Investors’ response to the #MeToo movement: does corporate culture matter?

Our study relies on using market returns to calibrate investor reactions to events surrounding the #MeToo movement; therefore, a natural question to ask is whether investors, on average, punish firms when they announce sexual harassment complaints. Borelli-Kjaer et al. (2021) examine the price impact of sexual harassment scandals for a broad sample of international firms reporting such scandals between January 2005 and February 2019. They find that market value, on average, falls 1.5% in response to the announcement of a sexual harassment scandal, thus validating the view that the market places a tangible cost on firms engaging in these types of behavior. In Appendix 1, we corroborate Borelli-Kjaer et al.’s (2021) findings by documenting an overall negative price response to 92 sexual harassment scandal announcements affecting NASDAQ and NYSE-listed U.S. firms that overlap with our sample. Thus, on average, investors place a jaundiced eye towards firms engaging in sexual harassment.

From a legal standpoint, sexual harassment is a form of employment discrimination that falls under Title VII of the Civil Rights Act of 1964. However, its inclusion as a violation of the Civil Rights Act was not immediate. Administratively, the Equal Employment Opportunity Commission (EEOC) began considering sexual harassment to be an action prohibited by Title VII in 1980,¹² a view upheld by the U.S. Supreme Court in a 1986 unanimous decision in the case of Meritor Savings Bank v. Vinson (see Tippett (2018) for a discussion of Meritor and subsequent Supreme Court rulings that refined and explained the conditions behind the existing law). In addition, many states have their own laws prohibiting sexual harassment—for example, the New Jersey Laws Against Discrimination and the Pennsylvania Humans Relations Act.

However, not all sexual misconduct incidents are publicly revealed by the firm, by the victim(s), or through a publicly documented initiation of a sexual harassment lawsuit. For example, firms historically have used nondisclosure agreements and arbitration agreements to suppress or hide the public disclosure of these violations, thus keeping their sexual harassment complaints within house (Tippett, 2018). More commonly, many complaints go unreported or are not vigorously pursued by the firm. According to a 2016 survey by the Merit Systems Protection Board (MSPB, 2018), only 11% of employees who endured any form of workplace sexual harassment filed a formal complaint; more common survey responses included the employee ignoring the incident, asking for the harassment to stop, or changing jobs.

There are substantive nonlegal costs associated with sexual misconduct incidents, both revealed and unreported. These costs include being absent from work, reduced productivity, job turnover, and the opportunity cost of manager time in handling complaints (Deloitte, 2019; Rizzo et al., 2018; Sandroff, 1988). In dollar terms, Sandroff (1988), in a survey of individuals across 160 Fortune 500 companies, estimated the average organizational costs of workplace sexual harassment to be $6.7 million per firm. Deloitte (2019), using survey data from the Australian Human Rights Commission (2018), reports an estimated $1053 cost per employee (for all employees, not just the targets) for Australian firms, with these costs stemming from absenteeism ($297), reduced productivity ($171), staff turnover ($336), and manager time ($250). Notably, Deloitte finds strong evidence that staff turnover encompasses not only the perpetrator and the person being harassed, but also bystanders who witness the incident.

If, as we contend, the #MeToo Movement increased investors’ concerns about the future costs of sexual harassment incidents within a firm, then the abnormal returns around the events surrounding this movement would be a reflection both of these costs and the perceived likelihood of future sexual misconduct existing within the firm.

Kreps (1990) defines corporate culture as an intangible asset designed to meet unforeseen contingencies as they arise (see also Camerer and Vepsalainen (1988)). Basically, because a firm cannot contract on unforeseen circumstances, it can create a culture consonant with shared beliefs, assumptions, and values that help employees understand which behaviors are and are not appropriate (Grennan, 2019; Schein, 1990). Importantly, corporate culture evolves over time (Cheng & Groysberg, 2021; Schein, 1990), suggesting both a stickiness in change and a historical perspective on what that culture is.

