Introduction

We examine whether gender-diverse boards prioritize product market concerns over capital market incentives when proprietary costs are high. The product market competition literature discusses the trade-offs between capital market incentives and product market concerns regarding the disclosure of proprietary information (Li, 2010).Footnote 1 The cost of disclosing proprietary information (referred to as a proprietary cost) is that competitors strategic response to the good news may harm the issuing company’s competitive position (Li, 2010) in the product market and hence is a matter of concern (referred to as a product market concern). On the other hand, the benefit of disclosing proprietary information is a short-term increase in the firm’s stock valuation (referred to as a capital market incentive) because proprietary information is value-relevant in the capital market (Boone et al., 2016), and its nondisclosure results in an adverse selection problem for firms (Verrecchia & Weber, 2006).Footnote 2 In a nutshell, in the short term, there is a conflict between product market and capital market incentives relating to the nondisclosure of proprietary information when firms face high competitive pressure.

Studies on corporate disclosures identify proprietary cost as a disincentive for voluntary disclosure (Li, 2010) and show that firms facing high proprietary costs reduce voluntary disclosures to avoid potential losses from knowledge spillovers (Guo et al., 2004; Li, 2010). Therefore, firms benefit via the redaction or decreased disclosure of proprietary information that is harmful within their product markets (Glaeser, 2018; Verrecchia & Weber, 2006). While the focus of prior research has been on proprietary cost-based motives for the redaction of information (Boone et al., 2016; Glaeser, 2018; Verrecchia & Weber, 2006), a recent research provides evidence on the agency cost-based motive (Bao et al., 2022). Bao et al. (2022) find that managers file confidential treatment (CT) requests with the Securities and Exchange Commission (SEC) to opportunistically redact information from material contract filings to hide bad news. They show that the opportunistic redaction of information has adverse consequences for firms, such as a higher stock price crash risk and poorer future performance, and that this effect is more pronounced for firms with lower external monitoring (Bao et al., 2022). A different strand of studies find that gender-diverse boards provide effective monitoring (Adams & Ferreira, 2009; Adams et al., 2010) and suggest that they are more likely to promote and protect shareholder interests (Lai et al., 2017; Levi et al., 2014). However, prior studies do not consider the role of gender-diverse boards in examining the trade-off between nondisclosure and disclosure of proprietary information in maximizing product market or short-term capital market incentives, respectively. We are, therefore, motivated to fill this gap in the literature.

An important fiduciary duty of the board, which is related to loyalty, is the confidentiality of information (Kenton, 2023). We argue that corporate boards can benefit from female directors when proprietary costs are high because their personality traits are different from those of male directors. More specifically, women are more ethical (Arlow, 1991; Deshpande, 1997) and risk-averse (Riley Jr and Chow 1992; Sunden & Surette, 1998) than men. Thus, we argue based on the theory of groupthink (Janis, 1972) that gender-diverse boards limit groupthink in the decision-making processes of corporate boards (Kamalnath, 2017). We predict that gender-diverse boards ethically and carefully redact or reduce the disclosure of proprietary information when proprietary costs are high in the following ways: (1) they ethically use CT orders to redact proprietary information that must be disclosed and (2) voluntarily disclose lower levels of proprietary information than all-male boards. In the short term, this would result in higher adverse selection for firms with gender-diverse boards. Nevertheless, adverse selection problems will presumably be resolved in the longer run when nondisclosures report superior performance as a result of their enhanced competitive advantage.

Our research is relevant and important because new rules adopted by the SEC in 2019 permit firms to “file redacted material contracts without applying for confidential treatment” (Bao et al., 2022, 30). Previously, firms were required to apply for CT orders before they could redact information from material contracts (Bao et al., 2022). The new rule increases concerns that firms may redact information to hide bad news. Our research provides important insights for regulators, firms, and shareholders by examining an antecedent for the ethical nondisclosure of proprietary information that is gender-diverse boards.

The focus of the existing literature on board gender diversity has been on reducing information asymmetry/improving transparency by increasing voluntary disclosures of firm-specific information (Gul et al., 2011) and, hence, favors capital market incentives. It is possible that the finding that gender-diverse boards increase voluntary disclosures (Gul et al., 2011) may relate to nonproprietary information or low proprietary cost settings. For instance, Glaeser (2018) shows that firms substitute reduced disclosure of proprietary information by increased voluntary disclosure of nonproprietary information when they heavily rely on trade secrecy. Huang et al. (2017) provide evidence that competition from existing rivals reduces voluntary disclosure due to high proprietary costs. Therefore, it remains empirically unclear whether firms with gender-diverse boards continue to prioritize capital market incentives by increasing corporate disclosures more than similar firms with all-male boards when proprietary costs are high.

Using a matched sample of United States (US)-listed firms with gender-diverse boards and all-male boards, we find that gender-diverse boards are associated with higher adverse selection costs when proprietary costs are high. We use the presence and proportion of female directors on the board to measure board gender diversity. Using the proportion of female directors addresses the issue of tokenism (Kanter, 1977).Footnote 3 We use product market competition and CT orders as proxies for high proprietary costs. As a robustness check, we also use an exogenous increase in competition. We address endogeneity issues such as self-selection bias by using a matched sample, correlated variable problem by controlling for board governance, and other endogeneity concerns by also using a firm-fixed-effects model that controls for time-invariant firm characteristics. Further, we find that, in the long run, firms with gender-diverse boards (compared to firms with all-male boards) that initially experience high adverse selection when their firms face high competition are rewarded with higher buy-and-hold abnormal returns.

Our results are economically significant. The results suggest that firms with gender-diverse boards experience about a 78 to 102 basis point increase in adverse selection costs above the sample average when their firms face high competition relative to similar firms with all-male boards. Further results suggest that firms with gender-diverse boards that initially experience adverse selection in a competitive environment experience 11.06% higher long-run buy-and-hold abnormal returns than similar firms with an all-male board. Therefore, the net gain for firms with gender-diverse boards is around 10% of their stock price in three years, despite the short-term adverse selection costs of about one percent of the stock price.

We make the following contributions. First, we contribute to the corporate disclosure literature. Prior research documents that firms facing competition from existing rivals reduce voluntary disclosure due to high proprietary costs (Huang et al., 2017). Although the decision to disclose proprietary information is considered strategic and ultimately requires the approval of corporate boards (Adams & Ferreira, 2007),Footnote 4 the role of boards in such decisions is unclear from prior studies. Recently, Bao et al. (2022) provide evidence that managers redact information from material contracts for opportunistic reasons (agency cost-based motive) to hide bad news. Jones (2007) concludes that managers of firms facing high proprietary costs disclose substantial amounts of proprietary information about their firm’s research and development activities. Further, Boone et al. (2016) provide evidence that initial public offering firms that redact proprietary information experience greater underpricing. We contribute by identifying board gender diversity as a determinant of a firm’s decision for nondisclosure of proprietary information for proprietary cost-based motives. To the best of our knowledge, our paper is the first to document this evidence.

