The vast majority of household wealth in the U.S. is held in illiquid assets, primarily housing, making households vulnerable to unexpected income shocks. To rationalize this preference for illiquidity, we build a life-cycle model where households are tempted to consume their liquid wealth but can use illiquid housing as a savings commitment device. The importance of temptation and commitment is identified using data on consumption, liquid assets, and housing wealth over the life-cycle. Our model matches observed portfolio choices and gives rise to a high demand for illiquid housing partially driven by the need for commitment. Preference for illiquidity has important implications for the consumption response to unexpected income shocks. Our model is able to replicate the recent empirical evidence that MPCs remain high in response to large shocks, a finding that cannot be explained by current heterogeneous agent models, but that has great significance for fiscal stimulus targeting.