The corporate culture we examine is the implicit attitude toward the inclusion or exclusion of women within the firm’s workplace. Firms with inclusive workplaces based on gender or race have been shown to be more innovative towards their customers’ needs (Jain-Link et al., 2020); more likely to create a learning culture, i.e., one which emphasizes flexibility and independence among its employees (Cheng & Groysberg, 2021); and more likely to act in the interests of a broader set of stakeholders, i.e., not just shareholders (Chen et al., 2021). Conversely, exclusive workplaces—defined by Cheng and Groysberg (2021) as organizations where differing perspectives are silenced, ignored, or neglected—are more likely to struggle with managing, hiring, and retaining qualified and diverse employees (Cheng & Groysberg, 2021).

One role of the board of directors is to “set the framework of values” (FRC, 2018), thus shaping its corporate culture. However, according to two surveys by Graham et al. (2016, 2022), boards do not directly choose a firm’s culture. Instead, they embody the firm culture and, accordingly, influence it through their actions—for example, via their choice of CEO (see also Sandford, 2014). In recognition of this embodiment, law firms (e.g., Akin Gump, 2019) and accounting consulting groups (e.g., Klemash & Dettmann, 2019) increasingly have counseled boards to consider the oversight of their firms’ corporate cultures as an important priority.

Our assumption that investors use board composition as a partitioning variable in assessing a firm’s corporate culture hails from Camerer and Vepsalainen (1988), who define a visible firm culture as one that can be seen from outside the firm. Since board composition is visible to outside investors, we propose that the market uses this composition in assessing the inclusivity or exclusivity of firm culture as it relates to gender. Further, consistent with Schein (1990), who states that corporate cultures evolve over time, we expect the market to consider long-term trends in gender composition as better indicators of the firm’s corporate culture vis-à-vis the most recent year.

As in any event study, our inferences depend critically on the proper identification of events. To avoid subjectivity in our selection of dates and the potential for bias, we use the #MeToo event timeline maintained by the Chicago Tribune for our analyses.¹³ This timeline remains the top search result from Google and Bing search engines (using the search terms of “#MeToo” and “timeline”), underscoring the awareness and influence of this source.

Our analyses focus on the events in the first nine months of the #MeToo timeline, i.e., the 37 event dates from October 2017 through May 2018 (see Appendix 2 for the dates and headlines of each event). We begin on October 5, 2017 [E1], when allegations of sexual harassment by Ashley Judd against Harvey Weinstein of Miramax were reported in the news, and we conclude with the indictment of Harvey Weinstein on May 25, 2018 [E37]. This timeline allows the market to evaluate the growing momentum of this social movement.

We assemble a sample of U.S. public companies with available stock return data from CRSP, financial statement data from Compustat, and board composition data from BoardEx. As shown in Table 1, we begin with the 5385 firms listed in the Compustat-CRSP merged database as of 2016, which represents the last full year of available data prior to the start of the #MeToo movement in 2017. Removing 884 foreign firms and 1097 firms with missing daily return data at any point during the nine-month sample period, we arrive at 3404 firms with available data for our market reaction tests. Although the BoardEx coverage expanded considerably in the past decade, we still lose 525 of these firms due to a lack of BoardEx data. After removing another 276 firms with either inconsistent BoardEx data (20 firms) or missing Compustat data (256 firms), we have a sample of 2603 firms with all available data at the end of 2016. We further add the restriction that each firm has available BoardEx data over the years 2012 through 2016 inclusive, which allows us to create subsamples of firms with multi-year inclusions and exclusions. After excluding the 578 firms with missing years, we arrive at our final sample of 2025 firms available for our tests.

Panel A of Table 2 provides descriptive statistics for the sample. The typical (based on mean or median) firm in our sample has nine directors on its board, which includes one woman. As shown in Fig. 2, women’s board representation steadily increased in the years leading up to the #MeToo movement. While nearly 40% of firms in our sample excluded women from the board room in 2012, only 27% did so as of 2016. Nevertheless, the percentage of firms welcoming just one woman into the board room each year held steady over this same time period, as approximately one-third of boards included only one woman from 2012 through 2016.

