University of Amsterdam, Amsterdam Business School
MSc Finance, Corporate Finance
Master Thesis
Female CEO compensation, risk preference and firm performance
Jie Xu
07-2018
Statement of Originality
This document is written by Jie Xu, who declares to take full responsibility for the contents of this document.
ACKNOWLEDGEMNTS
I would like to offer my profound gratitude to my supervisor, Dr. Tomislav Ladika, for providing me with continuous support and guidance throughout the master program with his expertise and patience. Also, I am indebted to all the other professors, tutors and my friends for their help and encouragement.
ABSTRACT
The surge in the number of female CEOs has aroused public attention during the recent decades, and compensation gap between female CEOs and male CEOs is one of the most discussed. In this study, I adopt the latest data retrieved from Compustat and CRSP database to study the changes of female CEOs’ compensation, and the relationship among compensation, risk preference and firm performance. The results are based on US firms for the period of 2007-2016. I find that female CEOs are on average compensated with higher level of compensation and lower level of equity-based incentives. These findings are associated with female CEOs’ risk preferences and firm performances. I employ two different propensity score matching approaches and sub sample test to reconfirm that firm led by female CEOs do not perform well in the long run and female CEOs tend to be risk averse, holding more cashing compared to their counterparts. These findings might be the reasons that female CEOs receive less equity-based incentives.
CONTENTS
ABSTRACT ... 4
CONTENTS ... 5
LIST OF TABLES ... 6
1. Introduction ... 7
2. Literature review and hypotheses ... 9
2.1 Changes in female compensation ... 9
2.2 Gender differences in risk preferences ... 10
2.3 Gender differences in firm performance ... 11
3. Data and research methodology ... 13
3.1 Data collection ... 13
3.2 CEO compensation ... 13
3.3 Risk prefernce ... 14
3.4 Firm performance ... 15
4. Empirical analysis and results ... 17
4.1 Descriptive statistic ... 17
4.2 Logistic regression results ... 19
4.2.1 Compensation ... 19
4.2.2 Cash holding ... 22
4.2.3 Firm performance ... 25
4.3 Robustness check ... 25
5. Conclusions and discussions ... 27
References ... 29
Appendix A: Variable definitions ... 31
LIST OF TABLES
Table 1 Descriptive statistics of the number in a given year………. 15
Table 2 Descriptive statistics of CEO characteristics………..17
Table 3 Gender differences in compensation ………..19
Table 4 Gender differences in cash holding ……….21
Table 5 Relation between compensation and firm performance………22
Table 6 Robustness test results for equation (1)………...31
Table 7 Robustness test results for equation (2) ..………33
1. Introduction
Female CEOs have been aroused public attention during the recent decades due to the momentous reform in the structure of corporate governance. The number of female CEOs, especially in large institutions, indeed presents a rapid growth, however, the top management is still dominate by male CEOs. According to CNBC news, there are only 32 female CEOs in the Fortune 500 companies by the end of fiscal year of 2016, which reached an all-time high by then. Besides the superior number, male CEOs also take the advantage in compensation compared to their counterparts.
Intensive literature identifies the inseparable connection between compensation and firm performance. It is pivotal for a firm to keep their profits level stable or even upgrading so that it could benefit their stakeholders and shareholders in both short-term and long-term. To evaluate firm’s profitability, one of essential features is firm performance, in which CEOs plays an indispensable and conclusive role. Therefore, firm performance is considered as one of main benchmarks when considering CEOs’ payment structure. Given the dominance of masculine power in higher management positions, female candidates are hampered to climb the corporate ladder up. (Keloharju, Knüpfer and Tåg, 2017)
Besides the incomparable number in male and female CEOs, the contextual issues which would have influence on the compensation, along with the concerns on gender differences, such as age, tenure, education level and marital status, are considered by management. (Mohan and Ruggiero, 2003) Those differences also could be derived from gender-based characteristics, for example, Shery Sandberg, the COO of Facebook, said that women do not take enough risk, men are just “foot on the gas pedal”, which indicates gender difference in risk preference. Numerous studies identify the differences in risk taking behavior by gender. (Cobey et al., 2013) Risk tolerance might one of the factors lying in the gender differences of compensation, which separates the male from female CEOs in terms of financial decision making, such as cash holding, research and development expenses, investment strategies, and merge and acquisition activities. Those behaviors further the impact on the firms’ performance and revenues both in a short-term and in a long-term.
The aim of my research is to shed a light on the changes in female CEOs compensation, also on the relationship among compensation, risk and firm performance. Therefore, I conduct three regressions to analyze these relationships. In specific, I first estimate the interaction between female CEOs and total compensation using equation (1). I substitute total compensation to salary, bonus and equity-based incentive, respectively, to further test how the other three types of payments would be different to male CEOs’ payment. Furthermore, considering cash holding as a proxy of risk level, I employ equation (2) to test the gender differences in risk level. Finally, I analyze how the compensation in prior year would affect the firm performance in a given year. The outcomes are estimated by equation (3).
The results may be biased by endogenous problems. In specific, firms with diversified corporate culture and relatively large size may have bigger chance to include a female into the top management. Similarly, female who are considered as more risk averse than their counterparts could take a position in cash-rich or mature firm. To address this endogeneity concerns and self-selection bias, I employ propensity score matching method, which is also used by Sah (2015), Huang and Kisgen (2013), to control the sample.
The empirical findings in this study are robust to controlling for a variety of firm related and personal characteristics, also with industry and year fixed effects. These results are estimated in multiple robustness tests as well.
The rest of the paper is organized as follows, section 2 literature review and hypotheses, section 3 data and research methodology, section 4 empirical analysis and results, and section 5 conclusions and discussions.
2. Literature review and hypotheses
2.1 Changes in female compensation
One of the reasons that female CEOs has been the focus, is that women are moving rapidly into the managerial ranks, increasing from 15 percent in 1961 to 27 percent in 1981. Nevertheless, the income statistics do not indicate that managerial women are getting ahead (Bartlett and Miller, 1985).
Along with the economic environment changes and employment-population ratio trends, the needs for diversity in the workplace force institutions to include female into top management. (Athanasopoulou, et al., 2017) This transformation indeed modifies the situation in which female are hard to climb up to the senior executive ranks. However, women still face an uphill battle with challenges, which is known as “glass cliff”. (Sabharwal,2013) This phenomenon indicates that women may inherit jobs that risk challenging their leadership capability. (Ryan and Haslam, 2007) In specific, female executives are more likely to get promoted to precarious positions. As the result, female CEOs have shorter tenures compared to their male peers (Glass and Cook, 2014). The surge in female participation in senior executive position does not alleviate the discriminatory issue in compensation payment and organizational gender biases play a major role in limiting the advancement and compensation of female executives.
However, contradict to prior empirical studies which test the existence of compensation gap in private and other sectors, Elkinawy and Stater (2011) suggest that this renumeration difference is expected to be diminishing. Given the fact that female CEOs, on average, tend to achieve higher education than peers, and intensive literature argue that education level raise both work productivity and consequently earnings. (Adams et al.,2007) Therefore, it is not a surprise if female CEOs are compensated with a better payment. Furthermore, according to a study in Sweden, female CEOs are hard to climb into top management without trading off their family time, (Keloharju et al., 2017) in this case, female may be compensated more in terms of hourly payment.
Hypothesis 1: Female CEOs receive higher level of total compensation than male CEOs
do.
