Assessing competition in hospital care markets: the importance of accounting for quality differentiation.

Quality differentiation is especially important in the hospital industry, where the choices of Medicare patients are unaffected by prices. Unlike previous studies that use geographic market concentration to estimate hospital competitiveness, this article emphasizes the importance of quality differentiation in this spatially differentiated market. I estimate a random-coefficients discrete-choice model that predicts patient flow to different hospitals and find that demand responses to both distance and quality are substantial. The estimates suggest that patients do not substitute toward alternative hospitals in proportion to current market shares, implying that geographic market concentration is an inappropriate measure of hospital competitiveness.

1. Introduction

* Firms compete on the basis of quality in many industries, including the hospital care industry. Hospital care is vertically differentiated by quality, and horizontally differentiated according to geographic location. Travel is costly, particularly when medical care is sought on an emergency basis. Consequently, there is a tradeoff between travel time and quality, and this tradeoff gives hospitals market power.

The hospital care industry has undergone considerable change since the early 1980s. Concerns about rising health care costs have led to a shift toward prospective reimbursement systems with reductions in reimbursement levels, while advancements in medical technology have reduced the length of inpatient stays and enabled more procedures to be done on an outpatient basis. The fall in demand for inpatient care, together with the increased pressures for cost containment, has changed the competitive environment of hospitals. These changes resulted in a consolidation of the hospital care industry throughout the 1990s.

A key issue that has arisen in attempts to assess the effects of these changes in market structure and competition, as well as in hospital antitrust cases, has been market definition (Gaynor and Haas-Wilson, 1999; Gaynor and Vogt, 2000). Previous studies have used geographic market-concentration measures to estimate the effect of the competitive environment on the behavior of individual hospitals. To evaluate the impact of these changes, however, we also need to account for quality differences between hospitals. In contrast with previous studies, I examine the predicted changes in individual demands due to quality changes, quantifying the importance of quality differentiation given the geographic differentiation. Concentration measures based on geographic spatial definitions do not capture the impact of competition on firm behavior in vertically differentiated markets. In such industries, the price competition literature has used estimates of the demand function to quantify the ability of firms to set prices (Bresnahan, 1989). This methodology cannot be directly applied to the hospital industry, but it is adapted to analyze spatial and quality competition in this market.

This article uses Medicare claims data that provide a record of all patients over 65 who suffered a heart attack. The dataset is a rich source of information not only about individual patients and their hospital choices, but also about the treatments and outcomes of patients admitted to particular hospitals. This is supplemented by data from the American Hospital Association (AHA) annual survey of all hospitals in the United States. These data sources give considerable individual-level information about both demanders and suppliers, enabling a detailed empirical analysis. The data also allow the analysis to focus on the hospital care for Medicare patients. The costs of inpatient care are insured under Medicare, and reimbursements to hospitals are determined by the Centers for Medicare and Medicaid Services (CMS), previously the Health Care Financing Administration (HCFA). (Since I use data from 1994, before the change occurred, I will use HCFA instead of CMS throughout.) Consequently, price competition is absent from this market, providing an opportunity to focus on quality competition, and abstract away from prices, which are difficult to measure in the non-Medicare hospital market.

* Estimating demand for hospital care. I begin the analysis by estimating a hospital-choice model with the above data, using a random-coefficients discrete-choice framework. With the detailed individual-level data, I am able to estimate a choice model that includes much individual variation. For the purposes of this article, it does not matter whether it is the patient, his family, or his physician who makes the choice, and I will refer to the decision maker as the "demander." The estimates of the demand model show that the location and quality of the hospital are key determinants of hospital choice, highlighting the importance of both spatial and quality differentiation of firms in this market. To capture the different aspects of hospital quality, the estimation includes both input and outcome measures of quality. Patient characteristics are found to affect not only the quality-distance tradeoff, but also the valuation of these different aspects of quality.

The estimates of the random-coefficients logit suggest that 14% or more of the demanders appear to care little about distance. This result is surprising, given the importance of getting immediate medical attention after having a heart attack. The obvious interpretation of this finding is that demanders who are found to care little about the distance of the hospital from the patient's home are likely to be away from home at the time the heart attack occurs. This interpretation is consistent with the additional finding that the age of the patient shifts the distribution of the coefficients, since we expect that the older patients are less likely to be away from their homes. I compare my estimates of hospital choice to those obtained using the logit, which is widely used to model hospital choice. By failing to allow for this unobserved difference between patients' valuation of distance from their homes to the hospital, the estimates of the logit suggest that patients are more willing to travel than implied by the estimates of the random-coefficients logit.

* Implications of demand estimates for assessing competition. There is a large literature that has looked at competition in the hospital market. Most studies use geographic market concentration to measure competitiveness. The importance of quality as well as geographic differentiation implies that any measure of a hospital's competitiveness or market power should account for both dimensions of differentiation. To illustrate this, I use the demand estimates to examine the effect of unilateral changes in hospital quality, in terms of the number of additional patients the hospitals would be able to attract.

The results suggest that increasing different aspects of quality substantially raises the predicted demand of the average hospital in my data, highlighting the importance of quality differentiation of hospitals. The diagnosis I focus on is one in which medical care is sought urgently, so that quality is likely to be an even more important determinant of choice for diagnoses where travel is less costly. Hence, quality differentiation is likely to have even greater effects on competition in the treatment of other diagnoses.

The gains in demand due to adopting new technologies are found to be sensitive to modest changes in the valuation of quality relative to distance, pointing to the importance of accurately measuring the quality-distance tradeoff. When a hospital closure is considered, patients are not found to substitute toward alternative hospitals according to prior market shares, implying that current market shares would not accurately capture the market power of hospitals. This suggests a need to reexamine antitrust policies, as well as assessments of the changes in the hospital industry, which have hinged on market-definition issues based on current market shares.

The theoretical literature does not offer an unambiguous prediction about the effect of hospital competition on the welfare of consumers. In most product-differentiated industries, less competition increases the ability of firms to raise prices, so that consumers are worse off. This may not be the case for hospital care. The informational problems in the demand for health care imply that physicians, acting as imperfect agents for their patients, determine demand for medical care services. Also, consumers purchase health insurance because of the uncertainty in demand, so that it is insurers, not consumers, who are the payers in this market. These features lead to the hypothesized "medical arms race" in which hospitals do not compete on prices but instead for physicians, who are argued to favor hospitals with technologically advanced medical equipment. Consequently, hospitals purchase costly equipment to increase their attractiveness to physicians and their patients, who are in turn insensitive to prices since hospital charges are covered by insurers or the government in the case of Medicare and Medicaid. Thus, such nonprice competition implies that costs would be highest in markets where competition is most keen, the reverse of what one would expect in most other industries. There may also be wasteful duplication of facilities, and even unnecessary use of procedures. These issues are especially relevant because the adoption of costly new medical technologies has been found to be an important factor in explaining the rising health care costs (Newhouse, 1992; Cutler and McClellan, 1996).