Our upcoming tests focus on the differential reaction to the #MeToo movement based on firms’ inclusive versus exclusive cultures. Accordingly, we identify subsamples of firms based on the presence or the absence of women in their board rooms. Specifically, we narrow our focus to firms that, as of the start of the #MeToo movement, had traditionally excluded women from their board. For contrast, we also identify firms that had already embraced the inclusion of women on their board, as evidenced by the presence of three or more women.

We select three women as our threshold for three main reasons. First, this number avoids the tokenism documented by prior work (Adams & Ferreira, 2009; Farrell & Hersch, 2005). Indeed, the steady trend of the percentage of firms with just one woman on the board (as shown in Fig. 2) is indicative of tokenism. Second, a threshold of three women follows critical mass theory of group dynamics. Kanter (1977) argues that a minority subgroup’s degree of influence within any full group is felt only when the size of that group reaches a certain dimension; Kanter refers to this as a “critical mass” theory. Erkut et al. (2008), Konrad et al. (2008), and Torchia et al. (2011) examine this theory on gender representation within a firm’s board of directors. Using survey data of women directors, they present evidence that achieving a critical mass of at least three women on the board enhances the board’s working dynamics in general and also its outlook on firm innovation. As Erkut et al. (2008) note, “One woman is the invisibility phase; two women is the conspiracy phase; three women is mainstream” (p. 227). Finally, the threshold of three women is consistent with the new California law requiring all California-based firms with boards of at least six directors to have a minimum of three women directors by the end of the 2021 calendar year.¹⁴

Because we are interested in measuring the culture of the firm leading up to the advent of the #MeToo Movement, we use the gender composition of the firm’s board over the five-year window 2012–2016 to categorize a firm as being exclusive or inclusive of women. Specifically, we categorize a firm as being EXCLUSIVE if over the full five years, its board was comprised entirely of men; symmetrically, we categorize a firm as being INCLUSIVE if over the same time period, its board continuously had at least three women directors. Our approach reflects a more stable, long-term board environment.

Over the five-year period, we find that 481 firms enter the #MeToo time frame without ever having included a woman on their board, while 122 firms enter 2017 with a persistent critical mass of at least three women directors. In terms of industry composition, as shown in Panel B of Table 2, both the EXCLUSIVE and INCLUSIVE subsamples include many financial firms—with the exclusive sample concentrating in the business equipment industry and the inclusive sample concentrating in the wholesale/retail sector.

As shown in Panel C of Table 2, comparing firm characteristics between the subsamples of EXCLUSIVE and INCLUSIVE firms produces a number of significant differences. INCLUSIVE firms tend to be larger (SIZE and SALES), better performing (ROA), and less volatile (RETURN VOLATILITY), and have higher growth (lower BOOK-TO-MARKET), higher leverage, and larger, more independent boards (# of DIRECTORS and % INDEP DIR). Consistent with other papers (Ahern and Dittmar 201; Kim & Starks, 2016; Matsa & Miller, 2011), we control for a number of these factors in our upcoming multivariate tests. In addition, our tests consider various approaches to benchmarking firm performance, all of which aim to control for differences across subsamples. Moreover, to explore whether unobserved (and, thus, uncontrolled) differences between these two subsamples—unattributable to the unfolding #MeToo movement—explain our findings, we calculate pseudo-returns based on random event dates.

We begin by presenting univariate evidence to corroborate our approach to identifying EXCLUSIVE versus INCLUSIVE firm cultures. Table 3, Panel A tests for differences in executive characteristics among firms, while Panel B tests for differences in external evaluations of firm culture across firms. In both panels, sample sizes vary depending on data availability for each measure.

