The Adequacy of Household Saving

Eric M. EngenWilliam G. Gale Cori E. Uccello

WP#2000-1

Abstract

This paper provides a new examination of the adequacy of households’ saving for retirement. We develop a stochastic, life-cycle simulation model in which people save both for retirement and as a precaution against uncertain future earnings and uncertain lifespan. The model shows that, even among observationally-equivalent households, there will be a distribution of optimal wealth-to-earnings ratios, rather than a single target level. The existence of this distribution fundamentally changes the interpretation of data on wealth accumulation because it implies that even low levels of wealth can be consistent with optimizing behavior.

Using data from the HRS and the SCF, we find that more than half of married households where the husband works full-time have observed wealth-earnings ratios that exceed the median simulated wealth-earnings target for households with the same characteristics, and that the model understates wealth accumulation among households with high wealth-earnings ratios. Both results suggest wealth accumulation is adequate for a majority of households. However, among households with low wealth relative to earnings, there is mixed evidence of under-saving at the 5th and 25th percentile of the wealth-earnings distribution. We also examine differences between households with high and low wealth-earnings ratios, the evolution of the adequacy of saving between 1983 and 1995, and the sensitivity of the results to numerous factors.

Our central conclusion is that the characterization of any undersaving problem depends crucially on the specification of the null hypothesis that describes optimal, or benchmark, saving behavior. In addition, we interpret our findings as consistent with the view that inadequate saving is not as serious a problem as has been touted in the past. We also show that most previous studies that have been interpreted as showing inadequate household saving can be reconciled with our findings.