High residential mobility is a distinguishing characteristic of American society. The majority of people move from one location to another at least once in their lives, and individual movement affects all geographic areas, urban and rural. Policymakers use government migration data to predict migration trends to implement or improve public policies. Because official migration data are released with a significant lag, it is difficult for policymakers to adjust housing and labor market policies for current and future migration trends. The Internet has become a default channel for information search about relocation and the beacon for estimating migration intentions. Because search engines provide real-time information, these online datasets can be exploited to predict current and future migration trends.

In this dissertation, I review internal migration in the United States over the past two decades and the determinants of the recent decline in migration rates. I then demonstrate how online search data can be used to measure migration intentions in origin states and help predict current and future interstate migration flows. I employ a combined panel data set of Internal Revenue Service (IRS) migration data and Google Trends data covering all 50 states and the District of Columbia. The data set contains a large set of macroeconomic indicators for each origin and destination state and migration flows between pairs of states. Using a gravity approach with the balanced bilateral migration data, I show strong additional predictive power for interstate migration flows with Google Trends as compared to the baseline models that do not include online search data as covariates. My results imply that real-time online search data is effective in prognosticating actual moves and can be essential to managing and designing contingent migration policies.