2.2 Gender differences in risk preferences
Risk taking is expected to be pivotal in a business context as multiple financial decisions involve this concept. (Perryman, Fernando and Tripathy, 2016) Bowman (1980) mentions in his earlier research that the relationship building on risk preference and firm performance could be the result of management structure and also decisions on financial strategies. Pervasive literature indicate that gender differences are related to risk preferences and attitudes, and those studies provide consistent evidence that men are more likely to make risky decisions and have a higher level of risk tolerance than women are. Croson and Gneezy (2009) document fundamental differences between men and women that women are more risk averse than men. Those differences include the number of merge and acquisition activities, leverage level, and research and development expenses. Consequently, those financial strategies would affect firm values, then further influence CEOs’ payment.
Risk tolerance, which represents the level of individual’s willingness to take risk in order to get profits in return, is used as a fundamental dimension to better understand individual difference in risk preferences. (Bucciol and Zarri, 2015) To be specific, financial risk tolerance is considered as a measurement of an attitudinal input into the financial decision-making process. (Gilliam et al., 2010) This element matters and is economically important, especially for the people sitting in the financial decision-making positions, like CEO: if CEO is risk-seeking, or in other works, has high level of risk tolerance, then he or she is more likely to invest in risky investment, and even willing to keep investing on projects with nonpositive net present value. On the other hand, if the CEO is risk-averse, then he or she will abandon the investment opportunities even with positive net present values. Those financial decisions are concerned by both shareholders and stakeholders. From shareholder’s point of view, they put more attention on the outcome of the financial decisions which resulting in the change of their holding shares’ value, even with the costs of stakeholders’ interest. Therefore, they may prefer to back the executive with aggressive ambitious without or less considering the level of risk. Though stakeholders may concern the long-term growth
opportunities and care more about the risk, they have limited voice in terms of nomination and designing the compensation for CEOs. Consequently, female CEOs’ risk averse behavior would have an impact on the compensation packages.
Prior literature (Kuan, Li, and Liu, 2012, Adhikari, 2018) link the firm risk with cash holding, since the motivation of increasing cash holding is for precautionary saving, which is considered as a riskiness financial decision. Therefore, the cash holding measures the risk level as proxy in this study to identify female riskiness. Hypothesis 2 is as followed:
Hypothesis 2: The firm which led by female CEO has higher level of cash holdings than
the one led by male CEO.
2.3 Gender differences in firm performance
Including a female into senior executive positions constantly arouses concerns since it has respect to the interest of both shareholders and stakeholders in terms of firm performance. The inseparable relation between payment and firm performance motivates researchers to test the relation between CEOs’ characteristics and firm value. Adam, Hermalin and Weisbach (2010) proposed the endogenous nature of this relation that it is difficult to distinguish if financial decisions that made by executives increase firm value or highly valued firms simply tend to have investment preferences. Therefore, Ahem and Dittmar (2012) present new evidence on the relationship between firm value and the top management structure by exploiting a nature experiment in which the certain number of female must represent on the board of directors. In board structure, it creates an unprecedented exogenous change to corporate boards, and it shows that this compulsory requirement leads to a substantial decline in Tobin’s Q, which may explain the existence of compensation gap.
Firm performance is widely used as a benchmark to decide CEOs’ payment. However, due to the differences in risk preferences lying between female and male CEOs, their performances may vary in terms of short-term performance and long-term performance, in addition, they may be compensated in a distinct way. From CEOs’ prospective of views, they might focus on stock market or firm long performances if they are compensated with relatively higher level of long-term payment. Likewise, company will compensate CEOs with more long-term payment as a motivation, if CEOs are risk averse. I thereby employ a
regression analysis to test how female CEOs’ payments would affected by firm performance. I expect that long-term compensation would have ex ante positive effect on long-term performance.
3. Data and research methodology
3.1 Data collection
I employ data from the S&P ExecuComp Annual Compensation database for information about CEO age, whether the CEO served as a director during the fiscal year and executive compensation related information, such as annual total compensation, equity-based incentives, salary, bonus and options. Firm related data are obtained from two resources. Frim stock price is retrieved from Center for Research in Security Prices (CRSP) database and financial ratio calculate from CRSP is winsorized at the 1% level. Other company financial data, such as total current asset, total common equity and cash, is retrieved from Compustat database. I collect the CEO education information from Capital IQ, and manually input the data since the table downloaded from Capital IQ cannot be merged with the table downloaded from the ExecuComp database in Stata. The data used in this paper include observations of US-based firms over a ten-year period from 2007-2016, using four-digital standard industrial classification (SIC) code from Compustat Annual database to identify. I take the year of 2016 to end the sample period because of the incomplete data in 2017. Consistent with large literature, this study excludes utilities (Standard Industrial Classification (SIC) codes 4900-4949) and financial firms (SIC 6000-6999) since the CEO’s compensation is either relatively stable and not significantly affected by the firm performance, or fluctuate and significantly affected by the firm performance, economic environment and regulations. The nominal variables are expressed in millions (U.S. dollars).
I apply several restrictions to build my sample. After removing observations with either non-positive sales, operating costs, book value, equity and total asset, or missing data on CEO salary, total compensation, and firms with inactive status, the final sample set consists of 10726 firm-year observations, including 1256 firms.
3.2 CEO compensation
To investigate the features and gender differences in compensation, I use two economic techniques to conduct this study, which are inspired by Sah (2015). The dependent variable in
equation (1), Compensation, is defined as natural logarithm of total compensation, (natural logarithm of TDC1 from Execucomp database) including salary, bonus, other annual compensation, total value of restricted stock granted, total value of stock options granted, long-term incentive payouts, and all other annual compensation. This variable is widely used by previous literature, such as Faulkender and Yang (2010), and Bugeja et al. (2012)
In terms of independent variables, Gender is a dummy variable that equals to 1 if the CEO in specific firm is female, and 0 otherwise. Age is the Chief Executive Officer age. Firm size is determined as the logarithm of the total asset as the control variable. Tenure is the number of years that Chief Executive Officer has been on the CEO position. Duality is a dummy variable that takes the value 1 when the CEO is the chairperson of the board and 0 otherwise. (Guoli, 2014) MB is the ratio of firm market value to book value, which is an indicator to growth opportunity. Control variables includes year and industry to account for potential unobservable time-invariant specific effects, which may be correlated with dependent variable.
Other Compensation includes natural logarithm value of Salary, Bonus, Options and Equity-based Incentives, respectively, to further analysis how the CEOs are compensated by short-term payment or long-term payment. According to Balkin, Markman and Gomez-Mejia’s (2000) paper, short-term remuneration consists of annual salary and bonus, and long-term remuneration is related to equity-based incentives. Following their model, I use Salary, Bonus, Options and Equity-based Incentives as dependent variables.
