The mediating effect of Risk-taking:
Proof from the banking industry
Renske Nijholt
s2512297
Submitted to
Supervisor: Dr. M.A. Lamers Co-Assessor: Prof. Dr. B.W. Lensink MSc. International Financial Management
Faculty of Economics and Business University of Groningen
June 8th, 2018
ABSTRACT
This study investigates the mediating effect of risk-taking in the relationship between board gender diversity and performance. Apart from that, the moderating effect of gender inequality on board gender diversity and performance and on board gender diversity and risk-taking is tested. Looking at an international sample of commercial banks, the impact of female board representation on performance is determined. The main findings of this study include that board gender diversity negatively impacts performance. This effect is strengthened by gender inequality. No evidence is found for the moderating effect of gender inequality on board gender diversity and risk-taking. Also, there is not enough support to establish the mediating effect of risk-taking on board gender diversity and performance. So, in the end, more women on the board of directors in the banking industry leads to lower performance, which is even lower in countries with high gender inequality.
1. Introduction
In the past years gender equality has become a popular topic. With movements such as ‘He for She’, ‘Time’s Up’, and the ‘Women’s March’ more attention is drawn to decreasing the gender gap. If these initiatives succeed in breaking the glass ceiling, what would this mean for business?
Men and women portray different traits. In general women are more agreeable and open, while men are more assertive and competitive (Costa, Terracciano & McCrae, 2001; Croson & Gneezy, 2009). This leads to leadership styles that vary between gender. Furthermore, several studies have showed that females are in general more risk-averse than men (Byrnes, Miller & Schafer, 1999; Barber & Odean, 2001).
So, what if this is applied to the banking industry, in which the level of risk-taking is a very important variable. Will the presence of more women in high positions, such as the board of directors change anything? If expectations about women taking less risks hold, what would this mean for the performance of banks? Therefore, this paper looks into the mediating effect of risk-taking in the relationship between board gender diversity and performance in banks. Furthermore, the effect of gender inequality is be used as a moderator to investigate whether this variable strengthens or weakens the relations between board gender diversity and performance and on board gender diversity and risk-taking.
that directors make decisions based on personal factors (Cyert & March, 1963). In order to avoid groupthink, a diverse group of directors is preferable for sound decision making.
Several studies also look into the effect of risk-taking on performance. In their empirical study Miller and Bromiley (1990) find that income risk negatively effects performance. Palmer and Wiseman (1999) find that a higher risk impacts the volatility of income, and this decreases the performance of the firm (Shapiro & Titman, 1986; Cornell & Shapiro, 1987).
Gender inequality can be a source of female discrimination. (Human Development Reports) Women in countries with high gender inequality need to be more competitive to gain a seat on the board of directors. Because of this competition, these women are more power loving (Adams & Funk, 2012), which makes their impact on performance stronger compared to women in gender equal countries.
To the best of my knowledge there no research is done into the mediating effect of risk-taking. Furthermore, previous research focuses on one country (mainly US), whereas this study is based on an international sample of commercial banks with yearly observations over the period 2012-2017. The main findings include a negative impact of board gender diversity on performance, which is strengthened by higher gender inequality.
The rest of this paper is organized as follows. First, a literature review addresses relating theories and discusses previous studies. After that, the methodology of this study is explained. This is followed by the results and the discussion of the results. Finally, the paper is concluded, and some implications and limitations will be given.
2. Literature Review
This section elaborates on theories and literature related to corporate governance, board gender diversity, gender inequality, risk-taking and performance. Based on these theories and articles five hypotheses are formed.
2.1. Corporate Governance
can act on their own needs as well. Three theories can be applied to this issue, namely the Principal-Agent theory, the Resource Dependency theory, and the Upper Echelons theory. Each of these theories can be used to explain the relationship between diversity and risk. The Principal-Agent theory looks at ways of monitoring management to ensure goal congruence, which is easier with a more diverse board. The Resource Dependency theory is based on external factors that influence the firm and how different stages of a firm’s life call for specific management qualities, which are more likely to be found in a diverse board of directors. The Upper Echelon theory states that a manager’s background influences their decision-making, so in order to decrease groupthink and risk, a more diverse board is preferable. Since all these theories present arguments regarding board diversity in relation to risk-taking, they are now discussed in more detail.
First, the Principal-Agent theory. Eisenhardt (1989) states there is goal incongruence between the principal (firm’s owners) and the agent (managers). Therefore, the principal needs a way to monitor the agent. This can be done based on output control and behavioral control. Ouchi (1979) proposes there are three types of control that can be used, market, bureaucracy and clan. When applying a clan mechanism, the applicants are screened thoroughly, and only people who exactly fit the requirements get a position, leading to a homogeneous group. In this case, there is no need for close monitoring, since all agents have the same goals. However, in reality it is quite hard to find people fitting all requirements. Therefore, people who best fit the requirements are chosen, leading to a more heterogeneous group. The principals rely upon training and monitoring to ensure goal congruence. So, the more heterogeneous a group, the more monitoring is necessary. The monitoring of the agents comes with costs (Fama & Jensen, 1983). Therefore, there is always the trade-off between monitoring and costs. The monitoring of the agents is one of the functions of the board of directors (Sila, et al., 2016). In order to properly fulfill this role, scholars find that it is beneficial to have more diversity within the board (Carter, et al., 2010; Erhardt, et al., 2003; Adams & Ferreira, 2009).
performance. Hillman, et al. (2000) also investigate the Resource Dependency theory. Their research, conducted between 1968 and 1988, is based on the US airline industry. Their findings suggest that the board of directors should reflect the environment of the firm. The firm should identify the most important resources and linkages and find directors that can fulfill the resource dependency role. The more complex the environment is, the more directors are needed to do this. When the directors are not capable of fulfilling the resource dependency, the firm risk will increase. Therefore, it is important to have the right amount, and most importantly, the right directors.
