A. Public Reaction to the OSC Proposed Amendment

The public first became aware that the OSC was considering legislation on board gender diversity in July 2013, when the OSC released Staff Consultation Paper 58-401, which proposed “comply or explain” disclosure requirements to encourage greater gender diversity on corporate boards and in senior management. The OSC invited feedback by September 2013 and received 89 comment letters from a wide range of stakeholders, including corporations, advocacy groups, investment firms, professional associations, academics, and individuals (see Table 3). The proposal received broad public support. Overall, 82% of submissions expressed a positive tone, 18% were neutral, and none were negative.Footnote ¹⁰ Corporate responses were largely favorable, with 74% in support and 26% neutral. The highest level of support came from advocacy groups, with 95% supporting the proposal and only 5% expressing a neutral stance.

TABLE 3Comment Letters to the OSC about OSC’s Board Gender Diversity Rule

TABLE 3 Long description

The table is divided into two panels. Each panel contains six columns: Entity Type, Total Number of Comment Letters, Percentage Negative, Percentage Neutral, Percentage Positive, and Average Polarity Score.

Panel A. Comment Letters due September 2013:

* Academics and think tanks: 4 letters, 0% negative, 25% neutral, 75% positive, 0.146 polarity.

* Advocacy groups: 22 letters, 0% negative, 5% neutral, 95% positive, 0.147 polarity.

* Government and public bodies: 3 letters, 0% negative, 33% neutral, 67% positive, 0.103 polarity.

* Individuals: 17 letters, 0% negative, 24% neutral, 76% positive, 0.153 polarity.

* Investment firms: 7 letters, 0% negative, 14% neutral, 86% positive, 0.147 polarity.

* Professional associations: 13 letters, 0% negative, 15% neutral, 85% positive, 0.141 polarity.

* Corporations: 23 letters, 0% negative, 26% neutral, 74% positive, 0.133 polarity.

* Total: 89 letters, 0% negative, 18% neutral, 82% positive, 0.142 polarity.

Panel B. Comment Letters due April 2014:

* Academics and think tanks: 1 letter, 0% negative, 0% neutral, 100% positive, 0.107 polarity.

* Advocacy groups: 7 letters, 0% negative, 14% neutral, 86% positive, 0.163 polarity.

* Individuals: 9 letters, 11% negative, 11% neutral, 78% positive, 0.128 polarity.

* Investment firms: 10 letters, 0% negative, 50% neutral, 50% positive, 0.113 polarity.

* Professional associations: 18 letters, 0% negative, 6% neutral, 94% positive, 0.160 polarity.

* Corporations: 7 letters, 0% negative, 43% neutral, 57% positive, 0.088 polarity.

* Total: 52 letters, 2% negative, 21% neutral, 77% positive, 0.135 polarity.

In January 2014, the OSC released a revised proposal, incorporating a single addition: the inclusion of director term limits. Stakeholders were again invited to submit feedback, with a deadline of April 2014. The response remained overwhelmingly positive: 77% had a positive tone about the amendment, with 21% neutral and 2% negative. The consistently strong support across both consultation periods indicated early on that the proposal was likely to proceed to implementation.

Beyond stakeholder comment letters, media coverage of the proposed amendment was also broadly supportive, reflecting and reinforcing the broad public and institutional approval of the initiative. Table 4 presents an analysis of 211 articles published between 2012 and 2016 in three major Canadian news outlets—Globe and Mail, Financial Post, and Toronto Star—that specifically discuss board gender diversity and the OSC’s initiative. Applying sentiment analysis, we classify 62% of the articles as positive in tone, 38% as neutral, and none as negative. Media sentiment was particularly favorable in 2014, the year the revised proposal was released and a second round of comments was invited: 71% of articles that year conveyed a positive tone.Footnote ¹¹ This consistently supportive media environment suggests broad public acceptance of the amendment and reinforces the perception that its implementation was likely.

TABLE 4Media Coverage about Board Gender Diversity and OSC’s Rule

TABLE 4 Long description

The table consists of six columns: Year, Total Number of News Articles in the Press, Percent of News Articles in the Press Which are Negative, Neutral, or Positive, and Average Polarity Score.

