The choice to migrate is never easy. While there are a lot of economic theories about where and why people migrate, what we find is that real world behavior often doesn’t match what theories would predict. We see both people staying in places that don’t offer much in the way of opportunities, as well as people moving to places that may seem less than optimal. In order to figure out what factors in to individual’s decisions to stay or go, AAE graduate student Zack Barnett-Howell decided to develop a new theory about how people decide whether or not to migrate, given that they have access to limited information. His research tests whether a Bandit algorithm could serve as a better approximation for the way individuals make decisions about migration, and created a game called ‘Planet Bandit’ to do so. The game separates decisions about when and where to migrate, and allows for learning over time. Participants get monetary payoffs based on their outcomes, and are rewarded for making optimal migration decisions.
The game was played with groups of university students in Madison, Wisconsin and in Bahir Dar, Ethiopia. Results show that the Bandit algorithm does a good job of predicting behavior when people are making sequential decisions under uncertainty. Moreover, the results show that participants experienced learning between rounds, updating their behavior based on additional information and outcomes. The tested theory seems to do a better job of predicting migration behavior than other current theories.
Zack is a Ph.D. candidate in both Agricultural and Applied Economics and Political Science. His dissertation is focused on decision-making and optimization under uncertainty. His work on development economics is focused on the impact of climate change on subsistence farmers. He also has technical expertise in behavioral economics, machine learning and high-throughput computing.