Second, we contribute to the board gender diversity literature by documenting evidence that gender-diverse boards prioritize product market concerns over capital market incentives when proprietary costs are high to maximize long-term shareholder wealth. We add to prior research (Gul et al., 2011) that suggests that gender-diverse boards increase disclosures in favor of capital markets by showing that gender-diverse boards do not unconditionally increase disclosures. Our results suggest that gender-diverse boards refrain from disclosing proprietary information when their firms face high proprietary costs. As proprietary information is not disclosed when proprietary costs are high, firms with gender-diverse boards experience higher adverse selection than similar firms with all-male boards in the short term. However, these firms benefit from shareholder value creation and, hence, higher buy-and-hold abnormal returns in the long term.

Third, our findings have implications for regulators concerning the benefits of gender-diverse boards. Our research suggests that gender-diverse boards ethically and carefully redact information from material contract filings for proprietary cost-based motives as opposed to agency cost-based motives. Bao et al., (2022, 32) suggests “that firms use redaction to hide bad news. The new rules that permit redaction without formal confidential treatment requests may provide managers with more opportunities to conceal bad news by redacting information from material contracts.” Our work extends Bao et al. (2022) by revealing the importance of gender-diverse boards to firms’ decisions to ethically and carefully redact information from material contract filings for proprietary cost-based reasons. Also, gender-diverse boards can provide effective monitoring to curb managerial opportunism in relation to the use of redactions to hide bad news.

Furthermore, our research findings have implications for firms and shareholders in that they contribute to the debate on gender balance in the boardroom. More specifically, our findings suggest that gender-diverse boards ethically maintain the confidentiality of proprietary information for proprietary cost reasons as opposed to opportunistic reasons. As such, firms experience a short-term adverse selection cost in the capital markets. Nevertheless, firms benefit in the long run through shareholder wealth maximization, as board gender diversity helps create a competitive advantage. Our research, therefore, advocates the importance of board gender diversity for creating competitive advantage by prioritizing product market concerns over short-term capital market incentives. This finding suggests that reducing board gender diversity will be counterproductive for firms in highly competitive settings. Therefore, firms should be more supportive of gender-diverse boards.

The remainder of our paper is organized as follows. We review the relevant literature and develop our hypothesis in Sect. Literature Review and Hypothesis. Next, we discuss our research design and variable measurement in Sect. Research Design and Variable Measurement. Then, we discuss data, sample selection procedures, and descriptive statistics in Sect. Data, Sample Selection Procedures, and Descriptive Statistics. Subsequently, we discuss our main findings and additional results in Sects. Main Result and Additional Analyses, respectively. Finally, we conclude in Sect. Conclusion.

Literature Review and Hypothesis

Board Gender Diversity and Proprietary Information

Extant literature suggests that diversity at both individual and group levels is associated with a greater knowledge base, increased creativity, innovation, and quality strategic decision-making (Erhardt et al., 2003). As such, diversity is perceived as a competitive advantage for firms (Richard, 2000; Watson et al., 1993). Prior studies relate diversity to boards of directors and suggest that board diversity promotes value creation through the development of new products and markets and increases sensitivity to threats and opportunities from external environment (Certo et al., 2001; Filatotchev & Bishop, 2002; Ray, 2005).

More recent research applies the arguments concerning board diversity to the gender diversity of boards and argues that greater board gender diversity increases a firm’s competitive advantage (Campbell & Mínguez-Vera, 2008). Prior studies suggest that female directors have a greater impact on strategic planning; help their board execute its strategic function (Fondas, 2000); deal effectively with diversity in labor and product markets (Bilimoria & Wheeler, 2000; Mattis, 2000); have a greater understanding of consumer behavior, customer needs, and opportunities for companies to meet those needs (Kang et al., 2007); and are more innovative (Torchia et al., 2011). These attributes of female directors enable them to have a greater impact on strategic planning and contribute to board discussions, which lead to product innovations and other forms of creativity, giving their firm a competitive edge over similar firms with all-male boards. Hence, firms with gender-diverse boards have a competitive advantage because they possess substantial proprietary information generated through their value creation activities.

Decreased Disclosure of Proprietary Information and Adverse Selection Cost

In the short term, there is a trade-off between product market incentives and capital market incentives concerning the disclosure of proprietary information. Li (2010) states that the capital market incentive is to decrease the cost of capital or increase stock valuation, and the product market concern is that disclosures in favor of capital markets may adversely affect a firm’s product market competitiveness. As such, it is beneficial for firms to redact or decrease the disclosure of proprietary information that is harmful to their product markets (Glaeser, 2018; Verrecchia & Weber, 2006).

The US SEC allows firms to redact proprietary information from material contracts (by filing CT orders) to protect themselves against competitive threats if the information is proprietary (Bao et al., 2022). Prior studies show that firms that face high competition are more likely to redact the disclosure of proprietary information (Verrecchia & Weber, 2006) and that decreased disclosure of proprietary information reduces corporate transparency (Glaeser, 2018). Verrecchia and Weber (2006) also provide evidence that the redaction of proprietary information by firms facing high competition increases adverse selection because it provides a refuge or “safe harbor” for firms with “bad news” to pool with firms with “better news” by appealing to the rationale that the news is proprietary.

It follows that proprietary information is useful for valuing a firm’s future prospects (Verrecchia 2001), and its nondisclosure is likely to increase information asymmetry between investors and managers (Glaeser, 2018). Investors may not be able to determine the value of the undisclosed proprietary information and adversely select firms for investments. Although the nondisclosure of proprietary information would result in adverse selection for firms in the short term, these firms would benefit from preserved competitive advantage, and hence, be rewarded with shareholder value creation via higher returns in capital markets in the long term.

Board Gender Diversity, Decreased Disclosure of Proprietary Information, and Adverse Selection Cost

Consistent with groupthink theory (Janis, 1972), our study contends that board gender diversity limits groupthink in the decision-making processes of corporate boards (Adams et al., 2010; Kamalnath, 2017). Groupthink occurs when individuals’ intent to maintain group loyalty overrides all other factors, such as moral considerations, independent thinking, and effective decision-making. One of the consequences of groupthink is that it results in unethical decisions.Footnote 5 According to Adams et al. (2010), female directors are more inclined to display independent thinking because they are not part of the old boys’ network.

Directors’ fiduciary duty to a firm includes loyalty (i.e., ensuring fair processes and avoiding conflicts of interest) and care (i.e., paying attention and trying to make good decisions: the business judgment rule) (Black, 2001). Directors’ duty of loyalty requires them to maintain the confidentiality of a company’s information (Kenton, 2023).Footnote 6 “Leaks of confidential information—whether related to the company’s products, strategy, or internal affairs, and whether disclosed to the public or private parties—may give a company’s competitors a leg up and jeopardize relations with customers, suppliers, and investors” (Stokdyk et al., 2015, 74). Therefore, it is important for directors to safeguard their firms’ confidential information to protect the firm’s competitive edge and success (Stokdyk et al., 2015).

