Abstract
Obesity in the United States has become an epidemic with both health and economic consequences. Research has shown that being obese can lead to serious health consequences such as heart disease, Type 2 diabetes, various types of cancer and more (Kuchler and Ballenger, 2002). Some of these health consequences can be fatal if not treated properly. Being obese can also lead to a decrease in an individual’s lifespan (Fontaine et al., 2003). Furthermore, obesity has economic consequences such as an increase in health expenditures, and loss of productivity, which cost U.S. society approximately $117 billion a year (Kuchler and Ballenger, 2002). This thesis extends the research on this problem by explicitly considering if both fast-food and full-service restaurant densities can explain an individual’s Body Mass Index (BMI) and the probability of an individual being obese. This thesis also extends the literature by using a more thorough repeated cross-section analysis than past studies by including state and time fixed effects to control for omitted variables that may bias ordinary least squares estimators. The thesis also determines if fast-food and full-service restaurant densities are endogenous given the possibility that these eating establishments may locate in areas where the population has a high measure of BMI. The analysis reveals that fast-food restaurant density has a negative relationship with BMI and obesity. In contrast, full-service restaurant density is found to have a positive relationship with BMI and obesity. These results are found to have superior explanatory power than the results found in Chou et al. (2004) yet still not as significant as expected. Also, this thesis distinguishes between fast-food restaurant density and full-service restaurant density unlike Chou et al. (2004). The instrumental variables analysis revealed that endogeneity was a problem. Regardless, the results implied that the data and analysis led to more explanatory results than previous studies. Furthermore, both models analyzed in this thesis revealed that either fast-food or full-service restaurant densities or both are endogenous. Also, using fast-food restaurant density at the state level may be an improper indicator of fast-food restaurants since we make the assumption that fast-food density is the same throughout each state. Lastly, the difficulty in locating valid instruments suggests that further research could consider a search for valid instruments or try to estimate the models using alternative instrumental variable methods.