* 2012: 6 total articles; 0 percent negative, 33 percent neutral, 67 percent positive; average polarity score 0.111.

* 2013: 35 total articles; 0 percent negative, 37 percent neutral, 63 percent positive; average polarity score 0.107.

* 2014: 52 total articles; 0 percent negative, 29 percent neutral, 71 percent positive; average polarity score 0.113.

* 2015: 57 total articles; 0 percent negative, 28 percent neutral, 72 percent positive; average polarity score 0.127.

* 2016: 61 total articles; 0 percent negative, 57 percent neutral, 43 percent positive; average polarity score 0.094.

* Total: 211 total articles; 0 percent negative, 38 percent neutral, 62 percent positive; average polarity score 0.110.

B. How Did the Market React to the Policy Announcement?

In this section, we study market reaction to the OSC policy to understand how the principles-based policy was perceived by shareholders.

3. Announcement Returns

Panel A of Table 5 reports the average CARs in the (0,0), (0,+1), and (−1,+1) announcement windows, along with t-statistics and p-values for testing whether the average CARs are statistically different from 0, following Kolari and Pynnönen (Reference Kolari and Pynnönen2010).Footnote ¹⁴ While the average CAR during these windows is positive but not statistically significant, firms without a disclosed voluntary female representation policy and those with all-male boards exhibit positive and mostly statistically significant announcement CARs (at the 5% and 1% level, respectively). Specifically, the 2-day (0,+1) CAR is 1.4% for firms without a female director policy and 2.0% for firms with all-male boards.Footnote ¹⁵ Other firms do not exhibit significant announcement CARs. Panel B shows that 63% of firms without a female representation policy and 72% of those with all-male boards have positive (−1,+1) CARs, suggesting the results are not driven by outliers. Table 6 presents ordinary least squares (OLS) regressions of the CARs, focusing on the (0,+1) window following MacKinlay (Reference MacKinlay1997).Footnote ¹⁶ We include an indicator for firms without a disclosed voluntary female representation policy (models 1 and 2), an indicator for firms with all-male boards (models 3 and 4), and the proportion of male directors on the board (models 5 and 6). All specifications control for total assets and 1-digit SIC industry, with models 2, 4, and 6 also controlling for market-to-book assets, return-on-assets, and debt/assets. The coefficients on our key explanatory variables of interest align with the univariate analysis in Table 5. In models 1 and 2, firms without a disclosed female representation policy have CARs that are 1.2 percentage points higher than those that disclose such a policy. Similarly, in models 3 and 4, firms with all-male boards have CARs that are 1.2 percentage points higher than for those with female directors.Footnote ¹⁷

TABLE 5Univariate Analysis of CARs around the OSC’s Announcement

TABLE 5 Long description

Panel A reports Mean C A R for five groups across three time windows: (0,0), (0,+1), and (-1,+1).

1. All Firms (N = 274): Mean C A R ranges from 0.543% to 1.016% with p-values between 0.233 and 0.527.

2. No Female Director Policy in 2013 (N = 213): Mean C A R is 0.786% (p = 0.045) for (0,0) and 1.387% (p = 0.038) for (0,+1).

3. Female Director Policy in 2013 (N = 61): Mean C A R is 0.346% for (0,0) and becomes negative for (0,+1) at -0.277% and (-1,+1) at -0.401%.

4. All-Male Board in 2013 (N = 127): Shows statistically significant positive returns across all windows: 1.115% (p = 0.006), 2.001% (p = 0.003), and 1.290% (p = 0.031).

5. >0 Female Directors in 2013 (N = 147): Mean C A R is 0.319% for (0,0), 0.165% for (0,+1), and -0.103% for (-1,+1).

Panel B reports the Percentage of Firms with Positive C A R s for the (0,+1) window:

* All Firms: 58%

* No Female Director Policy: 63%

* Female Director Policy: 43%

* All-Male Board: 72%

* >0 Female Directors: 47%

TABLE 6Regression Analysis of CARs around the OSC’s Announcement

TABLE 6 Long description

A regression table with eight columns labeled 1 through 8. The dependent variable is C A R (0, +1).

