1. Introduction
Limiting the risk of default and financial distress should be an important aim of firms managed in the interests of their shareholders as bankruptcy and financial distress are costly to the shareholder.
Almeida and Philippon (
2007) estimate the cost of financial distress to be 4.5% of the pre-distress value of the firm; the impact on shareholders is generally much worse in the event of bankruptcy. However, it has been demonstrated that managers tend to be more risk-averse than their typically well-diversified shareholders and so, unless otherwise incentivized, managers may take actions that reduce firm risk at the expense of firm value (see for example
Amihud and Lev 1981;
Coles et al. 2006; and
Dittmann et al. 2017). The optimal level of default risk is likely therefore to be higher than the default risk preferred by the firm’s managers but not so excessive that it reduces firm value.
There is considerable existing research on gender and risk aversion.
Barber and Odean (
2001) demonstrate that women are more risk-averse than men in their financial and investment decisions.
Croson and Gneezy’s (
2009) literature review claims there to be convincing evidence that women are more risk-averse.
Teodósio et al.’s (
2021) literature review finds that the presence of female executives on top management teams and boards of directors reduces firm risk-taking policies. It might therefore be expected that the higher risk aversion of female executives may also result in lower credit default risk for the firm. However, they find that the presence of female executives on the top management team “has no impact on … Altman Z-Score” based on four studies reviewed that failed to find a linear relationship between the percentage of total executives who were female and z-scores. This study investigates whether the influence of female executives may be a better predictor of credit default risk than simply the presence of female executives on the top management team.
Further, the existing literature suggests that diverse groups make better decisions. These groups leverage their differences to develop more potential alternatives and thus provide more optimal decisions (
Cox and Blake 1991). While research has addressed the link between diversity in the top management team and decision-making, others have focused on improvements in firm performance from those decisions (
Sieweke et al. 2023;
Wu et al. 2022). Diverse top management teams may therefore be better at determining the level of credit risk that optimizes firm value without making bankruptcy likely.
This paper considers whether there is actually an inverse squared relationship between gender influence and credit default risk: if diversity improves group decision-making, then we should expect a concave relationship rather than an ever-decreasing function of female representation. Secondly, this paper considers
gender power (
Walton and Tribbitt 2023), a measure of the relative influence of female executives based on their relative compensation, rather than
gender diversity, the percentage of female executives present on the team.
Walton and Tribbitt (
2023) compare the impacts of gender diversity and gender power on firm risk and firm performance and find they are two distinct constructs, and that gender power may be a stronger predictor of the relationship on firm performance and firm risk alone.
Using a matched sample of 6,652 firm-year observations of 938 publicly traded US firms over the period 2010 to 2020, this paper finds a convex relationship between gender power and Altman Z-scores. This is consistent with gender diverse top management teams voluntarily bearing higher levels of credit default risk than teams where male executives or female executives have greater influence.
Altman (
2013) finds that firms with a Z-score of less than 1.2 have a 91% probability of bankruptcy. This paper finds that the likelihood of a firm being in this high-risk group is lowest when female influence is highest. This is consistent with the greater influence of more risk-averse female executives resulting in firm policies which reduce bankruptcy risk. Boards particularly concerned about bankruptcy may therefore wish to consider having top management teams with more influential female executives.
This paper contributes to the literature by demonstrating a statistically and economically significant concave relationship between credit default risk and gender power on the top management team that allows the realization of an optimal Z-score of the firm. This is consistent with the expected benefits of heterogenous groups making better decisions than homogenous groups. Previous literature has failed to find a relationship between gender representation and Z-scores. This paper also demonstrates that high levels of gender power are associated with a lower probability of a firm being in Altman’s high bankruptcy risk group of firms. Robustness checks show these effects to be independent of the gender of both the Chief Executive Officer (CEO) and the Chief Financial Officer (CFO).
These findings provide support to the existing literature on the impact of female executives on firm outcomes and the composition of groups. The relationship between high levels of female influence on the top management team and the risk of bankruptcy provides a potential new rationale for
Ryan and Haslam’s (
2005,
2007) “glass cliff” phenomenon: our findings are consistent with maximizing the influence of female executives as a rational response to the threat of imminent bankruptcy.
