This post has been cited by the Carnival of Financial Planning #99, hosted by The Skilled Investor.
Close readers of this blog may have noticed my recent interest in issues pertaining to antitrust and competition, particularly in health care. Though I study and publish on topics related to competition among private comprehensive and drug plans in Medicare, until recently I had not read much in the economics antitrust literature. I’m beginning to rectify that deficiency.
There is a large body of literature on the economics and policy of antitrust with respect to hospital mergers. Many of the ideas apply more generally so it is as good a place to start as any. How does an economist or a court measure and reason about the state of competition in a market for hospital services?
A summary of the issues pertaining to this question is provided in a 2002 paper by Capps, Dranove, Greenstein, and Satterthwaite in The Antitrust Bulletin (Antitrust Policy and Hospital Mergers: Recommendations for a New Approach 47(4), 2002) and other publications by the same authors around the same time. The 1990s saw a lot of hospital mergers, over 900 of them. The merging of hospitals can increase the market power the merged entity has with respect to insurers, permitting the extraction of higher payments for inclusion in health plan networks and being a hinderance to competition.
The Federal Trade Commission (FTC) and the Department of Justice (DoJ) lost six straight hospital antitrust cases in the 1990s. In all but one instance the definition of relevant geographic market was instrumental in the defeat. Hospitals claimed that the market was geographically large, encompassing many other hospitals. This caused the merged entity’s market share to appear low and buoyed the defendant’s claim that the merger did not increase market power to anticompetitive levels.
The approach taken in these cases to justify a market area definition was proposed by Elzinga and Hogarty in 1972 (The Demand for Beer, The Review of Economics and Statistics 54(20)). The Elzinga/Hogarty (E/H) approach is based on the flow of consumers (or patients or goods) across a market boundary. The market area is expanded until the inflow and outflow are below a cutoff (typically 10% of total market volume).
Capps, et al. make an excellent argument as to why the E/H approach can be irrelevant to issues of market power. It all comes down to selection bias. Thinking about the case of hospitals, what if patients that travel further (“travelers”) are fundamentally different than those that do not (“non-travelers”)? It is reasonable to expect that travelers for hospital care have a different taste for travel due to their condition. One is typically more willing to travel further for specialty care for certain conditions than for routine treatments for common conditions. Thus, travelers and non-travelers can be fundamentally different. The willingness of some to travel does not eliminate the market power that hospitals have locally with respect to non-travelers. The E/H approach is flawed.
Whoops! Too bad so many cases were decided on the basis of E/H methodology. What to do? Capps et al. propose three alternatives for the case of hospitals, some of which are more generally applicable (see their “Silent Majority” and “Option Demand Markets” papers [free working paper version of the latter]). All three approaches are based on statistical and econometric analysis of the causes and degree of substitutability of one hospital for another using logit models. From these models one can estimate own- and cross-price elasticities and, hence, markups (such things are explained in What Is the Source of Price Setting Power).
With a methodology to estimate markups, Capps, et al. can define a market based on the DoJ and FTC concept of a “small but significant non-transitory increase in price” (SSNIP). If the hospitals within a market can implement a SSNIP as a joint monopoly then they are highly substitutable when in competition and therefor are a relevant set of competitors. If not then the market is increased to include other hospitals until a threshold SSNIP is reached. Within a market so defined, the methodology of Capps, et al. can be applied to estimate markups that would occur under a merger of any subset of market participants.
Is this new methodology being accepted by courts in antitrust cases? As I learned at the 2009 AcademyHealth meeting there was a recent success. Aparently a court invalidated a merger of several northern Illinois hospitals on the basis of SSNIP and not E/H analysis. It is nice to see evidence of the relevance of progress in economics.
Later: See also Dranove, Sfekas. (2009). The Revolution in Health Care Antitrust: New Methods and Provocative Implications. The Milbank Quarterly 87(3):607-632.