Key independent variables and their coefficients (standard errors in parentheses):

* No Female Director Policy sub 2013: Model 1 is 0.012** (0.005); Model 2 is 0.013*** (0.005).

* All-Male Board sub 2013: Model 3 is 0.012** (0.005); Model 4 is 0.015*** (0.005); Model 7 is 0.014*** (0.005); Model 8 is 0.014*** (0.005).

* Male Director Ratio sub 2013: Model 5 is 0.055** (0.022); Model 6 is 0.065*** (0.024).

* All-Male Board through 2017: Model 7 is -0.011 (0.008); Model 8 is -0.007 (0.008).

Control variables include Institutional Ownership, Media Coverage, Log(Assets), Market-to-Book Assets, and R O A. All models include Industry fixed effects.

Model Statistics:

* Number of observations: Ranges from 234 to 274.

* Adjusted R-squared: Ranges from 0.078 to 0.127.

Significance levels: *** p < 0.01, ** p < 0.05, * p < 0.10.

Despite the insignificant CARs for firms with some female directors in Table 5, models 5 and 6 of Table 6 show that the CARs are increasing in the proportion of male directors on firms’ boards. Combined with the fact that firms with female directors added more women post-regulation, albeit at a lower rate than firms with all-male boards (see Supplementary Material Figure C.1), these results suggest the following: having female directors at the time of the event announcement may signal a firm’s willingness to add more women to its board, even without regulation. In contrast, all-male boards may be perceived as less likely to do so. This does not imply that the regulation has no effect on firms with female directors, but rather that the behavioral shift expected of them is smaller compared to firms with all-male boards.

The results in Tables 5 and 6 contrast sharply with the negative abnormal returns documented for firms subject to California’s SB-826 law mandating gender diversity, as shown by Hwang, Shivdasani, and Simintzi (Reference Hwang, Shivdasani and Simintzi2020) and Greene, Intintoli, and Kahle (Reference Greene, Intintoli and Kahle2020), though these findings have been challenged by Allen and Wahid (Reference Allen and Wahid2023). One might hypothesize, based on these prior studies, that the positive returns we observe reflect relief that the regulation would not force changes on firms.Footnote ¹⁸ If this were true, we would expect the positive abnormal returns to be concentrated among firms that continued to have all-male boards post-regulation. In models 7 and 8, we re-estimate models 3 and 4, adding an indicator for firms with all-male boards in 2013 that still had all-male boards in 2017. The coefficient on this indicator is negative and not statistically significant, contradicting this hypothesis. Coupled with our Section VI findings that the regulation led to an increase in female director representation, these results suggest it is highly unlikely that the positive CARs for firms with all-male boards reflect an expectation that the principles-based regulation would not lead to real change.

A potential concern is that the non-random assignment of firms to treated and control groups (e.g., all-male vs. non-all-male boards) might be correlated with our event returns. We address this issue by using two approaches suggested by Roberts and Whited (Reference Roberts, Whited, Constantinides, Harris and Stulz2013). First, we compare observable characteristics between firms with all-male boards and firms with female directors (see Supplementary Material Table C.2). Both groups have similar proportions of female top-five executives (6.6% vs. 6.8%), confirming that our results in Tables 5 and 6 are related to OSC rules on female directors, but not executives. We find similar patterns when we compare firms with and without a voluntary director gender diversity policy. Second, we conduct a placebo test using 500 “placebo” event dates around the actual event (see Supplementary Material Figure C.2 and Supplementary Material Table C.3). This test shows that the regression coefficients in Table 6 fall far in the right tail of the distribution of the placebo distribution, indicating that results are not driven by persistent differences between treatment and control groups and are unlikely to be due to chance.

In untabulated tests, we do not find statistically significant market reactions to female director nominations. Because standalone announcements of female director appointments are rare, we identify nominations using proxy filing dates; however, proxy statements bundle considerable market-relevant information, making it difficult to isolate the valuation effect of a director nomination. Moreover, many appointments were likely anticipated following the OSC’s proposed ruling and, for some firms, the adoption of female director targets.

