Book Description

What economic and political factors determine migration choices? In three papers this dissertation answers this question at the microeconomic, macro-political, and household levels. The first paper reduces the microeconomics of migration to the combination of an optimal stopping time problem over when to migrate, and a type revelation matching problem over where to migrate. Multi-armed bandits provide a solution to these problems under uncertainty, one that admits immobility as a rational response to noisy signals over the value of moving. A migration video game is used to test the model in an incentivized lab experiment in the United States and Ethiopia. Experimental results show that players' sequential decisions in the lab matched the optimal bound achievable by bandit algorithms. Providing extra information as a randomized treatment improved optimization by 20%. These findings suggest that inefficient migration trajectories are the result of limited and noisy signals rather than behavioral idiosyncrasies. The second paper recovers policies enacted by national governments to increase remittance inflows and emigration. Remittances and emigration provide regimes with increased stability by allowing them to reduce expenditures on public goods. Autocratic regimes stand to benefit the most, as increased exit by their population reduces the demand for domestic reform. A structural Bayesian model with dyadic remittance data shows a correlation between regime types and emigration policy, and that there exists significant variation within autocratic regimes with respect to these policies. Party-based autocracies rely on a notion of popular support, and are therefore relatively less likely than other autocratic regimes to implement policies to induce induce remittances and emigration. Autocratic regimes that have minimal ideology or concept of broad support --- personalist and oligarchic regimes --- are the most likely autocracies to engage in these policies. The third paper determines whether a decline in local economic conditions increases or decreases migration. An unexpectedly successful farmer's strike in Sikasso, a region in souther Mali, caused households in one area to miss one growing season and to lose a substantial amount of income for one year. Using time-series cross-sectional household survey data, we show that the drop in income produced by the strike reduced household migration by approximately 32% over six years. This suggests that households in the developing world are more constrained than motivated to migrate by their economic conditions, and that there exists a large set of households that may want to migrate but are unable to do so.