Industrial Organization of Medical CareNBER Reporter: Summer 2000
Arnould, Bertrand, and Hallock examine the effect of the competitive pressures introduced by health maintenance organization (HMO) penetration on the labor market for managers in not-for-profit hospitals. Using data on about 1,500 nonprofit hospitals from 1992-6, they find that top executive turnover rises after an increase in HMO penetration. Moreover, the increase in turnover is concentrated among the hospitals that have low levels of economic profitability and are highly leveraged financially. While the link between top executive pay and for-profit performance measures is very weak on average, it tightens with increasing HMO penetration: as HMO penetration increases, top executives are compensated more for improving the profitability of their hospitals. These results are consistent with the view that HMO penetration increases the importance of for-profit performance objectives among not-for-profit hospitals. Boards appoint new managers that are better able to compete in the new market environment and they reward incumbent managers for achieving for-profit goals. However, there is no evidence that not-for-profit boards directly reduce the incentive for their top managers to provide high-quality care. If anything, HMO penetration seems to lead hospital boards to encourage more efficiency in the fulfillment of the not-for-profit objectives.
Cutler and Seinfeld first address the impact of chief executive officers (CEOs) on hospital financial performance. Specifically, they ask if hospitals operated by managers with different backgrounds report different net profits, or if CEOs respond to crises in different ways. Their primary measures of managerial background are the educational degree of the highest-ranking executive in the hospital and the quality of the institution from which the degree was obtained. Their results suggest three primary conclusions. First, financial acumen matters. Chief executives with only a B.A. earn lower profits per adjusted admission than chief executives with advanced degrees. This is attributable to higher costs -- hospitals run by B.A.s have longer patient stays and more payroll expenses than hospitals run by people with other degrees. In addition, hospitals run by M.D./Ph.D.s have lower profits, probably because of lower revenues. Second, the quality of the educational institution attended matters. Graduates of top ten programs perform significantly better than graduates of non-top ten programs, particularly in cutting costs. Third, Cutler and Seinfeld find that these differences are most pronounced during "good times." Managerial background is related to performance when HMO enrollment is low. As HMO enrollment increases, the effect of different managerial background shrinks.
Athey and Stern analyze the productivity of technology and job design in emergency response systems (or 911 systems). The technologies include
Basic 911, whereby callers access emergency services using the single phone number 911, and Enhanced 911 (E911), where information technology is used to link automatic caller identification to a detailed database of address and location information. Both basic 911 and E911 may improve the timeliness of emergency response systems. Job design choices involve the use of Emergency Medical Dispatching (EMD), by which call-takers gather medical information, provide medical instructions over the telephone, and prioritize the allocation of ambulance and paramedic services. The authors evaluate the returns to these practices in the context of data from Pennsylvania counties in 1994-6; during that time (in response to statewide legislation), about half of the 67 counties experienced a change in technology, EMD, or both. The authors measure productivity using the health status of cardiac patients upon ambulance arrival, which could be improved by timely response. They also consider is the total hospital charges incurred by the patient, which may incorporate the benefits of precision in ambulance allocation. Their main finding is that E911 improves several health status indicators that are related closely to mortality. Scaled in terms of the probability of death from a cardiac emergency, E911 is associated with at least a 10 percent decrease on the baseline mortality rate (which is itself 7.4 percent). E911 also reduces total charges. The results on EMD are more subtle, though. Although the average benefit of EMD in this sample is negligible, technology adoption is more productive in counties without pre-existing EMD programs, at least for some productivity measures. This latter result suggests that EMD and E911 may be substitutes in providing timely emergency responses to cardiac patients.
Gentry examines the financial decisions of not-for-profit hospitals. These hospitals benefit from special tax rules that allow state authorities to issue tax-exempt bonds on their behalf. Relative to having access only to higher cost taxable debt, this access to tax-exempt debt affects the incentives for the investment and financing choices of not-for-profit hospitals. Hospitals may respond by increasing their investment in physical assets; however, they may also engage in tax arbitrage by using the tax-exempt debt while maintaining endowment assets. In Gentry's sample of tax (information) returns of not-for-profit hospitals from 1993 to 1995, the results are consistent with substantial tax planning. Of the $53.6 billion in tax-exempt liabilities of hospitals in 1995, as much as $28.1 billion could have been eliminated if hospitals spent their endowments instead of borrowing. Furthermore, after controlling for hospital size (in terms of revenues and operating assets), Gentry finds that endowment assets increase the propensity of not-for-profit hospitals to have tax-exempt liabilities and the amount of these liabilities. In contrast, endowment assets reduce the probability that a not-for-profit hospital has taxable debt; this suggests that the effect is related to the tax status of the debt rather than to the demand for credit.
