Content area
Full text
(ProQuest: ... denotes formulae omitted.)
Almost 40 percent of U.S. wealth belongs to households whose head is sixty-five or older.1 These older households tend to run down their wealth more slowly than the rate of decrease implied by a basic life-cycle model in which the time of death is known. An active research literature has risen around this "retirement-savings puzzle" (De Nardi et al., 2016). One potential solution to this puzzle is that older households are holding their assets because they fear incurring high medical expenses at the ends of their lives. Another potential explanation is that the elderly are accumulating assets to bequeath to their heirs.2 Determining which of these motives is more powerful has proven challenging, in large part because the two motives generate similar saving behavior (Dynan et al., 2002; De Nardi et al., 2016).
One strategy for differentiating these saving motives is to analyze dissaving decisions near the end of one's life. If older households are saving to cover end-of-life medical spending, we should observe at least some households incurring significant medical expenses in their final years of life and experiencing concomitant falls in their wealth.3 On the other hand, if medical expenses do not rise at the end of life and people are not concerned with leaving wealth to their heirs, we should see households consuming their assets more quickly as death becomes imminent.
In this paper, we take a first step toward such an analysis by documenting the dynamics of wealth and medical expenses around the time of a person's death, using data from the oldest cohorts of the Health and Retirement Study (HRS). To do so, we employ an event study approach. For each "treatment" household experiencing a death, we identify another, similar, "control" household that experienced a death six to ten years later. Using fixed effects regression, we then estimate the asset trajectories of the treatment households as their death approaches, along with the asset trajectories of their matched control households over the same time period. We find that in the six years preceding their deaths, the assets of singles decline on average, relative to those of survivors, by an additional $20,000 on average. Correspondingly, over the same time span, the assets of couples who lose a...