Randomised Controlled Trials in Economics

How economists borrowed the randomised trial from medicine to test policies, and what the 2019 Nobel prize-winning research on poverty found.

In medicine, new drugs are tested with randomised controlled trials. Patients are randomly assigned either to receive the drug, the treatment group, or not, the control group. Because assignment is random, the two groups are alike on average in every way except the treatment. Any difference in outcomes can be attributed to the drug.

Why randomisation works

Randomisation solves the counterfactual problem. We still cannot see what would have happened to any one person without the treatment. But the control group shows what would have happened, on average, to a group just like the treatment group. Motivation, health, income and everything else are balanced by chance.

Field experiments

From the 1990s onward, economists increasingly ran field experiments: randomised trials in real-world settings such as schools, villages and businesses. In 2019, Abhijit Banerjee, Esther Duflo and Michael Kremer won the Nobel prize for their experimental approach to reducing global poverty.

Their work and that of many others tested questions such as: Does giving out free bed nets increase their use, compared with charging a small price? Do extra teachers improve learning? Does microcredit raise household income? Some results surprised people. For example, several studies found that even small prices sharply reduced take-up of health products like bed nets, supporting free distribution.

Deworming in Kenya

In a well-known study in Kenya, Edward Miguel and Michael Kremer studied a programme that provided deworming medicine, phasing it into schools in stages. Pupils in schools that received treatment attended school more often. Follow-up studies years later found higher earnings for those who had been treated, though researchers still debate the size of the long-run effects.

Limits of trials

Randomised trials are powerful but not perfect. They can be expensive and slow. Some questions, such as the effect of a country’s monetary policy, cannot be randomised. Trials sometimes raise ethical concerns about withholding benefits. And a result found in one place may not hold elsewhere, a problem explored later in this module.

Thinking randomised trials settle every question

Randomised trials give strong evidence about the effect of a specific programme in a specific setting. They do not automatically tell us why it worked, whether it would work at a larger scale, or whether it would work in a different country.

Key takeaways

Randomised trials randomly assign people to treatment and control groups.
Randomisation makes the groups alike on average, so differences in outcomes show causal effects.
Banerjee, Duflo and Kremer won the 2019 Nobel prize for experimental work on poverty.
Trials are powerful but can be costly, ethically tricky and hard to generalise.