Firms can safeguard proprietary information by not disclosing it (if the disclosure is not mandatory) or applying for CT orders to redact proprietary information from material contract filings (if the disclosure is mandatory). “In light of firms applying for and receiving confidential treatment in the presence of materiality, there may exist substantial heterogeneity in the quality and quantity of information disclosed in the material contract portion of the firm’s financial statements” (Verrecchia & Weber, 2006, 793). Bao et al., (2022, 50) suggest “that not all confidential treatment requests are made to protect proprietary information, and managers redact information from material contracts to conceal bad news.”Footnote 7

We propose that gender-diverse boards, relative to all-male boards, ethically and carefully refrain from disclosing proprietary information by, for example, using CT orders when information can harm a firm’s competitive advantage. Our argument is based on gender-based differences in inherent traits relating to ethicality and risk-averseness, which can influence a director’s decision to redact proprietary information or reduce its disclosure. Studies suggest that women are more ethical (Arlow, 1991; Deshpande, 1997) and risk-averse (Riley Jr and Chow 1992; Sunden & Surette, 1998) than their male counterparts. Hence, female directors improve board processes and bring important skills, knowledge and competencies to the boardroom (Gul et al., 2013). Studies suggest that women apply higher ethical standards in decision-making, have a stronger sense of duty (Cohen et al., 1998), and have less tolerance for opportunism (Bernardi & Arnold, 1997; Krishnan & Parsons, 2008; Schminke & Ambrose, 1997). The enforcement of conflict-of-interest guidelines and codes of ethical conduct is greater when a company’s board is gender-diverse (Flynn & Adams, 2004). Research also suggests that men are more likely than women to violate policies regarding expense reports and engage in insider trading (Betz et al., 1989), while women exhibit greater risk aversion in financial decision-making (Riley Jr and Chow 1992; Sunden & Surette, 1998) and demonstrate less overconfidence than men (Lundeberg et al., 1994).

Research suggests that gender-diverse boards protect shareholder interests. For example, Lanis et al. (2017) show that gender-diverse boards reduce tax avoidance. Tax avoidance is not in shareholders’ interests because it can lead to severe consequences, such as an ultimate assessment of penalties and interest, a challenge by a tax authority, or a loss of reputation (Gallemore et al., 2014; Kubick et al., 2015; Rego & Wilson, 2012), despite the benefits of an increase in cash flow and after-tax net income (e.g., Rego & Wilson, 2012).

From an ethical perspective, since the board represents the interests of shareholders, it is arguably unethical for the board to approve the nondisclosure of proprietary information (e.g., via CT orders) to hide bad news. This is unethical and not in shareholders’ interests because the use of CT orders to hide bad news is opportunistic and it can lead to adverse consequences, such as higher stock price crash risk and poorer future performance (Bao et al., 2022). As gender-diverse boards are more ethical than all-male boards, they are more likely to correctly use CT orders for the purpose of safeguarding their firm’s proprietary information instead of for opportunistic reasons.

“From a legal perspective, Directors must first and foremost act in the best interests of the company, and not the potentially shorter-term interests of its shareholders” (Butlers 2023). However, if the shareholders are unhappy with the board of directors’ decision, then they have the power to either remove or replace the directors via voting (Butlers 2023). As such, “conflicts can arise between what actions Directors actually take to benefit the corporation in the long-term (which may negatively impact a shareholder) and what actions a shareholder wishes for Directors to take” (Butlers 2023). As such, directors who vote for the nondisclosure of proprietary information face the risk of dismissal from the board if incumbent shareholders collectively prefer short-term returns over long-term returns.

We argue that since women are more ethical and risk-averse, gender-diverse boards (relative to all-male boards) would more ethically and carefully weigh the costs and benefits of disclosing proprietary information to maximize shareholder wealth in the long term. As the body with the advisory role, the board faces litigation risk if investors are hurt by disclosures that result in knowledge spillovers. Since female directors are more risk-averse concerning litigation and reputation loss (Srinidhi et al., 2011), they are likely to be more conservative with corporate disclosures and refrain from disclosing proprietary information.

Moreover, since research shows that men are less ethical than their female counterparts (Arlow, 1991; Deshpande, 1997), as well as being overconfident (Barber & Odean, 2001; Huang & Kisgen, 2013) and risk-takers (Faccio et al., 2016), we argue that an all-male board may disclose some proprietary information, if not all, due to the following reasons. First, it is plausible that an all-male board might be overconfident that competitors would not be able to strategically respond to their proprietary information upon its disclosure. Second, it is plausible that even if an all-male board foresees some risk of knowledge spilling over to rivals from the disclosure of proprietary information, they may still disclose such information owing to their risk-taker traits. Additionally, based on Janis (1972) theory of groupthink, we expect that an all-male board may redact information from material contract filings for opportunistic reasons. Overall, we expect all-male boards to approve the disclosure of proprietary information more frequently that gender-diverse boards.

Hypothesis

Since women are more ethical and risk-averse than men, gender-diverse boards are more likely to overcome groupthink. Hence, we expect that firms with gender-diverse boards would (1) ethically use CT orders to redact proprietary information and (2) disclose less proprietary information than similar firms with all-male boards when they face high competition (or high proprietary costs). Accordingly, we posit that firms with gender-diverse boards are more likely to experience higher adverse selection costs than similar firms with all-male boards because investors may incorrectly presume that the latter group of firms is as good as the earlier group and rate them similarly for investment purposes. Our hypothesis is stated below:

H1: Firms with gender-diverse boards are associated with higher adverse selection costs compared to similar firms with all-male boards when the firms face high competition.

In light of the recent findings of Glaeser (2018) and Barth et al. (2023), the argument that the nondisclosure of proprietary information automatically leads to material adverse selection problems due to increased information asymmetry is debatable. More specifically, Glaeser (2018) demonstrates that managers respond to product market constraints on disclosing proprietary information by reporting more nonproprietary information. Further, in the context of CT orders, Barth et al. (2023) show that voluntary non-earnings disclosures substitute the redacted disclosure of proprietary contract information. Nevertheless, Glaeser (2018) finds that disclosure of more nonproprietary information to substitute the reduced disclosure of proprietary information has the total effect of reduced corporate transparency. To the extent that gender-diverse boards might limit the negative effects of not disclosing proprietary information in highly competitive settings by increasing other types of voluntary disclosure that are value-relevant in capital markets, we may not find evidence in support of our hypothesis.Footnote 8

Research Design and Variable Measurement

A concern of this study is the potential endogeneity issue caused by the self-selection bias associated with a firm’s decision to include women on the board. If gender-diverse boards deal effectively with diversity in product markets (Bilimoria & Wheeler, 2000; Mattis, 2000) and have a greater understanding of consumer behavior, customer needs, and opportunities for companies to meet those needs (Kang et al., 2007) then firms may appoint female directors to prioritize product markets, especially when proprietary costs are high. In addition, prior studies suggest that firms with gender-diverse boards could systematically differ from other firms with all-male boards (Lai et al., 2017). To minimize potential endogeneity concerns, we use a propensity score matched design to examine the effect of board gender diversity on the adverse selection cost of firms facing high relative to low competition.

