This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Spang, H. R. Articles by Arnould, R. J. Search for Related Content PubMed PubMed Citation Articles by Spang, H. R. Articles by Arnould, R. J. Related Collections Hospitals Consumer Issues

Health Tracking

MARKETWATCH

Hospital Mergers And Savings For Consumers: Exploring New Evidence

Abstract

 
This study analyzes changes in costs and prices from 1989 to 1997 for 1,767 short-term hospitals, including 204 hospitals involved in mergers; 653 hospitals that were rivals to these merging hospitals; and 910 nonmerging nonrival hospitals. We find that merging hospitals generally had lower growth in costs and prices compared with their rivals and also with nonmerging nonrival hospitals. We find that the presence and extent of these savings varied based on market and hospital conditions. However, our findings suggest that cost and price savings resulting from mergers may be smaller than estimated in earlier studies, especially through our comparison of merging hospitals with their rivals.

Traditional antitrust analysis of horizontal consolidations has focused on increased market power, which results from rising market share. Mergers have been considered illegal if they resulted in market power increases great enough to allow nontransitory increases in hospital prices. Rule-of-reason analysis has been applied to mergers, which means that the act of the merger is not illegal per se but that the legality of the merger depends upon the reasonableness (or lack thereof) of the predicted impact of the merger. More recently, courts have taken a more holistic approach to antitrust analysis, considering a merger’s overall effect on consumers’ welfare. This expands the rule-of-reason analysis to balance the welfare-enhancing effects of consolidation, such as increased efficiency, with welfare-reducing effects, such as the potential to control prices.

In recent years the courts have increasingly accepted potential cost savings as sufficient basis for allowing a merger. Some responsible for the enforcement of antitrust laws have even suggested that hospital mergers should be presumed beneficial for consumers unless they pose a severe threat to competition.² However, health markets are dynamic, and the balance of power between payers and providers can shift rapidly. It is important for researchers to continue to assess the impact of hospital mergers on consumers, by tracing their effects on costs and prices. In this paper we use costs to refer to total expenses per admission, and prices to refer to net patient revenues per adjusted admission.

One method for analyzing the impact of hospital mergers is to evaluate how costs and prices change following a merger and how changes compare with those of nonmerging hospitals with similar characteristics. It is important to evaluate how these changes vary by characteristics of institutions and markets, to help antitrust enforcers identify situations in which hospital mergers are more or less likely to be harmful to consumers. This approach is consistent with the current interpretation of rule-of-reason analysis of horizontal consolidations. Our analysis is based on mergers that occurred between 1989 and 1997. We extend earlier analyses of hospital mergers by enhancing the time horizon and expanding the analysis to study the behavior of rivals.³

Background. Studies of hospital merger effects generally fall into two main groups: cross-sectional and longitudinal. One type of cross-sectional study estimates the relationship between hospital costs or prices and market concentration. This relationship is then used to infer what would happen in the event of a merger.⁴ Longitudinal studies generally examine actual merging hospitals using time periods before and after merger to measure changes in hospital costs, prices, or other key economic variables, usually in comparison to similar nonmerging hospitals.⁵

Robert Connor and colleagues conducted an important longitudinal study in which they examined costs and prices for 3,500 hospitals over the nine-year period 1986 to 1994.⁶ They separated hospitals into merging and nonmerging categories and examined cost and price changes for each group, distinguishing these changes by market and hospital characteristics.⁷ They found that cost growth was 7.2 percent lower and price growth was 7.1 percent lower when all merging hospitals were compared with all nonmerging hospitals. In addition, they found that the extent of cost and price savings varied across different subgroups of hospitals. Subsequent multivariate analysis by Connor and colleagues confirmed many of these descriptive findings.⁸

