Three Essays on Health Care and Industrial Organization
United States health spending grew dramatically over the last several decades, reaching approximately 18% of Gross Domestic Product in 2015. Much of these rising costs can be attributed to consolidation of health care providers, the provision of generous insurance plans, and innovations in medical technologies and pharmaceuticals. This dissertation studies how innovation in insurance plan design and pharmaceuticals affect health care prices, consumer behavior, consumer welfare, and the labor market. It also looks at whether certain policy interventions might lead to increased efficiency in these health care settings. In the first chapter, I study employer incentives to offer narrow-network health plans to their enrollees, and the welfare effects of switching to these plans. To do so, I estimate a model of supply and demand for health insurance offered by a large benefits administrator in Massachusetts, where I endogenize the product menu offered to consumers with respect to hospital and physician networks. I then use these estimates to study how the employer's number of products, networks, premiums, and consumer welfare would respond to a hypothetical tax on expensive health plans, in the style of the ACA "Cadillac Tax." I find that consumers' plan choices are driven primarily by inertia rather than by the value of the plan's network, and that this inertia causes the employer to continue offering plans that add little value in a strict price-versus-provider-choice tradeoff. A 60% tax on health plans in excess of $6,000 annually would cause the employer to drop each of its broad plans in favor of more narrow-network products, resulting in an approximately 21% reduction in health spending, or $76 per-member-per-month (pmpm). Consumer welfare from being moved to this new menu of products would decrease an average of $58 pmpm. I conclude that incentivizing employers to eliminate broad network products through a tax has the potential to increase social welfare. While broad-network insurance plans are one source of rising health costs, spending on pharmaceuticals has also increased over the last thirty years, accompanying large gains in life expectancy. The second chapter of this dissertation addresses the question of whether the spending on medical care has been worth the cost. Using discrete choice methods from the industrial organization literature, I, along with coauthors Claudio Lucarelli and Sean Nicholson, construct a series of quality-adjusted price indices for colorectal cancer treatments, a condition for which the average price of treatment increased dramatically from 1993 to 2005, largely due to the approval and widespread use of five new drugs over that period. Of note, we estimate a price index using parameters from a pure characteristics model, a demand technique which drops the idiosyncratic error term that produces undesirable switching patterns in traditional logit models. We find that the naive price index for these treatments greatly overstates the true price increase. In contrast, indices that account for the fact that consumers value the quality gains from pharmaceutical innovation show much more modest price increases. We also find that the magnitude of the price increase varies with modeling assumptions. Traditional logit models tend to overstate the value of product innovation, whereas the pure characteristics model implies a more substantial price increase. These results suggest the importance of modeling assumptions when constructing quality-adjusted indices. Chapter 3 turns its focus to the labor market, and examines the effects of state mandated health benefits on job transitions and job separations. Specifically, it looks at variation in state mental health parity legislation throughout the 1990s and 2000s to assess whether the addition of a high-cost benefit mandate has led to any significant displacement effect or treatment effect. While prior studies on mental health parity have focused primarily on health outcomes and the probability of having insurance following a mandate, few have paid attention to labor market outcomes. Those that did primarily reported the effects on levels of employment, ignoring potential effects on job flows. I exploit state variation in mandate passage by using restricted-access data from the Medical Expenditures Panel Survey (MEPS). The state identifiers in this data allow me to estimate changes in both employment and job transitions for the population most likely impacted by mental health mandates: employees with previously held mental health diagnoses. Finally, I provide initial estimates of the effect of the Mental Health Parity and Addiction Equity Act of 2008 on these labor market outcomes. I find that mental health parity has had little effect on employment levels, consistent with prior literature, but has caused a decrease in job separations, primarily employment to employment transitions. I present evidence that at least a third of this decline comes from a reduction in involuntary transitions, implying that mental health parity has a positive effect for individuals with untreated illnesses. In addition, I find that mandates led to increased expenditures for mental health services, further suggesting that the involuntary decline was likely due to a treatment effect of the mandates.
Economics; Health Care; Health Insurance; Industrial Organization; Labor Economics; Pharmaceuticals; Physician Markets
Barwick, Panle J; Kleiner, Samuel A
Ph. D., Economics
Doctor of Philosophy
dissertation or thesis