Three Essays on the Impacts of Public Policy on Behavioral Health

Abstract

This dissertation examines the impacts of public policies on the behavioral health of adults in the United States. The first essay entitled, “The Impact of Food Advertisements on Changing Eating Behaviors: An Experimental Study”, examines how three types of food advertising (healthy food, unhealthy food, and anti-obesity) impact consumers’ caloric and nutrient content selected in a lunch menu. The analysis is based on an economic experiment conducted with 186 adult non-undergraduate student subjects, each of which were randomly placed into either the control group or one of four treatments: (1) healthy food advertising, (2) anti-obesity advertising, (3) unhealthy food advertising, and (4) mixed (all three food) advertising. The results indicate that healthy, anti-obesity, and mixed food advertising reduced intakes of total calories, fat, sodium, and carbohydrates. Similarly, anti-obesity, healthy, and mixed food advertising results in increasing the probability of selecting more healthy items and fewer unhealthy items from a menu. Healthy food advertising has a stronger impact than anti-obesity or mixed food advertising. The second paper, “Food Stamps, Food Insufficiency and Health of the Elderly”, evaluates the efficacy of the Food Stamp Program (FSP) on improving the well-being of elderly Americans. The overarching objective of this study is to determine whether and how elderly health status is affected by FSP participation, food insufficiency and other determinants. To carry out this goal, first a theoretical framework is developed to ascertain why so few eligible elderly households participate in the FSP, and how food intake affects health status. In addition, the model examines the main determinants of food insufficiency and how FSP participation and food insufficiency are linked to each other and then to health status. The data utilized in this study are a subset of the Health and Retirement Survey (HRS) from the year 2002. State-specific FSP criteria are used to determine the eligibility of elderly households in the sample. The method used to examine these linkages is a two-step econometric model with two instrumental variables for the endogeneity of food stamp program participation and food insufficiency. In the first step, a simultaneous multivariate Probit model of endogenous FSP participation and food insufficiency equations is estimated. Based on these results, probabilities of FSP participation and food insufficiency are predicted in Step One for use in Step Two. In Step Two, an Ordered Probit of health status is estimated as a function of the predicted FSP participation and predicted food insecurity, controlling for other determinants of health status. The estimating procedure extends the Murphy and Topel’s (1985) standard error correction method to the case of two predicted explanatory variables. After correcting the standard errors, some coefficients lose their significance indicating the importance of the standard error correction procedure. Specifically, without the correction, FSP participation is found to worsen food insufficiency, but this relationship becomes insignificant after the correction. Conversely, being food insufficient significantly worsen health status with and without the correction procedure. The results suggest that FSP net benefits, though increasing food purchasing power, are inadequate to help elderly to achieve the minimum threshold of food intake that could significantly improve health status. The third paper, “The relationship between Unemployment and Obesity: Evidence from NLSY 97 Survey Data”, investigates simultaneous relationship between unemployment and obesity in the U.S. The unemployment and obesity probit equations are estimated simultaneously using an instrumental variables approach to deal with the problem of endogeneity. The mother’s BMI and unemployment insurance are instrumented for obesity and unemployment, respectively. The results reveal unemployment significantly increases the likelihood of obesity, but not vice versa. This significant finding raises concerns on potential obesity-related health problems on unemployed individuals for policy makers. Although the results found no statistical evidence of weight discrimination, it is inconclusive that normal or underweight individuals would decrease their risks of unemployment due to no statistical difference in the probability of unemployment between obese and non-obese individuals.