Our study refines and expands the work of Connor and colleagues. We updated the number of years included in their study and implemented three key changes in the approach. First, we eliminated rural hospitals from the sample. These hospitals face market pressures that differ from those in urban areas, which may affect operational decisions. Second, we excluded from our nonmerging group those hospitals that were affiliated or became affiliated with a multihospital system that had an increase in local market power.⁹ This specific type of system transaction may present opportunities for efficiency gains and increased market power that are similar to those created through a hospital merger. By eliminating hospitals involved in these types of transactions, we can better gauge merger effects because we contrast merging hospitals with a group of institutions that have not experienced the potential cost and price influences precipitated by ownership consolidation. Third, we separated the nonmerging hospitals into two groups: nonmerging rival hospitals and nonmerging nonrival hospitals. The former were facilities not involved in a merger but located in a market where a merger occurred, so they face the same market pressures that the merging hospitals face.

Data And Methods

Top Data And Methods Results Implications For Public And... NOTES

 
Our study examined short-term U.S. hospitals between 1989 and 1997. We examined American Hospital Association (AHA)–identified mergers in which the assets and operation of two or more previously separate hospitals were joined under one remaining operating entity. Our analysis focused on mergers in which the hospitals were located in the same metropolitan statistical area (MSA).¹⁰ Hospital data came from the AHA Annual Survey of Hospitals and Medicare Cost Report (MCR). We included hospitals that reported expense and revenue data to the AHA annual survey or MCR in 1989 and 1997.¹¹ Market data were calculated using AHA annual survey and Area Resource File (ARF) data. In total, the data were available for 1,767 hospitals: 204 hospitals involved in 108 mergers; 653 rival hospitals; and 910 nonmerging nonrival hospitals.

The market characteristics examined were the Herfindahl- Hirschman Index (HHI) of market concentration, health maintenance organization (HMO) penetration, and MSA population size. The HHI equals the sum of the square of market shares based on adjusted hospital admissions for all hospitals in a given market.¹² Markets were separated into high- and low-concentration groups using the threshold values of the Department of Justice and Federal Trade Commission (namely, an HHI equal to or greater than 1,800). The HMO penetration variable measures the percentage of the local population covered by HMO plans, using measures developed by Douglas Wholey and colleagues that correct for home office reporting.¹³ We distinguish high and low HMO market penetration using the median value of this variable for all hospitals in our sample. The MSA size variable is categorical, based on the number of residents in an MSA. Finally, the hospital characteristics included staffed beds, adjusted inpatient admissions, occupancy rate, percentage of Medicare and Medicaid patients, Council of Teaching Hospitals (COTH) membership, for-profit ownership, and government ownership.

Our analysis focused on hospital costs, prices, and organizational and market characteristics in the premerger year 1989, and how these changed from 1989 to 1997. In one analysis we replicated the comparisons of Connor and colleagues, contrasting merging hospitals with all nonmerging hospitals. We then repeated this analysis, separating nonmerging hospitals into rival and nonrival categories. Cost was defined as total expenses per adjusted admission. Price was defined as net patient revenue per adjusted admission.

Results

Top Data And Methods Results Implications For Public And... NOTES

 
Market area and hospital characteristics. HHI levels were lower in the markets of merging hospitals and their rivals compared with nonmerging nonrival hospitals in the baseline period of 1989, as we would expect from the focus of antitrust enforcement on high-concentration markets and the simple fact that markets with more hospitals are more likely to have mergers (Exhibit 1). Average HMO penetration was also higher for the merging and rival hospital markets in the 1989 base year as compared with the nonmerging nonrival hospitals, although HMO growth rates were consistent across all three hospital groups. Finally, nonmerging nonrival hospitals tended to be located in smaller markets: More than 75 percent of them were located in MSAs with fewer than a million people, compared with 60 percent of merging hospitals and only 35 percent of nonmerging rival hospitals. Generally, these findings support our decision to separate nonmerging hospitals into rival and nonrival groups.

