The Draw-Down of Retirement Savings

03/28/2012
Featured in print Bulletin on Aging & Health

The Baby Boomers have begun to reach retirement age. For the oldest boomers in particular, the window to accumulate retirement savings is closing (if not already closed), and attention is now shifting to how boomers are using their savings in retirement. The baby boom generation is of interest not only because of its size, but also because it may be the first for which personal retirement accounts such as IRAs and 401(k)s could play an important role in retirement financial security.

In The Composition and Draw-Down of Wealth in Retirement (NBER Working Paper 17536) researchers James Poterba, Steven Venti, and David Wise present new evidence on what resources are available to households as they enter retirement and how households use those assets in early retirement. The authors make use of data from the Health and Retirement Study (HRS) for their analysis.

The authors first look at the balance sheets of households with a head aged 65 to 69. The typical household has total non-annuitized wealth of about $220,000. About 80 percent of these households are homeowners, and primary home equity accounts for the largest share - roughly 30 percent - of non-annuitized wealth. Just over half (52 percent) of households have personal retirement accounts (PRAs), which account for roughly 21 percent of non-annuitized wealth; financial assets outside of PRAs account for another 23 percent of wealth and are held by most households (87 percent).

Not surprisingly, there are considerable differences in asset holdings across households, especially for financial assets. While the typical household has total financial assets (both inside and outside PRAs) of just $52,000, households at the 90th percentile of financial asset holdings have over $700,000, or more than 13 times as much. For home equity, households at the 90th percentile have about five times as much wealth as the median household ($585,000 vs. $120,000), while the equivalent ratio for Social Security wealth is only two ($643,000 vs. $315,000).

Next, the authors examine households' current annuity income - that provided by Social Security and defined benefit (DB) pension plans - and calculate the potential additional annuity income they could obtain by annuitizing their financial (non-housing) wealth. They find that relatively few could purchase an annuity that would generate an additional $10,000 in annual income - only about one-third of all households could do so, and only 40 percent of those households in the top quartile of current annuity income (those with current annuity income of over $30,000). Interestingly, while there is a positive correlation between housing equity and annuitizable wealth, there are nonetheless many households with little annuitizable wealth who have substantial home equity.

Turning to the question of how households draw down wealth in retirement, the authors show that there is little use of home equity early in retirement to support consumption or purchase other assets or annuities; rather, households tend to hold home equity until they experience a traumatic event such as one spouse's death or entry into a nursing home. For non-housing assets, singles and couples who do not experience death or divorce tend to have constant or slightly rising assets from one survey wave to the next, while couples that do experience one of these events see their assets drop sharply.

Exploring linkages between health and wealth, the authors find that there is a strong correlation between these factors, not only at given point in time but also in how they evolve over time. Specifically, the authors find that net worth rises with age for healthier households (those in the top three quintiles of initial health status), but is flat or more slowly increasing for less healthy households (those in the lower two quintiles). While there are many potential explanations that need to be explored more fully, the authors "conclude from these patterns of wealth evolution that if anything, past studies of the cost of poor health in late life under-estimate the risks that households face from adverse health shocks."

These findings have several important implications. First, for many households "discussions of whether to pur-chase an annuity or draw down wealth in another fashion are largely moot; the amount of retirement support that their savings will provide is very limited." For example, nearly half (43 percent) of households would not be able to make the $25,000 minimum investment typically required to purchase an annuity even if they liquidated all of their financial assets.

Second, for the minority of households who reach retirement with substantial assets, late-life financial planning is complex and multi-faceted. Households face three main risks: longevity, uninsured medical expenses, and poor asset returns. While most of these households have enough resources (including financial and housing wealth) to meet the expected cost of medical care, the small risk of very high out-of-pocket costs may nonetheless be an important factor in financial planning. Households in the top half of the wealth distribution do not spend down financial assets in the early decades of retirement, suggesting they can meet their regular consumption needs with their current annuity income and interest or dividends on their savings.

Third, home equity "may substitute for other forms of insurance against living longer than expected or facing unanticipated medical costs." For most households, housing equity exceeds financial assets, and households "appear to treat their houses as a source of reserve wealth that can be tapped in the event of a substantial expense, for example a health care need." The availability of housing equity may help to explain the limited demand for private annuities.

Finally, the authors note that their findings "underscore the heterogeneity in household circumstances at retirement and suggest the difficulties of applying a 'one-size-fits-all' approach to all retirees. A household's preferences regarding different payout streams may depend on its wealth, its planned future expenditures, and the range of uncertain potential outlays that it faces." Noting the financial pressures on programs like Social Security and Medicare, the authors conclude that the need to forecast government policies may be one of the most difficult challenges facing retirement-age households.


The authors are grateful to the National Institute on Aging (grant P01-AG005842), to the Social Security Administration (grant 5-RRC080984-00-03-00), and to the National Science Foundation (Poterba) for research support. David Wise received support for this research from the National Institution on Aging (grant P01- AG005842 and P30-AG012810). Poterba is a trustee of the College Retirement Equity Fund, and of the TIAA-CREF mutual funds; TIAA-CREF is a provider of retirement service and annuity products.