The observed health decline near the end of life coincides with less curative (e.g. hospital stay, doctor visits), and more comfort (e.g. nursing home) care, which accelerate both the fall in wealth, and the timing of death. We investigate whether these dynamics jointly result from a closing down the shop decision i.e. a depletion of the health stock is optimally selected (and eventually accelerated), leading to states characterized by indifference between life, and death. Towards that aim, we expand, structurally estimate, and simulate a life cycle model of financial, and health expenses with endogenous mortality exposure (Hugonnier et al., 2013).Under economically plausible, and statistically verified conditions, we find that, unless sufficiently rich and healthy, agents will optimally select expected depletion of their health capital, and associated increase in death likelihood. Moreover, we identify a wealth and health locus below which agents accelerate their health depletion. Importantly, wealth is also expected to decline for all, such that all surviving agents eventually enter the closing down phase.JEL classification: D91, D14, I12 1 optimality of these joint end-of-life health and wealth dynamics. More precisely, we look at conditions under which health, and wealth depletion is optimally selected to fall towards a region associated with very high mortality risk, and indifference between life and death. Put differently, agents optimally choose to let go, i.e. to close down the shop. The conditions under which they do are associated with lower wealth, such that richer individuals delay the health depletion. However, we show that under reasonable assumptions, wealth depletion is also optimally selected, such that agents eventually enter the closing-down phase. We further identify threshold effects whereby the depletion of the health capital is initially slowed down, before being accelerated.This paper has two main contributions: theoretical and empirical. First, we build upon a rich life cycle model developed in Hugonnier, Pelgrin and St-Amour (2013) in order to identify optimal depletion, and acceleration. This model encompasses a health investment setup with endogenous exposure to death risk, and exogenous sickness shocks that further depreciate the health capital. In addition to investing in their health, agents can buy actuarially fair insurance against health shocks, and save in risky, and risk-less assets. Agents also earn income, part of which is fixed (e.g. social security), and part which is health-dependent, reflecting their physical ability to work. Finally, preferences are characterized by subsistence consumption, as well as by generalized attitudes towards the various sources of risk (mortality, morbidity, and financial), and towards intertemporal substitution. Importantly, they also guarantee strict ex-ante preference for life, so that agents have no proclivity for death over life.We rely on the closed-form solutions of that model in order to characterize the optimal dynamics for health, and ...