Using Income Data to Predict Wealth

Amy Stubbendick for outstanding research assistance in the work reported here. Barry Johnson has long been an essential player in the work underlying this paper, and he provided essential data used in the analysis. The opinions expressed here are the responsibility of the author alone and do not necessarily reflect those of the Board of Governors of the Federal Reserve System. Most often, economists are interested in understanding household wealth as a reflection of past saving behavior. As a stock, wealth represents the cumulation of all past saving, transfers, and net shocks to income and consumption. The level of wealth implicitly reflects preferences about risk and intertemporal substitution, expectations about future income and expenses, life expectancy, family structure, institutional factors such as credit availability, and possibly more psychological factors such as cognitive abilities to make choices about the future and desire for autonomy or control. However, inherent in the nature of wealth, there is also a structural relationship between its value and investment returns, though the returns may be difficult to measure, irregularly distributed through time, or even conceptually ambiguous. This second type of functional relationship may be of interest to those who study portfolio allocations and to other who have a particular need to project wealth from a given pattern of income—for example, the Office of Tax Analysis in the Treasury where wealth is projected from income flows reported on tax returns, and at the Federal Reserve where wealth projected from income is a key factor in the sample design for the Survey of Consumer Finances (SCF). This paper attempts to contribute to the understanding of the relationship between income and wealth using data from the SCF, the Individual Tax File (ITF) at the Statistics of Income Division of the IRS (SOI), and information from Forbes magazine about the wealth of the 400 wealthiest people in the U.S. Although the SCF sample over-represents wealthy households, it specifically excludes very prominent individuals, including members of the " Forbes 400, " whose data might be impossible to protect sufficiently to include in a public dataset. The SCF is stratified by an index defined in terms of income flows which is intended to proxy for households' wealth. If this index is functioning as intended, one would expect that the Forbes group would have the very highest values of the index. However, examination of the 1998 SCF sample indicated that a number of …