A course I teach develops two ideas:
1. Poverty is a lack of money. Giving people money relieves them from poverty.
2. However, in many cases people face specific constraints that prevent them from doing the best they can.
In (2), the constraints I deal with are “static market failures” that arise because of a problem with the way society is distributing resources. For instance, perhaps one group is super entrepreneurial and would develop fantastic technologies if they only had the money, but the money is going to an entrenched, unproductive elite. Or, people would be willing to pay for health insurance, but a market for health insurance does not exist.
A key lesson I learnt from Jeff Hammer was that if we could only (a) agree to give people cash and (b) resolve, or at least alleviate, some of these market failures, life would be much better for the poor. Incredibly, and despite more than 100 years of consensus among economists, alleviating market failures has proven incredibly hard. Monopolies (a market failure that restricts trade) are roaring back and climate change (a market failure we call externalities) is devastating our planet.
So, generating the evidence that we can improve the lives of the poor by solving market failures and distributing cash is one of the most important topics we can tackle today. It is also an incredibly difficult job because the evidence has to be strong enough to battle an adversarial and entrenched elite. For instance, how would you counter these questions:
(a) If you give poor people money, they would just drink it away? (a pure economist would respect people’s preferences, but you can see where that leads…)
OR
(b) It is correct that the private market for education or health may not work well because consumers do not have information, but will solving this constraint really be enough? Isn’t the real problem that schools and households need our direct support?
OR
(c) If governments could alleviate these constraints, things would be better. But we have to balance market failures against government failures and we all know that the government is corrupt and inefficient, so why invest in the government?
Perhaps surprisingly, development economics has made fundamental contributions to each of these problems over the last 20 years. There has been *a lot* of work on such GIANT problems, not just tweaks around the edges. More strikingly, progress on these problems owes a lot to the use of RCTs. RCTs have allowed us to say that the outcomes we see are we see are NOT a reflection of the different preferences of the poor, but the constraints they face. Why? Because the people for whom we alleviated the constraint were exactly the same as the group where we did nothing. The RCT vs. non-RCT debate is entirely the wrong one—the question is one of the contributions of the work, not the methods used.
I will give three examples, all using RCTs:
Example 1: What happens when you give people cash.
Example 2: Does solving market failures improve outcomes, even in non-traditional sectors like education?
Example 3: Is the government corrupt and inefficient and can nothing be done to fix that?
Example 1: If governments gave poor people cash, would they just drink it away?
It is worth highlighting, once again, that an economist takes no view on what people do with the cash they receive. A preoccupation with how the poor use cash grants requires us to assume that the person giving the money (say, the government) has a different welfare function from those of the people it is giving the money to. It is unclear why this should be so, but that argument never holds water….
So, a development economist, Paul Niehaus, set up GiveDirectly, which has become a powerhouse in the philanthropy community with critical expertise on a single, focused task—giving people money while keeping administrative costs low. They are now doing some of the most fundamental experiments in the world around the role of cash and the alleviation of poverty. Many of these RCTs allow us to understand precisely what happens when you give cash to the poor.
There are four main results that have now been shown to hold across multiple studies.
RESULT 1: When you give people cash, their consumption of all kinds of things improve—from spending on education and health to decreasing debt.
RESULT 2: When you give people cash, they do not spend more on “temptation goods” like tobacco and alcohol.
RESULT 3: When you give people cash, they do not become lazy welfare recipients. In fact, they tend to work more.
RESULT 4: When you give cash to a lot of people in a community, the entire community benefits. This result, known as the “government multiplier” shows that giving $100 increases local activity so that total “GDP” increases to $250, give or take.
These results have completely changed the discussion around the form of social protection programs.
Example 2: Solving market failures improves outcomes—even in non-traditional sectors like education.
The dominant `problem’ of education in the 1990s was insufficient demand. Apparently, people did not want to send their children to school—or so it seemed. But, around the turn of the century, my co-author Tahir Andrabi noticed that villages in rural Pakistan were seeing tremendous growth in private schooling. Instead of insufficient demand, we thought that the question would quickly turn to the role of the government in (what we call) “education markets,” with people demanding higher quality schools and private schools arising to address that demand. Taking an economist’s viewpoint, we set out to see if education outcomes could be improved just by solving market failures.
To demonstrate this approach, we introduced the new idea of `market-level randomizations.’ Under a market-level randomization, instead of working with children or schools, the researcher works with an entire market (a village in our case) with multiple schools. The researcher then introduces a new innovation in some markets, randomly chosen, but not others. We then assess what happens, which could include a whole range of complex reactions from schools improving, children switching schools, private schools shutting down or new private schools entering.
Our first market-level randomization took on the idea that the lack of information was a market failure in the education sector. We tested children in Grade 3, and then in randomly selected villages, we provided `report cards’ detailing the child’s performance as well as that of the school, compared to other schools in the village.
After 2 years we found significant improvements in the test scores of children in the report card villages (compared to the control) and a significant decline in the fees of the private schools. These results were precisely in line with an economic model of asymmetric information, first articulated by George Akerlof in his famous `lemons’ paper. And they showed something quite remarkable—you could improve the quality of education while simultaneously decreasing its price. That is exactly what solving a market failure should do, since these are the equivalent of $-bills lying on the pavement.
We have not publicized these results outside the academic community because we did not know what the longer-term outcomes would be. Instead, have been intensively tracking the children who received the report cards and the markets where the report cards were given for the next 15 years, with a final round in 2017-2018.
