| This dissertation consists of three essays on empirical issues in health economics. The first essay considers the selection-efficiency trade-off in competitive health insurance markets, where insurers face incentives to exploit unpriced heterogeneity by selecting low-risk individuals ("cream-skimming"). The German Social Health Insurance does not adjust payments to sickness funds for geographic differences in costs, thereby generating incentives for funds to select against relatively more expensive areas, such as West Germany. In an audit study I present funds with fictive applicants from different locations and infer recruitment efforts by measuring callback for letters, emails and phone calls. The findings suggest that sickness funds are less responsive to requests from West German applicants, a result consistent with cream-skimming in this market.;The second essay evaluates the impacts of two school nutrition policies, California's state-wide beverage policy and Los Angeles Unified School District's food-and-beverage standards of 2004, on adolescent dietary behavior and obesity. Two large datasets on physical measures and food intake facilitate the construction of reliable control groups, including a "synthetic" control unit consisting of unaffected districts that are reweighted to closely resemble Los Angeles in the pre-intervention period. Both policies are found ineffective at reducing the prevalence of overweight or obesity. However, the district policy decreased consumption of its key targets, soda and fried foods.;The third essay examines the reliability of self-reported data in empirical analysis. Self-reported data is prone to systematic measurement error that may be constant or change in response to external events. The essay illustrates these issues with data on self-reported and measured overweight/obesity status, and BMI, height and weight z-scores of public school students in California from 2004 to 2006. In the cross-section, the prevalence of overweight/obesity is significantly lower in self-reported data relative to measured data. A district nutrition policy changed the reporting bias differentially in the treatment and control districts, so that program evaluations could find spurious positive or mill impacts of the intervention. |