A patent only protects an innovator from others producing the same product, not from others producing better products under new patents. Therefore, the source of destruction of innovative returns falls into two categories: uncreative destruction following patent expiration and creative destruction by new and superior patents, both before and after patent expiration. Lichtenberg and Philipson attempt to estimate the relative magnitudes of creative and uncreative destruction in the U.S. pharmaceuticals market. Twenty percent of the drugs approved during 1950-93 were obsolete -- no longer on the market -- by 1999. The authors estimate that creative destruction, most of which occurs while the drug is under patent, costs the innovator at least as much as uncreative destruction, which cannot occur until the drug is off patent. The reduction in the present discounted value of the innovator's sales from creative destruction appears to be at least as large as the reduction from uncreative destruction and may be much larger -- perhaps twice as large.
Despite the widespread attention paid to the determinants of equilibrium ownership structures and organizational form in private, for-profit firms, there has been much less attention paid to equilibrium choices of not-for-profit versus for-profit status. Hassett and Hubbard examine this choice for firms in the U.S. hospital industry which conforms well to recent theoretical advances stressing the importance of contractual failure in explaining organizational form. In addition, the frequency of conversions among private for-profit, private not-for-profit, and government organizational status allows the authors to empirically examine the key predictions of these theories and to compare those predictions to explanations based on altruistic or tax-related explanations of ownership status. Four principal findings emerge. First, among not-for-profit hospitals, government hospitals have higher relative input use, serve less-well-off markets, and have more bed-days represented by Medicaid patients. Second, consistent with a role for noncontractible quality, not-for-profit hospitals with a relatively high share of revenues devoted to wages are less likely to convert to for-profit status; for-profits tend to acquire not-for-profits in markets in which households have relatively low levels of educational attainment or income. Third, consistent with the noncontractible quality explanation, not-for-profits are more likely to shed not-for-profit status when there is no other government or not-for-profit hospital in the same market. Fourth, other transitions are also broadly consistent with the movement toward an equilibrium distribution of ownership status with variation explained in part by proxies for variation in preferences for noncontractible quality.
Currie and Fahr examine the effects of the penetration of managed care on the provision of charity care by hospitals, using data on all hospital discharges in California between 1988 and 1996. Their estimates suggest that higher managed care penetration rates shift the composition of Medicaid and uninsured patients across public and private hospitals. In response to higher managed care penetration, public hospitals end up with lower shares of Medicaid births and higher shares of uninsured and other Medicaid patients. They also end up with higher fractions of the charity caseload being admitted from the emergency room, which may indicate that the patients are sicker. In contrast, private hospitals reduce the fraction of the charity caseload that is admitted from the emergency room, with the biggest reduction coming in the "other Medicaid" category.
Competition and prospective payment systems (PPS) have been used widely in an attempt to control health care costs. Though much of the increase in medical costs over the past half-century has been concentrated among a few high-cost users of health care, PPS may provide incentives for selectively reducing expenditures on those users relative to low-cost users, and this pressure may be increased by competition. Chung and Meltzer use data on hospital charges and cost-to-charge ratios from California in 1983 and 1994 to examine the effects of competition on costs for high- and low-cost admissions both before and after the establishment of the Medicare Prospective Payment System (MPPS). Comparing persons above and below age 65 before and after the establishment of MPPS, the authors find that competition is associated with increased costs before MPPS in both age groups but with decreased costs afterwards, especially among those above age 65 who had the highest costs. Chung and Meltzer conclude that the combination of competition and PPS may result in incentives to selectively reduce spending among the most expensive patients. This raises important issues about the use of competition and PPS to control costs; it implies at a minimum the need to carefully monitor outcomes for the sickest patients under PPS in competitive environments.
