Competition Policy in Selection Markets

This article is part of a Chronicle. See more from this Chronicle

Neale Mahoney, Andre Veiga, Glen Weyl, Oct 15, 2014

One of the oldest arguments against both competition and the policies promoting it is the problem of cream-skimming. In selectionmarkets, like insurance and finance, where some customers are cheaper to serve than others, competitors have an incentive to poach the most lucrative customers from their rivals, the “cream.” As Rothschild & Stiglitz and de Meza & Webb famously showed, this form of competition often causes severe problems, as competing firms distort their product quality or price in order to attract the cream. Such concerns were a leading part of debates over public utility regulation and the antitrust defense of AT&T, as highlighted by Faulhaber, and have been well known in economics since the work of Rothschild and Stiglitz. However cream-skimming has never made it into the models economists use to evaluate mergers and other competition policy issues.

This article reviews a pair of recent papers in which we have begun to fill this lacuna. In particular, we have found that in many realistic cases there can be too much competition in selection markets and that, even when this is not the case, many standard intuitions of competition policy are reversed by the presence of selection.

While all these markets feature selection, the patterns of selection are quite different across them. Insurance markets often feature adverse selection, where those most eager to purchase the product are the least desirable customers. Conversely, financial markets often have advantageous selection: the riskiest customers are the least likely to apply for credit and are attracted only by the teaser rates and low down-payments that firms use to attract other, more profitable customers.

In this article we discuss the severe, but very different, problems created for competition policy in these two environments. To do so, we draw on calibrated models of sub-prime automobile lending and health insurance that we developed in related papers. We then turn to the implications of our analysis for practical competition policy.