C. Incentives for Firms to Comply Early

The high visibility and favorable framing of the proposed OSC regulation may have prompted firms to adjust their governance practices in advance of formal implementation, particularly for those sensitive to reputational concerns. Consistent with this view, the market reaction results indicate that little new economically material information was released after the July 30, 2013 proposal. Among the major regulatory milestones, only the July 30, 2013 proposal generates statistically and economically significant abnormal returns, while we observe no significant reaction to the Jan. 16, 2014 revision or the Oct. 15, 2014 formal adoption (see Section IV.B.5 and Supplementary Material Table C.4). This pattern suggests that the informational content of the amendment was largely incorporated at the initial proposal, consistent with firms beginning anticipatory adjustments prior to formal implementation.

By adopting governance changes in anticipation of the proposed amendment, firms could influence the evolution of emerging best practices, signal readiness for regulatory change, and mitigate the risks associated with a last-minute compliance rush. Some firms may have also aimed to influence the final form of the regulation by demonstrating the practical feasibility of disclosure-based requirements, thereby positioning voluntary compliance as a credible alternative to more prescriptive interventions, such as mandatory quotas. This concern was salient in the policy debate: several comment letters expressed concerns that the proposed amendment might be insufficient and suggested that quotas could be necessary to achieve more meaningful progress. By taking early action and nominating female directors during the spring 2014 proxy season, firms also positioned themselves to secure qualified female candidates ahead of anticipated competition following the rule’s formal adoption.

A. Assessing Compliance with the OSC Amendment

To analyze firms’ choices of whether to comply or explain with disclosures on board diversity, we manually collect data from annual proxy statements for 2011–2018 of firms’ disclosed policies regarding female board representation. For each firm, we record whether it has a diversity policy (item 2), whether it considers women in the director identification or selection process (item 3), and whether it has implemented a target for the number or percentage of women directors (item 5). We categorize firms into three distinct groups based on their adherence to these three aspects of the OSC amendment (see Panel A of Table 1).

Full Compliers meet all three key provisions—they disclose a gender diversity policy for board nominations (item 2), consider the representation of women in the director identification and selection process (item 3), and set formal targets for the representation of women on the board (item 5). These firms show the highest level of engagement with the regulation, complying with all provisions regarding board diversity.

Partial Compliers satisfy items 2 and 3 but do not adopt a formal target. These firms indicate that they take gender into account in the director identification and selection process, acknowledging gender diversity as a factor in board appointments. However, they stop short of setting quantitative benchmarks or targets.

Explainers fulfill the minimum level of compliance. They disclose a gender diversity policy but do not adopt a formal target, nor do they consider the representation of women in the director identification and selection process. Their disclosures offer explanations for non-adoption rather than demonstrating active integration of gender diversity into governance practices.

B. Incentives for Compliance Choice

Full compliance, particularly when firms adopt more ambitious female director targets, sends a stronger signal to investors, analysts, and regulators that the firm is genuinely committed to gender diversity. This can enhance its attractiveness to long-term institutional investors and strengthen public messaging around diversity leadership. However, full compliance requires a strong internal commitment and may require structural changes. We posit that full compliers are more likely to be larger firms with lower relative compliance costs and firms with greater access to qualified female directors.

Partial compliance enables firms to signal a commitment to good governance and avoid the perception of inaction without adopting quantitative targets that might constrain board decisions. For these firms, such targets may be seen as overly prescriptive or may generate internal tensions, including concerns about tokenism or challenges to merit-based selection. In some cases, partial compliance may also reflect a desire to align with peers that have not yet widely adopted gender targets.

