The interstate migration rate in the United States has been declining steadily for at least 15 years. We bring together theory and data to explore the sources and implications of this decline. We show that a wide range of long-term demographic changes -- including shifts in the age structure, education levels, and female labor force participation -- account for essentially none of the decrease. Rather, the decrease in migration affects all parts of the population, though it is concentrated among the young and highly educated. These findings rule out many popular theories of the decrease in migration. However, the empirical results fit well with a simple search model in which migration rates change because the world is becoming "flatter," in the sense that the variance in amenities across locations is becoming smaller. If job search is undirected, migration rates rise as the world flattens, because people are closer to being indifferent about where they live and smaller differences in wage offers are needed to justify moving. However, if search is directed, migration rates fall as the world flattens, because people are closer to being indifferent about where they live and hence will not direct their search toward distant locations.
