Association for Public Policy Analysis & Management

Economic research consistently finds increases in earnings and income volatility in the U.S. from the 1970s to the 1980s, and again in the 2000s.  Yet, most research on family income and child development continues to use static measures of income. Even prominent studies of the timing of poverty during childhood do not model income trend or volatility explicitly. This gap prevents social science from speaking to the salience of increasing economic instability for children and to the wisdom of social policies designed to stabilize or improve income. In this paper, I use multiple files from the Panel Study of Income Dynamics (PSID) to conduct two types of analyses. First, I use the PSID cross-year individual file from 1971 to 2011 to examine time trends in income instability during childhood.  The trend analysis uses simple descriptive statistics to examine how income dynamics during childhood have changed since the 1970s.  I examine income trends for children overall, and by the education level of the family head (a proxy for socio-economic status). In addition, I disaggregate the total income trends to examine whether any changes over time can be attributed to changes in employment and earnings, changes in public assistance, or both. Second, I use the PSID individual, family, and Child Development Supplement (CDS) files to examine the associations between income dynamics and child outcomes. I begin with sample of children in the PSID-CDS who were 0 to 12 years of age at the time of the 1997 PSID core survey. The total sample size is 3,394, with 1,344 unique children observed in at least one wave. I estimate regression models predicting one of six outcomes measures, as a function of income level, variability, and trend, controlling for a set of child and family characteristics. The dependent variables include two cognitive test scores, two parent-reported measures of child behavior, and two parent-reported measures of child health.  

Results from the trend analysis suggest a general improvement over time in the average level of income for families of children (at all ages), but that improvement has been experienced exclusively by families at the top end of the income distribution.  In addition, income variability has also increased over time and this change has been particularly dramatic for the families with children at the bottom of the income distribution.  Consistent with prior studies, higher average income is associated with higher child math and reading test scores, and with fewer problem behaviors.  Income variability, in contrast, is associated with lower math and reading test scores, and a decrease in positive social behaviors.  There is not a statistically significant relationship between variability and problem behaviors, although the direction of the coefficient is positive (more variability associated with more problem behaviors).  The average yearly growth rate of income is not statistically related to any outcomes, and none of the income measures are strongly associated with health outcomes.