Equation 1
3.3 Risk prefernce
Firm’s cash holding is considered as a benchmark for corporate risk level. (Sah, 2015). Fisher and Yao (2017) indicate that female have significantly lower level of risk tolerance
than male have and are more likely to have uncertainty about income and net worth. Following the model that Bates, Kahle, and Stulz (2009) used in their paper, I choose cash holding as dependent variable, which is defined as the ratio of cash and short-term investments to total asset. CF is cash flow which is operating income before depreciation subtract interest and expenses, income tax, and common dividend, then scaled by total asset. R&D is research and development expenses divided by sales. Capex is capital expenditures scaled by total asset. Leverage is the sum of long term debt and debt in current liabilities divided by total asset. Dummy variable Payout takes the value 1 if a firm pays a common dividend in a given year and 0 otherwise. Acquisition is the ratio of acquisition to total asset. CFV is cash flow volatility, which is calculated as the annual standard deviation of firms’ quarterly ratio of cash flow to assets. (Gormley and Matsa, 2016)
Equation 2
3.4 Firm performance
To analyze the relation between the CEO gender and firm performance, I employ two different proxies as firm value. ROA is return on asset and is defined as the ratio of net income to total asset. This measurement is widely used in extensive literature. (Khan and Vieito,2013; Finkelstein and Boyd,1998) Tang, Wu & Zhang (2013) indicate that ROA is a cash-related variable and highly related to firm short-term performance. Therefore, combining with the result of the equation (2), I consider ROA as an indicator to gender difference to firm short-term performance. The second proxy, Tobin’Q, is defined as the sum of the market value of the stock and the book value of debt divided by the book value of total assets. (Tobin, 1969) In contrast, Tobin’Q is used to measure the firm long-term performance and growth opportunity.
MV is market value, consisting of the product of close price and common shares outstanding. STV is stock volatility that equals the square root of the sum of squared daily
returns over the year. To adjust for differences in the number of trading days, the raw sum is multiplied by 252 and divided by the number of trading days. (Gormley and Matsa, 2016)
4. Empirical analysis and results
4.1 Descriptive statistic
Table 1 provides descriptive statistics on the final sample by gender. On average, female CEOs account for relatively small proportion of higher management, especially in the year of 2008, only 7 women took position in chief executive officer, which constitute less than two percentage across the sample in the given year. The number of female CEO peaks in 2014, reaching to 31 out of 769 year-firm observations. Nonetheless, even the number of female CEOs in 2016 is as three times as it is in 2007 and has been steady growing, female CEOs still have limited power in higher management positions.
Table 1
Descriptive statistics of the CEO number in a given year
This table reports the number of female CEO and male CEO separately in a given year. The data is retrieved from Compustat. The female CEO proportion is calculated as the number of female CEO in a given year scaled by total number of CEO hired in given year across the sample. All percentages are presented in decimal format.
Female CEOs Male CEOs
Year Observations Observations Female CEOs %
2007 8 328 2.44 2008 7 367 1.91 2009 9 394 2.28 2010 13 427 3.04 2011 14 482 3.27 2012 23 560 3.55 2013 28 637 4.11 2014 31 738 3.87 2015 30 863 4.20 2016 28 861 3.25
Table 2 reports summary statistics for CEOs’ characteristics. CEO’s personal characteristics consist of age, tenure, education level, total compensation, salary, bonus, options and equity-based incentives. Other related information consists of firm size, market value, annual sales and cash. The Table 2 also references the results of the samples T-tests, to compare the means of each variable for companies where the CEO is a male or female.
Female CEOs are just one year younger than the male CEOs, with 55 years old on average. The gender difference in ages is decreased by 75% compared with the result from Adams, Gupta and Leeth’paper (2007), in which they indicate that female CEOs are nearly four years younger than their counterparts. This difference may derive from the sample selection since they start the data from 1992 to 2004. As mentioned in the introduction, due to the policy change and public attention on female CEOs, the chance of being in that position for female is larger than over 10 years before, so this difference is unsurprised. Female executives are on average hired as CEO for 5.8 years, which is almost 2.3 years less than male CEOs, and this difference is statistically significant at 1% level. In terms of education level, although Keloharju, Knüpfer and Tåg (2017) employ exceptionally rich data from Sweden and conclude that female, on average, hold higher education level than their counterparts, Table 2 reports that female CEOs are lower than male CEOs, with 1.4 and 1.7 respectively. The two different results may be due to two reasons: sample selection and measurement in education level. Their sample is based on Sweden and from 1992 to 2011, in contrast, this study is based on US companies. Besides that, due to the limited data source, observations with education level information are significant fewer than the final sample. In the case of measurement in education level, Keloharju, Knüpfer and Tåg (2017) employ a more sophisticated classification method in their study, so it may end up with different results.
In terms of remuneration, on average, female CEOs receive higher salary but lower bonus compared with male CEOs, and the differences in salary and bonus are statistically significant at 5 percent level. Female CEOs are compensated with lower level of equity-based incentives, and the result is statistically significant at 5% level. However, there are no significant differences in total compensation between female CEOs and male CEOs. These findings are similar to earlier literature, such as Elkinawy and Stater (2011), Bugeja, Matolcsy and Spiropoulos (2012). Firms that hired female as CEOs show no significant differences in size, market value and size with firms that hired male CEOs, although the sales are higher and statistically significant at 5 percentage level.
It is worth noting that firms led by female CEOs tend to have a higher level of cash and lower level of research and development activities. Those statistic results provide preliminary indication that female CEOs are more likely risk averse than male CEOs and tend to make
conservative financial decisions. There are also notable differences in firms’ performances. Firms with female CEOs, on average, have better performances in short-term, which is implied by measuring ROA, however, they have worse performances in the long run, which implied by Tobin’Q. These results may also affect the payment decision, and I will discuss this relation in the later study.
Table 2
Descriptive statistics of CEO characteristics
This table displays the statistic description of differences in female and male CEO characteristics during the sample period, from 2007-2016. The sample is retrieved from Compustat and consists of 10726 firm-year observations. The last column reports the t-statistics for differences in means of female CEOs and male CEOs. All firms and CEO’s characteristics are measured at the end of fiscal year. Definitions of all variables are reported in Appendix A. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Female CEOs Male CEOs
Mean Observations Mean Observations T test-mean difference
Age 54.99 310 56.08 10416 0.23 Tenure 5.79 310 8.26 100324 3.23*** Education 1.39 130 1.74 4367 5.77*** Total compensation 8.35 310 8.18 10401 -0.47 Salary 6.64 310 6.51 10357 -1.10* Bonus 5.19 49 5.70 2027 1.66* Options 6.82 167 7.03 6075 0.54 Equity incentives 8.09 171 8.12 4735 0.18* Size 7.79 310 7.57 10416 -0.60 Market Value 7.74 310 7.63 10416 -0.18 Sales 7.89 310 7.46 10395 -1.62* Cash 5.16 309 4.92 10408 -0.50 R&D 0.04 200 0.39 6721 0.51* ROA 0.043 310 0.040 10416 -0.07*** TQ 1.87 310 2.02 10416 0.33*
4.2 Logistic regression results 4.2.1 Compensation
Table 3 reports the results of female compensation features. According to the results form Table 1, there is a conspicuous disparity in numbers of female CEOs and male CEOs. Therefore, to better test the relation between gender differences and compensation, I use two
sets of samples to estimate results. First, I use the entire sample to estimate the regression outcomes for firms with male CEOs and female CEOs by including the binary variable gender in equation (1). The regression results are present in the first four columns. Second, to address endogeneity concern and self-selection bias that mentioned in the earlier part of this study, I employ a propensity score one-to-one matching approach to proceed a second screening to improve the sample validity. The treatment group in propensity score matching approach is firms with female CEOs, on the other hand, the control group is firms with their counterparts. After several tests with different sets of controls, I execute the matching by using a logit regression of gender binary variable on executive ages, tenure, total current compensation, which includes annual salary and bonus, leverage, market to book ratio, and also industry and year fixed effects, without replacement. Column (5), (6), (7) and (8) report regression results for female executives and matched male executives’ differences in compensation. The presence of excess controls leads to a dramatic drop in the number of observations, falling from over 9000 to less than 500.