Propensity Score Matching Design

In our propensity score matching, we match a firm that has at least one female director with another firm with only male directors and the same propensity to engage female directors. Since the two sets of matched firms are similar in the characteristics that determine the likelihood of engaging female directors, differences in adverse selection cost could be attributed to board gender diversity. The advantage of using the propensity score matching approach over other methods of controlling for endogeneity is that we do not have to identify a clear source of the exogenous variables (Roberts & Whited, 2013). In this approach, the propensity score is computed using a propensity score model that predicts the probability of a firm engaging female directors based on observable firm characteristics (Lai et al., 2017; Roberts & Whited, 2013). However, a potential drawback of this approach is that any firm characteristic that is not captured by the controls in the propensity score model but is correlated with board gender diversity could still differ across the matched sample and, thus, could still affect the results. We address this potential omitted variable problem by also controlling for firm fixed effects and year fixed effects.

To match our sample of firms with gender-diverse boards to similar firms with all-male boards, we use a propensity score model based on prior studies that identify the determinants of female directorships (Hillman et al., 2007). Using this model, we first estimate a conditional logistic regression. Second, using the coefficients from the logistic model, a propensity score for each firm-year observation is determined, and then each firm with a gender-diverse board is matched to a unique firm with an all-male board within the same year, without replacement and with the neighboring propensity score based on a caliper width of 0.5 percent. Consistent with prior studies (Lai et al., 2017; Srinidhi et al., 2011), we use the following logit model to estimate the propensity scores.

$$\mathrm{Pr}\left[FDir=1\right]=\alpha +{\beta }_{1}ROA+{\beta }_{2}Size+{\beta }_{3}FirmAge+{\beta }_{4}SaleGrth+{\beta }_{5}NBSEG+{\beta }_{6}TotRisk+{\beta }_{7}Q+{\beta }_{8}Ret+{\beta }_{9}Vwretd+{\beta }_{10}CGboard+{\beta }_{11}CEOPower+{\beta }_{12}DirAge+{\beta }_{13}DirTenure+{\beta }_{14}Directorships{\sum }_{j}{c}_{j}{Years}_{j}+{\sum }_{k}{c}_{k}{Industries}_{k}+\varepsilon$$
(1)

In the propensity score model, the dependent variable is FDir, which is coded 1 for firms with at least one female director on the board and 0 otherwise. We control for the firm’s return on assets (ROA), the natural logarithm of total assets (Size), the number of years the firm appears on the Compustat database (FirmAge), the year-to-year percentage change in sales over the three prior years (SaleGrth), the number of business segments (NBSEG), the standard deviation of daily stock returns over the fiscal year (TotRisk), Tobin’s Q (Q), stock return over the fiscal year (Ret), value-weighted market return over the fiscal year (Vwretd), board governance index computed using principal component analysis (CGboard),Footnote 9 CEO power (CEOPower),Footnote 10 the average age of directors (DirAge), the average tenure of directors (DirTenure), and the average number of outside directorships held by the directors of the firm (Directorships). We also control for industry (Industries) and year (Years) fixed effects.

The untabulated results from the first-stage logit model (1) show a Pseudo R2 of 30.53%. Also, the model is significant at the one percent level, and it has reasonable discriminatory power as indicated by the area under the Receiver Operating Characteristic (ROC) curve, that is, 0.8454 (Hosmer & Lemeshow, 2000). We use the coefficients from the logistic model to determine a propensity score. Prior to matching, we examine the descriptive statistics of the difference between the propensity score for firms with and without female directors. The untabulated statistics show that the mean and standard deviation of the difference in propensity scores are 0.3023 and 0.0329, respectively.

Next, we use the propensity scores to match each firm with a gender-diverse board to a unique firm with an all-male board. Upon matching, we re-examine the descriptive statistics of the difference between the propensity scores for the matched sample. The untabulated statistics show that the mean and standard deviation of the difference in the propensity scores after matching are 0.0001 and 0.0001, respectively. These statistics suggest that the propensity scores are significantly and sufficiently lower for the matched sample, suggesting that it is a viable match. Further, using the matched sample, we compare the mean values for the control variables between firms with and without gender-diverse boards and find that the mean values are statistically indifferent between the two samples, suggesting that the matching procedure is successful.

Upon matching our sample firms with gender-diverse boards to similar firms with all-male boards, we conduct univariate and multivariate tests to examine our hypothesis that firms with gender-diverse boards, compared to similar firms with all-male boards, are associated with higher adverse selection costs, when the firm faces high competition.

Measurement of Adverse Selection

We measure adverse selection as the adverse information component of the effective bid-ask spread following the methodology of Lin et al. (1995). Lin et al. (1995) define the signed effective half spread as the transaction price (Pt) minus the quote midpoint (Qt) at time t, \({\mathcal{Z}}_{t}={P}_{t}-{Q}_{t}\). The quote midpoint is the average of the bid and ask price at time t, \({Q}_{t}=({A}_{t}+{B}_{t})/2\), where \(A\) is the ask price and \(B\) is the bid price. To estimate the adverse information component (\(\lambda\)), we estimate the following ordinary least squares (OLS) regression for each firm-year.

$${Q}_{t}-{Q}_{t-1}=\lambda {\mathcal{Z}}_{t-1}+{\mathcalligra{e}}_{t}$$
(2)

In our regression model (2), we use the natural logarithm of the quote midpoint and transaction price to yield a continuously compounded rate of return for the change in quote midpoint and a relative effective spread for the independent variable (Ness et al., 2001). To test our hypothesis for the effect of board gender diversity on adverse selection cost, we express the dollar adverse selection component as a percentage of the stock price to represent the adverse selection cost of transacting (ASCOST), consistent with Ness et al. (2001) and Hegde and McDermott (2004).

Adverse Selection Cost Model

We use the following model to test our hypothesis.

$$ASCOST=\alpha +{\beta }_{1}FD+{\beta }_{2}COMP+{\beta }_{3}FD\times COMP+{\beta }_{4}CGboard+{\beta }_{5}CEOPower+{\beta }_{6}DirAge+{\beta }_{7}DirTenure+{\beta }_{8}BoardSize+{\beta }_{9}LNPRICE+{\beta }_{10}LnAnalysts+{\beta }_{11}LNSDVOL+{\beta }_{12}LnME+{\beta }_{13}LNRISK+{\beta }_{14}LNDEBT+{\beta }_{15}LNPPE+{\beta }_{16}LNMKTBK+{\beta }_{17}RDDUM+{\sum }_{j}{c}_{j}{Years}_{j}+{\sum }_{k}{c}_{k}{Industries}_{k}+\varepsilon$$
(3)

In the above pooled cross-sectional regression model, ASCOST is an estimate of the adverse selection cost component of the effective bid-ask spread scaled by the firm’s stock price, computed following the methodology of Lin et al. (1995). Our main variable of interest to test our hypothesis is the interaction between FD and COMP, where FD is either FDir or PropFDir. Also, we use alternative proxies for proprietary cost including COMP, NPCM, and CTorder. To support our hypothesis that firms with gender-diverse boards are associated with higher adverse selection costs when the firm faces high relative to low product market competition compared to similar firms with all-male boards, we expect a positive coefficient on the interaction term (FDir × COMP), that is, \({\beta }_{3}>0\), and the sum of the coefficients on FDir and FDir × COMP to be positive and significant, that is, \({\beta }_{1}+{\beta }_{3}>0\).

