Elderly households hold most of their wealth in housing, maintain high levels of wealth throughout retirement, and often leave bequests. The value of their houses are subject to large shocks. To what extent do these shocks aect their savings, consumption, and bequests? Answering this question requires separating precautionary savings, bequest motives, and the desire to remain in one's home. I develop and estimate a structural model of retirement savings decisions with realistic risks, housing, and heterogeneity in bequest preferences. I exploit policy changes to the taxation of housing and bequests to separately identify the different motives for holding wealth. Estimates show approximately half of retirees have no bequest motive. House price changes are quantitatively important, with 1/4 of increases passed on to future generations. I use the estimated model to evaluate means-tested programs insuring retirees' LTC expenses. I find exemptions providing marginal liquidity have larger insurance value than fully eliminating LTC expense risk per pound it costs the government.