FGLS Regression Model

There are two set of regression equations are tested in the study. One is to capture the major effects of profit performance of the acquiring banks prior and post

5 “while capturing the return to owners is an attractive feature of this measure, a disadvantage … .”

mergers and acquisitions. The main performance measurement is the change of ROA. The second test is using change of expense ratio (ratio of non- interest expenses to assets) to determinate major effects which may improve efficiency performance of acquiring banks before and after mergers and acquisitions.

The first model is to test the determinants to improve the profitability. The regression model is used to incorporate variables reflecting the extent of operating figures and a number of control variables. The estimation of FGLS regression model is in the following form:

The dependent variable is calculated as the change in ROA for one-year before and one- year after, one-year before to three-year after, and one-year before to five-year after the mergers and acquisitions.

In order to test major indicators to improve the profitability performance through mergers and acquisitions, three independent variables identified in hypothesis 3 are used. The increase in net interest margin (NII/TA) is expected to improve the profitability before and after mergers and acquisitions. The second variable is non- interest revenue to assets (NIR/TA). The non-interest revenue to assets is to test how existed assets to generate revenues other than interest income. The increase in this variable is predicted to improve the profitability through action of mergers and acquisitions. The variable of non- interest expenses to total assets (NIE/TA) presents the expenses other than interest paid. The reduction of this expense ratio is likely to enhance the ROA of acquiring banks before and after deals. These three

independent variables are expressed as change of each variable from one-year before

to one-year after, three-year after, and five-year after the mergers and acquisitions.

The independent variables are included to control major factors likely to influence bank performance. A firm size variable (SIZE), use the log of the total assets, at the year of mergers and acquisitions is included to account for possible efficiencies associated with size, and to control for the certain size of the bank at the year of mergers and acquisition. The reason to take log of the total assets is that there are huge differences of dollar value of the total assets among different samples.

Most of studies used the firm size one- year prior mergers and acquisition as the control variable. There are few reasons not to use it in this study. First, analysis of most studies focuses on the combined data both before and after mergers and

acquisition. Since this study focuses on acquiring banks and only obtains the data of acquiring companies, the figures would not be accurate to use total assets one-year before deal. Second, since many banks have continued some other deals of mergers and acquisitions after observed year, it would be difficult to separate the consolidated financial data of acquiring firms of each deal. Therefore, the firm size, book value of total asset, of the observed year is used as a control variable. The dummy variable of the area (AREA) acquiring banks located is another control variable. This

variable accounts for the fact that each areas or continents may have special culture, economic environment, and government regulations, and all these factors may

influence the performance of the bank. The number of targets the acquirer purchased in the observed year (TARGETS) is also included.

The second model is to test the determinants to improve the efficiency. The regression model is used to incorporate variables generating non-interest expense figures and a number of control variables. The estimation of FGLS regression model

+

The dependent variable is calculated as the change in expenses ratio, ratio of non- interest expenses to assets (NIE/TA), for one-year before and one-year after, one-year before to three- year after, and one-year before to five- year after the mergers and acquisitions.

Three possible determinants mentioned in hypothesis 4 are tested through regression model. First one is administrative expense ratio (ADM/TA). The reduction of administrative expenses ratio is estimated to improve the efficiency level of acquiring banks after mergers and acquisitions. The ratio of loans to assets (LOAN/TA) is another variable used. A sign on this variable is not predicted since level of loan activity may present the aggressiveness of the banks to increase return, at the same time, the expenses may increase to maintain the portfolio. The last variable is non-interest revenue ratio. The increase in this ratio is expected to enhance the efficiency performance. These three independent variables are expressed as change of each variable from one- year before to one-year after, three- year after, and five-year after the mergers and acquisitions. The same control independent variables as equation (1), the firm size (SIZE), the area the acquiring banks located (AREA), and the number of targets acquirer purchased at the year of mergers and acquisitions (TARGETS) are used to control the major factors which may influence the bank performance. Table 4 presents detail definitions of variables used in regression.

Most of independent variables used in both equation (1) and (2) are ratio relative to assets rather than the absolute value. This is because the change in absolute value of each variable may not reflect the real situations. The fluctuation of total assets or

other variables may directly influence the administrative costs, net interest income, or other financial data. For instance, the reduction of administrative cost may not indicate the efficiency improvement since some branches may be sold out to cause the reduction of total assets. Also, since the data obtained in this study are consolidated financial data after mergers and acquisitions and financial data of acquiring firms only before mergers and acquisitions, comparison of each variable by absolute value may mislead the result. In order to compare the financial data of acquiring banks before and after mergers and acquisitions at the same standard with more accurate result, the ratio is used in this study.

Table 4 Variable Definitions

Variables Definitions

ROAt Return on assets, net income/total assets, change from one year prior to t year post mergers and acquisitions

TAt

NIE / Non-interest expenses over total assets, change from one year prior to t year post mergers and acquisitions

TAt

NII / Net interest income over total assets, change from one year prior to t year post mergers and acquisitions

TAt

NIR / Non-interest revenue over total assets, change from one year prior to t year post mergers and acquisitions

TAt

ADM / Administrative costs over total assets, change from one year prior to t year post mergers and acquisitions

TAt

LOAN / Loans to total assets, change from one year prior to t year post mergers and acquisitions

SIZE Take log of total assets at the year of mergers and acquisitions

AREA Dummy variable indicates the continent the acquiring banks located

TARGETS Dummy variable indicates numbers of targets purchased by acquirer at the year of mergers and acquisitions

V、Empirical Result