Forecasting Earnings Using Age-Earnings Profiles And Longitudinal Data

Age-earnings profiles are widely used by forensic economists to assist in estimating future earnings, especially in situations where a person has little or no "track record" of earnings. Furthermore, age-earnings profiles are the subject of a sizable literature in both labor economics (e.g., Easterlin, Freeman, Ghez and Becker; Hanoch and Honig; Heckman and Robb; Murphy and Welch; Polachek and Siebert; Ruggles and Ruggles; and Welch) and forensic economics (e.g., Bell and Taub; Brookshire and Mathis; Bryan and Linke; deSeve, Dillman, Lane and Glennon; Lewis and Mullett; Nelson and Patton). With few exceptions (e.g., Ruggles and Ruggles, and Hanoch and Honig), this literature relies on age-earnings profiles produced from cross-section data collected in the March Current Population Survey (CPS) and published in the Money Income, P-60 series from the U.S. Bureau of the Census. This paper has three major purposes: (i) to discuss some of the major issues, problems and pitfalls in the use of age-earning profiles; (ii) to provide tables comparing age-earnings profiles from the Census Bureau (most of which is provided here in greater detail than the data previously published in the Bureau’s P-60 Series, Money Income) with those constructed from longitudinal data from the Panel Study of Income Dynamics (PSID); and (iii) compare forecasts of earnings using cross-section age-earnings profiles with forecasts using longitudinal data on earnings from the PSID that follow individuals for up to 22 years.