an econometric study of the impact of economic variables on adult obesity and food assistance program participation in the NLSY panel

Over the past thirty-five years, the U.S. adult obesity rate has more than doubled from roughly 15% to 35%, reflecting a general diffusion of obesity across all segments of the adult population (United States Department of Health and Human Services). The objective of this research is to identify the factors that influence adults\u27 healthy weight, as reflected in body mass index (BMI) or being obese (having a body mass index of 30 or larger), the Food Stamp Program (or Supplemental Nutrition Assistance Program) participation, and the relationship of these two in longitudinal panel data. The panel data was obtained by merging the individual-level national data for the U.S. adults from the National Longitudinal Survey of the Youth, 1979 Cohort (NLSY79), with external price data obtained from the American Chamber of Commerce Research Association (ACCRA) Cost of Living Index. Six rounds of NLSY79 survey were extracted at a 4-year interval from 1986 to 2006. Using the geocode information, the secondary data on local food, drinks and health care prices and labor market conditions were merged with the data on adults in the NLSY79. We used three improved economic and econometric models to examine the effect of FSP (or SNAP) participation on women\u27s BMI and likelihood of being obese. First, least squares instrumental variable (IV) estimation of our benchmark model suggests that women in households that currently participate in the FSP (or SNAP) have a higher BMI and a higher probability of being obese. Other things equal, participation in the FSP increases a woman\u27s BMI by about 1.1%, and also increases her probability of being obese by about 2.6 percentage points. However, concerns are sometimes raised about least squares IV estimates being inconsistent because no account is taken of individual (or household) fixed (or random) effects. Second, a new model of lifetime utility maximization is developed with perfect foresight, and the equations for BMI (and obesity) and FSP participation are estimated using the least squares estimator incorporating IV (for FSP and wage rate) and individual fixed effects. Results from this fitted model suggest that if a woman is in a household that decides to participate in FSP participation, it reduces her BMI by 15.67% and her probability of being obese by 56.33 percentage points. Moreover, the estimates of the individual fixed effects have a frequency distribution that approximates a normal, and for a significant part of the sample, the individual fixed effects accounts for most of the explained variation in ln(BMI) and the probability of being obese. Third, we next consider a model of lifetime utility maximization with updating and autocorrelation of BMI or the probability of being obese. These results suggest that if a woman is in a household that participates in the FSP program, it reduces here BMI by 1.12% and her probability of being obese by 3.76 percentage points, which is significantly lower than the results from the second model. These latter two models have considerable appeal relative to the benchmark econometric model. Hence, we conclude that women in households that participate in the FSP participation have a lower BMI and a lower probability of being obese. Also, we conclude that individual-fixed effects play a large role in understanding obesity in women. These are key findings of this study