Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Because of the rising prevalence of obesity and its link to a wide range of chronic medical conditions, many researchers have sought to quantify the costs of obesity to the health care system. For example, Thompson and Wolf (2001) identified 18 such studies published between 1990 and 2000, 10 of which reported the burden of obesity on national health care systems. More recently, Finkelstein, Fiebelkorn and Wang (2003) estimated that annual medical expenses in the U.S. were on the order of $93 billion in 2002 dollars, half of which were financed by Medicare and Medicaid. Research by Sturm (2002) suggests that these costs outrank those due to smoking and drinking. In addition, Thorpe et al. (2005) linked rising health care costs to an increase in treated disease prevalence between 1987 and 2002 that was driven to a significant extent by medical conditions clinically linked to obesity. Most of these studies take a "reduced form" approach to estimating obesity-attributable medical costs. Medical expenditures due directly to obesity (in the form of bariatric surgeries or weight loss medications, for example) are relatively small compared to those on chronic conditions linked to obesity, such as diabetes and hypertension. In reduced form models the coefficient on BMI or obesity captures expenditure effects through these related medical conditions. However, is there are unobservable individual characteristics correlated with both obesity and medical expenditures that confound such estimates. For example, obese individuals may exercise less or eat unhealthy foods. These are also risk factors for conditions such as diabetes and hypertension, and if they are not controlled for their deleterious effects on health will be attributed to obesity. In order to control for such omitted individual behaviors we use an instrumental variables approach to determining obesity-attributable medical expenditures. We focus on expenditures by adults with biological children 10 year of age or older, and use child body weight as an instrument for adult body weight. Because of the genetic link between parents and their children this instrument is sufficiently powerful. In addition, studies have been unable to identify any effect of the common household environment on body weight (Hewitt 1997; Grilo and Pogue-Geile, 1991). While individual environment has been show to influence body weight, the environment that is common to all household members has not. Furthermore, adoption studies have demonstrated that the correlation between the body weight of children and their biological parents is the same regardless of whether the children grew up with their birth parents or in a different household (Vogler et al., 1995; Stunkard et al., 1986; Sorensen and Stunkard, 1993). We estimate our models using data from the 2000-2005 Medical Expenditure Panel Survey, which contains a rich array of health care information. In order to test the validity of our approach we determine whether obesity has an impact on expenditure categories that should be relatively unrelated to body weight (e.g. eye diseases other than glaucoma), after instrumenting.