Stephen Done and Dr. Eric Eide, Economics
Over the past three decades, there has been a significant increase in the percentage of obese persons in the United States. Among adults, rates have more than doubled, while nearly tripling for children over the same period (Anderson, Butcher, & Levine, 2003). Now the second leading cause of death in the United States, obesity contributes to heart disease, stroke, high blood pressure, diabetes, depression, and many types of cancers (Rashad, Grossman, & Chou, 2005). Economic costs, in the form of health care and lost productivity, amounted to $12.7 billion in 1994 alone (Tomson, Edelsburg, Kinsey, & Oster, 1998). It is worrisome, considering these costs, that more children are becoming obese. Such costs are likely to follow the children throughout adulthood, especially as habits formed early in life can make it difficult for people to reduce their weight to a level more pleasing to them (Bednarek, Jeitschko, & Pecchenino, 2003).
The focus of my research has been to better understand factors that contribute to obesity in children. In general, food consumption and obesity are closely linked. While adults largely have control over the food they purchase and eat, children are limited to what food is available at home. Food choices for children are directly affected by household income, and economic theory suggests that participation in social welfare programs such as Women, Infants, Children (WIC) and food stamps that skew normal food purchasing habits. Thus, such programs are likely to alter the foods children would otherwise eat.
Analysis of the report of Anderson et al. (2003) on the change in distribution of obesity suggests that much of the change in the average weight of U.S. residents over time has been led by those who were already slightly overweight. This suggests that the economic factors that have lead the increase in childhood obesity primarily affect groups that were more overweight than average, making it difficult to analyze using traditional statistical techniques. Most such methods focus on the average person in a population rather than those above or below average. To get around this problem, there is a statistical technique known as quantile regression, which gives the effect of the factors of interest on people other than the average. Analysis focused on data from the National Health and Nutrition Examination Survey, published annually by the Centers for Disease Control and Prevention (CDC).
I used two measures of obesity: the Body Mass Index (BMI) (weight in kilograms divided by height in meters squared), and waist circumference. The results from both measurements were quite similar.
Analysis focused on children aged 2 to 19 years old. Factors controlled for included race, country of birth, language spoken in the home, age, school attendance, school meal availability, consumption of school meals and school meal price. The variables of interest were family income, WIC participation, and receipt of food stamps. Initial analysis revealed little information, but after separating the genders, quite interesting dynamics were apparent.
A worrisome result of the current study concerns boys who are in the 95th percentile of BMI or waist measurement. The analysis suggests that the waist size of boys in that group who receive food stamps is 15.5cm (6.1in) larger than those who do not. Similarly, the BMI is 10 points higher for boys in the same quantile who receive food stamps than for those who do not. These numbers are abnormally high, and there may be an unobserved factor affecting boys in that group that is correlated with the usage of food stamps, resulting in an erroneous result. Further analysis is needed to shore up these results.
Similar effects were not seen for WIC. On the contrary, WIC appears to make girls who are already in high weight percentiles less likely to become more overweight, and has no statistically significant effect on boys in the same category.
It appears that the structure of food programs is important in the food consumption decisions of children. While WIC limits the types of food purchased, food stamps are much less restrictive. The results seem to suggest that, at least for boys, the less restrictive nature of food stamps allows a much higher than normal intake of food, resulting in an increased prevalence of obesity.
There are still many questions to be answered. The results need to be verified across years and across surveys. This may reveal whether the abnormal results for boys are robust. Perhaps a similar study conducted using the Child Development Supplement (CDS) of the Panel Study on Income Dynamics (PSID), published by the University of Michigan, would help in this regard. The CDS provides much more information on the family than the NHANES offers. In addition, the two available years of the CDS include the same children, so trends over time can be more readily examined.
Perhaps the most difficult portion of the project was finding suitable variables to use as controls. The NHANES is meant more for national health statistics than for analysis of choice. Thus, there is limited information on education levels, income, and other factors that may have large effects on choice. This is one of the reasons why further analysis of child food choice using another survey, such as the CDS or the PSID would be helpful in corroborating my results.