Measuring Competition in Banking: A Disequilibrium Approach

Abstract

The Rosse-Panzar revenue test for competitive conditions in banking is based on observation of the impact on bank revenue of variation in factor input prices. We identify the implications for the H-statistic of misspecification bias in the revenue equation, arising when adjustment towards market equilibrium is partial and not instantaneous. In simulations, fixed effects estimation produces a measured H-statistic that is severely biased towards zero. Empirical results for the banking sectors of six developed countries corroborate our principal finding, that a dynamic formulation of the revenue equation is required for accurate identification of the H-statistic.

Keywords: banking, competition, revenue test, dynamic panel estimation

JEL Classification: L11, N20