Prices and costs. Exhibit 2 examines price and cost changes from 1989 to 1997 for merging and nonmerging hospitals, where the latter group combines rival and nonrival hospitals. Overall, cost and price increases in our study period were smaller (around 20–30 percent) than those reported by Connor and colleagues for 1986–1994 (around 75–80 percent). This likely reflects growing pressure on the hospital industry to contain costs and prices due to growth in managed care, selective contracting, and tougher payer negotiations on price discounts and contract terms.¹⁴

Comparison of merging, nonmerging, rival, and nonrival hospitals. Exhibit 3 separates the nonmerging hospital group into rival and nonrival categories. This permits a better interpretation of merger effects by comparing merging hospitals to rival hospitals that are located in the same markets and thus face similar conditions. As such, the comparisons of cost and price changes for these latter hospitals effectively hold market pressures constant. Overall, merging hospitals had cost growth that was thirteen percentage points lower than that of nonmerging nonrival hospitals, but only 6.6 percentage points lower than that of rival hospitals. The pattern for price savings was similar.

We also found that cost and price growth of merging hospitals was lower in markets with high HMO penetration, but only when merging hospitals were compared with nonmerging nonrival hospitals. Cost and price growth was very similar when merging hospitals were compared with rivals in such markets. Thus, although Connor and colleagues concluded that hospital mergers in markets with high HMO market share generated substantial merger-related cost and price savings, our results suggest that these savings were modest or nonexistent when merging hospitals were compared with rival (nonmerging) hospitals in the same markets.

Patterns of cost and price growth for hospitals grouped by organizational characteristics were similar to those in Exhibit 2. Generally, however, merger cost and price savings were lower when merging hospitals were compared compared with nonmerging rival hospitals.

Implications For Public And Private Policy

Top Data And Methods Results Implications For Public And... NOTES

 
Given the ebbs and flows of the hospital market, it is important for researchers to continually reassess potential hospital merger effects. In our study we found that merger-related cost and price savings exist, but the presence and extent of these savings vary based on market and hospital conditions. These savings appear to be highest in competitive markets and also when low-occupancy, nonteaching, or nonprofit hospitals merge. These findings parallel those of Connor and colleagues; together, both studies provide compelling evidence that horizontal mergers under certain circumstances hold potential for beneficial cost and price effects.

However, our results suggest that merger-related savings are likely to be smaller than originally anticipated. Our analysis separated nonmerging hospitals into two groups: nonrival hospitals and rival hospitals. In all comparisons, merger cost and price savings (if present) were lower when merging hospitals were compared with rival institutions, not simply with all nonmerging institutions. This suggests that there are likely unobserved differences in the merging and nonmerging markets that affect the cost and price behavior of all hospitals. Thus, comparisons of merging hospitals with rival institutions provide a better basis for identifying merger-related cost and price savings.

Another important difference in findings between our study and that of Connor and colleagues relates to potential merger effects in markets with high HMO penetration. The earlier study found that merger-related cost and price savings were greater in such markets. We found that this was only true when merging hospitals were compared with nonmerging nonrival hospitals. When we compared merging hospitals with their rivals in these markets, we found that merger-related cost savings were modest (only 2.3 percentage points) and that both merging and rival hospitals had nearly identical price growth. Our findings suggested that merger-related cost and price savings were actually higher in markets with low HMO penetration when merging and rival hospitals were compared. It is possible that hospitals in high-penetration markets have already been forced to engage in more cost and price restraints than those in markets with less HMO penetration, thereby leaving more room for cost and price constraints resulting from mergers in the latter markets. An alternative explanation is that rivals in markets with low HMO penetration may have felt less need to contain cost and price growth; thus, the effects of efficiency-generating actions of merging hospitals may have been more readily apparent in these markets. Further research is needed to parse out these actual effects.

These findings have important implications for antitrust policy. First, our findings suggest that no "magic formula" can be used to sort mergers into the good and the harmful. There does not appear to be an appropriate "one size fits all" policy. The level of competition in markets in which mergers occur has been and should continue to be an important determinant of the potential impact of a merger. However, because merger-related cost and price savings appear to be more modest than anticipated, continuing antitrust scrutiny of hospital mergers remains important. Cost savings can indeed result and, in some instances, may be quite large, but anti-trust enforcers must be diligent in assessing the underlying arguments and the market-wide effects of these transactions.

