Until this year, I didn’t really know what experimental economists did. Despite having two experimentalists in my department, one of whom I spend a lot of time with, the whole concept of observing people’s decisions in a controlled environment but not controlling for anything about the individuals seemed totally silly to me. Aren’t you people worried about unobserved heterogeneity? I think my initial outburst was probably a bit oversimplified and I’m getting a better idea of what it means to do economics experiments now, but I’m still a bit skeptical.
After I excitedly read and recounted a paper on experiments and corruption the other day, a colleague of mine at Gettysburg allowed me to sit in on his energy economics class and experiment. Rim runs an experiment in every class after students present papers they’ve read and he presents the relevant theory on some aspect of energy economics. The day I visited, we talked about gasoline markets, differential pricing for differently located gas stations, and profit margins for owners of capital at various stages of the crude-to-consumer pipeline.
Then we got to play video games for the rest of class. I mean, we did an experiment. Half of us were set up in the experiment as gasoline suppliers and half as gas station owners. Each owner had two stations and a dedicated supplier who would set the price for each station. Importantly, the suppliers could dictate different prices for each station. When the game started, cars would enter the matrix and choose where to buy gasoline, usually closer to where they entered, but some were willing to drive if the price difference was great enough. Both suppliers and station owners could change the price as often as they wanted and consumer behavior was dictated by a random draw.
It took me a minute to get the hang of things. For instance, I didn’t realize for the first few rounds that the cost of crude (I was a supplier) was changing because I was so busy trying to figure out what the equilibrium price was for the two different gas stations. It was fun, though, to see how prices evened out and occasionally got out of control. It turned out that the inside-town gas stations had about 30-cent lower prices than the outside-town gas stations. This seems to me like a function of the parameters of the experiment, not really predictable behavior, but perhaps I’m wrong and it’s a function of the interaction of the parameters with behavior.
I think it’s asking a lot to assume that random college kids who know they’re getting paid a few bucks are going to act like gasoline station owners or suppliers. I know that experiments in other fields work like this too, in some way. The idea is to try to control everything you can in order to isolate the behavior or reaction or change. It’s what I try to do with econometrics; it’s what others try to do when they create models as well.
Perhaps it’s just indicative of my general skepticism of my own field, but I’m excited to learn more about it, too.