The leading variable, gender, is significantly positive in relation to total compensation in one-to-one matched sample, meaning that female CEOs receive higher level of compensation. In detail, female CEOs receive 16.5 percentage points more in total compensation than male CEOs within the same industry in a given year, and the magnitude of increase in matched sample is almost as twice as in the entire sample. This outcome is consistent with Hypothesis 1 and also with the result in Smith and Watts’s paper (1992), indicating that the gender compensation gap is diminishing and even shows a opposite trend. It is also interesting to notice that the negative coefficients on gender and age in column (4) and (8) indicate the lower equity-based payment for female CEOs relative to male CEOSs, and these estimations are qualitatively significant in both sample. The intuition might be that female are likely risk-averse and willing to focus on relatively safe financial segment, say cash holding, instead of stock market. (Almenberg and Dreber, 2012). Equity-based payment also is considered as a proxy of long-term incentive, logically, it would be linked with long-term performance. This inseparable connection will be discussed in the next two parts.
Furthermore, these four diverse types of payment all show positive relationships with firm size and these results are statistically significant at the 1% level, both in the entire sample and.
Table 3
Gender differences in compensation
This table presents the regression analysis of the gender affection on CEOs’ compensation package of equation (1). The dependent variables are total compensation, salary, bonus and equity-based incentives in the given year, respectively. Gender is binary variable that takes the value 1 if the firm is led by a female, 0 otherwise. Other explanatory variables are age, firm size, tenure, duality, lag values of market to book ratio, leverage and cash, since the joint effects of these firm-related financial features are highly related to the CEOs’ compensation in the next year. The results in column (1), (2) and (3) are estimated from the entire sample, and results in column (4), (5) and (6) are estimated from propensity score matching sample. All firms and CEO’s characteristics are measured at the end of fiscal year. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Entire sample One-to-one matched
(1) (2) (3) (4) (5) (6) (7) (8) Total
compensation
Salary Bonus Equity Incentives
Total compensation
Salary Bonus Equity Incentives Gender 0.0903 0.0397 0.0718 -0.2411** 0.1646** -0.0631 0.4579 -0.2798* (0.065) (0.078) (0.312) (0.094) (0.073) (0.131) (1.050) (0.164) Age 0.0026 0.0065*** 0.0126** -0.0075** 0.0038 0.0098 -0.0440 -0.0277** (0.002) (0.002) (0.006) (0.003) (0.005) (0.010) (0.064) (0.013) Size 0.3752*** 0.1092*** 0.2757*** 0.5263*** 0.4320*** 0.1328** 0.1343 0.5856*** (0.011) (0.014) (0.042) (0.020) (0.034) (0.060) (0.380) (0.081) Tenure -0.0060*** -0.0025 0.0068 0.0190*** -0.0136* -0.0005 -0.0920 0.0476** (0.002) (0.002) (0.006) (0.003) (0.008) (0.013) (0.098) (0.019) Duality -0.1092** -0.2603*** 0.0834 -0.440*** -0.2715*** -0.1519 -1.204 -0.3248 (0.043) (0.052) (0.135) (0.066) (0.094) (0.169) (0.920) (0.214) Lag MB 0.0539*** -0.0508*** -0.0148 0.1812*** 0.0888*** 0.0304 -0.0291 0.2254*** (0.009) (0.011) (0.033) (0.016) (0.030) (0.053) (0.250) (0.063) Lag Leverage 0.3224*** 0.3173*** 0.8554*** -0.2083** 0.2032 0.4916 -2.2228 1.3020*** (0.055) (0.066) (0.217) (0.086) (0.178) (0.319) (2.625) (0.449) Lag Cash 0.0366*** 0.0127 0.0318 0.0094 0.0028 0.0044* 0.7489** -0.0310 (0.010) (0.012) (0.035) (0.016) (0.032) (0.057) (0.321) (0.077) Observations 9,457 9,323 1,657 4,471 482 482 56 236 R-squared 0.412 0.174 0.509 0.517 0.850 0.438 0.903 0.818 Industry FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes
in the matched sample. Besides that, female also receive higher compensation for holding more cash and increasing leverage, however, these coefficients are significant only in entire
sample. Tenure shows an opposite effect on total compensation and equity incentive. In specific, one year longer in CEO’s tenure leads to 0.6 percentage point and 1.3 percentage points decrease in compensation, within the entire sample and matched sample respectively, but 1.9 percentage point and 4.7 percentage points increase in equity-based incentives. Salary and bonus are generally not affected by those variables in matched sample, except the positive relation with cash
4.2.2 Cash holding
To analyze the relation between gender differences in cash holding, I use the model in Sah (2015), and Bates, Kahle and Stulz (2009), which accounts for various controls on cash holding. Table 5 presents panel regression results of equation (2). Similar to Table 3, the first three columns show the results using entire sample, while column (4) to column (6) present results of screened sample. I separately test the equation (2) since some variables, such as R&D, acquisition and payout, are not available for the entire sample. After including these variables, the number of observations substantially drops from over 10000 to around 6000, and from 519 to 295, respectively.
The coefficients on the major dummy variable, gender, indicate that female chief executive officers tend to hold more cash than their counterparts, especially without including additional firm investment-related variables, such as research and development, acquisition and payout. The results in column (1) and (3) are statistically significant at 1 percentage level. The outcome is consistent with Hypothesis 2 that female CEOs are risk averse than male CEOs and more likely to hold more cash for precautionary purposes. Leverage shows a negative impact on cash holding, and results in all columns are significant at 1% level. With the negative relationship between cash flow volatility and cash holding, it further confirms that female CEOs are more sensitive to cash and tend to use riskiness cash to execute investment decisions.
Consistent with Sah (2015), I also identify that cash holding increase with research and development, and market to book ratio. However, different from earlier literature, I find that payout also holds a positive relation with cash holding.