Ryan et al.’s (
2016) survey of research into the “glass cliff” finds that the phenomenon is not universally accepted.
These findings also have practical implications for firms’ management selection decisions. CEOs and boards concerned that executives may implement policies that reduce credit default risk at the expense of firm value can consider adjusting the balance of influence between female and male executives. CEOs and boards sensitive to concerns of bankruptcy might consider increasing the influence of female executives in the top management team.
2. Literature Review and Hypothesis Development
2.1. Managerial Risk Aversion and Agency Costs
There is a long literature considering how differences in the risk aversion of managers and shareholders can result in managers taking actions which are harmful to the firm’s shareholders.
Amihud and Lev (
1981) propose that management’s desire to reduce their own “employment risk” by exploiting the risk reduction benefits of diversification as an explanation for conglomerate mergers. This reduction in risk does not benefit shareholders as they can choose for themselves whether to hold a diversified portfolio and there is a widely acknowledged reduction in firm value, the “diversification discount”, associated with this type of merger (
Lang and Stulz 1994;
Berger and Ofek 1995;
Servaes 1996).
Dittmann et al. (
2017) demonstrate that risk-averse managers will want to reduce firm risk even if it reduces firm value and argue therefore that shareholders will want to incentivize risk-taking.
Coles et al. (
2006) demonstrate that executive compensation plans that reward risk-taking can successfully cause managers to implement riskier corporate policies.
Faccio et al. (
2016) find that female CEOs are more risk-averse than male CEOs and the firms they lead implement policies that avoid risk. This literature suggests that risk-averse managers will prefer to implement policies that result in levels of risk lower than their risk-neutral shareholders would prefer.
2.2. Impact of Diversity on Group Decision-Making
There is a plethora of research considering the impact of diversity within the top management team on firm-decision-making. Utilizing elements of upper echelon theory, along with an information perspective, these scholars suggest that team members with differing characteristics bring unique and novel perspectives that increase the number of alternatives considered by the team. From this position, more optimal decisions might result (
Cox and Blake 1991;
DeBode et al. 2024;
Díaz-Fernández et al. 2019;
Jansen and Searle 2021;
Williams and O’Reilly 1998).
Patrício and Franco (
2022) present a review of literature describing how different dimensions of diversity affect team performance. When examining gender diversity within the top management team,
Huse and Solberg (
2006) suggest that female executives consider issues more thoroughly than their male colleagues; while
Bilimoria (
2000) states that women tend to add diverse ways of thinking based on their listening skills and empathy.
Arora (
2022) suggests that gender diversity can promote healthy debate and improve group problem-solving skills.
If diverse groups do make superior decisions, it might be expected that top management teams where influence is not predominantly exercised by one gender may be more comfortable accepting higher levels of credit default risk than teams where only one gender has significant influence.
2.3. Risk Aversion and Gender
Croson and Gneezy’s (
2009) review of the literature on gender differences in preferences find that women are more risk-averse than men.
Teodósio et al. (
2021) highlight an extensive existing literature demonstrating that female executives reduce corporate risk-taking. Extant research has examined the relationship between female executives in the top management team and firm risk. These papers generally apply upper echelon theory (
Hambrick and Mason 1984) predicting that the firm’s policies and outcomes will reflect the characteristics and experiences of the executives making the decisions: the higher risk aversion of female executives should therefore result in lower firm risk. Many studies find that increasing gender diversity of the top management team or female leadership is associated with lower levels of firm risk (
Baixauli-Soler et al. 2015;
Ozdemir and Erkmen 2022;
Perryman et al. 2016;
Saeed et al. 2024;
Faccio et al. 2016). It might therefore be expected that increased gender power on the top management team will reduce credit default risk and reduce the probability that the firm is classified as being at high risk of bankruptcy according to Altman’s Z-score.
However, there are some potentially conflicting results.
Kolev and McNamara (
2022) find no evidence to support or reject the hypothesis that gender diversity impacted firm risk.
Berger et al. (
2014) find that the portfolio risk at German banks increases with the addition of female executives. Similarly,
Scarlata et al. (
2024) find that, as the share of females on a decision-making team increases, the risk profile of investments of venture philanthropy firms also increases.