Our new papers on the long-term effects of report cards will become available over the next 6 months, but here is a sneak preview:
RESULT 1: The children in 2003 who received the report cards were more likely to go on to high school, graduate high school and go to college.
RESULT 2: The children in 2003 who received the report cards were also more likely to earn higher wages in the labor market.
Putting these two results together alone generates incredibly high returns to the intervention. But a third result raises the returns even more:
RESULT 3: The villages where we gave report cards in 2003, continued to have lower prices and costs and higher test scores eight years later in 2011. That is, we managed to reverse a common finding of declining productivity in education—we are now producing better learning at lower costs than before.
These results are incredibly gratifying. We did nothing to support either schools or parents—we only lifted a constraint that they faced in the schooling markets. But exactly as an economic model of information would suggest, lifting these constraints led to large gains, suggesting both that solving market failures can increase the productivity of education and that, once we lift these constraints, people in villages can figure it out for themselves.
Example 3: Yes, the government often does not work well—but we can fix that
At the turn of the century, many of us were very pessimistic about the ability of our governments to fix problems. You have to understand where we were coming from. I grew up in a very well-to-do neighborhood in Delhi. Without fail from the time I was 7 till I was 18 the electricity would go in the middle of the night and after tossing and turning for 2 hours, my father and I would drive to the electricity office, find a repairman and make it back home after they fixed the problem. Funeral processions would be carried out in cities for phones that had been `dead’ for a long time with no hopes of resurrection. Not surprisingly, then, many of the papers we wrote at that time highlighted the severe problems with government functioning and services in all kinds of sectors.
But this does not mean that state capacity would not develop over time, or that it would not improve, both as our nascent governments figured out how to work better and newer technologies became available.
This is where Karthik Muralidharan’s work has been so important. Muralidharan has been an incredible force working with the Indian government on very large programs to see if these programs can generate better returns for poor people. In many of his studies, he works with the government to randomize programs across entire Indian districts, some as large as small countries. His main result is that you can improve the efficiency of governments by 10X through these programs. Notable examples include:
RESULT 1: Biometric payments in NREGA increase what workers receive—and their wages in the private market.
RESULT 2: Informing officials that farmers would be called to see if they received money from a government program reduced the fraction of farmers who did not receive money by 7.8%.
RESULT 3: Adding a half-time worker to government-run preschools massively increased test scores among children.
Muralidharan has also been clear about programs that did not work. For instance, biometric ID requirements in a social protection program reduced payments to legitimate beneficiaries, and a large school management program had zero impact on the functioning of the schools.
Conclusion
My point here is not to claim that all development economics is great or that we are always looking at the giant, contextually relevant questions or that RCTs are the way forward. Those who have read my previous posts know that I am quite concerned about many aspects of the discipline, and therefore feel that I should be working hard to advocate for change.
But to understand the flaws in a discipline, you have to understand the strengths. And the examples I have illustrated hopefully go to show that development economists ARE working on some of the biggest and hardest problems in the world today and ARE making significant progress. AND this progress is being made using RCTs in highly imaginative ways.
The flaws in the discipline seem to be a more recent turn towards work that is not contextually grounded, but bolstered through an unearned epistemic authority.
But that is a topic for another day.
Read More on cash transfers
Egger, Dennis, Johannes Haushofer, Edward Miguel, Paul Niehaus, and Michael Walker. "General equilibrium effects of cash transfers: experimental evidence from Kenya." Econometrica 90, no. 6 (2022): 2603-2643.
Crosta, Tommaso, Dean Karlan, Finley Ong, Julius Rüschenpöhler, and Christopher R. Udry. Unconditional cash transfers: A Bayesian meta-analysis of randomized evaluations in low and middle income countries. No. w32779. National Bureau of Economic Research, 2024.
Evans, David K., and Anna Popova. "Cash transfers and temptation goods." Economic Development and Cultural Change 65, no. 2 (2017): 189-221.
Read more on market-level experiments in education
Andrabi, Tahir, Jishnu Das, and Asim Ijaz Khwaja. "Report cards: The impact of providing school and child test scores on educational markets." American Economic Review 107, no. 6 (2017): 1535-1563.
Long-term results from this experiment coming soon!
Andrabi, Tahir, Jishnu Das, Asim I. Khwaja, Selcuk Ozyurt, and Niharika Singh. "Upping the ante: The equilibrium effects of unconditional grants to private schools." American Economic Review 110, no. 10 (2020): 3315-3349.
Andrabi, Tahir, Natalie Bau, Jishnu Das, Naureen Karachiwalla, and Asim Ijaz Khwaja. "Crowding in private quality: The equilibrium effects of public spending in education." The Quarterly Journal of Economics 139, no. 4 (2024): 2525-2577.
Read more on improving public sector efficiency
Muralidharan, Karthik, Paul Niehaus, and Sandip Sukhtankar. "General equilibrium effects of (improving) public employment programs: Experimental evidence from India." Econometrica 91, no. 4 (2023): 1261-1295.
Muralidharan, Karthik, Paul Niehaus, Sandip Sukhtankar, and Jeffrey Weaver. "Improving last-mile service delivery using phone-based monitoring." American Economic Journal: Applied Economics 13, no. 2 (2021): 52-82.
Muralidharan, Karthik, Paul Niehaus, and Sandip Sukhtankar. "Identity verification standards in welfare programs: Experimental evidence from India." Review of Economics and Statistics 107, no. 2 (2025): 372-392.
Ganimian, Alejandro J., Karthik Muralidharan, and Christopher R. Walters. "Augmenting state capacity for child development: Experimental evidence from India." Journal of Political Economy 132, no. 5 (2024): 1565-1602.