The effects of mergers on bank efficiency: stochastic risk-return frontier analysis

Abstract

This dissertation is motivated by the need to evaluate the impact of mergers and acquisitions on bank efficiency. Over the past two decades, significant research effort has been devoted into measuring the efficiency of financial institutions, especially commercial banks. Even though many studies found that measured inefficiencies in these financial institutions were high (at least 20% of the total banking industry cost and about half of the industry's potential profits), there is still no consensus about the sources of the differences in measured efficiencies. Differences might be due to various efficiency concepts employed, and different measurement methods used to estimate efficiency within each concept. There is also not an agreed theory as to the potential efficiency correlates. This dissertation assesses bank efficiency under the following considerations: alternative profit efficiency concept, stochastic frontier analysis, flexible translog functional form, generalized normal-truncated normal distribution, and explicit model for inefficiency effects. Chapter 1 provides background of the study, statement of the problem and hypotheses to be tested. This chapter also outlines the mergers and acquisitions trend that significantly contributed towards the reduction of the number of banks in the U.S. from a maximum of 14,496 banks during 1984 to 7,887 banks during 2002. These mergers and acquisitions were propelled by the changing market conditions, regulatory changes, and globalization of financial institutions across the U.S. borders. Chapter 2 put this study into context in terms of the previous literature on bank efficiency. Financial ratio studies concluded that mergers were not associated with significant improvements in cost ratio or any other measure of firm performance. Of the X-efficiency literature, studies that examined merger benefits from the cost or input side also found no benefits on average from mergers. These studies led to several implications that could pave way for future research: first, studies on the implications of bank mergers should move away from simple financial ratios analysis. Second, studies should not only determine factors that predict merger efficiency gains or losses, but should also determine factors that explain differences in measured efficiencies among banks. Third, studies should employ the profit function rather than the cost function in all merger efficiency analysis. Last, following Shaffer (1993) and Berger and Humphrey (1992a) assertion, studies should analyze if merger efficiency effects are time-dependent. Chapter 3 not only presents the theoretical framework underlying the efficiency studies but also present the analytical foundation of producer theory and efficiency measurement. This chapter explicitly presents a stochastic frontier framework adopted in the study and two distributional assumptions (normal-half normal and normal-truncated normal distributions). The normal-truncated normal distribution, due to its flexibility and its accommodation of a wider range of distributional shapes, which includes ones with nonzero modes, is an adequate representation of the data in this study rather than the normal-half normal distribution adopted by other bank efficiency studies. Chapter 4 presents the specification of the profit frontier and the inefficiency effects models. This chapter presents the alternative profit efficiency concept and its specification. The alternative profit efficiency concept bridges the gap between the standard cost and profit functions, and is more relevant if the assumptions underlying the cost efficiency and the standard profit efficiency do not hold. This chapter also presents an explicit model for inefficiency effects. Most empirical studies investigated the determinants of technical inefficiency among firms in the banking industry by regressing the predicted inefficiency effects observed from an estimated frontier upon a vector of firm-specific factors in a second-stage analysis. This study outlines the shortcomings of the two-stage analysis thereby presenting a single-stage maximum likelihood technique that simultaneously estimate the frontier function and the inefficiency effects model. We also model the inefficiency error component, ξi, as independently distributed such that ξi is obtained by truncation (at zero) of the normal distribution with mean Ziδ and variance σ2 rather than the normal-half normal distribution adopted by other previous bank efficiency studies. Chapter 5 presents the empirical results of the study. After estimating the profit function, we found that the inefficiency component should be incorporated into the profit frontier. That is, the auxiliary equation (inefficiency effects model) plays an important role in the estimation of the profit function. This also suggests that we should concentrate on maximum likelihood estimation. We found that the flexible translog is a better representation of the data than the Cobb-Douglas form. In addition, the normal-truncated normal distribution is an adequate representation of the data than the normal-half normal distribution. The measured efficiencies were higher compared to inefficiencies summarized in Berger and Humphrey (1997). We also found that the inefficiency effects variables maintain their explanatory power on average when measuring inefficiency with respect to both risk and return. Also, banks that increased their risk-adjusted performance improved their alternative profit thereby becoming more technically efficient. The diversification hypothesis also holds. Chapter 6 presents the summary and conclusion of the study. This study indicated that mergers and acquisitions that take advantage of the diversification opportunities after a merger improve their efficiency. However, we take into consideration that this was a cross-sectional study. Therefore, future studies along these lines should employ a panel data to test the robustness of our results and methodology.