Wealth Accumulation across the Adult Life Course: Stability and Change in Sociodemographic Covariate Structures of...

This paper reports microlevel Tobit regression analyses of sociodemographic covariates of the life course accumulation of total household net worth data in eight waves of five distinct panels-spanning over 6 years from late 1984 through early 1991-of the Survey of Income and Program Participation (SIPP). It is found that the quadratic age-wealth relationship predicted by Modigliani's Life Cycle Hypothesis is evident in aggregate age-median wealth profiles as well as in the micro data for households with positive net worth. However, when adult status attainment variables are entered into the regression models either by themselves or in combination with marital/family status variables, the age of household head at which net worth begins to decline is far beyond the typical retirement age. In addition, the traditional criterion variables of sociological status attainment theory-educational attainment, occupational status, and earnings-are found to be positively associated with household net worth, although the net effect of occupational status generally is not statistically significant and the earnings effect is nonlinear. Further, consistent with status attainment theory, householder minority status (black, Hispanic) is negatively associated with the accumulation of net worth. It is found that both single male and single female householder status are negatively associated with the accumulation of household net worth (relative to married couple households) as is the size of the household (measured by the number of children under age 18 present). Separate logistic regression analyses show that households with zero and negative net worth are more likely than households with positive net worth to be black and have low earnings. Higher levels of educational and occupational status attainment reduce the probability of zero net worth but not the probability of negative net worth. Male- and female-headed households and households headed by Hispanics also are more likely to have zero net worth, but not negative net worth. The estimated sociodemographic covariate structures of household net worth are found to exhibit substantial stability across both waves and panels in the SIPP-although effects of the 1990-1991 recession are detectable in estimates for the 1990 panel. Possible applications of the estimated models in demographic projections of household net worth are suggested.