Table 4
Gender differences in Cash Holding
This table reports regression results of gender differences in cash holding. Based on equation (2), the depend variable in this table is cash holding, and independent variables are gender, size, cash flow, capital expenditures, leverage, market to book ratio, research and development, acquisition, payout and cash flow volatility. Definitions of all variables are reported in Appendix A. The results in column (1), (2) and (3) are estimated from the entire sample, and results in column (4), (5) and (6) are estimated from propensity score matching sample. Due to the missing values in R&D, acquisition and payout, observations are significantly less than those in regression that only control gender, size, cash flow, capital expenditures. leverage and market to book ratio. All firms and CEO’s characteristics are measured at the end of fiscal year. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Entire sample One to one sample
(1) (2) (3) (4) (5) (6) Gender 0.0373*** 0.0198 0.0196 0.0606*** 0.0548** 0.0553** (0.011) (0.015) (0.015) (0.018) (0.026) (0.026) Size -0.0364*** -0.0396*** -0.041*** -0.0420*** -0.0553*** -0.0559*** (0.001) (0.002) (0.0018) (0.005) (0.008) (0.008) Cash Flow 0.2214*** 0.2578*** 0.253*** 0.4434*** 0.5992*** 0.6013*** (0.005) (0.006) (0.006) (0.066) (0.082) (0.082) Capex -0.1647*** -0.2702*** -0.2883*** -0.3342* 0.04226 -0.0190 (0.043) (0.080) (0.080) (0.172) (0.318) (0.323) Leverage -1.0217*** -1.0108*** -1.0127*** -0.9928*** -0.9622*** -0.9604*** (0.009) (0.015) (0.015) (0.042) (0.066) (0.066) MB 0.9994*** 0.9993*** 1.0001*** 0.9940*** 0.9823*** 0.9852*** (0.001) (0.002) (0.002) (0.007) (0.010) (0.010) R&D 0.0006** 0.0007** 0.0003 0.0005 (0.001) (0.000) (0.001) (0.001) Acquisition -0.0416 -0.0379 -0.1185 -0.1192 (0.030) (0.030) (0.128) (0.129) Payout 0.0144** 0.0140** 0.0338 0.03043 (0.006) (0.006) (0.027) (0.028) CFV -0.2042*** -0.4702 (0.028) (0.424) Observations 10,103 6,041 6,007 519 295 294 R-squared 0.586 0.687 0.687 0.692 0.992 0.992
Firm FE Yes Yes Yes Yes Yes Yes
Industry FE Yes Yes Yes Yes Yes Yes
Table 5
Relation between compensation and frim performance
This table reports the regression outcomes of the relation between renumeration and firm performance. According to equation (3), dependent variables are ROA and Tobin’Q. Independent variables are gender dummy variable, age, tenure, duality, size, cash, firm market value, leverage, stock volatility, and lag value of compensation and equity-based incentive. To calculate stock volatility, I use data from CRSP which is winsorized at the 1% level. Similar to the method used in equation (1) and equation (2), I focus on how the compensation that CEOs paid in given year would affect the firm performance in the next year, so I use lag value of compensation and equity-based incentive in the regression. Definitions of all variables are reported in Appendix A. The results in first four columns are estimated from the entire sample, and results in the last four columns are estimated from propensity score matching sample. All firms and CEO’s characteristics are measured at the end of fiscal year. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Entire sample One-to-one matched
(1) (2) (3) (4) (5) (6) (7) (8)
ROA ROA TQ TQ ROA ROA TQ TQ
Gender 0.0019 0.0120* -0.0098 0.0261 0.0177 0.0223 -0.0665* -0.1225* (0.015) (0.010) (0.058) (0.073) (0.021) (0.025) (0.096) (0.148) Age -0.0011*** 0.0005* -0.0047*** -0.0005 0.0016 0.0008 -0.0126* -0.0032 (0.001) (0.000) (0.002) (0.002) (0.001) (0.002) (0.006) (0.012) Tenure 0.0009** 0.0002 0.0028* 0.0019 -0.0029 -0.0020 0.0073 0.0214 (0.001) (0.001) (0.002) (0.003) (0.002) (0.003) (0.009) (0.017) Duality -0.0025 0.0020 -0.0067 -0.0468 0.0233 0.0410 -0.0912 -0.0060 (0.011) (0.007) (0.042) (0.052) (0.027) (0.029) (0.124) (0.171) Size -0.0401*** -0.0396*** -1.4022*** -1.2042*** -0.0473*** -0.0080 -1.4263*** -1.6742*** (0.0042) (0.003) (0.016) (0.022) (0.015) (0.018) (0.070) (0.106) Cash -0.0038 -0.0005 0.0468*** 0.0672*** -0.0128 -0.0192 0.1104*** 0.1864*** (0.002) (0.002) (0.009) (0.013) (0.010) (0.012) (0.042) (0.0683) MV 0.0687*** 0.0504*** 1.3735*** 1.0795*** 0.0748*** 0.0457** 1.3372*** 1.4784*** (0.004) (0.003) (0.015) (0.021) (0.0140) (0.0182) (0.063) (0.107) Leverage -0.0339** -0.0097 1.3675*** 1.3444*** -0.0558 -0.0143 1.3378*** 1.3992*** (0.013) (0.009) (0.050) (0.068) (0.058) (0.077) (0.262) (0.450) STV -0.1832*** -0.1445*** 2.0217*** 1.5116*** -0.1612** 0.0051 1.6392*** 1.6053** (0.027) (0.020) (0.103) (0.148) (0.081) (0.119) (0.366) (0.695) Lag Compensation -0.0041* 0.0191** 0.0169 0.0855 (0.002) (0.010) (0.014) (0.061)
Lag Equity Incentive 0.0003** 0.0408* -0.0107 0.0131** (0.001) (0.011) (0.009) (0.053) Observations 7,898 3,618 7,898 3,618 404 197 404 197 R-squared 0.147 0.311 0.678 0.667 0.557 0.444 0.846 0.862 Industry FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes
4.2.3 Firm performance
The coefficients of main dummy variable, gender, provide evidence that female CEOs are better at short-term performance, which is reflected by ROA. Nevertheless, long-term performance, which is represented by Tobin’Q, is barely satisfactory. As predicted, compensation plays a pivotal role in motivating CEOs. The positive coefficients estimating on equity-based incentives confirm that the higher equity-based payment in prior year, the better firm will perform in the long run. Combining with this result reported in Table 5, it is reasonable to explain why female CEOs are compensated with lower level of equity-based payment, since firms led by female CEOs have worse performances.
Table 5 also provides evidence that market value contributes to the increase in firm short- and long-term performance. Long-term performance also increases with cash, leverage and stock volatility. On the contrary, leverage and cash might lead a decrease in short-term performance. Both two payments do not have significant influence on firm short-term and long-term performance in propensity score matching sample. I do not find any supportive evidence in the relation between frim short- and long-term performance and duality.
4.3 Robustness check
Endogeneity concerns may impact estimation validity and interpretation. For instance, female CEOs may choose to join a mature firm which riches in cash and are less likely to participate in risky investment, or, CEOs might adjust their investment strategies to cope with fluctuations in the economic environment. To address such self-selection bias, I use two different matching approaches and alternative sample period to reconfirm the results.
First, given the fact that the entire sample presents a notable gender disparity in numbers of CEOs, I use the same control variables, which are executives’ age, tenure, current total compensation, firm’s market to book ratio, leverage, and also fixed year, industry effects, conducting three-to-one matching approach to achieve similar objectives. One advantage of this approach is to expand sample capacity from maximum 500 observations to above 2000 observations, which is better to reflect the original male to female ratio in the entire sample. Second, following (Sah, 2015), I use alternative propensity score one-to-one matching
methods with less rigorous restrictions, in which I only control firm size and industry to match firms with CEOs in different gender. Robustness test results estimated three models by adopting above two matching approached are reported in Table 6 Panel A, first four columns of Table 7, and Table 8 Panel A. The outcomes of re-estimations are qualitatively similar to the main conclusions.
Considering that the sample period covers the 2008 global financial crisis, there are possibilities that firm performance, CEOs’ compensation, and firm financial decisions would be affected. Hence, I assess the robustness of main results by alternative the sample period to re-estimate three equations. I use the latest five-year time span as new sample period, from 2012 to 2016, to mitigate financial crisis’s influence on dependent variables and other control variables. Besides the change in time span, other controls and method remain the same. Results are presented in Table 6 Pane B, last two columns of Table 7, and Table 8 Panel B. Few results are changed by changing the sample period. In terms of the compensation, new result shows that gender difference in equity-based incentives diminishes. Results of Table 7 provide evidence that female CEOs are not sensitive to cash holding any more. These changes imply that the financial crisis does have impact on female CEOs compensation and risk preference, however, do not affect the main conclusions in this study.
5. Conclusions and discussions
This paper decomposes the gender differences in CEOs’ compensation and potential relation among CEOs’ compensation, CEOs’ risk preference, and firm performance.