Kessler and McClellan develop new evidence on the effects of hospital ownership and other aspects of hospital market composition on health care productivity. They analyze longitudinal data on nonrural elderly Medicare beneficiaries hospitalized for new heart attacks from 1985 to 1994. Their analysis differs from previous studies in several ways. First, they consider effects on both resource use and health outcomes. Second, they model individual patients' hospital choices as a function of exogenous patient, area, and hospital characteristics, particularly relative distances. Based on these estimates, the authors construct area-based measures of the "density" of hospitals of different types that depend only on these exogenous characteristics; the authors also include zip code fixed effects. This allows them to avoid the problem of patient selection into different types of hospitals on the basis of unobservable characteristics. It also allows them to examine "spillover" effects of ownership and other hospital characteristics -- for example, the effects of changes in ownership on the behavior of other hospitals in a market. The authors find that spillover effects of ownership are quantitatively important. Moving from essentially no presence of for-profit ownership in a market to any nontrivial presence is associated with significant reductions in hospital expenditures in less competitive markets. The "direct" effect of for-profit ownership appears to be less important, suggesting that within a market, for-profit and not-for-profit behavior is not that different. Greater teaching hospital presence and public hospital presence also have important spillover effects, particularly in less competitive markets. Hospital systems reduce expenditures, particularly in more competitive markets. At all levels of market competition, these differences in market composition generally have limited if any consequences for patient health. Differences in ownership and other aspects of market composition influence hospital productivity, mainly through spillover effects.
Barro undertakes an empirical examination of the Massachusetts hospital market reorganization experience. This paper builds on anecdotal evidence from an earlier case study paper on hospital mergers written with Cutler. In this paper, Barro empirically tests the conclusions from the case studies. He finds that consolidation facilitates partial closure through the reduction in the number of hospital beds, but does not lead to a greater likelihood of complete facility closure. The results on bed reduction, as well as results on staff reduction following consolidation, support the consolidation-for-efficiency motivation. In addition, the theory is supported by the fact that consolidations are more likely between pairs that overlap in their markets. However, the most direct implication of that theory -- cost reduction -- is not found in the data. Correspondingly, the most direct implication of the consolidation for market power hypothesis -- revenue increases -- also are not found. The evidence most in support of the market power hypothesis is indirect. The empirical results concerning efficiency gains, bed reduction, and market overlap are much weaker in explaining consolidations between hospitals that are farther apart from one another. That suggests a different motivation for those consolidations, which comprise one-third to one-half of all consolidations in Massachusetts.
In what way do for-profit and not-for-profit hospitals differ? Silverman and Skinner consider one dimension, the "upcoding" behavior of hospitals, that is, the shifting of a patient's diagnostic related group (DRG) to one that yields a greater reimbursement from the Medicare system. Upcoding has figured prominently in recent federal lawsuits against for-profit hospitals. More importantly, though, upcoding behavior provides a valuable window into the economic behavior of hospitals. The authors focus on hospital admissions involving pneumonia and respiratory infections; it is often difficult to distinguish between these diagnostic classifications, but the latter pays 50 percent more to the hospital. Between 1989 and 1996, the upcoding index rose from 21 to 33 percent among not-for-profits, but increased from 29 to 51 percent among for-profit hospitals (since then the index has fallen somewhat). There is no detectable difference in health outcomes among these groups. Finally, not-for-profit hospitals operating in heavily for-profit markets behaved much like their for-profit brethren, showing similar degrees of upcoding. In sum, for-profit hospitals were more likely to engage in upcoding and in turn may have influenced upcoding behavior by not-for-profit hospitals.
Advances in medical technology have been identified as a major driver of increases in health care costs, but increases in managed care activity may alter the incentives associated with the acquisition of new technologies and with their use. Baker and Phibbs discuss mechanisms by which managed care could influence the adoption of new technologies. They also empirically examine the relationship between HMO market share and the diffusion of neonatal intensive care, a collection of technologies for the care of high-risk newborns. They find that managed care slowed the adoption of midlevel neonatal intensive care units (NICUs) but did not influence the adoption of the most advanced units. Slowing the adoption of midlevel units should have generated savings and also could have benefited patients, because health outcomes for seriously ill newborns are better in higher-level NICUs and reductions in the availability of midlevel units appear to increase the chance of receiving care in a high-level center. Thus, in this case slowing the adoption of a new technology may unambiguously improve welfare.
Duggan investigates whether the behavior of private not-for-profit hospitals is influenced significantly by the share of competitors that are organized as profit-maximizing firms. He shows that not-for-profits in markets with more for-profit firms provide a smaller share of care to the indigent in their community, but they are also more responsive to financial incentives to treat the indigent. Government-owned hospitals in markets with more for-profit hospitals bore a greater share of the indigent care burden prior to the change in incentives caused by DSH but have since experienced a much greater change in their patient mix.
These papers will be published in the Rand Journal of Economics. In advance of the publication of that journal, most of these papers will be available at Books in Progress.