Third, the Upper Echelons theory predicts that the performance of the firm is related to the background of the managers (Hambrick & Mason, 1984). Decisions are made based on behavioral factors, so they are not objectively taken to maximize operations (Cyert & March, 1963). Therefore, these decisions reflect upon the maker of the decisions. Janis (1972) discusses that a homogeneous group leads to groupthink. Thus, decisions are made reflecting a very similar group of people. It is harder to solve problems with a homogeneous group as compared to a heterogeneous group (Filley, House & Kerr, 1976). Therefore, a heterogeneous group is able to make better decision in varying circumstances, which leads to less risk.
Thus, all three applicable theories come to the same conclusion. A more diverse board of directors is able to make better decisions and creates less risk.
One part of diversity in the board of directors is gender diversity. Research is done to look into the level of risk-taking by women compared to men. There are several studies claiming that in general women are more risk averse than men (Eckel & Grossman, 2008; Sapienza, Zingales & Maestripieri, 2009; Bernasek & Shwiff, 2001; Croson & Gneezy, 2009). This is in line with the findings of Barber and Odean (2001) who study the financial sector. This study, conducted in the US, compares common stock investments of over 78,000 households in the period between February 1991 and January 1997. The authors find that the size of the investment portfolio is equal between gender. However, compared to women, men have a higher turnover, lower their return more due to excessive trading and hold more risky positions. In a study about risk tendencies it is also concluded that women show less risk-taking behavior than men (Byrnes et al., 1999).
questions to measure the respondent’s values and their level of risk aversion. It is concluded that women who acquire a seat on the board through competition are more achievement and power oriented than worker representatives. Furthermore, these women are less risk averse than any other board member. So, they find proof that women take more risk than men. Finally, Sila, et al. (2016), who study the board of directors in the US, find no evidence for an impact of gender diversity on the board equity risk. They state that the number of women on the board is a choice the board makes. This view is also supported by Schubert, Brown, Gysler, and Brachinger (1999). They find that the amount of risk taken depends on the decision frame, not the gender of the decision maker.
As is seen from the above discussion, there are opposite views regarding the relation between board gender diversity and risk-taking. However, since the majority of the research is in support of a negative relationship between gender diversity and risk-taking the first hypothesis will follow this line of research. Therefore, the following hypothesis is applicable:
H1: Banks with a more gender diversified board show less risk-taking behavior.
2.2. Risk and Performance
risk increases (Shapiro & Titman, 1986; Cornell & Shapiro, 1987). The volatility is not only leading to a higher default risk, higher costs are also involved. Since income is more volatile, firms may not be able to keep all their staff when income is low. However, the staff needs to be hired again when income levels increase. This means the firm will have more costs for hiring, firing and training their staff.
H2: Banks that show more risk-taking behavior have a lower performance.
2.3. Diversity and Performance
Looking at the overall relationship between diversity and performance, Adams and Ferreira (2009) find some mixed results regarding gender diversity and performance. In their study on US firms from all industries between 1996 and 2003 they find, on the one hand, that more diverse boards usually do a better job monitoring throughout the firm. Thus, they know better what is going on inside the company. In the end, this benefits firms that have weak governance. In these firms a more diverse board leads to increased performance (ROA). On the other hand, for the firms that do not have weak governance they do not find any evidence for the relationship between gender diversity and performance. Therefore, there is no evidence that a more diverse board leads to a general increase in performance. Campbell and Minguez-Vera (2008) study the relationship between board gender diversity and performance among listed firms in Spain. The sample includes 68 non-financial firms with observations during the period from January 1995 to December 2000. They define performance as firm value. They conclude that a higher proportion of female board members leads to a higher firm value. Francoeur, Labelle, and Sinclair-Desgangé (2008) look into gender diversity and performance in Canada. In their 2001-2004 study of the 230 largest public firms, they find that women are often appointed to leadership positions when a firm has problems, the so called ‘glass cliff’. Under these circumstances, a higher proportion of female officers leads to higher abnormal returns.
H3: Banks with a more gender diverse board leads to higher performance
2.4. Gender inequality
development. Higher inequality will make it harder for women to obtain a position in the board of directors. Adams and Funk (2012) find that women who have to compete for their board seat are, among others, more power loving than other women. The women in countries with a higher gender inequality have to work harder to get to the board of directors than women in more gender equal countries. Therefore, they fit with the power loving type. Therefore, female board members in countries with a higher gender inequality will take more control and affect operations more. Thus, the following hypotheses are established:
H4: Gender inequality strengthens the relationship between board gender diversity and performance.
H5: Gender inequality strengthens the relationship between board gender diversity and risk-taking.
2.5. Other variables
In this study the main focus is on the mediating effect of risk-taking on the relationship between board gender diversity and performance and the moderating effect of gender inequality. Nevertheless, there are several other variables that can have an impact on bank performance and should therefore not be forgotten. Support is already found for the mediating effect of innovation and reputation in the relationship between racial board diversity and performance (Miller & Triana, 2009). Furthermore, Johnson, Schnatterly, and Hill (2013) discuss the impact of not only gender diversity, but also firm age, education level of board members, and racial diversity of boards. Thus, these variables will serve as control variables in this research.
3. Methodology
3.1. Data and Variable Definitions
3.1.1. Sample
This study focuses on board gender diversity, risk-taking and performance in the banking industry. The sample for this study is based on a six-year sample (2012-2017) of active, listed banks which was retrieved from Orbis Bank Focus. The initial sample consisted of 2653 banks. Since the objective of the different types of banks can greatly differ (Houston & James, 1995), this research only takes into account commercial banks. This ensures that the same types of banks are considered, making the risk-taking levels of the banks comparable. Deleting all other types of banks leaves 1084 banks in the sample. Furthermore, the unconsolidated banks are deleted from the sample, leaving 717 banks. Since board gender diversity is one of the main variables in this study, only banks with complete information regarding this issue are included. Moreover, banks with less than three years of data on the main variables (net income, total assets, total equity) are deleted from the sample. After deleting these banks, a sample of 544 banks remains.
