Identifying the effects of “flights to safety” on asset prices using pure time-series methods is difficult because crises are infrequent. We develop a new cross-sectional identification approach, motivated by the insight that investors may differ in their “preferred habitats” within a broad asset class. We apply the method to the question of whether foreign capital is responsible for residential real estate price movements in global cities such as London and New York, especially during crises. Using large data sets of housing transactions, we find that foreign risk strongly affects London house prices. The effects are long-lasting, and are associated with both safe-haven effects and immigration.