This paper tests market power in the banking industry. Price-cost margins predicted by different oligopoly models are calculated using discrete-choice demand estimates of own-price and cross-price elasticities. These predicted price-cost margins are then compared with price-cost margins computed using observed interest rates and estimates of marginal costs. This paper is among the first to apply this methodology on a detailed, bank-level dataset from the retail banking sector. It extends on previous papers and illustrates the advantages of structural modelling by simulating a counterfactual merger experiment with a number of mergers, each of which involves two major banks, and studying the unilateral effect of the mergers on interest rates. This provides more evidence that concentration measures (such as the Herfindahl index) could be very misleading indicators of market power.
Introduction: In the banking literature, until recently, the two most popular methodologies to test market power were the Panzar-Rosse (1987) test which infers conduct from the individual firm’s input-output cost relationships, and Bresnahan’s (1982) conjectural variation (CV) model which focuses on market structure parameters. Degryse and Ongena (2005) summarizes the advantages of the two methods. The Panzar-Rosse test’s data requirement is low and and the data are readily available across different countries while the conjectural variation model nicely embeds different types of competitive behaviour. However, Hyde and Perloff (1995) finds, that the Panzar-Rosse test is very sensitive to the specification of the reduced-form revenue function and to which input factors of production are included. Corts (1999) and Nevo (1998) show that the CV methodology has problems related to the interpretation and identification of the theoretical conduct parameter.
Author: József Molnár
Source: Research Discussion Papers, Bank of Finland
Download URL 2: Visit Now