I. INTRODUCTION AND OVERVIEW
Health insurance contracts may restrict consumers' choice of medical provider in order to minimize moral hazard. For example, consumers who are insured for their marginal healthcare expenses have no incentive to enforce price discipline on their own. In this case, insurers may impose this discipline by restricting consumers' choice of medical provider to those who offer competitive prices. (1) This practice, known as "selective contracting" (SC), may have other benefits as well, including channeling consumers to efficient providers and achieving other efficiencies associated with (partial) vertical integration. SC has been utilized by insurers both in the United States and abroad. (2) However, it is not a universal feature of health insurance contracts.
In part, this is because reducing the choice set of medical providers imposes costs on consumers to the extent that consumers assign "option value" to choice (Capps, Dranove, and Satterthwaite 2003). For example, SC was widely employed in the United States by health maintenance organizations (HMOs) and other insurers in the 1990s and appears to have been an important cost control mechanism (Cutler, McClellan, and Newhouse 2000; Robinson and Phibbs 1989). However, most private U.S. health insurers moderated their use of SC during the past decade, at the same time that healthcare costs accelerated. Anecdotally, a consumer and regulatory "backlash" against provider choice restrictions forced insurers to expand choice and largely abandon the practice (Blendon et al., 1998; Vita 2001). Simultaneously, HMO enrollments declined.
The purpose of this article is to explore the economic tradeoffs of SC with a particular focus on directly measuring the value that consumers assign to choice in medical markets. I am interested in assessing whether the magnitude of any utility loss from choice restrictions can be offset by volume-based price discounts that insurers may receive from hospitals. My results bear on the larger question of whether an economic linkage can be drawn between consumers' demand for choice and their apparent rejection of what is called "tightly managed care." They also have implications for many healthcare reform proposals that necessarily imply restrictions on the consumers' choice sets.
I analyze health plan's SC of hospitals in the state of Florida between 1999 and 2003. Hospital SC restricts consumer choice in multiple ways including their choice of hospital. In addition, hospital choice restrictions imply restrictions over the use of specific physicians. (3) I start by constructing hospital networks for each of a set of plan choices and then estimate a discrete choice model of health insurance demand, conditional on each plan's restricted network. For these purposes I use the Capps, Dranove, and Satterthwaite (2003) measure of SC disutility that focuses on the "option value" of choice. I employ several different discrete choice estimators to estimate the model, including models with plan-metropolitan statistical area (MSA) fixed effects designed to control for unobserved plan attributes. This provides me with estimates of disutility resulting from choice restrictions. Subsequently, I project the estimated plan effects onto an instrumented value of plan price in order to estimate the price elasticity and also convert my disutility estimates into dollars.
My results show that consumers associated restrictive hospital networks with plan disutility during each of the years of the study period with choice elasticities that range from -0.10 to -0.32 in pooled estimates (greater effects are obtained in 2003 using models where I allow the estimates to vary by year). Using my instrumental variables methodology to recover unbiased price coefficients, I show that consumers associated a one standard deviation reduction in the size of a plan's hospital network with a reduction in plan value of between $62 and $118. …