CEOs make economical contributions to firms that their compensation should be commensurate with their effort. (Jung and Subramanian, 2016) Extensive literature support the existence of gender pay gap, indicating that female CEOs are, in general, compensated with lower level of remuneration relative to their counterparts. While in this study, I find that this gap is diminishing, and female are compensated more in terms of total compensation but less in equity-based incentives. I also test the female CEOs’ risk preference, which measured by cash holding, and the results are consistent with most literature, confirming that female CEOs tend to be risk averse and hold more cash at hand. Furthermore, the results indicate the positive relationship between long-term performance and long-term incentive. These findings, to some extent, explain why female CEOs are compensated with less equity-based payment, since female CEOs lead firms to worse situations in the long run.
This study contributes to the literature in two aspects. First, I provide statistics on female compensation, cash holding and frim performance by adopting the latest data, which covers the time span from 2007 to 2016, while the majority studies are based on the data in earlier time. The results in this paper are most consistent with studies from nearest Second, I use two different propensity score matching approach, one-to-one matching and three-to-one matching, to estimate three models.
There is no exception that every study has limitations. Due to the insufficient data on CEOs’ education level, I only provide descriptive statistics with this personal characteristic. Although I test the significance level of this control on female CEOs’ payment, I do not find any desirable outcomes. Board characteristics, such as board size and the number of female sitting in the board, would have potential effect on compensation and also investment strategies, since agency problem plays a pivotal role in the decision making for CEO’s compensation. In this case, it is desirable to include board related controls into models.
However, Sah (2015), Bugeja, Matolcsy,Spiropoulos (2012) and other studies have similar results without controlling such variable, so the results that I obtained should apply.
References
Adams, R. B., Hermalin, B. E., & Weisbach, M. S. (2010). The role of boards of directors in corporate governance: A conceptual framework and survey. Journal of economic literature, 48(1), 58-107. Adams, S. M., Gupta, A., Haughton, D. M., & Leeth, J. D. (2007). Gender differences in CEO
compensation: Evidence from the USA. Women in Management Review, 22(3), 208-224.
Adhikari, B. K. (2018). Female executives and corporate cash holdings. Applied Economics Letters, 25(13), 958-963.
Ahern, K. R., & Dittmar, A. K. (2012). The changing of the boards: The impact on firm valuation of mandated female board representation. The Quarterly Journal of Economics, 127(1), 137-197.
Almenberg, J., & Dreber, A. (2012). Gender, stock market participation and financial literacy. SSE (No. 737). EFI Working Paper Series.
Athanasopoulou, A., Moss‐Cowan, A., Smets, M., & Morris, T. (2017). Claiming the corner office: Female CEO careers and implications for leadership development. Human Resource Management.
Balkin, D. B., Markman, G. D., & Gomez-Mejia, L. R. (2000). Is CEO pay in high-technology firms related to innovation?. Academy of management journal, 1118-1129.
Bartlett, R. L., & Miller, T. I. (1985). Executive compensation: Female executives and networking. The American Economic Review, 75(2), 266-270.
Bates, T. W., Kahle, K. M., & Stulz, R. M. (2009). Why do US firms hold so much more cash than they used to?. The journal of finance, 64(5), 1985-2021.
Benmelech, E., Kandel, E., & Veronesi, P. (2010). Stock-based compensation and CEO (dis) incentives. The Quarterly Journal of Economics, 125(4), 1769-1820.
Bowman, E. H. (1980). A risk/return paradox for strategic management.
Bucciol, A., & Zarri, L. (2015). The shadow of the past: Financial risk taking and negative life events. Journal of Economic Psychology, 48, 1-16.
Bugeja, M., Matolcsy, Z. P., & Spiropoulos, H. (2012). Is there a gender gap in CEO compensation?. Journal of Corporate Finance, 18(4), 849-859.
Campbell, K., & Mínguez-Vera, A. (2008). Gender diversity in the boardroom and firm financial performance. Journal of business ethics, 83(3), 435-451.
CNBC, (2018. ) Just 24 female CEOs lead the companies on the 2018 Fortune 500—fewer than last year.
Retrieved from
https://www.cnbc.com/2018/05/21/2018s-fortune-500-companies-have-just-24-female-ceos.html Cobey, K. D., Laan, F., Stulp, G., Buunk, A. P., & Pollet, T. V. (2013). Sex differences in risk taking
behavior among Dutch cyclists. Evolutionary psychology, 11(2), 147470491301100206.
Cook, A., & Glass, C. (2014). Above the glass ceiling: When are women and racial/ethnic minorities promoted to CEO?. Strategic Management Journal, 35(7), 1080-1089.
Coxbill, A. L., Sanning, L. W., & Shaffer, S. (2009). Market Reaction to the Announcement of a Male-to-female CEO Turnover (No. 2009-13). Centre for Applied Macroeconomic Analysis, Crawford School of Public Policy, The Australian National University.
Croson, R., & Gneezy, U. (2009). Gender differences in preferences. Journal of Economic literature, 47(2), 448-74.
Elkinawy, S., & Stater, M. (2011). Gender differences in executive compensation: Variation with board gender composition and time. Journal of Economics and Business, 63(1), 23-45.
Faulkender, M., & Yang, J. (2010). Inside the black box: The role and composition of compensation peer groups. Journal of Financial Economics, 96(2), 257-270.
Finkelstein, S., & Boyd, B. K. (1998). How much does the CEO matter? The role of managerial discretion in the setting of CEO compensation. Academy of Management journal, 41(2), 179-199.
Fisher, P. J., & Yao, R. (2017). Gender differences in financial risk tolerance. Journal of Economic Psychology, 61, 191-202.
Gilliam, J., Chatterjee, S., & Grable, J. E. (2010). Measuring the perception of financial risk tolerance: A tale of two measures.
Gormley, T. A., & Matsa, D. A. (2016). Playing it safe? Managerial preferences, risk, and agency conflicts. Journal of Financial Economics, 122(3), 431-455.
Huang, J., & Kisgen, D. J. (2013). Gender and corporate finance: Are male executives overconfident relative to female executives?. Journal of Financial Economics, 108(3), 822-839.
Jung, H. W. H., & Subramanian, A. (2016). CEO talent, CEO compensation, and product market competition.
Keloharju, M., Knüpfer, S., & Tåg, J. (2017). What Prevents Female Executives from Reaching the Top?. Khan, W. A., & Vieito, J. P. (2013). CEO gender and firm performance. Journal of Economics and Business,
67, 55-66.
Kuan, T. H., Li, C. S., & Liu, C. C. (2012). Corporate governance and cash holdings: A quantile regression approach. International Review of Economics & Finance, 24, 303-314.
Lemaster, P., & Strough, J. (2014). Beyond Mars and Venus: Understanding gender differences in financial risk tolerance. Journal of Economic Psychology, 42, 148-160.
Mohan, N., & Ruggiero, J. (2003). Compensation differences between male and female CEOs for publicly traded firms: a nonparametric analysis. Journal of the Operational Research Society, 54(12), 1242-1248.
Perryman, A. A., Fernando, G. D., & Tripathy, A. (2016). Do gender differences persist? An examination of gender diversity on firm performance, risk, and executive compensation. Journal of Business Research, 69(2), 579-586.
Ryan, M. K., & Haslam, S. A. (2007). The glass cliff: Exploring the dynamics surrounding the appointment of women to precarious leadership positions. Academy of Management Review, 32(2), 549-572. Sabharwal, M. (2013). From glass ceiling to glass cliff: Women in senior executive service. Journal of
Public Administration Research and Theory, 25(2), 399-426.
Sah, N. B. (2015). Essays on the impact of CEO gender on corporate policies and outcomes. University of South Florida.