These apparently contradictory results are consistent if there is in fact a non-linear relationship between gender power and risk outcomes. Risk outcomes will improve as female influence moves closer to the optimal level of gender power and then deteriorate as further increases in female influence increase gender power away from its optimal level.
2.4. Measurement of Gender Power
The studies cited above measure gender diversity as the percentage of female executives in the top management team.
Mooney (
2022) highlights that the mere presence of female executives on the top management team does not necessarily result in their influencing firm decisions.
Walton and Tribbitt (
2023) aim to better understand the impact of female executives by differentiating between the impact on firm risk and firm performance of gender diversity (the percentage of female executives within the top management team), and the impact of gender power (the influence of female executives within the top management team).
Walton and Tribbitt (
2023) measure gender power as the percentage of the top management team’s total compensation paid to female executives. This is motivated by measures of individual executives’ power in the prior literature.
Finkelstein (
1992) uses relative compensation as a proxy for the structural power of individual executives.
Bebchuk et al. (
2002) argue that an executive’s compensation reflects the executive’s power (the “managerial power explanation”).
Walton and Tribbitt (
2023) apply this explanation to argue that an executive’s share of total compensation is a reflection of the executive’s relative influence and the share of compensation paid to female members of the top management team therefore reflects their relative influence in the top management team. Consistent with this interpretation, they find that gender power and gender diversity have distinct and sometimes opposite effects on systematic risk (measured by beta) and firm performance (measured by return on assets).
This study applies
Walton and Tribbitt’s (
2023) measure of gender power to examine the relationship between the influence of female executives within the top management team and the firm’s credit default risk.
2.5. Measuring Credit Default Risk and Probability of Bankruptcy
Altman (
1968) proposed a measure of financial distress based on the interaction of five financial ratios calculated from data available from the firm’s financial statements.
Altman (
1983) proposes revised versions of the Z-score so they could be applied to both public and private firms and would have improved accuracy across industries. The Altman Z-score has been determined to be a reliable determinant of such a firm default (for example,
Altman et al. 2017). The Z-score is negatively related to credit default risk: higher values are associated with lower risk. The Z-score is used to group firms into: (1) firms that are highly likely to declare bankruptcy; (2) firms that are very unlikely to declare bankruptcy; and (3) all other firms (the “gray zone”). In
Altman’s (
2013) sample, 91% of firms with a Z-score below 1.2 subsequently declared bankruptcy and 97% of firms with a Z-score above 2.9 did not subsequently declare bankruptcy.
The Z-score therefore reflects credit default risk, but a firm with a score greater than 1.2 does not necessarily have an excessive risk of imminent bankruptcy. Top management can implement policies that simultaneously reduce the firm’s Z-score (in other words, increase its default risk) and increase firm value. For example, reducing working capital, increasing payout ratios and increasing financial leverage can all improve firm value but each will reduce the firm’s Z-score. It is therefore unlikely that the policies that maximize the firm’s Z-score (thereby minimizing its default risk) will also maximize firm value. A top management team working in the best interests of the shareholders should manage these trade-offs to maximize firm value without being at high risk of default.
Since the original Altman Z-score (
Altman 1968) was first proposed, many alternative models for predicting financial distress have been developed, for example, based on hazard models, logistic regressions, and non-parametric methods (see
Almaskati et al. 2021, for a review of the main categories of bankruptcy prediction models). This paper uses the revised version of the Z-scores described in
Altman (
1983) as they are simple to calculate with readily available data, widely used and easily understood by both academics and practitioners. Furthermore, the Z-score continues to perform well in predicting bankruptcy (
Altman et al. 2017;
Almaskati et al. 2021).
2.6. Hypotheses
Based on the above review, this paper will test the following hypotheses concerning gender power in the top management team, credit default risk, and probability of bankruptcy.
H1. There is a concave relationship between female influence within the top management team and the credit default risk of the firm.
If diversity improves group decision-making, we would expect that more diverse top management teams will better manage the trade-off between credit default risk and firm value resulting in higher levels of default risk when top management teams are not dominated by one gender. As there is a negative relationship between credit default risk and Z-scores, if this hypothesis is correct, we expect to see a convex relationship between gender power and Z-scores with Z-scores being highest at the lowest and highest levels of gender power.
