臺大管理論叢 NTU Management Review VOL.29 NO.1

267 NTU Management Review Vol. 29 No. 1 Apr. 2019 3.3 Main Regression Specification We used the following regression specification, modified from Loutskina (2011), to test the relation between bank liquidity and CEO incentives. Bank Liquidity i,t = β 0 + β 1 CEO Incentives i,t -1 + β 2 Loan Commitment i,t -1 + β 3 Deposit Base i,t -1 + β 4 (Loan Commitment i,t -1 × Deposit Base i,t -1 ) + β 5 Bank Capital i,t -1 + β 6 Share of Deposit Financing + β 7 Business Cycle i,t -1 + μ i,t -1 (1) All regressions included firm and year dummies as control variables, and standard errors were adjusted for heteroskedasticity and potential clustering of the residual at the firm level. 4. Empirical Results 4.1 Sample Statistics Table 1 presents the summary statistics of all research variables during the sample period. Because the research variables in this study were compiled from a combination of different datasets and hand collected data, the number of observations for each variable varied with data availability. To maximize the sample size and validate the empirical results, we required that the sample banks have data for CEO compensation, liquidity ratio, loan commitment and deposit available, and they were then included in the further empirical analyses. 8 The mean (median) liquidity ratio and the ratio of transactions deposits to the total deposits (Deposit Base) of sample banks was 14.70% (12.81%) and 7.31% (5.43%), respectively. Both numbers were smaller than the ones reported in Kashyap et al. (2002). The divergence between these two studies may be due to different research periods. Commercial banks were allowed to extend into investment banking and other financial services businesses after 2000. The operational flexibilities of the banking business may have resulted in decreased liquidity ratio and changed the revenue breakdown after the 2000s. The mean (median) loan commitment ratio of the sample banks was 14.63% (13.74%), quite similar to the one reported in Kashyap et al. (2002). 8 Some sample firms disclosed the aggregate amount of CEO compensation, but they did not report details of each component. Therefore, the number of observations in each component of CEO incentives was less than the number of observation for firms with available CEO compensation.