• Market concentration and entry

    Imagine a highly concentrated market like, say, that for hospital services or health insurance. Does the fact that the market is highly concentrated encourage or deter entry by other firms? This is a harder question than it might seem.

    By “highly concentrated” I mean, roughly, that most of the total market share is locked up by a small number of firms. At the extreme is a monopoly, one firm with 100% of the market share. A more technically useful notion of market share is the Herfindahl index (hereafter denoted H), which is the sum of squared market shares.  H is proportional to the degree to which the prices firms charge exceed marginal costs in homogeneous goods markets, those for which products produced by one firm are perfect substitutes for those produced by another. The higher H the higher the price markup.

    However, few markets are perfectly homogeneous, but some are closer than others. Actually, hospital and health insurance markets are very far from homogeneous. One hospital can be quite different from another in many ways. It matters where you get your surgery! Insurance products differ in many dimensions. A high-deductible plan is not the same as a plan with first dollar coverage, for instance. Despite this, the FTC and DOJ still use H as a measure of market concentration in these industries. Thus, so do health economists, and so will I.

    Back to the main line: what is the effect of H on entry? If a prospective entrant into the market–a new hospital, a new health insurer–comes along and observes the market’s structure (H), what does it do? Enter or not?

    Well, if higher H implies higher markups over marginal cost that suggests a profit opportunity. A market with high H might therefore attract entrants, driving H and markups down to more competitive levels. That’s right, but not complete. It isn’t how a health economist would think. When considering a causal relationship, like H’s effect on prices and entry, one should always ask the next questions, “What affects the thing that we think affects the outcome? What causes it to be high or low?” In this case, “What affects H? Why would it be high or low?”

    High H means high market concentration. But if high market concentration means high profits, which as argued above should invite entry pushing H lower, how has H remained high? What is causing the (presumed) market equilibrium at high levels of concentration?

    An answer is barriers to entry, which could include factors that would cause prospective rivals to face higher costs (see my “Market predation” post on this). Another is product differentiation, which includes reputation effects. In a widely cited 1997 paper on banking markets, Amel and Liang explain,

    [M]arket concentration may be an entry barrier to the extent that it reflects superior product differentiation or a first-mover advantage of incumbents. In the case of banking, if transactions costs of changing banks are high (Sharpe, 1996) or if incumbent firms can establish valuable reputations, then higher market concentration will be negatively correlated with entry.

    Thus, from reasoning alone the effect of H on entry is ambiguous. It depends on whether there are entry barriers or not. One has to go to the data to learn, empirically, which way it goes for a given market at a given time. Amel and Liang did just that for banking markets and found that the effect of H on entry in banking markets (weakly) supports the entry barrier hypothesis.

    Estimated coefficients on [H] are significant in only two [of eight] equations, and in both cases the coefficient is negative. …[T]he negative coefficients are more consistent with the hypothesis that high market concentration acts as an entry barrier rather than an indicator of the ability to collude. These results are generally consistent with previous empirical studies (e.g., Rose, 1977 and 1979; Hanweck, 1971; and Amel, 1989).

    What about the health insurance or hospital industries? Does high market concentration invite entry or not? The fact that H has been increasing suggests not. We’ve already hypothesized about possible entry barriers in a prior post. The truth of the matter can only be revealed by a sound empirical analysis, however.

    Share
    Comments closed