Smith Jr, C. W., & Watts, R. L. (1992). The investment opportunity set and corporate financing, dividend, and compensation policies. Journal of financial Economics, 32(3), 263-292.
Tang, Y., Wu, J., & Zhang, L. (2013). Do anomalies exist ex ante?. Review of Finance, 18(3), 843-875. Tobin, J. (1969). A general equilibrium approach to monetary theory. Journal of money, credit and
Appendix A: Variable definitions
Dependent variables Definitions
Compensation Natural logarithm of total compensation consists of salary, bonus,
other annual compensation, total value of restricted stock granted, total value of stock options granted, long-term incentive payouts, and other annual compensation. Calculated from Compustat using ln(tdc1).
Salary Calculated from Compustat using ln(salary)
Bonus Calculated from Compustat using ln(bonus)
Option Calculated from Compustat using ln(option_awards)
Equity-based Incentive Calculated from Compustat using ln(eip_unearn_val)
Cash Holding The ratio of cash and short-term investments to total asset.
Calculated from Compustat using che/at
ROA Return on Asset. Calculated from Compustat using oibdp/at
ROE Return on equity. Calculated from Compustat using ni/ceq
TQ The sum of the market value of the stock and the book value of debt
divided by the book value of total assets. Calculated from Compustat and CRSP using ((at-ceq) + (csho*prcc))/at
Explanatory variables Definitions
Age Age of the Chief Executive Officer.
EDU The education level of Chief Executive Officer, equals to 1 if the
CEO holds the bachelor degree or equivalent degree and below bachelor degree; equals to 2 if CEO holds Master or equivalent degree; equals to 3 if CEO holds degree above Master degree. Data retrieved from Capital IQ database.
Size Calculated from Compustat using ln(at)
Tenure Tenure is the number of years that Chief Executive Officer has been
on the CEO position.
Duality Duality is a dummy variable that takes the value 1 when the CEO is
the chairperson of the board and 0 otherwise Calculated from Compustat using execdir.
MB Ratio of market value to book value Calculated from Compustat
and CRSP using ((at-ceq)+(csho*prcc_f))/at.
Lag MB The value of MB one year prior to the given year.
Cash Calculated from Compustat using ln(ch).
Lag Cash The value of Cash one year prior to the given year.
Leverage Long-term debt plus debt in current liabilities to book value of assets.
Calculated from Compustat using (dltt+dlc)/at.
Lag Leverage The value of Levergae one year prior to the given year.
CF Cash Flow. Calculated from Compustat using
R&D Ratio of research and development expenses to sales. Calculated from Compustat using (xrd/sale).
Capex Ratio of capital expenditures to book value of assets Calculated
from Compustat using (capx/at).
Payout Dummy variable takes the value 1 if a firm pays a common dividend in
a given year and 0 otherwise Calculated from Compustat using dvc.
ACQ Acquisition activity. Calculated from Compustat using aqc/at.
CFV Cash Flow Volatility is the annual standard deviation of firms’
quarterly ratio of cash flow to assets.
MV Market Value of Equity of the firm for that year Calculated from
CRSP using prcc*csho.
STV Stock Volatility is the square root of the sum of squared daily
returns over the year. To adjust for differences in the number of trading days, the raw sum is multiplied by 252 and divided by the number of trading days. Calculated from CRSP database.
Appendix B: Robustness test results
Table 6
Robustness test results for equation (1)
This table presents robustness test results based on equation (1), which focuses on the female compensation package. Panel A presents outcomes from two different propensity score matching methods. In first four columns, I use the same controls, which are CEOs’ age, tenure, total current compensation (salary and bonus), firms’ market to book ratio, leverage, and industry classification as well, but I substitute one-to-one matched to three-to-one matched. Correspondingly, observations are significantly increased by this change. Second, in the last four columns, I use a set of less rigorous restrictions to conduct another propensity score one-to-one matching that I only employ year and industry controls. Panel B presents the robustness test results using a different sample period. Due to the potential affection of 2008 global financial crisis on the results, I use the latest five-year period, from 2012-2016, to test the consistency. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Panel A
Three-to-one matched (same restrictions) One-to-one matched (relax restriction) (1) (2) (3) (4) (5) (6) (7) (8) Total
Compensation
Salary Bonus Equity Incentives
Total Compensation
Salary Bonus Equity Incentives Gender 0.1112 0.0798 0.1242 -0.2394** -0.0591 -0.1182 0.6822 -0.1934 (0.069) (0.092) (0.397) (0.109) (0.055) (0.080) (0.652) (0.143) Age -0.0010 -0.003 0.0242** -0.0042 -0.0096 0.0027 -0.0874 -0.0019 (0.003) (0.0037) (0.011) (0.005) (0.006) (0.009) (0.0771) (0.016) Size 0.3852*** 0.1383*** 0.2952*** 0.5423*** 0.4182*** 0.1094** 0.202 0.6902*** (0.018) (0.025) (0.079) (0.032) (0.0345) (0.050) (0.425) (0.093) Tenure 0.0086* 0.0236*** -0.0052 0.0205** -0.0089* -0.0028 0.0562 0.0169 (0.005) (0.007) (0.021) (0.009) (0.005) (0.007) (0.056) (0.013) Duality -0.1602*** -0.1872** -0.3882 -0.4724*** -0.1532 -0.1032 0.219 -0.2813 (0.062) (0.0834) (0.249) (0.101) (0.111) (0.163) (0.668) (0.262) Lag MB 0.0571*** -0.037 -0.0848 0.2655*** 0.1336*** 0.0124 -0.0926 0.3201*** (0.018) (0.0238) (0.069) (0.034) (0.025) (0.037) (0.166) (0.091) Lag Leverage 0.1962** 0.3352** 0.1301 -0.0967 0.0863 0.3863 -0.7951 0.8272* (0.099) (0.133) (0.418) (0.185) (0.163) (0.239) (2.988) (0.477) Lag Cash 0.0396** 0.0321 0.0233 -0.0181 -0.0129 -0.0479 0.3522 -0.1413* (0.016) (0.021) (0.066) (0.025) (0.033) (0.049) (0.409) (0.081) Observations 3,491 3,479 532 1,752 564 564 69 317 R-squared 0.479 0.188 0.576 0.560 0.760 0.340 0.893 0.665 Industry FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes
Panel B
(1) (2) (3) (4)