H2. Increases in the influence of female executives within the top management team will reduce the probability of a firm being classified as high default risk.
As women are on average more risk-averse than men, we expect that the probability of firms being at high risk of bankruptcy will decrease as the influence of female executives increases. If this hypothesis is correct, we should expect to see a negative relationship between gender power and the probability of having a Z-score of less than 1.2.
3. Methodology
3.1. Data Samples
The hypotheses were tested using samples created from Compustat’s Execucomp database which contains compensation data of key executives of publicly traded American firms from their DEF14A annual proxy filings with the Securities Exchange Commission (SEC).
Our complete sample consists of all observations of firms in the S&P1500 during the period 2010 to 2020, excluding financial firms and utilities which are typically highly regulated. Firm-year observations with missing data for the variables considered were also excluded. This resulted in 10,118 firm-year observations and 1,014 unique firms. The key data for the complete sample are summarized in
Table 1a Summary Statistics – Complete Sample.
During this period, female representation amongst the named executive officers was relatively rare; there was: a female Chief Executive Officer (CEO) in only 4.2% of firm-year observations, a female Chief Financial Officer (CFO) in only 11.6% of firm-year observations, and at least one female named executive officer in only 40.8% of firm-year observations. To allow for systematic differences between firms with no female named executive officers and firms with female named executive officers and to reduce endogeneity issues, we constructed a
matched sample. For every firm-year observation with at least one female named executive officer, we selected an observation with no female named executive officers with the same two-digit Standard Industrial Code (SIC), from the same year and with the closest size as measured by total assets. The matched sample was created with replacement. The resulting matched sample consists of 6,652 firm-year observations and 938 unique firms. The key data for the matched sample are summarized in
Table 1b Summary Statistics – Matched Sample.
Table 1c shows the differences-in-means for economic characteristics between firms with no female named executive officers (gender diversity = 0%) and firms with at least one female named executive officer (gender diversity > 0%). In the complete sample, there are large and statistically significant differences between the two groups. On average, firms with female named executive officers are: larger (as measured by total assets and sales); have lower EBIT/total assets (X3) and shareholders’ equity/total liabilities (X4) ratios; and lower total asset turnover ratios (X5). The matching process succeeded in reducing the systematic differences between the two groups: the differences-in-means in the matched sample are both lower and less statistically significant than in the complete sample.
3.2. Variables of Interest
The key variable of interest is
gender power, which is calculated, following the methodology of
Walton and Tribbitt (
2023), as the percentage of total compensation of a firm’s named executive officers paid to female executives.
Walton and Tribbitt (
2023) argue that an executive’s value to the firm and hence their relative influence within the firm will reflect the compensation paid to the executive. In order to test whether the impact of the influence of female executives is continuously increasing or whether there is a trade-off as gender diversity decreases, we created the variable
gender power squared which, as the name suggests, is calculated as the square of the gender power variable.
Table 1a shows that, due to the high proportion of firms with no female named executive officers, for the complete sample the mean
gender power is only 7.9% and the distribution is heavily skewed. By construction, the matched sample is less skewed and has a mean
gender power of 11.9%.
All these variables of interest were calculated using data from Compustat’s Execucomp.
3.3. Dependent Variables
To measure credit default risk, we primarily used the 1983 revised version of Altman’s Z-score, which can be applied to both public and private firms, the
Z’-score (1983).
Altman (
2013) details the calculation of the revised Z-score and finds it to accurately predict bankruptcy: 91% of firms in the high-risk group subsequently declared bankruptcy and 97% of low-risk firms did not declare bankruptcy. Z-scores are inversely related to credit default risk: credit default risk decreases as the Z-score increases. As well as the
Z’-score (1983), we consider the indicator variables
High Risk (which has a value of 1 if the
Z’-score (1983) is less than 1.2, suggesting a high probability of bankruptcy, and 0 otherwise) and
Low Risk (which has a value of 1 if the
Z’-score (1983) is greater than 2.9, suggesting a low probability of bankruptcy, and 0 otherwise).
