3.1 Univariate analysis
Table 5 shows the results of t-test, which was used to compare pre-merger and post-merger bank performance. Results show that both the mean value of difference between pre-merger and post-merger ROA and Tobin‟s Q are not significantly different from zero. This means that Japanese banks cannot use M&A to gain profit, which is consistent with prior studies (Hagendorff and Keasey, 2009; Beccalli, 2009; Piloff, 1996).
We also examine whether M&A can improve cost efficiency, and find that the mean value of difference between pre-merger and post-merger Costs/Assets is significantly smaller than zero (-0.0113). This shows that the M&A of the bank can create cost synergy, which is also consistent with previous literature (Kwan and Wilcox, 2002; Huizinga et al., 2001; Beccalli, 2009).
Table 5 Paired t-test of bank performance measures
The data consists of 20 Japanese commercial banks that have M&A activity. We run paired t-test and take pre-merger and post-merges average over five year bank performance measures. If the pre-merger and post-merger data are available for less than five years, we take the average over the maximum years for which we can observe the data. Bank performance measures include ROA, Tobin‟s Q, and Costs/Assets. ROA is income before extraordinary items divided by total assets. The Tobin‟s Q is (Total assets - Book value of equity + Market value of equity) divided by book value of total assets. Costs/Assets is total interest plus noninterest expenses divided by assets. *, **, *** indicate significance at 10%, 5%, and 1% levels, respectively.
ROA Tobin's Q Costs/Assets results are shown in Table 6. In first stage, the regression model includes only size and market share. Then, we add bank attributes to regression model in second stage. We use bank
Table 6 Regression on bank performance measures
This table reports regressions results for Equation (1) and Equation (2). The dependent variables are bank performance measures. Bank performance measures include ROA, Tobin‟s Q, and Costs/Assets. ROA is income before extraordinary items divided by total assets. The Tobin‟s Q is (Total assets - Book value of equity + Market value of equity) divided by book value of total assets. Costs/Assets is total interest plus noninterest expenses divided by assets. M&A dummy is a dummy variable that equals 0 before the bank‟s M&A and 1 following the M&A. The dummy equals 0 for all period for banks that did not undergo M&A. The bank attributes include E/A, BADL/NII, LOAN/DEP, NINT, and OE/A. E/A is capital ratio of equity to total assets. BADL/NII is credit risk measured by loan loss provisions divided by net interest income. LOAN/DEP is the ratio of total loans to total customer deposits. NINT is the ratio of non-interest income to total operating income. OE/A is other costs (i.e., total costs excluding interest, staff and other overhead payments) to total assets. Size is natural log of bank‟s total assets in t-1. Market share is natural log of bank‟s market share in t-1. *, **, *** indicate significance at 10%, 5%, and 1% levels, respectively. The values of t-statistics are in parentheses.
Model excluding bank attributes Model including bank attributes
ROA Tobin's Q Costs/Assets ROA Tobin's Q Costs/Assets
However, we can see both Q in two models decreases significantly after M&A. We find that Japanese bank‟s profit performance may remain the same or even get worse after M&A.
Besides, results also show that Costs/Assets decrease significantly after M&A. We find mergers and acquisitions can improve Japanese bank‟s cost performance. Therefore, we conclude that Japanese banks cannot improve their profit performance through M&A but can
create cost synergy. This finding is consistent with previous literature.
Then we run regression analysis to identify short-term and long-term effect on bank M&A.
The post-merger 13 exhibits short-term effect, which measures the adjustments made during the transition. And the post-merger 3 shows the long-term effects of mergers and acquisitions.
The regression results are shown in Table 7.
Results show that the coefficients on post-merger 13 and post-merger 3 are insignificant on ROA, indicating ROA does not significantly change both in short-term and long-term period after M&A. However, the coefficients on post-merger 13 and post-merger 3 are negative significantly on Q. We find that mergers and acquisitions have not only short-term effect, which is the adjustments made during the transition, but also long-term effect on Q. From aspects of ROA and Q, we find that mergers and acquisitions cannot improve Japanese bank profit performance or even make it worse both in short term and in long term.
Results also show that the coefficients on post-merger 13 and post-merger 3 are negative significantly on Costs/Assets. We find that mergers and acquisitions have short-term cost cutting effect. At the same time, we expand bank‟s size and have long-term cost reduction effect through M&A; thus, achieve economic scale.
In addition, results show that a significantly positive relation exists between capital ratio (E/A) and bank profit performance. It shows that banks with high capital ratio have better performance, which is consistent with “signaling hypothesis.”3 We also find that banks with higher credit risk (BADL/NII) have better performance (higher ROA and Q; lower Costs//Assets). And banks with higher loan activity (LOAN/DEP) have worse performance (lower ROA and Q; higher Costs//Assets). Besides, banks with a more fee-based activity (NINT) have positive effect on bank performance (for Q). The finding is consistent with a previous study that states that increases in non-interest income have been associated with higher profits (DeYoung and Rice, 2004). The coefficients on technology and innovation (OE/A) have mixed performance results.
Smaller banks, as measured by natural log of lagged assets, are generally associated with better performance than larger banks (statistically significantly lower ROA). This negative association is consistent with the results of EU bank efficiency studies.4 In contrast, banks
and lower Costs/Assets).
Table 7 Regression on bank performance measures
This table reports regressions results for Equation (3). The dependent variables are bank performance measures.
Bank performance measures include ROA, Tobin‟s Q, and Costs/Assets. ROA is income before extraordinary items divided by total assets. The Tobin‟s Q is (Total assets - Book value of equity + Market value of equity) divided by book value of total assets. Costs/Assets is total interest plus noninterest expenses divided by assets.
Post-merger 13 is a dummy variable that equals to 1 from first to third year after M&A and post-merger 3 is also a dummy variable that equals 1 in all years after the third year, otherwise the two dummy variables are equal to 0. The two dummy variables equal 0 for all period for banks that did not undergo M&A. The bank attributes include E/A, BADL/NII, LOAN/DEP, NINT, and OE/A. E/A is capital ratio of equity to total assets. BADL/NII is credit risk measured by loan loss provisions divided by net interest income. LOAN/DEP is the ratio of total loans to total customer deposits. NINT is the ratio of non-interest income to total operating income. OE/A is other costs (i.e., total costs excluding interest, staff and other overhead payments) to total assets. Size is natural log of bank‟s total assets in t-1. Market share is natural log of bank‟s market share in t-1. *, **, *** indicate significance at 10%, 5%, and 1% levels, respectively. The values of t-statistics are in parentheses.
ROA Tobin's Q Costs/Assets