2.1 Data source
We identify all Japanese bank acquisitions between 1993 and 2008 reported in the Securities Data Corporation (SDC) M&A database. Twenty-four Japanese commercial banks engaged M&A from 1993 to 2008. The accompanying individual accounting data for each of the merged companies were taken from Compustat. Referring to Table 1, there are 114 commercial banks in Japan, but only 96 are available in Compustat.
The sample M&A announced between 1993 and 2008 are majority acquisitions, which resulted in the acquirer having a stake of at least 50% in a target institution.
To be included in the sample, the acquirers must have been involved in only one merger or acquisition activity. Four banks involved more than one merger or acquisition between 1993 and 2008 were deleted. Thus, our effective sample total comprises 92 commercial banks in Japan and 20 banks that participated in one merger or acquisition during the data period, all of which are domestic M&A.
2.2 Variables
Univariate analysis and regression model are applied to investigate how M&A affect bank performance. The three different bank performance measures in our model are return on assets (ROA), Tobin‟s Q, and Costs/Assets. All variables are defined and shown in Table 3.
Table 3 variable definition
This table presents descriptions of variables used in this study. The sample period is from 1993 to 2008. All the financial data computed for the variables are from Compustat (GV). The market premium comes from Securities Data Corporation (SDC).
Dimension Variable Definition
bank performance measures
ROA ROA return on asset
Tobin's Q Q (the book value of total assets minus the book value of equity plus the market value of equity) to the book value of total assets
Costs/Assets Costs/Assets total expenses to total assets
M&A indicator
M&A dummy Dummy indicating the years following a bank's M&A. Equals 0 before the bank's M&A and 1 following the M&A.post-merger 13 Dummy variable that equal 1 from first to third year after M&A
post-merger 3 Dummy variable that equal of 1 in all years after the third
bank attributes
Capitalization E/A Total capital to total assets
Credit risk BADL/NII Loan loss provisions to net interest revenues Loan activity LOAN/DEP Customer loans to customer deposits
Fee-based activity NINT Non-interest income to total assets Technology and
innovation
OE/A Other expenses to total assets
Size
Size natural log of total assets in t-1 for each bankMarket share
Market share market share in t-1 for each bankWe employ a regression method to determine how M&A affect bank performance. We construct a dummy variable (M&A dummy) that equals to 0 before M&A of banks, and equals to 1 following M&A. The dummy is equal to 0 for all period for banks that did not undergo M&A. To distinguish short-term effect and long-term effect of M&A, we construct
post-merger 3 shows the long-term effects of mergers and acquisitions.
We also use a variety of bank attributes to define the features of banks. These indicators include measures of capital structure, risk exposure, type of activities, and financial innovation. Among the explanatory variables, size and market share are included because these variables are expected to be important determinants of bank performance. The summary of statistics is reported in Table 4.
Table 4 descriptive statistics
assets. We calculate Tobin‟s Q (hereafter, Q) as the market value of equity plus the book value of debt (computed as the book value of assets minus the book value of equity) divided by the book value of total assets. This definition of Q has been used in various studies (La Porta et al., 2002; Doidge, Karolyi, and Stulz, 2004).We use ROA and Q to measure profit performance and Costs/Assets to measure cost performance. The variable Costs/Assets is total interest plus noninterest expenses divided by assets. Both interest and noninterest expenses are included because bank management may substitute between providing depositor services and interest payments in attracting funds and because both contribute equally to the goals of the organization (Berger et al., 2005).
The capital adequacy level is measured as the ratio of equity to total assets (E/A).
Practitioners, analysts, and regulators have attached great importance to this variable in recent years. From a prudential regulatory perspective, bank capital has become the focal point of
bank regulation (Vives, 2000).
The effect of changes in the capital adequacy level on performance depends on the theory of the banking firm. According to the “signaling hypothesis,” commercial banks specialize in lending information to problematic borrowers (Berger, Herring, and Szego, 1995). Because bank managers usually have a stake in the capital of the bank, “it will prove less costly for a
„good‟ bank to signal better quality through increased capital than for a „bad‟ bank.”1 Therefore, banks can signal favorable information by merging with banks with larger capital ratios, indicating a positive correlation between capital and earnings, and suggesting a positive relationship between capital structure dissimilarities and performance (Acharya, 1988). Conversely, Ross (1977) argues that lower, rather than higher, capital ratios signal positive information because signaling good quality through high leverage would be less onerous for a “good” bank than for a “bad” bank.2
Credit risk strategy is measured as the level of loan loss provisions divided by net interest revenues (BADL/NII). Regarding loan and deposit profiles of banks, the ratio of total loans to total customer deposits (LOAN/DEP) is included because it provides a proxy for the use of relatively low-cost deposits in relation to the amount of outstanding loans (Altunbas and Marques, 2008).
Traditionally, banks rely on interest income as their primary revenue source. Today, rather than traditional core deposit business, banks have other activities that generate non-interest income, This non-interest income includes fees from the sale of mutual funds and insurance policies, fees from securitization activities, income from loan servicing, fees from providing trust services, and income from providing cash management services. Thus, the ratio of non-interest income to total operating income (NINT) indicates the importance of fee-generating activities versus more traditional lending (DeYoung and Rice, 2004).
Finally, the strategy of banks in terms of technology and innovation is measured as other costs (i.e., total costs excluding interest, staff, and other overhead costs) to total assets (OE/A) (Altunbas and Marques, 2008).
We include the natural log of lagged assets and the market share to account for differences in bank size and market power.
2.3 Empirical models
First, we use univariate analysis (t-test) to compare pre-merger and post-merger bank performance variables.
We take the pre-merger average of the bank over the five years prior M&A. If the pre-merger data available is for less than five years, we take the pre-merger average over the maximum years for which data are available. Focarelli and Panetta (2003), Focarelli and Pozzolo (2005), and Rhoades (1998) show that a 2–3 year post-merger period is needed to determine whether there are any post-merger gains. We take the average of the post-merger bank performance measures over the (at most) five years after M&A and take the difference between the pre-merger 5-year average and the post-merger 5-year average.
We perform the t-test for the null hypothesis, where the differences between pre-merger and post-merger bank performance measures have a mean equal to zero.
We then run two stage regressions to examine how M&A influence bank performance. The bank performance measures specified as the dependent variables are ROA, Tobin‟s Q, and Costs/Assets.
The regression model includes only size and market share in first stage, which is shown in Equation (1). In second stage, we add bank attributes to regression model. We use bank attributes to represent bank characteristics, including capital structure, risk exposure, type of activities, and financial innovation. Bank attributes include E/A, BADL/NII, LOAN/DEP, NINT, and OE/A, which are defined in Table 3. Equation (2) is a complete regression model, which includes bank attributes, size, and market share.
Finally, 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