The evidence in Table 3 consistently supports the notion that sorting firms based on the presence or absence of women in the board room is effective in identifying firms with exclusive versus inclusive cultures. In Panel A, we use ExecuComp data to identify the five highest-compensated executives in each firm, which reduces our sample to 1173 firms.¹⁵ None of the EXCLUSIVE firms have a woman CEO in 2016¹⁶; this contrasts with 18% of the INCLUSIVE firms. When comparing the incidence of having any woman executive, we find that 76% of the EXCLUSIVE firms and 41% of INCLUSIVE firms had no women executives in 2016. Equally striking, when looking across 2012–2016, 67% of EXCLUSIVE firms had no women executives over the entire five-year period; in contrast, 31% of INCLUSIVE firms had the same lack of women in their executive suites. These findings are consistent with Matsa and Miller (2011) who, using an earlier period (1997–2009), find similar associations between a firm having women on its board and the gender of its top five executives. Thus, the trend connecting the exclusivity or inclusivity of women on the board of directors with the gender make-up of the C-suite appears to span an almost 20-year period, beginning in 1997 (Matsa & Miller, 2011) and ending in 2016, the year prior to the #MeToo movement.

In Panel B, we turn from the executive suite to external evaluations of firm culture. Again, in support of the premise that gender representation on the board provides a signal about the culture of the firm, we detect significant differences in the likelihood that the firm is recognized by Glassdoor or Fortune on their lists of “Best Places to Work” at any point during 2012 through 2016, with INCLUSIVE firms appearing more frequently. These findings are consistent with Au et al. (2021), who employ a textual analysis on online job reviews from Glassdoor.​com and Indeed.​com to determine a measure of sexual harassment within a firm’s workplace. Using these data, they find a negative association between seven-year (2011–2017) stock returns and the prevalence of sexual harassment within a firm.

In Panel B, we also document superior diversity and inclusion (D&I) scores, as measured by composite scores compiled by TruValue Labs and Arabesque. Both companies maintain proprietary databases with the aim of uncovering ESG data that offer insights into various dimensions of diversity and inclusion at the firm level. As the panel shows, our measure of board-based inclusivity correlates with the higher likelihood of the firm having higher D&I scores. Additional (untabulated) analyses contrasting specific datapoints collected by Arabesque indicate that inclusive firms differ from exclusive firms in a number of key aspects that collectively suggest broad cultural differences. For example, as compared to exclusive firms, inclusive firms are more likely to (1) offer more flexible work schedules, (2) provide child daycare services, (3) favor internal promotion, (4) set performance targets/objectives based on diversity and equal opportunity, (5) have formal policies to drive diversity and equal opportunities, and (6) have formal policies against forced or child labor.

The evidence presented in Table 3 corroborates our use of board gender representation as a signal of firm culture. In addition, these findings are consistent with several survey papers (e.g., Graham et al., 2016, 2022) that document that directors believe they can influence their firm’s corporate culture through their actions and behavior.

We turn our attention to assessing investors’ reactions to the #MeToo movement, beginning with an initial examination of cumulative abnormal returns (CAR) associated with the #MeToo movement events listed in Appendix 2. The CAR is the summation of the abnormal return around day 0 for each of the 37 event dates. We use seven alternative models as our benchmark for returns: (1) the Fama-French five-factor model (FF5), (2) the Fama-French-Carhart four-factor model (FFC4), (3) the Fama-French three-factor model (FF3), the CAPM model using (4) equally weighted and (5) value-weighted market returns (CAPM_EW and CAPM_VW), and the Daniel et al. (1997) model using (6) equally weighted and (7) value-weighted benchmark returns (DGTW_EW and DGTW_VW). We describe each approach in detail in Appendix 3. As shown in the first column of Table 4, depending on the model used to cumulate abnormal returns, we detect a significantly negative overall market reaction to the event dates of the #MeToo movement. In particular, we document negative overall market reactions ranging from −0.32% for the DGTW_VW model (insignificant, t = −0.97) to −1.98% for the CAPM_EW model (significant, t = −5.11). Recall, however, that our prediction focuses on the relative market reaction to the #MeToo movement depending on the existing culture of the firm.

To examine whether investor reactions to #MeToo movement vary based on gender representation on the board, the latter columns of Table 4 contrast the CAR for the subsample of exclusive firms with the CAR for the subsample of inclusive firms. Regardless of the model we use to calculate expected returns, we find that exclusive firms experience significantly negative abnormal returns as compared to the significantly positive abnormal returns enjoyed by inclusive firms. For example, using the FF5 model, we find that the sample of exclusive firms earn a CAR of −3.25%, whereas the sample of inclusive firms have a CAR of 2.33%; testing for a difference in the means produces a t-stat of 2.66 (p < 0.01). Thus, on a univariate basis, we find distinct differences in market reactions to the #MeToo movement based on whether a firm’s board excludes or includes women, with firms with zero board gender diversity, on average, shouldering the lion’s share of the negative abnormal stock return reaction.

