whose problems he argues are not just inherited but exacerbated by the pace of change.
Global knowledge, local context. Dhaliwal separates rules that can be changed from genuine context differences. A programme that works in India but "won't work in Ghana" because teachers there work six hours rather than six hours fifteen minutes is a rule problem, and solvable. A classroom of 1:20 versus 1:60 is a real contextual difference worth worrying about.
More VoxDev Talks episodes

The development economics of AI: Lessons and questions. Oliver Hanney and Deena Mousa take stock of what an entire series of conversations revealed about where AI helps in development, and where the evidence runs thin.
Related reading on VoxDev.org

AI and development economics: Early evidence and how to keep up, a running VoxDev reading list on the impacts of AI in low- and middle-income countries that includes J-PAL's AI Evidence Playbook among its resources.

If you walk down a street in a low-income country, count the businesses you can see, you will miss many entirely. Someone who cooks at home and sells at the roadside never appears in a registry. What we know about businesses in developing countries has always been incomplete. 
In this week's VoxDev Talk, Marcela Eslava (Universidad de los Andes Bogota) talks to Tim Phillips about two decades of research into why firms in developing economies stay small, grow slowly, and rarely break through. A worker in a developing economy is about three times more likely to be running, or working in, one of these very small businesses than a worker in a high income economy, and far more likely to be self-employed, a one-person firm. 
Weaker human capital and less access to technology make it harder to start a business. Costlier credit, labour regulation that makes it difficult and expensive to employ staff, and taxes on formal firms make it harder to grow. Some of these distortions are even introduced by well-meaning policymakers, such as low taxes for small firms - which then effectively tax growth.
The research behind this episode:
Eslava, Marcela. 2026. "Firm Size and Dynamics in Less-Developed Economies." Annual Review of Economics, volume 18. Review in advance; changes may still occur before final publication.
To cite this episode:
Phillips, Tim, and Marcela Eslava. 2026. "Why businesses stay small in emerging markets." VoxDev Talks (podcast).

About the guest

Marcela Eslava is Professor of Economics and Dean of the Faculty of Economics at Universidad de los Andes in Bogota¡, and President of the Latin American and the Caribbean Economic Association. Her research spans firm dynamics, productivity and demand at the firm level, labour informality, credit constraints, and the regulations that shape how businesses grow in developing economies.
Research cited in this episode

The Lucas-Hopenhayn framework. The workhorse model of firm dynamics, built on Robert Lucas's 1978 account of occupational choice and Hugo Hopenhayn's 1992 model of entry and exit; it links the size of the average firm to the productivity of potential entrepreneurs and to distortions that stop the most efficient firms from scaling. Eslava's review extends it to explain why firms are smaller in poorer economies.
Nonemployers and micro employers. Own-account and self-employed workers plus the smallest employers; in Eslava and co-authors' sample of 51 economies they account for over 90% of employment in lower-middle-income economies and around 20% in the United States. Their dominance is the main reason average firm size is so low in poorer countries.
Correlated distortions. Taxes, regulations and credit frictions that fall more heavily on high-productivity firms; because they hold back the firms that should be growing, they push labour towards less productive businesses and drag down aggregate output. They show up in the models as residuals, the part of the size gap that productivity and quality cannot explain.
Size-dependent policies. Rules that switch on above an employment threshold, such as the extra workplace obligations that apply to firms above 50 employees in France, or above 20 in Peru; Eslava describes comparing firms just above and just below a threshold as one way to prise open the black box of distortions and measure what a specific rule does.
Demand versus cost efficiency. Work on Colombian manufacturing decomposing a firm's success into technical efficiency and the appeal of its products; the demand side contributes many times more to differences in revenue than cost efficiency does. Eslava's point is that trimming costs buys survival, while product quality and market access drive real growth.
The data-coverage puzzle. Early firm-level studies found average firm size falling as countries got richer, the reverse of what later work established; the anomaly came from datasets that captured only the large, registered firms in poor economies and missed the vast base of tiny ones. Better censuses, and household and employment surveys that count people rather than firms, corrected the picture.
More VoxDev Talks episodes

What have we learned about the informal sector? Gabriel Ulyssea and Mariaflavia Harari on the causes and consequences of informality, the close cousin of micro-enterprise that runs through this conversation.
Related reading on VoxDev

Can variation in firm growth explain the development gap? Colombia vs the US, by Marcela Eslava, John Haltiwanger and Alvaro Pinzón, on how a deficit of superstar plants and an excess of surviving underperformers shape Colombia's development problem.
The links between capabilities and export dynamics in developing countries, on how firm capabilities, product quality and market knowledge shape which firms manage to export.