FDir is coded 1 for firms with at least one female director on the board and 0 otherwise. PropFDir is the proportion of female directors on the board. We measure the level of product market competition (COMP) each firm faces using the excess price–cost margin (PCM). PCM is the industry-adjusted price–cost margin for each firm every year measured following Peress (2010) and Kubick et al. (2015). We compute each firm’s PCM for each year as the operating profit divided by sales and then adjust the firm-level price–cost margin by the value-weighted (based on sales) average price–cost margin of a given two-digit SIC industry-year.Footnote 11 Lower (higher) values of the excess price–cost margin represent lower (higher) product market power and higher (lower) product market competition for a firm. We code COMP as 1 (0) to represent high (low) proprietary costs or high (low) product market competition if the firm-year is in the bottom two (top) tercile of PCM for the industry-year. We also use the continuous measure of PCM, which we multiply by -1 so that higher values represent high product market competition (NPCM).

Further, we use the redaction of proprietary information as an alternative proxy for proprietary cost. Glaeser (2018) uses a text search to identify redacted 10-K filings. However, Glaeser (2018) also states that using a text search to measure the redaction of proprietary information results in a type I (II) error rate of 15 (23) percent. Glaeser (2018) suggests the use of CT orders to measure redactions, but these data have only been available since 2007. We combine the text search of Glaeser (2018) and CT orders to measure redactions. We obtain the data for CT orders from Wharton Research Data Services (WRDS) SEC Analytics Suite from 2007 to 2016 and merge it with our matched sample. For years between 1998 and 2006, we search the firm’s 10-K filings for redaction keywords (“confidential treatment,” “confidential information,” “redacted,” “CT order,” “FOIA,” “rule 406,” or “rule 24b-2”) to identify firms likely to have a CT order. We also merge the text search data with our matched sample. Our measure for the redaction of proprietary information (CTorder) is, therefore, coded 1 for 2007 to 2016 if the firm has a CT order, and 1 for 1998 to 2006 if the firm has a successful text match for redaction keywords, and 0 otherwise.

Since prior studies show that board governance influences a firm’s information environment through effective monitoring (Srinidhi et al., 2011), we control for board governance (CGboard), CEO power (CEOPower), the average age (DirAge) and tenure (DirTenure) of directors, and board size (BoardSize). We also control for the determinants of adverse selection cost identified in prior studies (e.g., Hegde & McDermott, 2004; Ness et al., 2001). The determinants include the natural logarithm of the following variables: average stock price (LNPRICE); number of analysts following (LnAnalysts); the standard deviation of daily trading volume (LNSDVOL); the market value of equity (LnME); the standard deviation of daily returns (LNRISK); financial leverage (LNDEBT); the ratio of property, plant and equipment to total book assets (LNPPE); and market value to the book value of assets (LNMKTBK). In addition, we control for the intensity of research and development efforts. Since many firms do not report research and development expenditures (Hegde & McDermott, 2004),Footnote 12 we control for a dichotomous variable, which is coded 1 if the firm reports research and development expenditure, and 0 otherwise (RDDUM). We also control for year (Years) and industry (Industries) fixed effects.

Data, Sample Selection Procedures, and Descriptive Statistics

Data and Sample Selection Procedures

We obtain financial statement data from Compustat, stock returns data from CRSP, and data relating to directors and board governance from ISS (formerly RiskMetrics) for the fiscal years between 1998 and 2016, inclusive. We start our sample from fiscal year 1998 due to the non-availability of director tenure data prior to 1998 in ISS. The sample selection procedure is provided in Table 1.

Table 1 Sample selection procedure
Full size table

The initial sample of female directors, after merging data from ISS with Compustat and CRSP, consists of 26,299 firm-year observations. We exclude 3337 firm-year observations from the Banking and Financial Trading industries (SIC codes 6000–6199 and 6200–6799) due to their regulated nature and 6240 observations with missing data on the control variables. This results in 16,722 firm-year observations or 2213 unique firms that are used for the propensity score matching. Upon matching each firm-year observation with a gender-diverse board to the most similar firm with an all-male board based on the propensity score matching approach, we end up with a final sample of 6636 firm-year observations or 1711 unique firms. Our final sample comprises 3318 firm-year observations with gender-diverse boards and 3318 matched firm-year observations with all-male boards.

Descriptive Statistics

We report the descriptive statistics on variables used in our second-stage regression model (3) and compare the statistics between firms with and without female directors in Table 2. Panel A provides descriptive statistics on the dependent and test variables. The average adverse selection cost for the full sample is 5.35% of the stock price and is larger for firms with gender-diverse boards (5.46) compared to firms with all-male boards (5.24). On average, the proportion of female directors on the boards is 7.40% for the entire matched sample, and it is 14.79% for firms with gender-diverse boards. The average excess price–cost margin (NPCM) is significantly higher for firms with gender-diverse boards relative to firms with all-male boards, suggesting that firms with gender-diverse boards face higher product market competition than firms with all-male boards.

Table 2 Descriptive statistics on variables used in the second-stage main regression models for the propensity score matched sample
Full size table

Panel B of Table 2 provides the descriptive statistics on the control variables used in the adverse selection cost model (3). The statistics on most board governance variables show that they are mostly not significantly different between firms with and without gender-diverse boards. This is because our sample is matched on board governance and other firm characteristics. We find that firms with gender-diverse boards have, on average, one more director than similar firms with all-male boards. Descriptive statistics further suggest that firms with gender-diverse boards have a lower average stock price, number of analysts following, standard deviation of daily trading volume, and market value of equity. The statistics also show that firms with gender-diverse boards have a higher ratio of property, plant and equipment to total book assets and research intensity. The average value of the standard deviation of daily returns, financial leverage, and market value to the book value of assets is not statistically different between the sample of firms with and without female directors.

Since we argue that firms with gender-diverse boards have higher adverse selection costs when they face high product market competition, we split our matched sample into a subsample of firms that face high product market competition and a subsample of firms that face low product market competition to examine the univariate statistics on adverse selection cost. Untabulated statistics show that within the subsample of firms facing high (low) competition, firms with gender-diverse boards have higher (lower) average adverse selection cost compared to similar firms with all-male boards, significant at the ten (one) percent level. These univariate results provide some preliminary support for our hypothesis.