Next, in Table 5 we test whether these differing reactions remain in a multivariate setting. To some extent, our seven alternative approaches to benchmarking abnormal returns control for differences in firm characteristics (e.g., SIZE and BOOK-TO-MARKET ratios). Yet, as we noted in our discussion of Table 2, there are other fundamental differences in firm characteristics for firms with boards with zero women vis-à-vis firms with boards with three or more women. Based on the significant differences found in Table 2, we estimate cross-sectional regressions that control for additional covariates—Ln (BOARD SIZE), LEVERAGE, ROA, % INDEP DIR, RETURN VOLATILITY, and SIZE.¹⁷ Based on Table 2, Panel B, we also include industry fixed effects. The dependent variable is the #MeToo CAR multiplied by 100, and our variables of interest are the EXCLUSIVE and INCLUSIVE indicator variables.

Consistent with our univariate results, we observe contrasting coefficients for our EXCLUSIVE and INCLUSIVE indicators. That is, we observe significantly negative (positive) coefficients for the exclusive (inclusive) firms, suggesting that the dissimilar returns we documented in Table 4 remain after including additional controls for firm and industry characteristics. F-tests comparing the EXCLUSIVE and INCLUSIVE coefficients confirm significant differences between the two subsamples. For example, in column [1], using the FF5 model to benchmark returns, the F-test comparing the two coefficients is significant at the 0.001 level. Thus, the analysis in Table 5 indicates that the change in investors’ beliefs about the risks injected by the #MeToo movement vary predictably in the cross-section, based on investors’ perceptions of firm culture (as signaled by gender diversity in the board room).

To depict the dynamics of the cumulative market reaction that we observe in Table 5, we plot the cumulative differences in market reactions for the EXCLUSIVE and INCLUSIVE subsamples over the 37 #MeToo event dates. The plot in Panel A of Fig. 3 illustrates a striking contrast in market reactions between the two subsamples as the movement gained momentum. We observe a steady increase in the size of the coefficient for the INCLUSIVE firms through the first nine event dates and continued growth of the coefficient throughout the movement. More importantly, considering the pattern for the EXCLUSIVE firms, we note a symmetric decrease in the coefficient for the EXCLUSIVE firms. These findings highlight the importance of simultaneously considering both the presence and absence of women in the boardroom. In addition, the growing contrast in returns between the two groups over the full nine-month timeframe underscores the value of using an extended timeline instead of narrowly focusing on the start of the movement.

Collectively, this picture suggests that the unfolding of the #MeToo movement continuously revised investors beliefs about the costs of fostering a culture that excludes women vis-à-vis one that creates an inclusive work place.

Despite our efforts to appropriately benchmark return performance and to control for observed differences between the two subsamples, a natural question remains as to whether unobserved differences between the two subsamples—unattributable to the unfolding #MeToo movement—drive these documented differences. To address this question, we test whether the same set of results emerges when we replace the #MeToo event dates with 37 randomly generated “pseudo” event dates from the same time period. Specifically, we re-estimate the analysis provided in Table 5, replacing the dependent variable with the cumulative abnormal returns associated with 37 machine-generated pseudo-events randomly drawn using the seed of “123” in Stata, computed using our seven alternative models to compute the benchmark returns. We also plot the cumulative market reaction over these 37 pseudo-event dates.

As shown in Table 6, when using these randomly generated pseudo-events, we no longer detect differences in the market response between firms with exclusive and inclusive cultures. In particular, the coefficients on our EXCLUSIVE and INCLUSIVE indicator variables no longer exhibit significance; nor do they contrast in sign. Further, in contrast to the results shown in Table 5, the F-tests for differences between the coefficients for EXCLUSIVE and INCLUSIVE no longer detect differences between the two subsamples. Moreover, as shown in Panel B of Fig. 3, plotting the cumulative differences in market reactions over the 37 pseudo-event dates does not produce the striking pattern shown in Panel A.