Data on the Gender Inequality Index (GII) is collected from the United Nations Development Programme's Human Development Reports. This report contains data on 159 countries over the period between 2012-2015.
The variables net income, total assets, equity and total liabilities are winsorised at the 0.05 and 0.95 level. Loans to total assets, ROAA and ROAE are winsorised at the 0.01 and 0.99 level. After winsorising the data, two banks in the sample do no longer have any deviation in their assets and net income over the six-year period. Consequently, the standard deviation of ROA which is required to calculate the z-score is zero. The z-score cannot be calculated using this number and therefore the two banks were removed from the sample. The final sample consists of 542 banks with 3250 company-year observations divided over 94 countries.
3.1.2. Dependent variable
relative to equity. To account for fluctuations in the amount of assets and equity they will be averaged over the year, giving the return on average assets (ROAA) and the return on average equity (ROAE) as the two main measurements for performance in this study.
3.1.3. Independent variable
The independent variable is board gender diversity. The first measurement of this variables consists of the proportion of female board members (Rose, 2007; Adams & Ferreira, 2009; Nguyen, Locke & Reddy, 2015). This variable looks at how many board seats are taken up by women as a percentage of the total seats on the board of directors. This measure looks at the effect of the size of the female director group on risk-taking and performance. However, the proportion of women can be affected by both a change in the number of women on the board of director or a change in the number of total board members. Thus, this measure is not always the appropriate metric. An alternative measure to the proportion of women on the board of directors is to measure for female presence in the board of directors. This is done in the form of a dummy variable (Carter, Simkins & Simpson, 2003, Rose, 2007). This dummy takes the value of 1 when there is at least one female board member and 0 if there is no female board member. So, for this measure the number of female board member is not taken into account.
3.1.4. Mediator
The mediator in this study is risk-taking. Unfortunately, there is no measure that can accurately measure the risk-taking behavior of banks in a quantitative way. However, the z-score, which measures bank risk can be used as a proxy (Pathan, 2009; Lepetit & Strobel, 2013; Laeven & Levine, 2009). Since the z-score measures riskiness and not risk-taking it is not a perfect measure, however, it is the best measure available and will therefore be used in this study. The z-score is based on the return on assets (ROA) plus the capital-asset ratio relative to the standard deviation of the return on assets (ROA) (Boyd, Graham & Hewitt, 1993). The higher the Z-score, the more stable the bank is perceived to be. So, a lower Z-score equals a riskier bank.
3.1.5. Moderator
participation (Human Development Reports, 2016). The index can range between 0 and 1 where zero represents total equality and one represents extreme inequality. So, a lower score represents a country with a smaller gender gap and thus more equality between men and women.
3.1.6. Control variables
Second, the GDP real growth rate is used to control for business cycle differences. The GDP real growth rate is the GDP growth rate relative to the inflation rate. When this is positive, the country is experiencing an upward cycle and performance is increasing. Therefore, performance is related to the GDP real growth rate, and a variable to control for this effect should be added to the regression equation.
Furthermore, some company control variables are used. The firm size, measured as the natural logarithm of total assets, controls for the effect of size on the performance of the bank (Bonin, Hasan & Wachel, 2005).
Board size controls for the difference in the size of the board of directors. As discussed, the
percentage of women on the board of directors depends on both the number of women and the number of board seats. Also, the size of the board has an impact on performance (Guest, 2009). This variable will control for this effect.
Generally, Liquid assets generate lower returns than other assets. Furthermore, liquid assets are less risky and less costly (Iannotta, Nocera & Sironi, 2007). In order to make liquid assets a more comparable variable it will be measured as liquid assets as a percentage of total assets. Finally, the amount of loans is used as a control variable. More loans in general increases risk (Berger, Kick & Schaeck, 2014). Therefore, a firm with relatively more loans will be less risky. To control for this fact, loans will be added to the regression equation. The variable is measured as total loans relative to total assets.
3.2. Regression model
The regression equation is tested using the four steps proposed by Baron and Kenny (1986), which is used in several other papers (Bear, Rahman & Post, 2010; Miller & Triana, 2009; Huselid, 1995; Chatman & Flynn, 2001). These steps are as follows: (1) establish that there is a relationship to be mediated, so a relationship between board gender diversity and performance. (2) Confirm that the independent variable, board gender diversity, relates to the mediating variable, risk-taking. (3) Showing the relationship between the mediating variable, risk-taking, and the dependent variable, performance. The independent variable should also be included in this regression, since the mediator might only have a relationship with the dependent variable because of the effect of the independent variable. (4) Finally, establish whether there is complete or partial mediation. The prior is the case when the regression in step 3 shows that the relationship between board gender diversity and performance is 0. If this is not the case, there is partial correlation. The overall mediating effect can then be calculated as the direct effect and the indirect effect of the mediating variable.
