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

265 NTU Management Review Vol. 29 No. 1 Apr. 2019 banks (424 banks). 3 We collected stock return data from the Center for Research in Security Prices (CRSP). 4 Data on interest rates were available daily from the Federal Reserve Board of Governors. To measure CEO incentives, we relied on the data collected from the ExecuComp database, which provides information about CEO compensation in terms of salary, bonus, and stock options granted since 1992. Because the ExecuComp database only tracks executive compensation in S&P 1,500 firms, we hand collect all compensation data regarding the measures of CEO incentives for non-S&P 1,500 banks in this study. 5 After all data was compiled, our sample contained 1,674 observations (209 banks) that include 433 observations (35 banks) of S&P 1,500 banks (denoted as S&P 1,500 banks) and 1,241 observations (174 banks) of non-S&P 1,500 banks (denoted as non-S&P 1,500 banks). 6 3.2 Variables 3.2.1 Variables of Interest 3.2.1.1 Liquidity Ratio The dependent variable was bank liquidity ratio, which was measured by liquid assets divided by total assets. Specifically, bank liquidity asset was the sum of Fed funds sold and securities purchased under agreements to resell, securities held to maturity, and securities available for sale. 3.2.1.2 CEO Incentives To test the impact of bank CEO incentives on the bank liquidity management, we followed previous studies such as Jiang (2009) and Fahlenbrach and Stulz (2011), and measured CEO incentives in five different categories: (1) CEO short-term incentive was 3 The ratios of liquidity, transaction deposits, and unused loan commitments were calculated using the data from the Call Report, and the sample size decreased dramatically once we combined them with the other research variables. This was because bank identification numbers used in the two different datasets are not identical, which made the data compilation difficult and reduced the sample during this process. Even after the individual bank information was checked manually, the situation remained unchanged. 4 We have 7,668 CRSP observations (1,022 banks), including 852 observations of S&P 1,500 banks (52 banks) and 6,816 observations of non-S&P 1,500 banks (970 banks) from years 1992 to 2012. 5 We have total 4,861 compensation observations (536 banks), including 1,045 observations of S&P1,500 banks (79 banks) and 3,816 observations of non-S&P1,500 banks (457 banks) from years 1992 to 2012. 6 Not every sample bank had sufficient data for analysis each year. Therefore, the number of observations for each bank in different years varied during the study period.