Beyond the Wage Premium: Fatherhood and Asset Accumulation

Abstract

This study examines the association between becoming a father and men's asset accumulation trajectories. It is the first study to include nonmarried fathers in addition to married fathers. We used data from the National Longitudinal Survey of Youth, 1979 cohort (the 1979 and 1985–2000 wave) and multilevel modeling techniques to examine the research questions. Findings revealed that the association between fatherhood and men's total asset accumulation depended on marriage. Upon becoming fathers, married men increased their rates of asset accumulation. By way of contrast, unmarried men who became fathers had a decline in their rate of asset accumulation, on average. Race also moderated the association between fatherhood and asset accumulation.

ACKNOWLEDGMENT

A grant from the Social Trends Institute supported the first author's efforts in this research.

Notes

¹We tested the possibility of threshold effects in the current study (not shown). They were not significant, so we do not discuss this issue.

²The initial NLSY wave in 1979 had 6,403 male participants. Of the 936 individuals we lost, most were excluded due to the requirement that they participate in at least one wave in 1985 or later. An attrition analysis (not shown) indicated that older participants, those with less income, and Whites were most likely to leave the sample before 1985.

Some individuals did not participate in all 10 waves of the NLSY between 1985 and 2000. These individuals were shown to have lower income, less education, and lower asset levels in 1985. White participants were also less likely to complete all 10 waves. We were less concerned about this pattern of differential participation because individuals still contributed information to the multilevel analysis even if they did not complete all 10 waves (Raudenbush & Bryk, 2003; Singer & Willet, 2003).

³Multilevel modeling allowed us to obtain unbiased estimates of change in assets and in the estimates of how other variables were associated with asset change. Multilevel modeling also allowed us to assess the within-person change in assets as men became fathers (Raudenbush & Bryk, 2003). Our hypotheses are more oriented toward measuring intraindividual (within-person) change than measuring interindividual (across-persons) differences between fathers and nonfathers. Thus, our analytic methodology reflects the hypotheses.

⁴We ran the analyses without the weights and the findings were unchanged.

⁵This graph does not imply that the average White would have more than $1,000,000 in assets by the age of 38. Like other studies that use assets as the outcome variable, we transformed assets using a base 10 logarithmic transformation. It is inappropriate to reverse transform predicted values from the equation as the association between the independent variables (e.g., age, fatherhood) and untransformed assets likely differ from the association between the independent variables and transformed assets.

⁶The graph for Hispanics and other race/ethnic minority participants would have looked the same as the White curve with lower intercepts at the beginning. We omit these trends to keep Figure 1 simpler.