Regression Equation Step 1: Firm Performance = 𝐵𝑜𝑎𝑟𝑑_𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡 (1) Step 2: Risk-taking = 𝐵𝑜𝑎𝑟𝑑_𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡 (2) Step 3: Firm Performance = 𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝐵𝑜𝑎𝑟𝑑_𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡 (3)
Furthermore, some control variables should be added:
Firm performancei,c,t = α0 + 𝛽1Board_Diversityi,t + 𝛽2Firm_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽3GDP_Growthc,𝑡 (4) + 𝛽4Board_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽5Loans_to_Assets𝑖,𝑡 + 𝛽6Gender_Inequalityc,t + 𝜀c,𝑖,𝑡
Firm performancei,c,t = α0 + 𝛽1𝑅𝑖𝑠𝑘_Taking𝑖,𝑡 + 𝛽2Board_Diversityi,t + 𝛽3Firm_ 𝑆𝑖𝑧𝑒𝑖,𝑡 (6) + 𝛽4GDP_Growthc,𝑡 + 𝛽5Board_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽6Loans_to_Assets𝑖,𝑡 + 𝛽7Gender_Inequalityc,t + 𝜀c,𝑖,𝑡
Finally, the moderating variable is added:
Firm performancei,c,t = α0 + 𝛽1Board_Diversityi,t +
𝛽2(Board_Diversityi,t*Gender_Inequalityc,t) + 𝛽3Firm_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽4GDP_Growthc,𝑡 (7) + 𝛽5Board_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽6Loans_to_Assets𝑖,𝑡 + 𝛽7Gender_Inequalityc,t + 𝜀c,𝑖,𝑡
Risk-takingi,c,t = α0 + 𝛽1𝐵𝑜𝑎𝑟𝑑_𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡 + 𝛽2(Board_Diversityi,t*Gender_Inequalityc,t) + 𝛽3Firm_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽4GDP_Growthc,𝑡 + 𝛽5Board_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽6Loans_to_Assets𝑖,𝑡 (8) + 𝛽7Gender_Inequalityc,t + 𝜀c,𝑖,𝑡
Firm performancei,c,t = α0 + 𝛽1𝑅𝑖𝑠𝑘_Taking𝑖,𝑡 + 𝛽2(𝑅𝑖𝑠𝑘_Taking𝑖,𝑡*Gender_Inequalityc,t) + 𝛽3Board_Diversityi,t + 𝛽4(Board_Diversityi,t*Gender_Inequalityc,t) + 𝛽5Firm_ 𝑆𝑖𝑧𝑒𝑖,𝑡 (9) + 𝛽6GDP_Growthc,𝑡 + 𝛽7Board_ 𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝛽8Loans_to_Assets𝑖,𝑡 + 𝛽9Gender_Inequalityc,t + 𝜀c,𝑖,𝑡
3.3. Descriptive statistics
The descriptive statistics, found in table 1, are based on the total sample of 542 banks. Looking at the mean of the return on average assets (1.289) suggests that the banks in general can manage their assets well and make a profit off them. This result is confirmed by the mean of the return on average equity, which is also positive (11.061). The Z-score (log), which measures the riskiness of the bank, shows that the average bank (3.729) is more stable than the median bank (41.627). Furthermore, the statistics for gender shows that the average bank has 13.2% female board members. However, the total percentage of women on board varies between 62.5% and 0%. When comparing this to the dummy, which takes the value of 1 when there is female presence in the board of directors, it can be seen that on average 66.5% of the banks have at least one female director. So, even though the percentage of women is only 13.2%, most board do have female representation.
at a similar distance from the mean (10.0176 and 19.842 respectively). The mean growth rate of the economy is 3.250%. This might seem high, however, this is due to a relatively higher number of banks in high growth countries such as Turkey and India. The size of the board for the banks in this sample is between 4 and 31 members, with an average of 11.010 members. The mean amount of loans to assets is 58.071% while the median is 61.488. Moreover, for the banks in this sample the average liquid assets to total assets ratio is 18.473% with a median of 15.820%. Finally, the mean gender inequality index is 0.333.
Comparing these numbers to the descriptive statistics of previous studies it becomes clear that the amount of women on the board of directors is higher in this sample. The mean proportion of female board members is 13.3% in this sample compared to 9.6% in Sila et al. (2016) and 8.5% in Adams and Ferreira (2009). The number of boards that has at least on female member is also slightly higher in this sample, 66.7% as compared to 63.8% and 61%. Furthermore, firm size is much lower in the previous study by Sila et al. (2016). The logarithm of assets is 7.398 in the previous study as compared to 15.041 in this study.
Table 1
Summary statistics
This table presents the descriptive statistics of all variables. The data sample consists of 3250 company-year observations of 542 banks over a period of 2012-2017. ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Variable definitions are presented in Table A.1. The variables ROAA and ROAE are winsorised at the 0.01 and 0.99 level.
Variable Mean Min Median Max Standard
deviation N ROAA 1.289 -3.448 1.101 7.314 1.479 2978 ROAE 11.061 -39.296 10.718 47.756 11.819 2978 Z-score (log) 3.729 -1.090 3.739 7.195 0.926 2978 % Female 13.332 0.000 12.000 62.500 13.059 3249 Dummy female 0.667 0.000 1.000 1.000 0.471 3243
Firm size (log) 15.041 10.176 15.196 19.842 2.467 2979
Table 2 shows the correlation matrix of the main variables. This table shows whether multicollinearity is present. Following Miller and Triana (2009), a correlation that is higher than 0.4could be a sign of multicollinearity. Looking at Table 2, only two of the correlations exceed this level. However, since these correlations are between ROAA and ROAE and between Female(%) and the dummy for female presence, multicollinearity will not be an issue since the correlating variables will not be used in the same regression. ROAE will be used as an alternative measure for ROAA, and the same can be said for the female dummy, which is an alternative measure for Female(%). The rest of the correlations are not over the 0.4 level. Therefore, it can be concluded that multicollinearity is not an issue in the dataset.
Table 2
Correlation matrix
This table presents the correlation matrix among the variables used in this research. The data sample consists of 3250 company-year observations of 542 banks over a period of 2012-2017. ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Report. Variable definitions are presented in Table A.1. The variables ROAA and ROAE are winsorised at the 0.01 and 0.99 level.
Variable ROAA ROAE
4. Empirical Results
To investigate the mediating effect of risk-taking in the relationship between board gender diversity and performance, a series of OLS regressions is conducted. First, an OLS regression is conducted to establish whether there is a mediating effect of risk-taking on board gender diversity and performance. Then, these outcomes are re-established using alternative measures and by including time fixed effects in the original model. Finally, the moderating variable is added to study the effect is has on the relations between board gender diversity and risk-taking and on board gender diversity and performance.