Table 1a and
Table 1b show that the mean
Z’-scores (1983) in the complete sample and matched sample are 2.11 and 2.14, respectively, with about 20% of observations classified as
High Risk and 25% classified as
Low Risk in both samples.
When investigating how gender power influences Z’-score (1983), we also consider each of the five ratios used to calculate the Z’-score (1983): X1 (Working Capital/Total Assets), X2 (Retained Earnings/Total Assets), X3 (Earnings Before Interest and Taxation/Total Assets), X4 (Shareholders’ Equity/Total Liabilities), and X5 (Sales/Total Assets).
Alternative versions of Altman’s Z-score are available (
Altman 1968,
1983). The
Z”-Score (1983) can also be applied to both public and private firms and was developed to improve applicability across industries. The
Z”-score (1983) is calculated using only X1, X2, X3, and X4 (details are shown in
Appendix A). It excludes X5 “because of a potential industry effect that is more likely to take place when this kind of industry-sensitive variable (asset turnover) is included” (
Altman et al. 2017). Our panel data regressions essentially control for such industry effects and so we use the informationally superior
Z’-score (1983). For robustness, we also include the results of regressions using both the original
Z-score (1968) and the
Z”-score (1983). These are shown in models 4 and 5 in
Table 2.
All of the variables were calculated using data from Compustat’s Fundamentals Annual database, which is constructed based on firms’ annual 10-K filings.
3.4. Control Variables
To address concerns that relationships between gender power and measures of default risk are actually driven by gender diversity or the influence of female CEOs or female CFOs, we also control for these factors.
Gender diversity is measured as the percentage of female named executive officers.
Female CEO is an indicator variable with a value of 1 if the CEO is female and 0 otherwise. This variable is included, as the CEO has overall executive responsibility for the firm and gender differences in risk aversion may result in different actions in firms led by female CEOs.
Female CFO is an indicator variable with a value of 1 if the CFO is female and 0 otherwise. This variable is included, as the CFO has particular responsibilities relating to financial policies.
Doan and Iskandar-Datta (
2020) find firms with female CFOs follow distinct financial policies associated with a reduction in agency costs.
Wang and Fung (
2022) find that female CEOs and female CFOs follow distinct policies that impact firm risk and firm value. These control variables allow us to separate the effects of these related variables.
Altman (
1983) highlights that firm age will impact the components of the Z-score; in particular, young firms are likely to have lower levels of retained earnings. To control for the impact of firm age on the Z-score and its five components, we include the variable
Log Firm Age which is calculated as the natural log of the firm’s age. Firm size is also likely to impact the probability of default as well as the components of the Z-score (total assets is in the denominator of
X1,
X2,
X3, and
X5). To control for any such impact, we include the variable
Firm Size (calculated as the natural log of the firm’s total assets).
Default risk and the components of the Z-score will likely be affected by the economic cycle. Most obviously, sales, operating profits, and total liabilities will respond to economic growth and recessions. We have included year-fixed effects in each of our regressions to control for such impacts.
3.5. Testing Hypothesis 1
To test whether there is a concave relationship between female influence on the top management team and credit default risk, we performed a panel data regression of the following equation:
If Hypothesis 1 is correct, then there will be a convex relationship between gender power and Z-scores, so β2 will be positive. β1 may be positive, negative, or zero but gender power is included in the regression so that the total effect of changes in gender power can be calculated. Controls include Female CFO in order to distinguish between the effects of the CFO’s gender on credit default risk and the effects of gender influence across the whole top management team. As explained above, Log Firm Age and Firm Size are also included as the literature suggests they should also directly impact Z-scores. Year fixed effects are included in all regressions to take account of differences in economic conditions between years that will affect all firms similarly (such as demand and interest rates). Firm fixed effects are included in order to reflect time-invariant differences between firms that are not controlled for in the regression. All estimates are clustered by firm and, to allay concerns about potential heteroskedasticity or within-panel serial correlation, p-values are calculated using robust standard errors.