Our findings suggest that the competitive impact of the level of HMO penetration may be more complex than expected. Connor and colleagues speculated that trends toward increased managed care penetration and consolidation of buying power through purchaser coalitions would increase the likelihood that cost savings would result from hospital mergers and that these savings would be passed on to consumers in the form of lower prices. Our findings may not be totally consistent with this interpretation and suggest the need for continued analysis.

The position that hospital mergers should be presumed beneficial for consumers unless they pose severe threats to competition is not well supported, given historical patterns of cost and price changes found in this analysis. The results suggest that some mergers do have beneficial results, but a variety of characteristics of hospitals and hospital markets must be carefully considered to reduce the probability of allowing mergers that may to be harmful to competition. Therefore, a rule-of-reason approach in assessing merger effects continues to be a well-advised approach for antitrust analysis and enforcement.

Study limitations. The research reported here has limitations, which can be addressed through continued research. First, as did the original study, we have assessed merger cost and price savings through a simple cross-tabular framework. Although this provides a straightforward approach for illustrating differences in price and cost growth across categories of hospitals, more sophisticated multivariate analyses would illuminate merger effects even further. Of course, an advantage of using this approach is that it allows direct comparison of our findings and observations with the original study of Connor and his colleagues.

Second, the complex intramarket behavior of rival and merging hospitals requires careful focused study. Merging and rival hospitals might engage in coordinated or oligopolistic behavior in highly concentrated markets. This type of behavior is likely to be less relevant for the period we have studied, given that our merging and rival hospitals were predominantly located in markets with moderate or low concentration (average HHI of 1,823 and 1,003, respectively). However, hospital markets have continued to consolidate and recently there has been less antitrust scrutiny of hospital mergers. Continued assessment of hospital consolidation and resulting market behavior is important to future policy debate.

Editor's Notes

 
Heather Spang is a consultant at Lexecon, Inc., in Chicago. Gloria Bazzoli is a professor of health administration, Medical College of Virginia/Virginia Commonwealth University. At the time this research was conducted, Bazzoli was research professor with the Institute for Health Services Research and Policy Studies of Northwestern University and Spang was a doctoral student at the University of Illinois at Urbana-Champaign. Richard Arnould is a professor and head, Department of Economics, University of Illinois.

This research was supported by a grant from the Agency for Healthcare Research and Quality, no. R01 HS09201-01. It was conducted in collaboration with the Health Research and Educational Trust of the American Hospital Association, which undertook aspects of the data analysis. The research was part of Heather Spang’s doctoral dissertation while at the University of Illinois.

NOTES

Top Data And Methods Results Implications For Public And... NOTES

 

