In this article, we first summarize the empirical analyses that economists have used to test for price effects following the mergers of providers in different geographic regions. Second, we outline the main mechanisms that could explain these findings. We highlight three mechanisms: (1) mergers change providers’ bargaining sophistication, (2) for some high value services, providers are substitutes at the point of service for patients over a wider geographic area than assumed in the cross-market studies, and (3) providers are substitutes for inclusion in an insurer’s network, even if they are not substitutes for patients at the point of service. Our review finds that some empirical evidence can help distinguish between these mechanisms, but further study is needed to disentangle these three factors. Likely, this will require a deeper understanding of how provider and insurer incentives change post-merger.
I. Empirical Findings
A series of recent articles have documented higher hospital prices following the merger of two hospitals that, at a first glance, are not substitutes for patients at the point of service. Further, some of this research also documents higher prices for hospitals that are not direct parties to the transaction after this type of merger. All of these articles use reasonable control groups in standard difference-in-differences approaches to identifying merger effects, and most of them look at data from across the United States. We begin by summarizing the main findings from this literature and then offer some comments. In our view, the findings in the empirical literature provide credible evidence that cross-market mergers systematically lead to higher prices.
A. Overview of Results
Matt Lewis and Kevin Pflum and Leemore Dafny et al. study the price effect of mergers between non-geographically proximate hospitals. However, they approach the question in different ways. In addition to using different data, the studies differ in how they define cross-market mergers, the treatment group they analyze, and how they define control groups.
Lewis and Pflum define a cross-market merger as an acquisition in which an independent hospital is acquired by a hospital system that does not include a hospital located within 45 miles of the acquired hospital. Their treatment group consists of any independent hospital that was acquired by an out-ofmarket system, and their control group consists of any hospital that did not join a system during their sample period. Using data from 2000–2010, they find that following this type of merger, the price of services at the acquired hospitals increased by 17 percent, on average. This qualitative conclusion remains when using distance cutoffs that define the treatment group of 75 and 90 miles. In addition to these effects on acquired hospitals, they find that prices increase by about 8 percent at third-party competitor hospitals that are within seven miles of the acquired hospital.
Using data from 1998–2012, Dafny et al. define two different types of cross-market mergers, and therefore two types of treatment groups. The first treatment group consists of hospitals involved in a merger in which at least one hospital on each side of the merger is located in the same state but none of the hospitals on the opposite side of the transaction are closer than 30 minutes by car (the “in-state” treatment group). The second treatment group consists of hospitals involved in a merger in which none of the hospitals on opposite sides of the transaction are: (1) located in the same state; or (2) closer than 30 minutes by car (the “out-of-state” treatment group). For both treatment groups, they exclude hospitals that are the major hospitals in the acquired system, so called crown jewel hospitals. They use two different control groups of hospitals outside of the markets where the acquisitions took place. In one, they look at only hospitals that are part of a system, and in the other, they look at all hospitals in those markets.
For the in-state treatment group, they find that prices increase by approximately 10 percent. This result remains after dropping all acquired hospitals from the in-state treatment group. However, they do not find any impact on the prices of services at hospitals in the out-of-state treatment group.
In a related, but distinct, finding, Matt Schmitt finds that mergers that increase multimarket contact between two hospitals systems raise prices by 6 percent in the markets in which no merger took place. That is to say, imagine system B’s acquisition of C in market 2, where system A has a hospital. Due to the acquisition, A and B now overlap in market 2. However, in market 1, where A and B previously had hospitals, there is no change in market structure from the merger. Schmitt finds that prices in market 1 increase by 6 percent on average from the increase in multimarket contact between the firms.
Continue reading the full text of this article in PDF format.
