The Decline in Household Saving: What Can We Learn From Survey Data?

In a global economy characterized by some as being awash in saving, Americans stand out for their devotion to consumption. The rate of private saving in the United States has declined precipitously over the past two decades. While the corporate component has surged during the current economic expansion, the household saving rate has continued to fall. Household saving has fallen from ten percent of disposable income in the first half of the 1980s to less than two percent in the first half of the current decade. This development should strike us as all the more surprising given the large number of baby-boomers who are in their peak saving years. Despite considerable empirical research, the source of the saving decline remains controversial; primarily because it's one-time nature makes the question of the causes difficult to resolve on the basis of macroeconomic correlations.

In this paper we examine the saving decline from the perspective of microeconomic survey data on the wealth position of American households. Can the surveys provide information on the nature and causes of the saving decline that are not evident in the macroeconomic information? It is an extension of a prior paper that focused more on the macroeconomic characteristics of the saving decline and provided some preliminary evidence from the microeconomic data (Bosworth, 2004). Unfortunately, the availability of panel data that could have been used to track the change in saving within specific subgroups of the population is extremely limited. Our analysis concentrates on data obtained from six Surveys of Consumer Finances (SCF) covering the period of 1983-2001. The SCF had a panel dimension only in the 1983-89 period. Thus, a primary objective of this study is to determine whether the construction of information on synthetic cohorts from successive cross-sectional surveys provides a useful substitute for panel data.

We conclude that the 1983-89 panel was a valuable addition to the empirical analysis of saving. It is particularly instructive in demonstrating the heterogeneous nature of saving behavior and the dominant role of high-income households. Unfortunately, the panel component of the survey was discontinued after 1989. Despite our use of a variety of different adjustments to the cohort-based wealth data, we conclude that it is not an effective substitute for a panel survey. The most substantial opportunity to improve our knowledge of the reasons for the decline in household saving would be to repeat the 1989 exercise by re-interviewing a portion of the households in each SCF survey.

The following section provides a brief overview of the major macroeconomic features of the saving decline, as reported in the 2004 paper. The remainder of the paper is devoted to an analysis of the Survey of Consumer Finances (SCF) which is the primary data source for the microeconomic analysis. It includes an overview and evaluation of the six surveys that have been produced since 1983. The SCF is a survey of wealth, not saving; so it is necessary to explain our methods for obtaining measures of saving from consecutive wealth surveys. The subsequent analysis is divided into two portions. The first focuses on a panel of households that participated in both the 1983 and 1989 SCFs. Only the 1989 survey had a panel dimension with a link back to 1983.1 We use the 1983-89 panel to explore some of the characteristics of the households that account for the bulk of the saving. However, because the panel does not exist over multiple surveys, it cannot be used to explore the change in the saving of a specific household over time. The second section reports on estimates of the wealth change and saving of synthetic cohorts as constructed from the six independent wealth surveys. As a result, we can observe changes in saving over an 18-year period that includes the years of largest decline in the aggregate household saving rate.