1. For a review of these issues, see M. Gaynor and D. Haas-Wilson, "Change, Consolidation, and Competition in Health Care Markets," Journal of Economic Perspectives (Winter 1999): 141–164.
2. C.A. Varney, "Efficiency Justifications in Hospital Mergers and Vertical Integration Concerns" (Remarks made before the Health Care Antitrust Forum, Chicago, 2 May 1995).
3. R.A. Connor et al., "Which Types of Hospital Mergers Save Consumers Money?" Health Affairs (Nov/Dec 1997): 62–74.
4. Examples include J.C. Robinson and H.S. Luft, "The Impact of Hospital Market Structure on Patient Volume, Average Length of Stay, and the Cost of Care," Journal of Health Economics (December 1985): 333–356; L.M. Manheim, G.J. Bazzoli, and M.W. Sohn, "Local Hospital Competition in Large Metropolitan Areas," Journal of Economics and Management Strategy 3, no. 1 (1994): 143–167; and W.J. Lynk, "Nonprofit Hospital Mergers and the Exercise of Market Power," Journal of Law and Economics (October 1995): 143–167.
5. Examples include Connor et al., "Which Types of Hospital Mergers Save Consumers Money?";; U.T. Sinay, "Pre- and Post-Merger Investigation of Hospital Mergers," Eastern Economic Journal (Winter 1998): 83–97; and J.A. Alexander, M.T. Halpern, and S.D. Lee, "The Short-Term Effects of Merger on Hospital Operations," Health Services Research (February 1996): 827–847.
6. Connor et al., "Which Types of Hospital Mergers Save Consumers Money?"
7. The definition of merger that Connor and colleagues use is based on that of the American Hospital Association, which distinguishes full-asset mergers, where separate institutions come together under one license, from system acquisitions, where hospitals retain separate licenses and administration even though they are owned by one parent. G.J. Bazzoli et al., "A Taxonomy of Health Systems and Networks: Bringing Order Out of Chaos," Health Services Research (February 1999): 1683–1717, noted that some systems are characterized by highly decentralized control in which affiliated hospitals have substantial say over the service and product lines they offer. Such decentralized control would not occur in full-asset mergers.
8. R.A. Connor, R.D. Feldman, and B.E. Dowd, "The Effects of Market Concentration and Horizontal Mergers on Hospital Costs and Prices," International Journal of Economics and Business 5, no. 2 (1998): 159–180.
9. In our study, system acquisitions were the purchase of a hospital by a multihospital system in a market where the purchasing system had a market presence (that is, owned at least one hospital) prior to acquisition. In this case, the market shares of the new and existing system hospitals combine into one and thus increase hospital concentration and market power for the hospitals involved. System acquisitions were identified using changes in AHA system rosters from year to year. We decided not to include hospitals involved in these types of system transactions in our "merger group," even though they shared the potential for efficiency and market-power effects, because this would represent a substantial deviation from Connor and colleagues’ original design.
10. This market definition deviates slightly from Connor’s use of Health Service Areas (HSAs). This difference is likely to have a minimal effect because HSAs generally are similar to MSA definitions in urban areas, and we exclude rural hospitals. See D.M. Makuc et al., Vital and Health Statistics: Health Service Areas for the United States, DHHS Pub. no. (PHS)92-1386 (Hyattsville, Md.: U.S. Department of Health and Human Services, November 1991).
11. Approximately 25 percent of hospitals do not report financial data in the AHA annual survey. To maintain as large and unbiased a sample as possible, we substituted variables from the MCR data when AHA annual survey data were missing. The hospitals that did not report financial data to the AHA were more likely to be for-profit, but there did not appear to be any other tendencies that would create bias in our sample or results. In addition, we ran correlations on the variables from the MCR and AHA data sets when hospitals reported to both, and the correlations were very high (usually around 0.97).
12. Hospitals within the same multihospital system were considered the same hospital for the purpose of measuring market share and, thus, HHI.
13. D. Wholey, R. Feldman, and J.B. Christianson, "The Effect of Market Structure on HMO Premiums," Journal of Health Economics 14, no. 1 (1995): 81–105.
14. An alternative explanation for this result might be regression to the mean rather than efficiency effects of consolidation. Regression to the mean is a phenomenon identified in health economics by David Dranove and Kenneth Cone. They contended that state regulators may be implementing rate regulation not only because hospitals had systematically high costs but also as a reaction to anomalistic high hospital costs that occurred in a given period. Because these anomalistic costs would return to normal through regression to the mean, the effects of state regulation on costs would necessarily be overly inflated. D. Dranove and K. Cone, "Do State Rate Setting Regulations Really Lower Hospital Expenses?" Journal of Health Economics 4 (1985): 159–165. We believe that the very nature of the hospital consolidations is inconsistent with regression to the mean. Hospital consolidations are extremely complicated and expensive ventures, taking years to plan and implement. Unusually high hospital costs for a time period are likely to return to normal as merger plans develop. Further, hospital administrators (unlike state regulators) have sufficient involvement in and knowledge of the institution to distinguish these phenomena.

                      What's this?