We study the role of preferred habitat in understanding the economic effects of the Federal Reserve’s quantitative easing (QE). Using high-frequency identification and exploiting the structure of the primary market for US Treasuries, we isolate demand shocks that are transmitted solely through preferred habitat channels but otherwise mimic QE shocks. We document large localized yield curve effects when financial markets are disrupted. Our calibrated model, which embeds preferred habitat in a New Keynesian framework, can largely account for the observed financial effects of QE. QE is modestly stimulative for output and inflation, but alternative policy designs can generate stronger effects.

We develop a two-country model in which currency and bond markets are populated by different investor clienteles, and segmentation is partly overcome by arbitrageurs with limited capital. Risk premia in our model are time-varying, connected across markets, and consistent with the empirical violations of uncovered interest parity and expectations hypothesis. Through risk premia, large-scale bond purchases lower domestic and foreign bond yields and depreciate the currency, and short-rate cuts lower foreign yields, with smaller effects than bond purchases. Currency returns are disconnected from long-maturity bond returns, and yet the currency market is instrumental in transmitting bond demand shocks across countries. (JEL E43, E44, E52, F31, G12, G15)