4.1. Ordinary Least Squares (OLS)
To identify the relationship between the main variables (board gender diversity, risk-taking, and firm performance), several OLS models are used. This relationship can be affected by endogeneity as a result of omitted variables or reverse causality.
The outcomes of the OLS regressions are found in table 3. As indicated, to establish the mediating effect of risk-taking, four steps need to be taken. The first step, to establish a relationship between the dependent and independent variables (board gender diversity and performance) is depicted in model (1) and model (2). Model (1) shows a negative, highly significant relationship between these variables with a coefficient of -0.704. This outcome means that when the share of female board members increases by 1, performance (ROAA) decreases by 0.704. When adding the control variables in this regression the relationship is no longer significant. However, when removing one of the control variables, namely the GII, the relationship is again significant and negative. The results of this regression can be found in table A.2. These mixed results make it hard to say for sure whether the relationship between board gender diversity and performance really exist. The relationship that exists in the first regression that only includes the dependent and independent changes when adding control variables. This could mean that the gender development index explains the fluctuation in performance, and not board gender diversity. For now, the relationship in model (1) and the relationship from table A.2 will be used as the basis for measuring the mediating relationship and further research is necessary to establish the effect of the GII.
in models (3) and (4) of table 3. In model (3) the variable for the percentage of female board members has a coefficient of -0.206, however is not significant. When adding the control variables to the model, the relationship between the proportion of female board members and the z-score becomes stronger (-0.613) and highly significant at the 1% level. This means that there is a negative relationship. When the proportion of female board members increases by 1, the z-score decreases by 61.3%, the bank becomes riskier. This outcome is different from what theory predicts. However, in the study by Adams and Funk (2012) they find results similar to the results of this study. They claim there is a difference between the general female and a female on the board of directors. Female directors, who have to compete for their board position, are more risk loving. These traits cannot be captured by the measures used in this study, since it only looks at gender.
Table 3
Pooled OLS: Mediating effect of risk-taking on the relationship between board gender diversity and performance
This table presents the results of regression (6) to establish the mediating effect of risk-taking. Columns (1)-(2) represent regressions (1) and (4), columns (3)-(4) represent regressions (1)-(2) and (5), and columns (5)-(6) represent regressions (3) and (6). The dependent variable is ROAA for columns (1), (2), (5) and (6), and the dependent variable is Z-score (log) for columns (3) and (4). ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAA (2) ROAA (3) Z-score (4) Z-score (5) ROAA (6) ROAA
Thirdly, the relationship between the mediating variable and the dependent variable (risk-taking and performance) is to be established while including the independent variable (board gender diversity). This can be seen in model (5) and (6). In model (5) both the measure for risk-taking and the measure for board gender diversity are negative (-0.095 and -0.724 respectively) and both significant at the 1% level. When adding the control variables only the coefficient of the z-score remains significant and changes from a negative effect to a positive one (0.121). These mixed and contradicting results show that there is no strong enough prove to establish the mediating effect of risk-taking on the relationship between board gender diversity and performance. The regression should be tested using alternative measures to gain more insight about the relationship
4.2. Robustness Check
In order to check the results and gain further insight into the relationships, the regressions were re-estimated. This is done to test whether the first series of regressions are accurate. First, alternative measures were used. This means that ROAA was replaced by ROAE and the percentage of female board members was replaced by a dummy indicating female presence in the board of directors. The results of these regressions can be seen in table 4.
Another way to check the results is to include time fixed effects in the model. Fixed effects allow for changes over time, while cross-sectional intercepts stay constant. The time fixed effects model captures time variation in the regression model. The results of this are shown in table 5.
Looking at table 5, the differences with the original model are even smaller. Most coefficients barely changed. Since the changes between both the alternative measures and the original measure are so small, it can be concluded that the results are robust.
Table 4
Pooled OLS: Mediating effect of risk-taking on the relationship between board gender diversity and performance with alternative measures
This table presents the results of regression (6) to establish the mediating effect of risk-taking. Columns (1)-(2) represent regressions (1) and (4), columns (3)-(4) represent regressions (1)-(2) and (5), and columns (5)-(6) represent regressions (3) and (6). The dependent variable is ROAE for columns (1), (2), (5) and (6), and the dependent variable is Z-score (log) for columns (3) and (4). ROAE is the return on average equity, and is calculated as the net income divided by average equity. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. Dummy female is a dummy variable that has the value of 1 when there is at least on female board member. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAE (2) ROAE (3) Z-score (4) Z-score (5) ROAE (6) ROAE
4.3. Moderating effect
Table 6 represents the outcome of regression equations (7)-(9), which include the moderation effect of the GII. Looking at model (1), the variable for the proportion of female board members is significant at the 10% level. With a coefficient of -1.075 an increase of the proportion of female board members by 1 would mean that the ROAA would decrease by 1.075. The moderating variable added to this relationship (%Female*GII) is also significant at the 10% level. Because this variable has a positive coefficient (3.045) it will strengthen the relation between the proportion of female board members and performance. Since it was established this relation is a negative one, the relationship will become more negative because of the moderating effect of GII. However, when comparing at the same relationships in model (3), the
Table 5
Pooled OLS with fixed period effects: Mediating effect of risk-taking on the relationship between board gender diversity and performance
This table presents the results of regression (6) to establish the mediating effect of risk-taking. Columns (1)-(2) represent regressions (1) and (4), columns (3)-(4) represent regressions (1)-(2) and (5), and columns (5)-(6) represent regressions (3) and (6). The dependent variable is ROAA for columns (1), (2), (5) and (6), and the dependent variable is Z-score (log) for columns (3) and (4). ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAA (2) ROAA (3) Z-score (4) Z-score (5) ROAA (6) ROAA
variable for the proportion of female board members is no longer significant. So, there is no relationship to be moderated.
