The Causal Effect of Family Income On Childhood Obesity

Background and Purpose: The prevalence and subsequent health impacts of childhood obesity are disproportionately concentrated among low-income and minority children. Many believe the correlation between income and childhood obesity is due to income's effect on some combination of the type of food people eat (or have access to) and whether they have the time and ability to engage in physical activity in safe, suitable environments. However, it is also possible that the correlation reflects the influence of unmeasured parental characteristics, also associated with family income, that directly influence childhood obesity and health. The purpose of this paper is to estimate a causal effect using a fixed-effects instrumental variable design that addresses the endogeneity of reported family income.

Methods: The instrumental variable approach leverages two sources of exogenous variation in family income to identify a causal effect: changing returns to maternal education at baseline and expansion of the Earned Income Tax Credit. Both sources of variation are shown to have a significant effect on changes in family income and are assumed to be independent of childhood obesity. The fixed effects instrumental variable design is applied to data on the children of the National Longitudinal Survey of Youth 1979. Wald and two-stage least squares estimates are used relate family income to childhood obesity.

Results: This study finds that a $10,000 increase in family income implies a 4.5% decrease in the probability of being obese (≥95th body mass index percentile). Relative to the national childhood obesity prevalence of 17 %, the results imply that income transfers could have a substantial effect on the prevalence of obesity in the United States.

Conclusions and Implications: The results of this analysis provide evidence of the protective effect of increased income. Although the research design used in this analysis improves upon existing estimates of the effect of family income, the assumptions upon which the causal effect is identified mean the results may not generalize to all policies that affect income including increasing minimum wages or non-labor market based cash-assistance programs (i.e. TANF, SSI).