Commentary

LABORATORY EXPERIMENTS IN MACROECONOMIC ENVIRONMENTS Experiments such as the ones undertaken in Arifovic and Sargent’s paper are rare. The conventional wisdom has been that controlled laboratory environments are infeasible for macroeconomic questions because we are attempting to understand how a very large number of individual households and firms interact to produce the prices and quantities we observe in the data. While this conventional wisdom is, of course, true at some level, there is more to the story. There are good reasons to take experimental macroeconomics very seriously. Mainly, we need to obtain laboratory confirmation of predictions from simple models before we can hope to correctly infer what forces are at work in large, industrialized economies. We economists often write down simple models in an effort to get some core economic intuition concerning topics of interest. The literature begun by Kydland and Prescott (1977) and continued by Barro and Gordon (1983), Rogoff (1985), Walsh (1995), and many others is one outstanding example. In these simple models, we presume to know how human beings would act when confronted with the environments we construct. Laboratory experiments can help us to understand whether such presumptions are warranted. In fact, the current paper calls those presumptions into question. The literature begun by Kydland and Prescott (1977) has been perhaps the most influential for central bankers during the last 25 years. It provides the leading explanation of why there is so much inflation among the industrialized countries of the world today.