nr 114 - Misspecifiation of the Panzar-Rosse Model: Assessing Competition in the Banking Industry

DNB Working Papers
Date 20 September 2006

This paper demonstrates that the level of competition in the existing Panzar Rosse (P-R) literature is systematically overestimated and that the tests on both monopoly and perfect competition are distorted. This is due to the use of bank revenues divided by total assets as dependent variable in the P-R model instead of unscaled bank revenues. We provide both theoretical and empirical evidence to illustrate the impact of the misspecification on the estimation of competition and the statistical tests on the market structure. Inclusion of scale variables as explanatory variables, which is common practice in the current literature, has a similar distorting effect. Our overview of the extensive P-R literature reveals that all 28 studies considered suffer from these types of misspecification. The empirical evidence provided in this paper is based on a large sample of more than 18,000 banks in 101 countries over 16 years. We find that monopoly cannot be rejected in 28% of the countries (against 0% under misspecification) and that perfect competition cannot be rejected in 38% of the cases (against 20-30% under misspecification). Keywords: competition, banking industry, Panzar-Rosse model, misspecification, market structure JEL Classification: C52, G21, L11, L13