3.6. Testing Hypothesis 2
To test whether differences in risk aversion between the genders result in a negative relationship between female influence on the top management team and the probability of the firm being classified as being at high risk of bankruptcy, we created a linear probability model using a panel data regression with fixed effects clustered by firm. We used a linear probability model rather than a logistics regression as we wanted to be able to calculate the total effect of changes in gender power. The regression was performed on the equation below:
If hypothesis 2 is correct, then β1 will be negative indicating that the probability of being considered high risk decreases as the female influence on the top management team increases. If the relationship is continually decreasing, then β2 will not be statistically different from zero. Similar controls were used as for the Z-score regression and for the same reasons explained above. Year fixed effects are included in all regressions to take account of differences in economic conditions between years that will affect all firms similarly (such as demand, interest rates, etc.). Firm fixed effects are included in order to reflect time-invariant differences between firms that are not controlled for in the regression. All estimates are clustered by firm and, to allay concerns about potential heteroskedasticity or within-panel serial correlation, p-values are calculated using robust standard errors.
4. Results
4.1. Credit Default Risk and Gender Influence
The results of the first set of regressions are shown in
Table 2. Models 1 through 3 show the results for regressions of Z-scores using the key variables of interest and control variables. The dependent variable in models 1, 2, and 3 is the
Z’-score (1983). To confirm that our results are consistent with alternative versions of the Z-score, models 4 uses the
Z”-score (1983) and model 5 uses the original
Z-score (1968).
In all models, the estimated coefficients for
gender power squared are positive and both economically and statistically significant. This is consistent with hypothesis 1: there is a concave relationship between credit default risk and female influence on the top management team. The estimated coefficients for
gender power are negative and also both economically and statistically significant.
Figure 1 below shows the combined impact on the Z-score of changing
gender power (and hence
gender power squared) when all other variables in models 1 and 2 are held constant at their mean values. As hypothesized, the Z-score decreases as
gender power increases above zero but then begins to increase as
gender power increases above the local minimum. The local minimum Z-score is 1.87 when
gender power is 48%. As credit default risk is inversely related to the Z-score, this is consistent with the hypothesis that there is a concave relationship between credit default risk and the influence of female executives on the top management team.
As a robustness check, model 2 includes gender diversity (the percentage of female executives on the top management team) and gender diversity squared (the square of gender diversity) to confirm that the identified effect is actually due to the relative power of female executives rather than simply the number of women present within the top management team. As discussed above, the existing literature has found an association between gender diversity and firm risk. Consistent with female executive power being the driver of this effect, the estimated coefficients to gender diversity and gender diversity squared are not significantly different from zero, suggesting that gender diversity does not have a direct impact on credit default risk. As described above, the existing literature has also found a relationship between having a female CEO and policies related to firm risk. Model 3 includes the Female CEO indicator variable as a control in order to identify any separate impact from that of gender power. Consistent with our hypothesis, the estimated coefficients for the gender power variables are still significant and the estimated coefficient for the Female CEO is not significantly different from zero. The estimated coefficients for gender power and gender power squared in models 4 and 5, which are alternative specifications of Altman’s Z-score, are also consistent with hypothesis 1.
In summary, the results of the regressions summarized in
Table 2 are all consistent with hypothesis 1 supporting the claim that there is a concave relationship between gender power on the top management team and the firm’s credit default risk.
4.2. Probability of Being at High Risk of Bankruptcy and Gender Power
Table 3 shows the results of the Linear Probability Model regression. Model 1 shows the estimated coefficients when
High Risk is the dependent variable. Model 2 shows the results of regressions using
Low Risk as the dependent variable. Both models use robust standard errors.
The estimated coefficients for
gender power in model 1 are positive and statistically significant. This result by itself is inconsistent with the hypothesis that the higher risk aversion of female executives should result in a negative relationship between gender power and the probability of being in the high-risk-of-bankruptcy group of firms. However, the estimated coefficient for
gender power squared is negative and statistically significant, suggesting an overall concave relationship between gender power and the probability of the firm being in the high-risk group.
Figure 2 shows the combined impact of changes in gender power. Once again, we see a non-linear relationship. The graph suggests that at low levels of gender power, the probability of a firm being in the high-risk group actually increases but at very high levels of gender power (above 60% for this sample), an increasing influence of female executives is associated with a lower probability of the firm being at high risk of imminent bankruptcy. This combined effect is consistent with the higher risk aversion of female executives resulting in policies and actions which reduce the likelihood of imminent bankruptcy.
