Differential Mortality, Uncertain Medical Expenditures, and the Savings Decisions of the Elderly

Featured in print Bulletin on Aging & Health

Household savings decisions are of great interest to policy makers, as they affect not only the household's well-being but also asset markets and the rate of economic growth. One of the most widely-used models of savings behavior, the life cycle model, predicts that people will accumulate assets during their working years and spend them during retirement. Yet the data indicate that many elderly keep large amounts of assets until very late in life.

One theory that has been advanced to explain this discrepancy is that elderly households retain their assets in order to protect against the risk of medical expenditures at older ages. Yet earlier studies that have used this theory to forecast household savings decisions predict a much more rapid decumulation of assets after age 70 than that observed in the data.

In Differential Mortality, Uncertain Medical Expenses, and the Savings of Elderly Singles (NBER Working Paper 12554), Mariacristina De Nardi, Eric French, and John Bailey Jones revisit the savings decisions of the elderly. The new angle the authors bring to the topic is to recognize that people have different life expectancies and face different risks of incurring large out-of-pocket medical expenditures based on their gender, age, income, and health status. By incorporating this heterogeneity into their model of savings decisions, the authors aim to more closely match the actual savings behavior of the elderly.

The authors use the Assets and Health Dynamics of the Oldest Old (AHEAD) dataset for their analysis. They focus on elderly singles in order to sidestep the added complications that arise when modeling the behavior of couples.

The authors' first task is to estimate the uncertainty in mortality and out-of-pocket medical expenditures as a function of gender, health, permanent income, and age. They have several notable findings. First, out-of-pocket medical expenditures rise very rapidly with age. For example, average spending for a woman in poor health rises from $1,200 at age 70 to $19,000 at age 100. Second, the share of income spent on health care rises with income, indicating that health care is a luxury good for the oldest old. For instance, a 95-year-old woman in poor health would incur $2,700 in medical spending if she were at the 20th percentile of the income distribution, vs. $16,000 if she were at the 80th percentile. Finally, life expectancy varies greatly for older individuals - while a 70-year-old man with poor health and low income can expect to live only 6 more years, a 70-year-old woman with poor health and high income can expect to live 17 more years.

Next, the authors construct a rich model of savings behavior that accounts for this heterogeneity in mortality and out-of-pocket medical expenditures. They then use their model to predict the savings of elderly individuals and compare this to people's actual savings. They find that their model fits the data much better than previous studies - in particular, they predict a slow rate of asset decumulation after age 70, as observed in the data.

There are several key findings from their model simulations. First, differences in average out-of-pocket medical spending across income groups are very important in explaining asset decumulation decisions. However, differences within income group (that is, within one's income group, the risk of having high vs. low expenditures) are much less important. Second, the longer life expectancies of high-income elderly are important in explaining their higher level of savings. Finally, the existence of programs such as Medicaid and Supplemental Security Income, which guarantee older individuals a certain level of consumption in retirement, depress savings for older individuals of all income levels. For example, the authors estimate that in the absence of these programs, assets for a person in the highest permanent income quintile would rise by about fifty percent, from $150,000 to $220,000.

In sum, the authors find that out-of-pocket medical expenditures are higher and more volatile than previously estimated, rise rapidly with age, and are a luxury good for the oldest old. Heterogeneity in medical expenditures and life expectancy play a significant role in explaining the savings behavior of the elderly. The authors conclude that any analysis of the effect of a policy change on the savings behavior of the elderly should be based on a model that incorporates the effect of the government-provided consumption floor and allows medical spending to vary by age and income.