In my last post, I explained that all credit card systems in the United States charge merchants a fee above marginal cost and that the systems use this revenue to stimulate card use. The Visa and MasterCard systems call this increment above cost their interchange fees, but all systems effectively do the same thing. At first blush, this sort of above cost pricing suggests serious antitrust concern. Aren’t card systems extracting supra-competitive profits from merchants? And even if they compete these profits away simulating card use, doesn’t overcharging merchants distort resource allocations? This post explains why the answers may be no.
In a typical market in which a producer competes for a single type of customer, an efficient price – one that will lead to an optimal consumption level – will generally approximate the marginal cost of production plus the profit necessary to attract investment to the industry. This pricing model is efficient because it maximizes short-run output consistently with the producer earning sufficient revenue to continue providing the product or service.
Credit card markets are not typical, however, in that card systems must price their products in a way that appeals to two inter-related sets of customers, merchants and cardholders. In such a two-sided market, a purely cost-based rule – such as “set the price charged to the cost-causer at marginal cost” – is unlikely to produce an optimal pricing structure. Although the economic literature is riddled with papers written by consultants for the card systems – not that there’s anything wrong with that – economists on all sides generally agree on this basic point. Compare Meg Guerin-Calvert and Janusz Ordover’s analysis with Michael Katz’s take, from which I quote above.
In a two-sided market, efficient pricing must take account of both total cost and the relative elasticities of demand between the two customer sets. If customers in such a market were charged the marginal cost of serving them, they would fail to internalize the benefits of their decisions to the customer set on the other side. If demand elasticities diverge to any significant degree, industry output under a marginal-cost pricing policy would be inefficiently low. To obtain an efficient output level, a producer must charge the customer set that is more sensitive to price less than marginal cost (effectively enabling those consumers to internalize the benefits to both sides of the market).
The classic example is the daily newspaper. Readers have many sources of news, including television, magazines, and the internet. Reader demand for newspapers is thus likely to be quite elastic, leading them to turn away from the morning paper if the subscription price were to approach the marginal cost of producing and delivering it. By contrast, advertisers perceive significant benefits in print advertising (so long as readership is high) and are thus willing to pay substantially above the newspapers’ marginal cost of printing and providing associated services. As a result, readers pay significantly below marginal cost and advertisers pay substantially more. Competition between newspapers and other media for advertising space still drives pricing, but not to marginal cost plus normal profit for each customer set.
This pricing pattern efficiently optimizes newspaper circulation, satisfying both the advertisers’ need for broad exposure and the readers’ need for information. Assuming that newspapers have little market power, both advertisers and readers would be worse off if pricing were forced into line with marginal cost. Were advertising fees to drop, and reader fees proportionally increased, prices would move toward marginal cost on each side of the market. Because reader demand is more elastic, however, readership would drop more than advertising would increase, and advertising rates would thus fall. As a result, the paper would (1) earn lower overall revenue; (2) be less valuable to advertisers because readership would fall; and (3) be less valuable to readers because the paper would have less revenue for newsgathering.
To the extent that the elasticity of demand varies significantly between merchants and cardholders, credit systems resemble newspapers. Assuming that merchants, like print advertisers, are willing to pay significantly above the marginal cost of credit card acceptance services, but cardholders, like newspaper readers, would be unwilling to pay the marginal cost of providing credit cards and associated services, then efficient credit card pricing should place a greater share of the costs on merchants.
One might question whether this analysis explains the array of interchange fees that the card systems now charge. Even if efficient pricing requires merchants to pay more, why should they pay more still for reward cards used primarily by wealthy customers who would arguably make roughly the same purchasing decisions with a simple basic credit card? The answer may be that price discrimination can result in competitive markets. Michael Levine’s work is instructive. He uses the example of cattle, which are generally sold as whole animals in a competitive market. The butcher’s cost of preparing particular cuts does not vary in any significant way, and those cuts are again sold in quite competitive markets. Yet, ultimate consumer prices vary considerably depending on the desirability of the cut of meat. Although the comparison between wealthy customers and filet mignon may be somewhat crude, the value of high spending consumers to merchants may justify higher card acceptance fees for the cards used by big spenders in much the same way that the desirability of the tender cuts leads to higher prices despite competition.
All this suggests that interchange fees and price discrimination based on type of card would exist in an efficient and competitive credit card market. But these pricing practices do not ensure that the market we have is in fact efficient and competitive. If above cost pricing and discrimination are not determinative, what can we look to in order to evaluate the competitive performance of credit card markets? My next post will comment on that issue.