Main Result

We report the results for the test of our hypothesis in Tables 3 and 4, using various proxies of board gender diversity and proprietary cost. For our regressions, we use two-dimensional clustering by firm and year and the cluster-robust t-statistics are used to compute the reported p-values. We provide the results using FDir (PropFDir) and COMP as the test variables in Panel A (B), and PropFDir and NPCM in Panel C of Table 3. The results in Column 1 are for a baseline regression without the interaction variable. The adjusted R2 is around 72% in the three panels, suggesting that the model has significant explanatory power. The coefficient on FDir is positive but not significant at the conventional level, but the coefficient on PropFDir (in Panel B and C) is positive and significant at the five percent level, suggesting that more gender-diverse boards are associated with more adverse selection costs on average. This is consistent with our expectation because we predict that firms with gender-diverse boards have higher adverse selection costs when they face high product market competition compared to similar firms with all-male boards.

Table 3 Regression results of adverse selection cost on board gender diversity and its interaction with proprietary cost proxied by high product market competition
Full size table
Table 4 Regression results of adverse selection cost on board gender diversity and its interaction with proprietary cost proxied by confidential treatment orders
Full size table

Column 2 of Table 3 provides the results after including the interaction term (i.e., FDir × COMP, PropFDir × COMP, PropFDir × NPCM) in the adverse selection cost model. The results in Panel A and B show that the main effect of board gender diversity is not significant, suggesting that the effect of gender-diverse boards on adverse selection costs is the same as that of all-male boards in less competitive environments.Footnote 13 Next, we find that the coefficient on the interaction term is positive and significant at either one or five percent level and that the sum of the coefficients on the main and interaction effect is positive and significant at either ten percent level (in Panel A) or one percent level (in Panel B). We continue to find consistent results when we also control for firm fixed effects as reported in Column 3.Footnote 14

The results in Table 3 support our hypothesis and suggest that firms with gender-diverse boards are associated with higher adverse selection costs relative to similar firms with all-male boards when the firms face high product market competition. The results are consistent with female directors overcoming groupthink by bringing a different perspective based on their ethical and risk-averse traits to the board discussions. Hence, gender-diverse boards, compared to all-male boards, prioritize product market concerns over short-term capital market incentives in favor of long-term shareholder interests by reducing or redacting the disclosure of proprietary information when proprietary costs are high.

The results are also economically significant. For example, referring to Column 2 of Panel A, the sum of the main and interaction effect is 0.1898 (= − 0.1502 + 0.3400), suggesting that, on average, a firm with at least one female director experiences an increase of approximately 19 percent in adverse selection cost relative to similar firms with all-male boards in a highly competitive environment. This translates into an increase of 102 (= 0.1898 × 5.3493) basis points in adverse selection cost above the sample mean. Similarly, referring to Column 2 of Panel B, the sum of the main and interaction effect of 1.6843 (= − 0.4035 + 2.0878) suggests that, on average, a one-standard-deviation increase in the proportion of female directors on a board would result in an increase of approximately 15 (= 1.6843 × 0.0867) percent in adverse selection cost. The 15% increase translates into an increase of 78 (= 0.1460 × 5.3493) basis points in adverse selection cost compared to the sample mean.

We provide the statistics on control variables in Panel A of Table 3 but do not provide these in other panels and tables to conserve space. We find that the coefficient on the proxies for board governance is mostly insignificant. This is likely because our sample is matched on board governance. The statistics on the other control variables show that LNPRICE, LnME, LNDEBT, and LNMKTBK are significantly negatively associated with adverse selection cost, consistent with larger firms with higher market values and leverages having lower adverse selection costs. The coefficient on LnAnalysts is positive and significant (in Column 3), suggesting that firms with more analysts following have larger differences in opinions (Diether et al., 2002) and higher dispersion (Liu & Natarajan, 2012); together, these increase adverse selection cost. The results show that LNSDVOL and LNRISK are significantly positively associated with adverse selection cost, consistent with riskier firms having higher adverse selection.

Further, in developing our hypothesis, we argue that firms with gender-diverse boards have higher adverse selection costs because the board refrains from disclosing proprietary information when facing high competition. Proprietary information can be disclosed either through mandatory or voluntary disclosures. It is possible to identify firms that choose not to disclose mandatory information, but it may not be possible to identify firms that decide to refrain from disclosing nonmandatory (voluntary) information. Firms can apply to the SEC, which can grant their request to withhold (or redact) proprietary information from investors in their material contract filings (Glaeser, 2018; Verrecchia & Weber, 2006). Since we argue that firms with gender-diverse boards experience high adverse selection costs when they face high product market competition because of the boards’ decision to refrain from disclosing proprietary information, we also examine whether firms with gender-diverse boards are associated with higher adverse selection costs if the firm redacts confidential information from its material contract filings. That is, we re-estimate the regression model (3) after replacing the interaction variable, COMP, with CTorder and report our results in Table 4.

The results in both panels show a positive coefficient on the interaction term (FDir × CTorder) that is statistically significant. The total effect of board gender diversity on adverse selection costs for firms that redact proprietary information is positive and significant at either one or five percent level. The results collectively suggest that firms with gender-diverse boards are associated with higher adverse selection costs compared to similar firms with all-male boards when they redact proprietary information from material contract filings.

Additional Analyses

An Exogenous Increase in Competition

As our first additional analysis we replicate our main test using an exogenous event to proxy for a change in product market competition. Recent studies suggest that large import tariff rate reductions trigger a significant exogenous increase in the competition faced by domestic companies (Huang et al., 2017; Valta, 2012; Xu, 2012). Research also suggests that tariff reductions in the US are adopted through a complex bureaucratic and political process and enacted as part of trade agreements (Huang et al., 2017). It is unlikely that a firm’s disclosure policy could effectively influence trade agreements, despite some firms’ possible lobbying in favor of some features of trade agreements. Similarly, the board composition of individual firms is unlikely to influence trade agreements because boards are responsible for advising and monitoring management and discussing issues that may affect shareholder interests. For example, a board would discuss how a firm should respond to a large reduction in tariffs. Hence, it is likely that tariff reductions are exogenous to firms’ disclosures or boards’ decisions, and the interpretation of our results is unlikely to be affected by reverse causality.

Prior studies argue that large tariff rate reductions increase competition because import tariffs represent a significant part of total trade costs (Anderson & van Wincoop, 2004) and tariff rates have significantly decreased in recent years (Huang et al., 2017). Import tariffs constrain the intensity of foreign competition because they impose additional costs on foreign companies that translate into higher prices for imported goods or services relative to the price of goods or services domestically produced. Thus, large reductions in tariff rates are likely to increase competition in US markets.Footnote 15

Proprietary costs are likely to increase as a result of the exogenous increase in competition. Huang et al. (2017) show that large reduction in tariff rates is associated with a significant decrease in the disclosure of management earnings forecasts. They state that tariff rate reductions are more likely to increase competition from existing foreign rivals than from potential foreign entrants (Huang et al., 2017). Hence, their findings are consistent with the theory suggesting that competition from existing rivals reduces voluntary disclosures due to increased proprietary costs (Huang et al., 2017; Verrecchia, 1983, 1990). In our context, we expect that an exogenous increase in competition due to large reduction in tariff rates is likely to increase proprietary costs. Therefore, gender-diverse boards are more likely to reduce or redact disclosure of proprietary information to protect their firms in the product market and maximize long-term shareholder interests. This is likely to result in an increase in adverse selection costs for firms with gender-diverse boards, compared to firms with all-male boards, after tariff rate reductions.

We measure an exogenous increase in competition using the list of industries with large tariff reductions and the respective tariff reduction year from Table 1 of Huang et al. (2017). For this test, we restrict our sample to between 1998 and 2011 because data for large tariff reductions beyond those years are not reported in Huang et al. (2017). To examine our hypothesis using large tariff reductions as an exogenous increase in competition, we use the following difference-in-differences (DID) specification.

$$DV=\alpha +{\beta }_{1}FDfirm+{\beta }_{2}POST+{\beta }_{3}FDfirm\times POST+CONTROLS+\varepsilon$$
(4)

The dependent variable (DV) is the adverse selection cost (ASCOST). CONTROLS represents the control variables from the adverse selection cost model (3).

Our variable of interest is the DID estimate that is represented by the interaction of FDfirm and POST (\({\beta }_{3}\)). \({\beta }_{3}\) captures the change in DV after an exogenous event for firms with gender-diverse boards (treatment firms) relative to the change in DV after the same event for firms with all-male boards (control firms) during the event year. FDfirm is coded 1 if the firm has at least one female director in all the sample years and it is coded 0 if the firm has an all-male board throughout the sample years. Firms that switched from having a gender-diverse board to an all-male board or vice versa are deleted from this test. POST represents an exogenous event and is coded 1 for industries based on three-digit SIC codes after the year in which the industry experienced a significant reduction in tariffs, and it is coded 0 for the same industry before the reduction in tariff. We identify large reductions in tariffs as our exogenous event because it led to a significant increase in competition and our predicted effect of board gender diversity on DV is conditional on product market competition.

Consistent with Huang et al. (2017) and Valta (2012), we also control for firm fixed effects and year fixed effects in an additional DID specification to remove the effects of other observable or unobservable variables that we have not controlled for in model (4). This method fully controls for time-invariant differences between treated and non-treated firms and aggregate fluctuations (Bertrand & Mullainathan, 2003). In this additional DID specification, FDfirm and POST are subsumed by firm and year dummies, and they are dropped from the regression.

The results are reported in Columns 1 and 2 of Table 5. The results in Column 1 (2), which excludes (includes) firm fixed effects, show a positive coefficient for the DID estimate, providing further support for our hypothesis. The results suggest that, relative to firms with all-male boards, firms with gender-diverse boards experience a significant increase in adverse selection costs after large reductions in US tariff rates. Based on the firm fixed-effect model, the average firm with a gender-diverse board that experiences a large tariff rate reduction experiences a 60.63 percent increase in adverse selection costs. Since the average adverse selection cost is 5.1383 for the DID sample, the 60.63 percent increase translates into an increase of 312 (= 0.6063 × 5.1384) basis points or 3.12 percent in adverse selection cost above the sample mean.

Table 5 Difference-in-differences (DID) regression results for adverse selection cost
Full size table

The results in Column 3 validate the parallel trend assumption. We re-estimate Eq. (4) with firm fixed effects upon adding three interaction terms, FDfirm × PRE3, FDfirm × PRE2, FDfirm × PRE1, consistent with prior studies (e.g., Dou et al., 2019). PRE3 (PRE2) [PRE1] is coded as 1 for industries based on three-digit SIC codes from three (two) [one] years prior to a significant reduction in tariffs and coded 0 otherwise. The insignificant coefficients on these three interaction terms suggest that the trends three years, two years, and one year prior to the large tariff reduction are indistinguishable between firms with gender-diverse boards and firms with all-male boards. Also, the results in Column 3 show that the DID term (FDfirm × POST) is positive and significant at the five percent level, suggesting that the increase in adverse selection costs occurs after the large reduction in tariff.

Confidential Treatment Orders

A critical link in the causal chain underpinning Hypothesis 1 is that gender-diverse boards disclose less proprietary information in the face of stiff product market competition. We next examine this channel by estimating a regression of CT orders on board gender diversity and its interaction with high product market competition using the following equation.

$$CTorder=\alpha +{\beta }_{1}FD+{\beta }_{2}COMP+{\beta }_{3}FD\times COMP+CONTROLS+\varepsilon$$
(5)

The dependent variable is the CT orders (CTorder). Our variable of interest is the interaction term (FD × COMP), where FD is either FDir or PropFDir, and COMP is coded 1 to represent high product market competition and 0 otherwise.

CONTROLS represents the control variables proxying for board governance and firm characteristics. Consistent with our adverse selection cost model (3) we control CGboard, CEOPower, DirAge, DirTenure and BoardSize. Further, we control for determinants of a redaction of proprietary information following Glaeser (2018). These include firm size (LnME), leverage (LEV), return on assets (ROA), market to book ratio (MB), stock return (RET), loss (LOSS), special items (SITEMS), research and development expenditures (R&D) and missing research and development expenditures (MISR&D). We also control for debt issue (DEBTISSUE), number of exhibits (LnNEXIB), beta (BETA), idiosyncratic volatility (IVOLATILITY), number of analysts following (LnAnalysts), Altman’s Z score (ALTMANSZ), analysts’ earnings forecast dispersion (DISP), long-term growth (LTG) and year and industry fixed effects. All variables are defined in Appendix A.

The results reported in Table 6 show that while the main effect of board gender diversity on CT orders is insignificant, the interaction between board gender diversity and high product market competition is positive and significant at the five or ten percent level.Footnote 16 The total effect is also positive and significant at the five percent level. These results collectively suggest that while there is no significant difference in less competitive environments between firms with and without female directors in terms of receiving CT orders, firms with gender-diverse boards are associated with significantly more CT orders in highly competitive settings, compared to similar firms with all-male boards. The results support our theoretical link between gender-diverse boards and adverse selection costs in highly competitive settings as they show that gender-diverse boards redact more proprietary information from material contract filings when facing high product market competition.

Table 6 Regression results of confidential treatment orders on board gender diversity and its interaction with high product market competition
Full size table

Long-run Abnormal Stock Returns

In our theoretical development, we argue that gender-diverse boards promote and protect shareholder interests and take fewer risks by refraining from approving the disclosure of proprietary information when proprietary costs are high. This decision unfavorably affects firms’ capital market positions and results in increased adverse selection compared to similar firms with all-male boards, as shown in our main results. Nevertheless, firms with gender-diverse boards should also benefit from an increased competitive edge in the product market and higher value for their innovation, which, collectively, should translate into higher future value for shareholders. Since nondisclosure of proprietary information prevents competitors from strategically responding to the incumbent firm’s strategies, the incumbent firm, with a gender-diverse board, maintains and improves its competitive edge. As a result, we expect that firms with gender-diverse boards have higher long-run buy-and-hold abnormal returns (BHAR) compared to similar firms with all-male boards, especially when the firm faces high competition and initially suffers adverse selection.

To examine our proposition, we examine the BHAR one, two and three years ahead for firms with gender-diverse boards compared to similar firms with all-male boards. We first match each firm with a gender-diverse board with a similar firm with an all-male board by selecting a control firm using the propensity score model (1) within the same year and the same stock exchange, without replacement and with the neighboring propensity score based on a caliper width of 0.5 percent. Next, we compute the BHAR for each firm with a gender-diverse board as follows:

$${BHAR}_{i,t}=\prod_{t=1}^{T}\left(1+{r}_{i,t}\right)-\prod_{t=1}^{T}\left(1+{r}_{c,t}\right)$$
(6)

In Eq. (6), ri,t and rc,t are the monthly returns of a firm with a gender-diverse board, i, and a control firm with an all-male board, c, respectively, in month t. Here, t is 12, 24 and 36 months for one-year, two-year, and three-year BHAR, respectively. The BHAR results are reported in Table 7.

Table 7 Long-run buy-and-hold abnormal returns of firms with gender-diverse boards
Full size table

The results in Panel A show that 1-year and 2-year BHARs are not significantly different for firms with gender-diverse boards compared to firms with all-male boards; however, the 3-year BHAR is significantly higher for firms with gender-diverse boards at the 10% level. Panel B provides the results for firms facing high product market competition, and it shows that firms with gender-diverse boards have a 7.30% higher 3-year BHAR than similar firms with all-male boards. For firms facing high product market competition that initially suffered adverse selection, the results in Panel C show that firms with gender-diverse boards experienced, on average, a 6.19% higher 2-year BHAR and 11.06% higher 3-year BHAR than those of similar firms with all-male boards. These results are statistically and economically significant.

In addition, we also observe that the BHARs of firms with gender-diverse boards are not significant when firms face high competition but are not initially adversely selected compared to similar firms with all-male boards (Panel D). The BHARs are mostly not significantly different between firms with and without female directors if the firms face low product market competition (Panels E–G). This result is consistent with the broad board gender diversity literature (e.g., Gul et al., 2011), which suggests that gender-diverse boards make significantly more voluntary disclosures compared to all-male boards. Based on our theory, this is likely when the firm has low proprietary costs. Hence, firms with gender-diverse board that face low competition and less adverse selection than a corresponding control firm do not experience a significant increase in BHAR because they are instantly rewarded by the capital market. However, such firms experience significantly negative BHAR because the corresponding control firms with male directors are rewarded in later years when they reveal more information to the capital market through their disclosures (e.g., three-year BHAR in Panel G).

These results collectively suggest that firms with gender-diverse boards have significantly higher long-run BHAR when the firm faces high competition and more adverse selection, potentially because the board refrains from disclosing proprietary information that would give the firm a competitive advantage in the product market. Thus, our results support our argument that gender-diverse boards promote and protect shareholder interests by prioritizing product market concerns when proprietary costs are high.

Other Results

In our final set of analyses, we replicate our main test using alternative measures of board gender diversity: a higher proportion of female directors, and the presence and proportion of independent female directors.

First, prior studies suggest that having only one female director on a board may often serve as tokenism (e.g., Bourez, 2005; Kanter, 1977; Kramer et al., 2006). Also, having only one female director on a board may not be sufficient to constitute gender diversity, especially for large boards. Lai et al. (2017) suggest that a higher proportion of female directors on a board proxies for a higher degree of board gender diversity and that the effect of at least two female directors corrects the issue of tokenism. In our main analysis, we also provide results using the proportion of female directors; however, in this additional analysis, we test the effect of board gender diversity measured as the proportion of female directors in the fourth quartile compared with the first quartile. We exclude the observations between the first and third quartiles of the proportion of female directors to compare more diverse gender boards with all-male boards.Footnote 17 The untabulated results collectively suggest that firms with a more gender-diverse board are associated with significantly higher adverse selection costs when the firm faces high product market competition.

Second, prior studies show that the majority of female directors are independent female directors, and they contribute to boards through their monitoring role because they have stricter ethical standards (e.g., Lai et al., 2017). We replicate our main test using the presence and proportion of female independent (nonexecutive) directors and find similar results (untabulated).

Conclusion

We examine whether gender-diverse boards prioritize product market concerns over capital market incentives when proprietary costs are high. Our results show that firms with gender-diverse boards are associated with higher adverse selection costs relative to similar firms with all-male boards when the firms face high product market competition or redact proprietary information from their material contract filings. Further results show that firms with gender-diverse boards, compared to similar firms with all-male boards, benefit in the long term by earning higher BHAR, and the net gain for these firms is about 10 percent of their stock price in 3 years.

The inferences that we draw from our analyses largely suggest that gender-diverse boards strategically prioritize product market concerns over short-term capital market incentives when proprietary costs are high. Collectively, our findings suggest that gender-diverse boards ethically and carefully maintain the confidentiality of proprietary information for proprietary cost reasons as opposed to opportunistic reasons. As such, firms experience short-term adverse selection costs in the capital markets. Nevertheless, firms benefit in the long run through shareholder wealth maximization, as board gender diversity helps create and maintain competitive advantage. Our research, therefore, advocates the importance of board gender diversity in creating competitive advantage by prioritizing product market concerns over short-term capital market incentives. This suggests that any reduction in board gender diversity will be counterproductive for firms in highly competitive settings.

We contribute to the literature on corporate disclosure and board gender diversity. Our research is the first to show that board gender diversity is a determinant of a firm’s decision against disclosing proprietary information and that such a decision is for proprietary cost-based motives. Recently, Bao et al. (2022) find that managers opportunistically redact proprietary information as a means to hide bad news rather than using it for its intended purpose (i.e., the proprietary cost-based motive). As such, our work extends that of Bao et al. (2022) by highlighting the importance of gender-diverse boards in fostering the ethical and careful redaction of proprietary information for proprietary cost-based reasons as opposed to agency cost-based motives.

We add to prior studies (e.g., Gul et al., 2011) by highlighting that gender-diverse boards do not unconditionally increase voluntary disclosures of firm-specific information. Our findings suggest that in the presence of high proprietary costs, gender-diverse boards ethically and carefully redact or reduce the disclosure of proprietary information to prioritize product market concerns and maximize long-term shareholder wealth. Our findings have implications for regulators and firms on the benefits of gender-diverse boards. As the US regulatory regime is moving toward new rules that permit redaction without formal CT requests, concerns that firms may redact information to hide bad news (Bao et al., 2022) are likely to increase. Our results inform regulators and firms on how gender-diverse boards can minimize this problem by ethically and carefully redacting proprietary information for its intended purpose: to maintain the confidentiality of proprietary information for proprietary cost-based motives.