A river dries up. The soil turns salty. A harvest fails. Farmers across the developing world feel climate change constantly and personally, they rarely blame their government, or demand action.
Guy Grossman (University of Pennsylvania) is one of three authors of a new review of the politics of climate change in the developing world. He tells Tim Phillips that almost all of the existing research on this topic is focused on rich countries, even though the developing world faces the worst of the damage,  and has the least capacity to absorb it, because in those countries the link between climate change and political action is more explicit.
Political solutions are needed: developing country income losses could run 60% higher than losses in wealthy countries, and climate change could push between 32 and 132 million people into extreme poverty within a decade. Grossman's review turns up a paradox in the public opinion data. Concern runs high even where formal climate literacy is low, because people experience the crisis through a failed harvest or a dried up well, not a scientific chart. This disconnect isn't neutral, because vulnerability isn't simply inherited. It is produced, by decisions about who owns land, whose villages get seawalls, and whose voice counts when climate money is handed out.
The research behind this episode:
Grossman, Guy, Audrey Sacks, and Alice Xu. 2026. "The Politics of Climate Change in the Developing World." Annual Review of Political Science 29: 101-126.
To cite this episode:
Phillips, Tim, and Guy Grossman. 2026. "Climate Change Politics in Developing Countries." VoxDev Talks (podcast).

About the guest

Guy Grossman is the David M. Knott Professor of Global Politics and International Relations in the Department of Political Science at the University of Pennsylvania. He founded and co-directs Penn's Development Research Initiative (PDRI-DevLab), and his research spans governance, forced displacement, political accountability, and conflict processes across the developing world, with a particular regional focus on Sub-Saharan Africa.
Research cited in this episode

Extreme poverty projections. World Bank economists Bramka Arga Jafino, Stephane Hallegatte, Julie Rozenberg, and Brian Walsh estimate that climate change could push between 32 and 132 million people into extreme poverty by 2030; the wide range reflects uncertainty over which emissions and development pathway the world follows. Read the working paper.

Afrobarometer. A long running, pan African survey network covering more than 30 countries. Grossman's review draws on it to show that only around four in ten respondents identify human activity as the main cause of climate change, even as concern about its effects runs far higher.
The attitudinal and accountability channels. Two frameworks political scientists use to trace how climate exposure might change political behaviour. The attitudinal channel asks whether living through a flood or a drought changes what someone believes about climate change; the accountability channel asks whether it changes their vote. Grossman finds evidence for both, but little that explains when concern turns into political pressure.
Maladaptation. The academic term for private adaptation that shifts harm onto someone else, such as a village embankment that protects one community by pushing floodwater into the next. Grossman uses it to illustrate why adaptation without government coordination can widen inequality rather than close it.
Ecuador land titling. Mark Buntaine, Stuart Hamilton, and Marco Millones's 2015 study of a titling programme in Morona Santiago found it did almost nothing to slow deforestation, because the state never backed the new titles with enforcement. Grossman cites it as evidence that representation without power tends to fail.
Indigenous managed land. Research led by Stephen Garnett finds that Indigenous peoples, roughly 6.2% of the world's population, manage more than a quarter of the planet's land surface, often protecting carbon sinks more effectively than formally designated protected areas.
More VoxDev Talks episodes

Financing climate adaptation: what works, what doesn't, and can carbon credits help to bridge the gap? Namrata Kala, Rohini Pande, and Catherine Wolfram pick up where Grossman leaves off, on who pays for adaptation when governments won't.
How the urban environment can adapt to climate change. Matthew Kahn and Siqi Zheng discuss how cities in the developing world can adapt their buildings and infrastructure as climate driven migration accelerates.
Related reading on VoxDev.org

Climate politics: understanding political inaction on climate change. Allan Hsiao and Nicholas Kuipers show that Indonesian politicians underestimate voter concern about climate and pollution, and that correcting their misperceptions does not, on its own, produce policy action; a real world case of the accountability channel breaking down.
Political representation and forest conservation? This finds that transferring formal political power, not just consultation, to India's historically marginalised Scheduled Tribes led to a measurable fall in deforestation.

This is an episode from VoxDev's new podcast series, Ideas in Development. This series has a separate podcast feed, where you can find every episode of Oliver Hanney’s conversations on evidence.
YouTube: https://www.youtube.com/watch?v=EacHFVRt9p4 
Apple Podcasts: https://podcasts.apple.com/us/podcast/has-development-economics-lost-its-way/id1866874059?i=1000775748550 
Spotify: https://open.spotify.com/episode/2Lcy3FrbBuoE2nj3cnhOAm?si=76aedb574426479e 
Audioboom: https://audioboom.com/posts/8924691-has-development-economics-lost-its-way 
Substack: https://ideasindevelopment.substack.com/p/has-development-economics-lost-its 
What should development economists be working on – and how does their work actually reach the people making decisions?
Rachel Glennerster, President of the Center for Global Development, whose career spans the research and policy sides of development, joins Oliver Hanney to discuss her proposal for a radical simplification of aid, why she feels the micro-macro debate is largely a false one, the messy but vital process of building consensus, and what impactful careers look like in economics.
In this wide-ranging conversation, we cover the Smart Buys evidence panels in education and how cross-disciplinary consensus gets built, her three-box framework for evidence-based policymaking, why AI tools move too fast for RCT-based procurement, and what it would take to fix development economics' concentration problem.