The GII was also added as a moderator for the relationship between the proportion of female board members and risk-taking. These results can be seen in model (2) of table 6. In this model the moderating variable (%Female*GII) is not significant. So, there is no evidence for a moderating effect in this model, and no evidence to support hypothesis 4.
Since removing the control variable GII changed the results for the relationship between gender diversity and performance in previous regressions, this is also tested in this regression. The results of the regression without the control variable GII are shown in model (4) and (5) of table 6. For model (4), which is similar to model (1), both the independent and moderating variable become more significant (1% level). The coefficients do not change in sign, the independent variable remain negative and the moderating variable remains positive. However, the coefficients do become larger. The negative effect of the proportion of female board members becomes -2.813, so an increase in the proportion of women on the board of director by 1 results in a decrease of ROAA by 2.813. The moderating effect also becomes larger, thus the negative relation becomes even more negative. Model (6) also shows this relationship. Again both the independent and moderating variable are significant, both at the 5% level. The coefficients decrease in size, but the sign does not change.
Model (5) also shows different results from model (3). Where there was no significant result for the independent variable in model (2), the variable is significant at the 5% level in model (5). Since the moderating variable is also significant at the 5% level, it can be concluded that based on this model the proportion of female board members negatively impacts the performance of the bank. This effect becomes more negative because of inequality.
The results of models (1)-(3) are also tested when including time fixed effects. These results are found in table A.3. These results are very similar to the results in table 6.
5. Conclusion
The main findings of this study are that female board members decrease performance of a bank. Contrary to what was expected this relationship is not positive but negative. There also is some evidence for a strengthening effect on this relation by gender inequality. Even though the results are mixed, there are some significant results that provide support for this result. It makes the negative impact of board gender diversity on performance even more negative. Finally, not enough evidence is found to establish the presence of the mediating effect of risk-taking on
Table 6
Pooled OLS with moderating effect of GII
This table presents the results of regression (7)-(9) to establish the mediating effect of risk-taking. Columns (1) and (4) represent regression (7), columns (2) and (5) represent regression (8), and columns (3) and (6) represent regression (9). Columns (4)-(6) are excluding the GII control variable. The dependent variable is ROAA for columns (1) and (3), and the dependent variable is Z-score (log) for column (2). ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAA (2) Z-score (3) ROAA (4) ROAA (5) ROAA
board gender diversity and performance. Finally, there is no evidence for a moderating effect of gender inequality on risk-taking.
5.1. Managerial Implications
The negative effect of board gender diversity in banks on the performance means that it might not be beneficial to appoint more women to the board of directors. Especially in countries that have higher gender inequality, since this variable makes the effect of board gender diversity even larger. The flip side is that more people are calling for an increase in the number of women in top-management positions, such as the board of directors ignoring the issue might affect the bank’s reputation. Banks should either make the tradeoff between decreasing their reputation and decreasing performance or find a way to work around this issue. A possible solution could be for banks to look into putting women in other top positions, while maintaining a male oriented board of directors to optimize both reputation and risk-taking factors.
5.2. Limitations
This study also has some limitations. First, since this study is very focused with regards to the industry, it is not generalizable to any other than the banking industry. Since every bank has its own characteristics, it will not work to apply the findings in the banking industry to, in example, the manufacturing industry. The risk-taking measures used in this research are specific to the banking industry, and other measures will be more suitable for other industries.
Next, the measures used in this study do not always fully depict the variable. The z-score is meant for establishing the riskiness of a bank, not the risk-taking. As discussed it is used for this purpose in other studies, this does not mean it is the perfect measure for this variable. So, some information can get lost due to not using a perfect measure. A similar problem occurs with the variables that measure female presence (female % and the dummy variable.). These variables measure based on gender, it does not take into account the personality of the board members. The variable cannot pick up on male traits depicted by female board member or the other way around, pick up on male board members with female traits.
Furthermore, the data on gender diversity was not complete. Many banks had to be deleted from the sample due to this reason. This affected the sample and might have caused some biased results as many of the deleted banks were from the same country, mainly China. Still banks from these countries remain in the sample, but China is a large economy thus the sample should contain more Chinese banks in relation to the total amount of banks to depict reality.
6. References
1. Adams, R. B., & Ferreira, D. 2009. Women in the boardroom and their impact on governance and performance. Journal of financial economics, 94(2), 291-309.
2. Adams, R. B., & Funk, P. 2012. Beyond the glass ceiling: Does gender matter?. Management science, 58(2), 219-235.
3. Altunbaş, Y., & Marqués, D. 2008. Mergers and acquisitions and bank performance in Europe: The role of strategic similarities. Journal of economics and business, 60(3), 204-222.
4. Barber, B. M., & Odean, T. 2001. Boys will be boys: Gender, overconfidence, and common stock investment. The quarterly journal of economics, 116(1), 261-292. 5. Baron, R. M., & Kenny, D. A. 1986. The moderator–mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6), 1173.
6. Bear, S., Rahman, N., & Post, C. 2010. The impact of board diversity and gender composition on corporate social responsibility and firm reputation. Journal of Business Ethics, 97(2), 207-221.
7. Berger, A. N., & Bouwman, C. H. 2013. How does capital affect bank performance during financial crises?. Journal of Financial Economics, 109(1), 146-176.
8. Berger, A. N., Clarke, G. R., Cull, R., Klapper, L., & Udell, G. F. 2005. Corporate governance and bank performance: A joint analysis of the static, selection, and dynamic effects of domestic, foreign, and state ownership. Journal of Banking & Finance, 29(8-9), 2179-2221.
