Obesity and Nutrient Consumption: A Rational Addiction?

Article excerpt


The Surgeon General estimates the annual direct and indirect costs of obesity at approximately $117 billion. Clearly, the search for an appropriate public policy response has gone beyond a public health interest to a national economic imperative. Existing research on the economic causes of the national "obesity epidemic" cites technological changes that have reduced the price of food at the same time that burning food, or expending calories through either work or leisure activities, has become more expensive (Lakdawalla and Philipson, 2002; Philipson, 2001; Philipson and Posner, 1999), the proliferation of convenient meal solutions through fast food restaurants, the effectiveness of antismoking campaigns, greater labor market participation and engagement in low-wage jobs and lower real-food prices (Chou, Grossman, and Saffer, 2004), or individuals' propensity to become addicted to the consumption of food (Cawley, 1999). Although these studies develop comprehensive models that incorporate potential explanations from both sides of the energy balance equation (i.e., weight gain = energy in-energy out), recent evidence on aggregate energy intake relative to physical activity levels suggest that a more careful analysis of food consumption is warranted. Consequently, this study investigates whether specific macronutrients or minerals (protein, carbohydrates, fat, or sodium) are indeed addictive, and if so, whether addiction results from rational economic decisions. (1)

Cutler, Glaeser, and Shapiro (2003) cite U.S. Department of Agriculture (USDA) statistics that document a remarkable rise in the total amount of calories consumed since 1980. Further, much of this increase is attributable to a rapid rise in the consumption of refined carbohydrates, from 147 pounds per capita per year in 1980 to 200 pounds in 2000 (USDA, 2002). This trend is somewhat alarming as refined carbohydrates are a nutrient that is typically associated with obesity. Over the same period, however, calories used through both work and recreational activities have remained relatively static (Cutler, Glaeser, and Shapiro, 2003). Significantly, according to recent estimates from the Centers for Disease Control and Prevention (2004), fully 30.5% of U.S. adults were obese in 2001, and 64.5% were either overweight or obese (Flegal et al., 2002) (2) On the surface, therefore, it appears as though the obesity epidemic is largely due to not only food consumption but consumption of particular types of foods, consumption beyond the point necessary to maintain a healthy lifestyle. If consumers are rational, utility-maximizing agents as economists assume, how can their demand for food be so clearly suboptimal from a health perspective? This study is the first to test whether consumers' "rational addiction" to specific macronutrients constitutes a viable explanation for the rising incidence of obesity in the United States.

To test the rational addiction hypothesis, we use a dynamic random coefficient (mixed) logit (RCL) model similar to Erdem (1996). This approach represents a dynamic extension of the static, attribute-based RCL models used by Berry (1994), Berry, Levinsohn, and Pakes ([BLP] 1995), Nevo (2001), Chintagunta (2002), and Chintagunta, Dube, and Singh (2002) to explain the demand for differentiated products in a high-dimension discrete choice environment. RCL models convey several advantages over traditional, multilevel demand systems for problems such as this. First, they are parsimonious representations of a complex decision process. Second, they do not suffer from the "independence of irrelevant alternatives" (IIA) problem of traditional logit models, which leads to unrealistic estimates of substitutability among products. Third, viewing different products as bundles of desired attributes allows the modeler to project demand from product space into characteristics space, thus greatly reducing the number of parameters to be estimated, Fourth, RCL models are consistent with consumer utility maximization, so response parameters estimated in an RCL context are assumed to represent optimal, rational economic responses (Berry, 1994; BLP, 1995; Nevo, 2000, 2001). …


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.