Total Compensation Salary Bonus Equity Incentives Gender 0.5652** 0.8121*** 4.6062*** 0.1782 (0.237) (0.308) (0.762) (0.175) Age 0.0236 0.0244 -0.0894 0.0145 (0.017) (0.023) (0.182) (0.015) Size 0.1992** 0.0449 1.8371*** 0.3721*** (0.091) (0.118) (0.309) (0.081) Tenure -0.0183 0.0167 0.3932 0.0513** (0.023) (0.029) (0.196) (0.0207) Duality -0.9772*** -0.9634*** 2.0974*** -0.5702*** (0.279) (0.363) (0.322) (0.207) Lag MB 0.0550 0.0292 -0.7597** 0.0785 (0.074) (0.096) (0.235) (0.054) Lag Leverage 0.3772 0.5792 -0.2801 0.9252** (0.541) (0.703) (1.117) (0.461) Lag Cash 0.0473 0.0753 -0.2759 0.0247 (0.077) (0.100) (0.293) (0.051) Observations 309 309 25 185 R-squared 0.441 0.361 0.999 0.795
Industry FE Yes Yes Yes Yes
Table 7
Robustness test results for equation (2)
This table presents robustness test results based on equation (2), which focuses on the female risk preference using cash holding dependent variable. The first two columns present outcomes from the sample that use same controls as earlier approach, which controls CEOs’ age, tenure, total current compensation (salary and bonus), firms’ market to book ratio, leverage, and industry classification as well, but I substitute one-to-one matched to three-to-one matched approach. As expected, observations are significantly increased by this change. Results in column (3) and (4) are estimated by using a set of less rigorous restrictions to conduct another propensity score one-to-one matching that I only employ year and industry controls. The last two columns present the robustness test results using a different sample period, from 2012-2016, to test the consistency. All firms and CEO’s characteristics are measured at the end of fiscal year. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Three-to-one matched One-to-one matched One-to-one matched (2012-2016) (1) (2) (3) (4) (5) (6) Gender 0.0329*** 0.0215 0.0387*** 0.0248 -0.0272 -0.0033 (0.012) (0.018) (0.010) (0.016) (0.025) (0.034) Size -0.0339*** -0.0370*** -0.0299*** -0.0417*** -0.0403*** -0.0472*** (0.002) (0.0031) (0.004) (0.007) (0.006) (0.011) Cash Flow 0.1211*** 0.2367*** -0.0172 0.2793** -0.0730 0.0929 (0.025) (0.036) (0.074) (0.110) (0.082) (0.1692) Capex -0.1292* -0.2003 -0.4624*** -0.5032** -0.2271 0.7102* (0.070) (0.140) (0.168) (0.246) (0.206) (0.420) Leverage -0.9922*** -0.9966*** -1.0562*** -0.9375*** -1.0273*** -1.0553*** (0.018) (0.028) (0.033) (0.057) (0.058) (0.081) MB 0.9982*** 0.9882*** 1.0083*** 0.9976*** 0.9962*** 1.0046*** (0.0035) (0.0047) (0.006) (0.008) (0.010) (0.014) R&D 0.0017 0.3142*** 0.1093 (0.001) (0.091) (0.0912) Acquisition -0.0369 0.0821 -0.1422 (0.053) (0.112) (0.151) Payout 0.0147 0.0453** 0.0884** (0.010) (0.022) (0.036) CFV -0.0946** 0.0671 0.5952* (0.039) (0.142) (0.329) Observations 3,767 2,244 582 336 285 163 R-squared 0.977 0.976 0.993 0.994 0.994 0.996
Industry FE Yes Yes Yes Yes Yes Yes
Table 8
Robustness test results for equation (3)
This table presents robustness test results based on equation (3), which focuses on relationships between female compensation and firm performance. Dependent variables are ROA, return on asset, and Tobin’Q. Panel A presents outcomes from two different propensity score matching methods. In first four columns, I use the same controls, which are CEOs’ age, tenure, total current compensation (salary and bonus), firms’ market to book ratio, leverage, and industry classification as well, but I substitute one-to-one matched to three-to-one matched. Second, in the last four columns, I use a set of less rigorous restrictions to conduct another propensity score one-to-one matching that I only employ year and industry controls. Panel B presents the robustness test results using a different sample period, from 2012-2016, to mitigate the affection from 2008 global financial crisis. Definitions of all variables are reported in Appendix A. Constants are included in all regressions but not reported. Standard errors are provided in parentheses. *, **, and *** indicate significant at 10%, 5%, and 1%, respectively.
Panel A
Three-to-one matched One-to-one matched
(1) (2) (3) (4) (5) (6) (7) (8)
ROA ROA TQ TQ ROA ROA TQ TQ
Gender 0.0070 0.0061 0.0336 0.0398 -0.0031 0.0046 0.1032 -0.1982*** (0.009) (0.008) (0.047) (0.057) (0.009) (0.009) (0.067) (0.062) Age -0.0003 0.0003 -0.0058*** -0.0026 0.0036*** 0.0030*** 0.0088 0.0116 (0.001) (0.001) (0.002) (0.003) (0.001) (0.0011) (0.008) (0.007) Tenure -0.0017** -0.0011 -0.0090** -0.0093* -0.0022*** -0.0005 0.0072 0.004 (0.001) (0.001) (0.0040) (0.005) (0.001) (0.0008) (0.0055) (0.005) Duality -0.0122 0.0021 -0.0365 -0.0926 0.01452 0.0171 -0.0128 0.1743* (0.010) (0.009) (0.049) (0.063) (0.020) (0.015) (0.147) (0.100) Size -0.0496*** -0.0573*** -1.1252*** -1.2828*** -0.0856*** -0.0633*** -1.6622*** -1.2612*** (0.0048) (0.005) (0.024) (0.033) (0.010) (0.011) (0.074) (0.072) Cash 0.0014 0.0012 0.0362*** 0.0490*** -0.0026 0.0006 -0.002 -0.0117 (0.002) (0.002) (0.012) (0.015) (0.0055) (0.005) (0.0404) (0.035) MV 0.0609*** 0.0633*** 1.1192*** 1.2462*** 0.1022*** 0.0673*** 1.7614*** 1.3002*** (0.004) (0.004) (0.022) (0.031) (0.0090) (0.010) (0.066) (0.065) Leverage -0.0197 -0.0090 0.9423*** 1.1033*** 0.0026 -0.0511 1.4572*** 1.1432*** (0.018) (0.017) (0.091) (0.119) (0.029) (0.034) (0.211) (0.229) STV -0.1662*** -0.0962*** 1.5232*** 1.3242*** 0.0468 -0.0287 2.1471*** 1.0924*** (0.027) (0.025) (0.133) (0.176) (0.055) (0.052) (0.407) (0.347) Lag Compensation 0.0018 0.0101 -0.0011 -0.0051 (0.003) (0.014) (0.006) (0.048) Lag Equity Incentive -0.0020 0.0144 -0.0009 0.0161 (0.002) (0.014) (0.004) (0.026) Observations 2,164 1,049 2,164 1,049 469 249 469 249 R-squared 0.381 0.570 0.753 0.833 0.478 0.661 0.814 0.915 Industry FE Yes Yes Yes Yes Yes Yes Yes Yes
Panel B (1) (2) (3) (4) ROA ROA TQ TQ Gender 0.0249 0.0275 -0.1671 -0.4440*** (0.024) (0.022) (0.118) (0.151) Age 0.0044** 0.0012 0.0270*** 0.0482*** (0.002) (0.002) (0.009) (0.011) Tenure -0.0033 0.0038* 0.0111 -0.0279* (0.002) (0.002) (0.010) (0.015) Duality -0.0302 0.0548* -0.1491 -0.0939 (0.033) (0.032) (0.159) (0.217) Size 0.0063 0.0084 -1.7761*** -1.9941*** (0.019) (0.016) (0.090) (0.109) Cash -0.0128* 0.0005 -0.0410 -0.0408 (0.008) (0.006) (0.038) (0.043) MV 0.0256 -0.0296 1.8579*** 2.0670*** (0.020) (0.019) (0.094) (0.131) Leverage -0.147*** -0.0810 1.4096*** 1.3132*** (0.0501) (0.067) (0.244) (0.451) STV -0.2321** -0.4352*** 3.7209*** 5.8691*** (0.111) (0.101) (0.539) (0.687) Lag Compensation -0.0108 0.1252* (0.013) (0.065)
Lag Equity Incentive 0.0014 0.1072**
(0.007) (0.047)
Observations 215 124 215 124
R-squared 0.544 0.723 0.920 0.954
Industry FE Yes Yes Yes Yes
