Field Experiments and the Practice of Economics 1 a Structural Model and a Randomized Experiment to Evaluate

when, some twenty-five years ago, I first started doing randomized controlled trials (RCTs), the most common reaction was one of puzzled tolerance. Colleagues and friends seem to admire the effort involved and could see that RCTs had the advantage of avoiding the then common wrangling about what is causal and what is not. But in the end they were skeptical that it was worth it. In part they were concerned for me – I had a successful career doing economics (in particular economic theory) the way it was done then. Why go down this particular rabbit-hole? But more importantly, as the more candid among them put it, “are RCTs economics?” The prize in economics has perhaps settled that question at one level. But the doubts that motivated it remain. Responding to them provides a useful way to bring out what RCTs have brought to the field. At the risk of some caricature, there are really four closely related questions: