Effects of pensions on savings: analysis with data from the health and retirement study

Abstract

This paper examines the composition and distribution of total wealth for a cohort of 51- to 61-year olds from the Health and Retirement Study (HRS), and the role of pensions in forming retirement wealth. Pension coverage is widespread, covering two-thirds of households and accounting for one-quarter of accumulated wealth. Social security benefits account for another quarter of total wealth.

As calculated from earnings records, the present discounted value of social security benefits is less than the present value of taxes paid. Earlier than many expected, social security is already a poor investment on average for this cohort on the verge of retirement. When pensions and social security are included, wealth accumulated by the HRS population to date is substantial. At their expected retirement date, using only the wealth accumulated by their mid-fifties, the HRS household with median replacement rate could finance a fixed, nominal two-thirds joint and survivor annuity replacing 79 percent of last earnings, and a real annuity replacing 52 percent of last earnings. Replacement rates for median earners are higher. Additional savings made over the seven years remaining until retirement will raise those replacement rates by about a fifth. When measured against a standard of adequacy based on average yearly earnings over the worklife, with adjustments made for the absence of preretirement savings, children, taxes, work-related expenses and other factors, these replacement rates appear adequate.

Lifetime earnings are measured for each individual in the HRS from social security earnings records augmented by self-reported earnings histories. When pensions and social security are counted in total wealth, the ratio of wealth to lifetime earnings declines from very high levels in the bottom ten percent of the earnings distribution, remains at roughly 40 percent from the 25th through 95th percentile of the lifetime earnings distribution, and then falls to 32 percent for those in the top five percent of the earnings distribution.

This result is consistent with the predictions of a simple, stripped-down life-cycle model. Also consistent is a finding that the ratio of wealth to lifetime earnings is no higher for those with pensions than for those without pensions. However, heterogeneity is quite important. Real estate and business wealth are a larger share of total wealth for those without pensions, reflecting the importance of self-employment in wealth accumulation.

Multivariate regressions relating total wealth to pension coverage and pension value, which standardize for sources of heterogeneity, suggest that pensions cause very limited displacement of other wealth, if any. Pensions add to total wealth by at least half the value of the pension, and in most estimates by a good deal more.

These findings are not consistent with a simple life-cycle explanation for savings. They also raise questions about whether pensions are fundamentally a tax avoidance device, allowing substitution of pension for nonpension savings.