Economists have accepted the view of personal nonproperty incomes as the returns to the quantity of human capital which the individual possesses. Numerous studies have estimated internal rates of return or rental rates to investments in human capital.' Several studies have used the basic human capital model in estimating the contribution of education to the growth in national income,2 while others have applied the concept to the analysis of the distribution of income.3 However, in the empirical estimation of parameters, most of these studies of growth and income distribution use either a single cross-section of data or else make separate estimates for each of a series of cross-sections. The main purpose of this paper is to estimate the rate of growth in personal income in the United States using a series of cross-sections in a single regression. A second purpose is to explore the effects of including a finite life correction in Johnson's earlier model, thus estimating internal rates of return rather than rental rates. Finally we examine the effects of alternative specifications of the on-the-job training (OJT) function.4 In general, these corrections are found to be minimal. After the biasing effects of exogenous growth is removed, the estimates of rates of return, depreciation, and OJT investment are in reasonable agreement with previous estimates. Moreover, the rates of depreciation are more reassuring than the very high rates previously estimated by Johnson (1969, 1970). The lifetime incomes by cohort implied by the parameter estimates exhibit some interesting and instructive peculiarities. The reduced estimate of the rate of depreciation results in net investment reaching zero at considerably older ages than with previous estimates. Because of the different rates of exogenous growth estimated for different schooling levels, the income patterns of those cohorts who were 20 years and 10 years of age in 1960 show lifetime incomes which do not increase monotonically with increasing schooling levels. We conclude that this must imply that the trends for middle-aged workers, which dominate the parameter estimates, cannot continue if * Assistant professor of economics and statistics, Southern Methodist University, and econometric analyst, Weyerhauser Company, Tacoma, Washington, respectively. The research for this paper was supported by the Department of Labor Manpower Development Grant 31-46-70-06 to Southern Methodist University. We are grateful to Theodore W. Schultz, William J. Haley, George Borts, and an anonymous referee for helpful comments on an earlier draft. However, we alone are responsible for remaining deficiencies. I See for example Gary Becker (1964, 1967), Giora Hanoch, Lee Hansen, Barry Chiswick, Becker and Chiswick, Johnson (1970), Jacob Mincer (1962). 2 See Theodore Schultz (1961), Edward Denison, and the critique by Mary Jean Bowman. In the applications of the human capital model to the analysis of growth, Bowman has emphasized the distinction between internal rate of return and rental rates to human capital. 3 See particularly Mincer (1970), and Mincer and Chiswick as well as the other papers in Schultz (1972). M Mincer (1970, 1971) claims some superiority in the fit to his data when he uses the exponentially declining form of the OJT investment function as compared to the linearly declining form used by Johnson (1969, 1970).