9. Berger, A. N., Hasan, I., & Zhou, M. 2010. The effects of focus versus diversification on bank performance: Evidence from Chinese banks. Journal of Banking & Finance, 34(7), 1417-1435.
10. Berger, A. N., Kick, T., & Schaeck, K. 2014. Executive board composition and bank risk taking. Journal of Corporate Finance, 28, 48-65.
11. Bernasek, A., & Shwiff, S. (2001). Gender, risk, and retirement. Journal of economic issues, 35(2), 345-356.
12. Bonin, J. P., Hasan, I., & Wachtel, P. 2005. Bank performance, efficiency and ownership in transition countries. Journal of banking & finance, 29(1), 31-53.
14. Boyd, J. H., Graham, S. L., & Hewitt, R. S. 1993. Bank holding company mergers with nonbank financial firms: Effects on the risk of failure. Journal of Banking & Finance, 17(1), 43-63.
15. Brooks, C., 2014. Introductory Econometrics for Finance. Cambridge University Press, Cambridge.
16. Byrnes, J. P., Miller, D. C., & Schafer, W. D. 1999. Gender differences in risk taking: A meta-analysis. Psychological bulletin, 125(3), 367.
17. Campbell, K., & Mínguez-Vera, A. 2008. Gender diversity in the boardroom and firm financial performance. Journal of business ethics, 83(3), 435-451.
18. Carpenter, M. A., & Westphal, J. D. 2001. The strategic context of external network ties: Examining the impact of director appointments on board involvement in strategic decision making. Academy of Management journal, 44(4), 639-660.
19. Carter, D. A., D'Souza, F., Simkins, B. J., & Simpson, W. G. 2010. The gender and ethnic diversity of US boards and board committees and firm financial performance. Corporate Governance: An International Review, 18(5), 396-414.
20. Chatman, J. A., & Flynn, F. J. 2001. The influence of demographic heterogeneity on the emergence and consequences of cooperative norms in work teams. Academy of Management Journal, 44(5), 956-974.
21. Cornell, B., & Shapiro, A. C. 1987. Corporate stakeholders and corporate finance. Financial management, 5-14.
22. Costa Jr, P. T., Terracciano, A., & McCrae, R. R. 2001. Gender differences in personality traits across cultures: robust and surprising findings. Journal of personality and social psychology, 81(2), 322.
23. Croson, R., & Gneezy, U. 2009. Gender differences in preferences. Journal of Economic literature, 47(2), 448-74.
24. Cyert, R. M., & March, J. G. 1963. A behavioral theory of the firm. Englewood Cliffs, N. J.: Prentice-Hall.
25. Eckel, C. C., & Grossman, P. J. 2008. Men, women and risk aversion: Experimental evidence. Handbook of experimental economics results, 1, 1061-1073.
26. Eisenhardt, K. M. 1989. Agency theory: An assessment and review. Academy of management review, 14(1), 57-74.
28. Fama, E. F., & Jensen, M. C. 1983. Separation of ownership and control. The journal of law and Economics, 26(2), 301-325.
29. Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organization as a reflection of its top managers. Academy of management review, 9(2), 193-206.
30. Filley, A. C., House, R. J., & Kerr, S. 1976. Managerial process and organizational behavior. Scott, Foresman.
31. Francoeur, C., Labelle, R., & Sinclair-Desgagné, B. 2008. Gender diversity in corporate governance and top management. Journal of business ethics, 81(1), 83-95.
32. Guest, P. M. 2009. The impact of board size on firm performance: evidence from the UK. The European Journal of Finance, 15(4), 385-404.
33. Hillman, A. J., Cannella, A. A., & Paetzold, R. L. 2000. The resource dependence role of corporate directors: Strategic adaptation of board composition in response to environmental change. Journal of Management studies, 37(2), 235-256.
34. Houston, J., & James, C. 1995. CEO compensation and bank risk is compensation in banking structured to promote risk taking? Journal of Monetary Economics, 36(2), 405-431. doi:10.1016/0304-3932(95)01219-2
35. Human Development Reports. 2016. Retrieved May 25, 2018, from http://hdr.undp.org/en/content/gender-inequality-index-gii.
36. Huselid, M. A. 1995. The impact of human resource management practices on turnover, productivity, and corporate financial performance. Academy of management journal, 38(3), 635-672.
37. Iannotta, G., Nocera, G., & Sironi, A. 2007. Ownership structure, risk and performance in the European banking industry. Journal of Banking & Finance, 31(7), 2127-2149. 38. Janis, I. L. 1972. Victims of groupthink. Boston: Houghton Mifflin.
39. Johnson, S. G., Schnatterly, K., & Hill, A. D. 2013. Board composition beyond independence: Social capital, human capital, and demographics. Journal of management, 39(1), 232-262.
40. Laeven, L., & Levine, R. 2009. Bank governance, regulation and risk taking. Journal of financial economics, 93(2), 259-275.
41. Lepetit, L., & Strobel, F. 2013. Bank insolvency risk and time-varying Z-score measures. Journal of International Financial Markets, Institutions and Money, 25, 73-87.
43. Miller, K. D., & Bromiley, P. 1990. Strategic risk and corporate performance: An analysis of alternative risk measures. Academy of Management Journal, 33(4), 756-779.
44. Miller, T., & del Carmen Triana, M. 2009. Demographic diversity in the boardroom: Mediators of the board diversity–firm performance relationship. Journal of Management studies, 46(5), 755-786.
45. Nguyen, T., Locke, S., & Reddy, K. 2015. Does boardroom gender diversity matter? Evidence from a transitional economy. International Review of Economics & Finance, 37, 184-202.
46. Ouchi, W. G. 1979. A conceptual framework for the design of organizational control mechanisms. In Readings in accounting for management control (pp. 63-82). Springer, Boston, MA.
