The Panzar-Rosse Revenue Test and Market Power in Banking

Abstract

The Panzar-Rosse H statistic is a commonly used measure of market power in banking. It is widely believed that H>0 is inconsistent with significant market power. The theoretical part of this study rigorously disproves this perception. Instead, the possibility of H>0 under conditions of substantial market power turns out robust to the timing of firms' actions, relative costs, choice of strategic variable, degree of product differentiation, strategy (static or dynamic), and degree of heterogeneity in firms' conduct (collusive versus fringe), and hence may be common in practice. The empirical part of this study analyzes a US banking duopoly to illustrate the problem. We find 'competitive' estimates of H but, consistent with a priori expectations, non-competitive outcomes according to two alternate measures of competition. Moreover, our bank-specific estimates of H are mutually inconsistent, suggesting additional problems with the Panzar-Rosse test.