In an untabulated analysis, we repeat the placebo test using ten alternative seeds to randomly draw the 37 pseudo-events. In each of these ten additional rounds of placebo testing, we do not detect significant differences in cumulative market reactions between the EXCLUSIVE and INCLUSIVE subsamples. As such, the contrasting patterns documented in Panel A of Fig. 3 do not manifest for these alternative pseudo-event dates. This offers further evidence in support of the conclusion that the documented return patterns for the #MeToo event timeline stem from the growing momentum of the cause.

Thus far, our tests have compared the absence of women to the presence of at least three women. Yet, a question remains as to whether the market reaction differs depending on the presence of a critical mass of women as opposed to a token presence.

According to Kanter (1977), a minority group cannot exert influence over a larger body of people unless its numerical size reaches a critical mass. Our choice of designating three as the minimum critical mass is based on the survey evidence provided by Erkut et al. (2008), Konrad et al. (2008), and Torchia et al. (2011), who conclude that having at least this many women serving on corporate boards exacts changes within their firms. In this section, we take the question to the data and see whether our results hold for alternative thresholds of at least one or two women directors, instead.

Specifically, we create two TOKENISM variables: an indicator representing boards with at least one woman from 2012 to 2016 inclusive, and a second indicator for boards with two or more women during the same timeframe. Using these two indicators, we re-run the regressions shown in Table 5 with each variable (in lieu of using our INCLUSIVE indicator). Column [1] in Table 7 mirrors our analysis in Table 5, documenting contrasting negative and positive coefficients for EXCLUSIVE and INCLUSIVE firms, respectively. Yet, the inclusion of our TOKENISM variables in the remaining two specifications does not produce the same pattern (columns [2] and [3]). That is, these alternative approaches to measuring the inclusion of women on the board do not detect significant reactions by investors.

The findings in Table 7, along with those reported earlier in Tables 5 and 6, are consistent with several views about the role of women on the board. First, the market does not reward firms for having a token woman on its board, as evidenced by the insignificant coefficient on the TOKENISM variables. Second, our findings corroborate lawmakers’ choice of three women as the threshold guiding the inclusion of women on the boards of California-based firms. Finally, our findings suggest that examining the impact of having women on boards might be more nuanced than merely creating an indicator for the presence of women on the board (as several papers do) (Adhikari et al., 2019). That is, our findings support the view that having three or more women directors constitutes a critical mass of women, as it relates to creating a corporate culture that values women within the workplace. They also raise a question as to whether identifying inclusivity as a percentage of the board is appropriate, particularly if one ignores board size. We examine this issue next.

In Table 8, we explore the extent to which our findings change when we use alternative approaches to identify EXCLUSIVE and INCLUSIVE firms. We begin by identifying INCLUSIVE firms as those meeting or exceeding a threshold percentage (instead of a number) of women on the board. Beginning in 2007, Norway mandated a gender quota for listed companies of 40% women. Other European nations followed, including Belgium (33%) in 2012, Austria (35%) in 2013, France (40%) and Italy (33%) in 2014, and Germany (30%) in 2015. During this time, other European countries and the European Union itself took a softer approach by providing recommended thresholds but not actual gender quotas. In column [1], we continue to identify EXCLUSIVE firms (n = 481) based on the historical absence of women on the board (dating back to 2012) but make our threshold for inclusivity more restrictive by focusing on the 40 firms where women held at least 30% of the board seats since 2012. Despite the reduced sample size on INCLUSIVE, we continue to document significant, contrasting coefficients for the EXCLUSIVE and INCLUSIVE subsamples, as well as a significant difference between the two coefficients (F-test Pr[EXCL = INCL] = 0.001).¹⁸