47. Palmer, T. B., & Wiseman, R. M. 1999. Decoupling risk taking from income stream uncertainty: A holistic model of risk. Strategic Management Journal, 20(11), 1037-1062.
48. Pathan, S. 2009. Strong boards, CEO power and bank risk-taking. Journal of Banking & Finance, 33(7), 1340-1350.
49. Pfeffer, J. 1987. A resource dependence perspective on interorganizational relations. In M. S. Mizruchi, & M. Schwartz (Eds.), Intercorporate relations: The structural analysis of business: 22-55. Cambridge, UK: Cambridge University Press.
50. Pfeffer, J., & Salancik, G. R. 1978. The external control of organizations: A resource dependence perspective. New York: Harper & Row.
51. Pearce, J. A., & Zahra, S. A. 1992. Board composition from a strategic contingency perspective. Journal of Management Studies, 29, 411-438.
52. Rhoades, S. A. 1998. The efficiency effects of bank mergers: An overview of case studies of nine mergers. Journal of Banking & Finance, 22(3), 273-291.
53. Rose, C. 2007. Does female board representation influence firm performance? The Danish evidence. Corporate Governance: An International Review, 15(2), 404-413. 54. Sapienza, P., Zingales, L., & Maestripieri, D. 2009. Gender differences in financial risk
aversion and career choices are affected by testosterone. Proceedings of the National Academy of Sciences, 106(36), 15268-15273.
56. Shleifer, A., & Vishny, R. W. 1997. A survey of corporate governance. The journal of finance, 52(2), 737-783.
57. Schubert, R., Brown, M., Gysler, M., & Brachinger, H. W. 1999. Financial decision-making: are women really more risk-averse?. American Economic Review, 89(2), 381-385.
58. Sila, V., Gonzalez, A., & Hagendorff, J. 2016. Women on board: Does boardroom gender diversity affect firm risk?. Journal of Corporate Finance, 36, 26-53.
7. Appendices
7.1. Table A.1
Table A.1
Variable definitions Variable
ROAA Net income divided by average assets
ROAE Net income divided by average equity
Z-score (log) Natural logarithm of the following: ROA plus the capital-asset ratio divided by the
standard deviation of ROA
% Female Number of female board members divided by total number of board members
Dummy female Takes a value of 1 when there is at least 1 female board member and 0 otherwise
Firm size (log) Natural logarithm of total assets
GDP real growth rate GDP growth rate relative to the inflation rate
Number of directors Number of total board members
Loans (%) Total loans divided by total assets
Liquid assets (%) Total liquid assets divided by total assets
GII Represents the gender gap in a country, between 0 (no gap) and 1
7.2. Table A.2
Table A.2
Pooled OLS: Mediating effect of risk-taking on the relationship between board gender diversity and performance
This table presents the results of regression (3) to establish the mediating effect of risk-taking. Columns (1)-(2) represent regression (1), columns (3)-(4) represent regression (1)-(2), and columns (5)-(6) represent regression (3). The dependent variable is ROAA for columns (1), (2), (5) and (6), and the dependent variable is Z-score (log) for columns (3) and (4). ROAA is the return on average assets, and is calculated as the net income divided by average assets. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. %Female is the proportion of female board members. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAA (2) ROAA (3) Z-score (4) Z-score (5) ROAA (6) ROAA
7.3. Table A.3
Table A.3
Pooled OLS with moderating and period fixed effects
This table presents the results of regression (7)-(9) to establish the mediating effect of risk-taking while including GII moderating and period fixed effects. Columns (1)-(2) represent regression (7), columns (3)-(4) represent regression (8), and columns (5)-(6) represent regression (9). The dependent variable is ROAE for columns (1), (2), (5) and (6), and the dependent variable is Z-score (log) for columns (3) and (4). ROAE is the return on average equity, and is calculated as the net income divided by average equity. Z-score is measured as ROA plus the capital-asset ratio relative to the standard deviation of ROA, the natural logarithm of this variable is used. Dummy female is a dummy variable that has the value of 1 when there is at least on female board member. Firm size is measured as the natural logarithm of total assets. GDP real growth rate represents the growth values. Board size is measured as the number of board members. Loans(%) is the amount of loans relative to total assets. Liquid(%) is the amount of liquid assets relative to total assets. GII represents the Gender Inequality Index. The data is retrieved from Orbis Bank Focus and the UN Human Development Reports. Corresponding standard errors are shown in parentheses. ***, ** and * represent statistical significance at the 1%, 5% and 10% level respectively.
(1) ROAA (2) ROAA (3) Z-score (4) Z-score (5) ROAA (6) ROAA
Constant 1.264*** (0.047) 1.802*** (0.401) 3.796*** (0.030) 2.394*** (0.249) 1.206*** (0.139) 1.511*** (0.412) %Female -3.866*** (0.379) -1.073* (0.610) 1.061*** (0.248) -0.665* (0.379) -3.886** (0.381) -0.992 (0.609) %Female*GII 12.042*** (1.027) 3.034* (1.745) -4.912*** (0.672) 0.174 (1.084) 12.123*** (1.042) 3.013* (1.741) Z-score 0.015 (0.035) 0.121*** (0.042) Firm size -0.057*** (0.018) 0.106*** (0.011) -0.070*** (0.019) GDP real growth rate 0.066*** (0.013) 0.036*** (0.008) 0.061*** (0.013) Board size -0.013 (0.008) -0.010* (0.005) -0.011 (0.008) Loans (%) -0.009*** (0.003) 0.005*** (0.002) -0.009*** (0.003) Liquid (%) 0.014*** (0.004) -0.004* (0.002) 0.014*** (0.004) GII 1.540*** (0.354) -1.211*** (0.211) 1.687*** (0.357) R-squared 0.068 0.142 0.028 0.165 0.069 0.147 Adjusted R-squared 0.067 0.136 0.027 0.160 0.066 0141