This finding is consistent with our first hypothesis. There is a positive impact of power being shared across a diverse group: executives accept a tradeoff between credit default risk and firm value (the benefits of diversity), but the higher risk aversion of female executives reduces the probability of bankruptcy for firms when power is concentrated amongst female executives.
Model 2 is included to understand whether the higher risk aversion of female executives also increases the probability of a firm being in the low-risk group of firms very unlikely to declare bankruptcy. The estimated coefficients for neither of the gender power measures are statistically significant, suggesting that the relative power of female and male executives has no impact on the probability of being in the low-risk group. This is a noteworthy result. It may be that when management teams are in exceptionally low-risk environments, it does not matter how power is distributed between risk-averse and non-risk-averse executives and there is no benefit from gender diversity: if a firm has a low risk of bankruptcy, the marginal effect of policies associated with lower default risk may be negligible. This is speculation, but understanding the reason behind this finding would be an interesting topic for future research.
4.3. Robustness and Endogeneity Concerns
There are always legitimate concerns about potential endogeneity in corporate finance research. This paper has aimed to reduce such concerns by: controlling for variables known to impact credit default risk; using firm fixed effects to control for time-invariant unobserved influences; using year fixed effects to control for unobserved influences that impact all firms but vary by year; and creating a sample by matching firm-year observations with no female executives that are most similar to the firm-year observations with some female executive representation. Nevertheless, although this paper finds convincing support for a concave relationship between gender power and credit default risk that is consistent with finance and organizational theories, it may not be a causal relationship.
To better understand the nature of the relationship, we therefore performed further panel regressions with the five variables used to calculate Z-scores (X1, X2, X3, X4, and X5) as dependent variables and the gender power variables as potential independent variables.
Table 4 summarizes the estimated coefficients for these regressions. The results suggest that there is a statistically significant relationship between
gender power or
gender power squared and only X2 (calculated as retained earnings/total assets) and X4 (calculated as the book value of shareholders’ equity/total liabilities).
Model 4 on
Table 4 suggests a convex relationship between X4 and
gender power with the highest values being where
gender power is very low or very high with a local minimum when
gender power is at 52% (in other words, power is about evenly shared between male and female executives). As X4 is inversely related to the debt-to-equity ratio (based on book values), it seems more likely that the gender power of the top management team influences the firm’s financial leverage than the other way around. Financial leverage increases the firm’s systematic and credit default risk, particularly at high levels. But financial leverage also increases expected equity returns. As theorized above, this is the sort of tradeoff that would be better supported by a heterogeneous top management team than a homogeneous team.
Table 4 shows that for model 2 only the estimated coefficient for
gender power squared is positive and statistically significant. If the coefficient for
gender power is indeed not significantly different from zero, then over the range of feasible values (0% to 100%), X2 will be a continuously increasing function of gender power. This is consistent with the probability of a firm being high risk being lowest at high levels of gender power.
5. Discussion
This study aims to better understand how gender power within the top management team might be related to a firm’s credit default risk. Utilizing a matched sample of 6,652 firm-year observations of American publicly traded firms between 2010 and 2020, we find a curvilinear relationship between the influence of female executives within the top management team and credit default risk. This is consistent with the exploitation of a trade-off between firm value and credit default risk by top management teams where power is more evenly shared between male and female executives.
Specifically, we find that the Z-score of a firm decreases as the influence of female executives begins to increase within the top management team. This direction indicates an increasing likelihood of credit default, but as the influence of these female executives continues to increase, we see a shift in the Z-score in a positive direction. We find this inflection point for this sample to occur when power is roughly evenly shared between male and female executives.
Further, we find an overall decrease in the probability of a firm being at high risk of bankruptcy as the gender power within the top management team increases, but that this probability first increases at lower levels of female gender influence. The overall negative association between the influence of female executives and the probability of being at high risk of imminent bankruptcy supports the notion that the risk aversion of female executives can significantly reduce the risk profile of the firms that they manage and thus provides a rationale for the phenomenon of the “glass cliff”. Interestingly, we find no significant relationship between gender power and the probability of firms falling within the low-risk range of Altman’s Z-score. This is an interesting area of future research to determine if other factors may impact this relationship.