In column [2], we return to our original thresholds for exclusivity and inclusivity, but we now limit our scope of consideration to the board gender diversity as of the end of 2016 and no longer factor in the board composition prior to 2016. Ignoring the historical perspective in our measurement increases the size of the EXCLUSIVE subsample by 15% (from 481 to 555 firms). Not surprisingly given the upward trend documented in Fig. 2, limiting attention to board gender composition as of 2016 leads to a striking increase in the number of firms identified as INCLUSIVE, from 122 to 316 (259%). As shown in column [2], using this more lenient approach to identifying INCLUSIVE cultures, the negative coefficient for the EXCLUSIVE firms remains robust, but the positive coefficient for INCLUSIVE firms no longer exhibits significance. This supports our choice of using a longer period of time, i.e., five years, to evaluate the culture of the firm, a choice consistent with the notion of corporate culture displaying a stickiness over time (Cheng & Groysberg, 2021; Schein, 1990). We, however, continue to detect the predicted significant difference across the two subsamples (F-test Pr[EXCL = INCL] = 0.006).

Several papers correlate corporate culture with the presence or percentage of women in the C-suite (Kunze & Miller, 2017; Tate & Yang, 2015), including Lins et al. (2022), who examine the association between women executives and investor response to the #MeToo movement. Accordingly, we consider the presence of women executives at the firm, as collected by ExecuComp, as an indicator of gender inclusivity. Ideally, we would like to replicate our methodology comparing C-suites with zero women over the same five-year period with C-suites with a critical mass of three or more women over the same time period. However, given that ExecuComp (1) covers a smaller sample of firms than BoardEx and (2) only lists the five highly compensated employees (as mandated by the 14A filing rules), our critical mass group comprises only four firms. Thus, we turn to alternative indicators of gender inclusivity.

In column [3], we require EXCLUSIVE firms to exclude women from both the board room and the C-suite dating back to 2012, and we require INCLUSIVE firms to have at least one woman on their board and at least one woman executive in the C-suite each year since 2012. This additional restriction based on executive gender diversity changes the EXCLUSIVE and INCLUSIVE samples to 122 firms and 200 firms, respectively. In column [4], we focus exclusively on executive characteristics in our measurement of culture. To do so, we identify EXCLUSIVE (INCLUSIVE) firms based on the absence of women (presence of at least one woman) in the C-suite dating back to 2012. This alternative approach ignores board gender diversity and results in sample sizes of 669 EXCLUSIVE firms and 247 INCLUSIVE firms.

As shown in columns [3] and [4], we no longer find the differing pattern in stock returns between the INCLUSIVE and EXCLUSIVE groups. Whereas the coefficients on EXCLUSIVE remain negative, they are no longer significantly different from zero; the coefficients on INCLUSIVE flip from positive to negative (albeit insignificantly different from zero). Further, testing for the differences between coefficients no longer produces significant differences between the two classifications.¹⁹

In the remaining specifications (shown in columns [5] through [8]), we replicate the analyses in columns [1] through [4] but replace the #MeToo event dates with the 37 randomly generated pseudo-event dates examined earlier in Table 6. Consistent with our earlier analyses based on pseudo-events, we no longer detect differences in the market response between firms with EXCLUSIVE and INCLUSIVE cultures—regardless of our approach to measuring culture. And, again, the F-tests for differences between the coefficients for EXCLUSIVE and INCLUSIVE no longer detect differences between the two subsamples.

The evidence presented in Table 8 indicates that our results are sensitive to alternative approaches to measuring corporate culture—particularly those based on executive characteristics. Yet, we caution readers against interpreting these findings as indicative of gender diversity in the C-suite being unimportant to a firm’s culture. As shown in Table 3, gender diversity in the C-suite and the boardroom are correlated, and both likely provide a signal of firm culture to investors. We further caveat the results shown in Table 8 by acknowledging the data limitations in the ExecuComp database, which reduce the power of these tests.

