Skip to content

Who Was the First to Use Randomization in Development Economics, and When?

Last updated on May 26, 2014

If you answered “Probably Esther Duflo, sometime in the early 2000s?” just as I did, you were wrong, just as I was.

As it turns, Immink and Viteri were the first to use randomization in a two-part article (part 1 is here; part 2 is there) testing the efficiency wage hypothesis published in the Journal of Development Economics in 1981:

This study represents an empirical test of the productivity consumption relation of the efficiency wage hypothesis, which is briefly discussed here. The research setting and design of the study are described. An energy supplementation program was generally effective in raising the daily energy intake and energy expenditure levels of Guatemalan sugarcane workers who were moderately energy deficient. Increased energy availability did not result in increased energy expenditure at work, or in an increased supply of work units. The results do not provide evidence of a strong productive-energy intake relationship among these workers.

Now, their randomization was not perfect, which they acknowledge themselves: they would have liked to assign individual workers to the high-energy treatment group at random, but instead decided to assign pre-existing groups of workers (which were not formed randomly) to the treatment and control groups.

Still, it is fascinating to see that randomization and the notion of internal validity were discussed in development economics more than thirty years ago, and that it took the field a good 20 years to rediscover randomization. This is similar to how Herman Wold was the first to take the idea of causality seriously in economics in 1945, yet it wasn’t until much later that economists followed suit.

ht: Traviss Cassidy via Facebook.