Economist and behavioral scientist. I study how people decide, and what follows for policy.
I founded the World Bank's behavioral science unit. For two decades my work has been on how people actually decide, and what policy design should do about it. Since 2023 I have also built the systems that put evidence in front of the people who use it, in more than 100 countries, and set up the measurement to know whether it worked.
My work sits between economics and psychology. A school with poor attendance and a clinic nobody uses often share a behavioral cause. That is usually where the useful question is.
As founder and head of eMBeD, the World Bank's behavioral science unit, I built an interdisciplinary team to diagnose what drives behavior, then design and test interventions. The work covers learning, public services, health, and the social norms around how we use natural resources, with governments and project teams in more than 100 countries. Behavioral design is now built directly into more than $10 billion of World Bank lending a year. The unit's methods, tools and training reach further, across the roughly $70 billion the Bank lends each year, shaping how projects are designed and how they speak to the people they serve. The unit sits inside the Bank's Development Impact Department.
Evidence is only useful if it reaches the people making the decision. Since 2023 I have also worked on that side of it through the Institute for Economic Development: knowledge diffusion, impact analytics, applied AI.
I build alongside the teams that run these systems, in Python and with the current model APIs and agent tooling, and I set up the evaluation before the launch rather than after. The unglamorous part, making a tool trustworthy enough that a ministry will actually use it, is most of the work.
The question is always the same. Why do people do what they do, and what would change it.
Why I build these the way I do: what I believe about AI for policy, in about four minutes.
An assistant grounded in the World Bank's report base that answers with page-level citations and abstains when the evidence is insufficient. I adapted it to that evidence base and set up its ongoing evaluation, including a peer-reviewed study of how it is actually used in the field.
A platform that turns research into structured courses, with AI built into both design and delivery. A personalized AI tutor is in development.
An AI-assisted, editorially controlled podcast distilling recent research into short briefings, published in several language editions. Research nobody has time to read does not get used.
Development projects depend on people acting, yet the communication meant to influence that is briefed as production and rarely measured. STORY works backwards from the outcome instead: segment who needs to act, target message and messenger, test before spending, then measure against the project's own results framework.
Built at eMBeD · now in pilot with country teams
From a platform supporting youth employment to a tested intervention that builds resistance to health misinformation.
The papers are in the record below. This is the rest: where I have tried to say the same things in plain language. All World Bank posts are on one page.
Notes from building evidence tools for governments. Trust is earned by what a tool refuses to say, and usage is not impact.
ACM CHI 2026, with Karnatak, Chatila, Pinzón Hernández, Yazdanfar and Dugas. What the first users taught us about citations, abstaining, and time saved.
Entertaining Change: next-generation media partnerships for social impact and gender equality. World Bank, Abuja, October 2025. Replay.
A long conversation with Merle van den Akker on eMBeD, what worked, and what I would do differently.
A short film about the unit and the way it works with governments.
World Bank Blogs, with Tomio, Rawlins, Vongthavilay and Pinzón Hernández.
A Load of BS, with Zeina Afif and Ana María Muñoz Boudet.
World Bank Blogs, with Sousa Lourenço and De Castro Zoratto. Policymakers have biases too.
World Bank Blogs, with Bidani, Menon, Nguyen and Afif.
World Bank Blogs, with Carolina Sánchez-Páramo. The short case for the whole approach.
Ten selected pieces first; the full record of 47 is one click away. Peer-reviewed articles, books and policy volumes, and current working papers. Where a working paper has since been published, only the published version appears.
Google Scholar · 6,300+ citations · h-index 32
I have spent twenty years on how people actually decide, and on designing things that work with the grain of human behavior rather than against it. Economics, psychology, experiments, and lately AI: the discipline follows the problem, not the other way around. If that is what you are working on, I would be glad to compare notes.
Renos Vakis is an economist and behavioral scientist at the World Bank, where he founded eMBeD, the Bank's behavioral science unit, and leads work at the Institute for Economic Development on applied AI, learning and knowledge diffusion. He builds tools that put evidence in front of policymakers at scale, and measures whether they work. Berkeley Ph.D.; co-chair of the Global Behavioral Evidence Network for Development.
Renos Vakis brings behavioral science to the forefront of policy design, shaping how governments address some of the world's most pressing development challenges. As head of the Mind, Behavior, and Development Unit (eMBeD) in the World Bank's Development Impact Department, he leads an interdisciplinary team working with governments and project teams in more than 100 countries: on human development, from school performance to healthier behaviors, mental well-being and aspirations; on government effectiveness, from service delivery bottlenecks and policy uptake to frontline accountability; and on climate, from adaptation to the social norms that govern how resources are used. Since 2023 he has also led work at the Institute for Economic Development on knowledge diffusion, impact analytics and applied AI, building the systems that scale behavioral solutions, accelerate experimentation, and make evidence usable by policy and practitioner communities. The work sits deliberately across disciplines, connecting economics, psychology and technology rather than treating them as separate problems. He holds a Ph.D. in Agricultural and Resource Economics from the University of California, Berkeley, and his research has appeared in The Economic Journal, American Economic Journal: Applied Economics, Science Advances and the Journal of Development Economics. He teaches behavioral insights for policy at the University of Cyprus and advises governments, universities and development partners worldwide. He also brings magic and music to the work, on the view that development practice should be engaging as well as rigorous.