Collectively, the evidence suggests that investors respond negatively to the events tracking the timeline of the #MeToo movement for firms with an absence of women on their boards between 2012 and 2016. We interpret these findings as suggesting that investors’ perceptions of culture (as signaled by board gender diversity) shape their response to the unfolding #MeToo movement. An alternative view of the evidence may be that the negative response potentially reflects investors’ fear that this movement will force women into the board room and, in so doing, push the firm out of its optimal board composition (Ahern & Dittmar, 2012; Greene et al., 2020; Levitt, 2021). To explore this possibility, we examine the market reactions to a subsequent appointment of a woman (or women) to boards for our sample of EXCLUSIVE firms that have historically excluded women. We examine appointments from October 2017 (the beginning of the #MeToo movement) through May 2020.

Table 9 documents the cumulative abnormal returns for three windows surrounding the announcement dates: day [0] only, days [0,+1], and days [−1,+1]. We use two methods to determine the event dates. In Panel A, we hand-collect new director appointment announcements by searching for the firms’ earliest press releases or third-party media articles using both firms’ names and the directors’ names as keywords. Of the 163 announcements contained in our sample, 154 encompass the appointment of one woman only and nine encompass the appointment of two women. In Panel B, we use BoardEx as our source for “role effective” dates (Green & Homroy, 2018). Accordingly, Panel B includes 194 appointments.

The multivariate regressions in Tables 5, 6, 7 annd 8 control for various firm and industry characteristics. In Panel A of Table 10, we provide evidence that our main finding—of a differential market reaction to the #MeToo Movement when comparing firms with exclusive cultures to firms with inclusive cultures—is robust to alternative approaches to constructing matched samples. As shown in the first column of Panel A, we continue to detect contrasting market reactions to the movement depending on the existing culture of the firm when we execute propensity-score matching (using the control variables included in our earlier multivariate analyses) to identify similarly situated control firms. In column [2], we find the results are robust to using entropy balancing to construct the matched sample. In untabulated analyses, results continue to hold when we execute a “hard matching” approach to identify the nearest neighbors in terms of SIZE, ROA, and INDUSTRY. Consistent with our earlier findings, we continue to detect no significant coefficients or differences between subsamples when we replace our #MeToo event dates with the pseudo-event dates (columns [3] and [4]).

In Panel B of Table 10, we conduct several additional robustness tests. In the first two columns, we consider potential confounding events. In column [1], we exclude 857 firms that announce earnings on any of the 37 #MeToo event dates. Consistent with the expectation that earnings announcement confounds are evenly distributed across the full sample, this reduces the exclusive and inclusive subsamples by 40% and 39%, respectively. Despite the considerable reduction in power associated with the reduction in sample size, we continue to detect significant differences in market reactions across the two subsamples, as evidenced by a significant F-test that manifests only when we examine the actual #MeToo event dates (as opposed to the pseudo #MeToo event dates, as shown in column [5]). In an untabulated analysis, we drop the #MeToo event dates that fall within the week surrounding an earnings announcement event and use the FF5 model to compute the average abnormal return on the remaining #MeToo event or pseudo-event dates. Again, we continue to document contrasting market reactions between the inclusive and exclusive firms along the MeToo event dates. In column [2] we exclude 18 firms with sexual harassment scandal revelations during the nine-month #MeToo period. We continue to detect significant differences in market reactions across the two subsamples, as demonstrated by a significant F-test that manifests only when we examine the actual #MeToo event dates (as opposed to the pseudo #MeToo event dates, as shown in column [6]).

We next turn our attention to controlling for firm size (using alternative approaches) and sales growth. In column [3], we document that our results are robust to the inclusion of the log of assets and sales growth in the regression. In column [4], we examine whether our results hold when we concentrate on boards with six or more directors only. The California gender diversity law that requires public companies headquartered in California to maintain a board with at least three women applies only to those companies with six or more directors. We find that our results are robust to eliminating the 177 firms with less than six directors.

Finally, we examine whether California firms are more likely to have boards with three or more women during our 2012–2016 timeframe, compared to firms headquartered in other states. In 2013, the California legislature passed a resolution encouraging (but not mandating) all publicly held firms in California to have a minimum number of women on their boards by December 2016.²⁰ We find that of the 122 INCLUSIVE firms, only 10 were headquartered in California. In the same vein, the inclusion of state fixed effects in our main analyses does not alter our findings (untabulated).