Discussion of “Bank Mergers and the Dynamics of Deposit Rates”

Abstract

Listen

Translate article

Craig and Dinger (2009) tackle the question of whether bank mergers lead to increased or decreased rates paid to customers. Two theories lead to conflicting predictions. The first theory is that mergers lead to increased market power and therefore decreases in rates paid to customers. The other theory is that mergers lead to economies and hence to higher rates paid to customers. Both theories have found empirical support in the past when tested using deposit rates. Craig and Dinger bring to this puzzle a new, more complete, high frequency data set and a new empirical methodology. They test both deposit rates and money market fund rates. Their empirical methodology accounts for the censoring of deposit and money market fund rates which rarely change. As a result of this stickiness simple OLS, as has been used in the past, is biased and may lead to incorrect inferences. Craig and Dinger show this clearly by presenting OLS and censoring-adjusted results. Craig and Dinger find that mergers lead to decreased deposit rates after incorporating their adjustment for data censoring. Their money market fund rate analysis is inconclusive. The recognition that the left hand side variables (rates paid) are censored because they rarely change is important. It may however diminish the usefulness of high frequency (monthly) data. In their data set deposit rate changes are only observed in ten percent of the observations. On the other hand many of the explanatory RHS variables, such as size, income, deposits to asset ratio, are only observed quarterly. Thus, in monthly regressions some of the RHS variables are repeated (i.e., censored). It might be cleaner to simply use quarterly observations, eliminating the RHS censoring problem, reducing the LHS censoring, and then employing the same methodology to adjust for the remaining censoring on the LHS variables. The authors earlier noted that the inconsistency in prior empirical results seemed to be related to the horizons examined in previous studies. Shorter term (following the merger) studies such as Berger and Hannan (1989) and Hannan and Prager (1998) found rates declined, while longer term studies such as Focarellia and Panetta (2003) found the opposite. Craig and Dinger use a simple piecewise linear spline to provide insight into the J Financ Serv Res (2009) 36:135–136 DOI 10.1007/s10693-009-0070-y