Comment on "Real Effects of Bank Competition"

Cetorelli (2004, this issue of JMCB) examines the relationship between bank concentration and (nonfinancial) industrial structure. This paper forms part of an interesting research agenda on a timely and important topic. It sits at the intersection of two literatures, one on changes in the structure of the banking industry and the other one on the effect of financial system design on economic growth. With the substantial changes in banking across many countries and the rapid changes in the financial systems of many emerging economies, discovering the extent of spillovers, if any, from changes in banking on the real economy can be significant. While the focus of this paper is Europe, the results can shed light on other areas as well.

There has been a rapid consolidation in banking throughout the world during the past few decades, especially during the 1990s (report of Group of Ten 2001). During the ten-year period 1990-99, there were over 2900 bank mergers in the U.S., with almost one in four banks a target, while in Japan, there were 108 mergers among large banks (49 in 1999 alone), or two mergers for every three banks. In Europe, the focus of this paper, there were over 1200 mergers, which is one for every six banks.

The consolidation has led to public policy concerns including the effect of a changed banking structure on borrowers. Much of the debate has focused on the quantity of small business lending (also know as small and medium enterprise, or SME, lending) and the effect of consolidation on small business loan interest rates (see, e.g., Berger, Rosen, and Udell 2003). The basic intuition in many papers is that being larger may reduce the ability or the incentives to establish relationships (see, e.g., Stein 2002). By extension, since relationships are most important for small businesses, as banks become larger, small business lending may fall (although this is not consistent with the findings in Berger, Rosen, and Udell 2003).

Cetorelli points out that the effect on borrowers may not be limited to small firms. By reducing competition among banks, consolidation influences all lending decisions by banks. As banking becomes more concentrated, banks may prefer to lend to large, established firms relative to smaller, start-up firms. Thus, changes in banking concentration can affect industry concentration in other fields. This has direct and indirect impacts on economic growth. The findings of a study like this can offer input on decisions of optimal financial structure.

Cetorelli argues that Europe is an excellent laboratory to examine the effect of bank consolidation because the European Union (EU) undertook a large deregulation during the 1990s. The Second Banking Coordination Directive allowed banks to branch freely across the EU, thus lowering barriers to entry. In theory, this should increase competition for EU banks. Thus, we have an opportunity to test the effect of changes in competition in banking on the real economy.

The basic approach in the paper is to look at how average firm size in an industry is related to interaction terms involving bank concentration, bank deregulation, the importance of the bank lending to the industry, and a dummy for EU membership as well as some control variables. In what follows, I briefly give the results in the paper as well as Cetorelli's interpretation of the results. I then discuss some alternative hypotheses and tests to distinguish the alternatives from the interpretations in the paper.

The first set of tests examines how market concentration in an industry is affected by bank concentration and the need for borrowed funds (external dependence) in that industry, both within the EU and outside it. The basic model tested is:

It may seem that bank concentration is missing from the right-hand side of the regression. However, Cetorelli uses a fixed effects approach that implicitly captures the impact of bank concentration that is independent of its effect on the interaction terms. Also note that in Equation (1), bank concentration is measured for the same year as average firm size. I return to...