Explainers may emphasize managerial discretion in board appointments, signaling a strategic resistance to external accountability mechanisms and prevailing normative expectations regarding board composition. For these firms, disclosing formal diversity policies may risk exposing a lack of progress or creating unwanted future accountability. In contrast, offering explanations enables them to retain control over the narrative without making binding commitments. Among smaller firms, tracking and reporting on board gender representation may be perceived as administratively burdensome. In contexts where investor scrutiny is minimal, the firm operates within male-dominated industries, or public visibility is low, the incentives to comply with diversity expectations are significantly reduced. Moreover, explainers may face greater frictions in appointing female directors—for example, due to fewer network connections to female candidates, a limited local pool of potential female directors, or higher recruitment costs. Under these conditions, the perceived benefits of aligning with normative standards may be outweighed by the costs. Moreover, some firms may view diversity initiatives as externally imposed, politically motivated, or fundamentally misaligned with their corporate identity and values.

C. Levels of Compliance

Panel B of Table 1 illustrates the evolution in compliance with the OSC’s board diversity disclosure requirements. Prior to the OSC’s initial public call for commentary on board diversity in July 2013, only about 12% of firms disclosed a diversity policy. However, in the wake of the OSC announcement, there was a notable upswing, with 23% of firms reporting the existence of a diversity policy by the end of 2013, sharply increasing to 95% by 2014. Compliance became nearly universal after the official effective date of the amendment at the end of 2014, with all but two firms disclosing a diversity policy. By 2017, all firms disclosed a diversity policy, with compliance of item 2 at 100%. Notably, though all firms disclosed a diversity policy, not all firms embraced board diversity. While some show a clear commitment to board gender diversity in their diversity policies, others are vocal in their reluctance to consider gender in board appointments, stating that they give precedence to the role of skills and experience in selecting board members. Such firms often maintain that they make board appointments based on merit without considering gender or racial attributes.

The trend for firms disclosing their consideration of women in the director identification or selection process (item 3) displays a similar upward trajectory over the sample period. Prior to the OSC’s announcement, only about 11% of companies took gender into account when identifying or selecting directors. Following the 2013 announcement, this figure rose to 21%, indicating a growing awareness and response to the upcoming implementation of the amendment. With the OSC’s policy implementation in 2014, compliance increased significantly to 89%. This high level of compliance persists through the end of the sample period, with 93% of firms disclosing their consideration of gender in the director identification or selection process in 2018.

Item 5 calls for companies to disclose whether they have set targets for female board representation. As shown in Panel C of Table 1, it was rare for firms to have gender targets in place before 2014. However, coinciding with and following the amendment’s introduction in 2014, target adoption rose from eight firms in 2013 to 40 firms in 2014, to 77 firms by 2017, and to 102 firms (41% of all firms) by 2018. This indicates a notable escalation in the embrace of quantifiable goals for the inclusion of women on corporate boards.

While the amendment did not mandate the form of these targets, whether as a percentage or a specific number, most firms that implement a target utilize a percentage. Common benchmarks include female director targets of 25%, 30%, and 33%. Over the entire sample period, the mean target among firms adopting targets is 28.9%, while the median target is 30%, with targets ranging from 12.5% to 50%. Panel C of Table 1 reports the evolution of the female director targets adopted by Full Compliers. As indicated by the mode, the most common target used before 2014 was 25%, while the most common target utilized after the amendment’s introduction is 30%. The average target is just under 29% each year since 2014, with the median at 30% since 2014. Notably, firms rarely scale back existing targets once adopted, suggesting that target adoption reflects persistent rather than symbolic commitments to board gender diversity.Footnote ²⁰

The targets adopted by Canadian firms under the OSC’s principles-based framework differ from the quota structures implemented in several other jurisdictions. Norway requires that at least 40% of board members be of each gender, while the European Union adopted a 40% quota for women on corporate boards set to take effect in mid-2026. California’s now-invalidated SB 826 mandate similarly imposed minimum numerical requirements based on board size, requiring two female directors for five-member boards and three female directors for boards with six or more directors. While the voluntarily adopted targets in our sample are on average somewhat lower than the quotas imposed under mandatory regimes, they nevertheless represent meaningful commitments to increasing female board representation. More broadly, the variation in targets under the OSC framework is consistent with the idea that a principles-based regime allows firms flexibility in determining target levels rather than converging toward a single mandated threshold.

Among Partial Compliers and Explainers that do not employ targets, 75% offer a rationale. The most frequently cited reasons are as follows:

• • Appointments are based on merit/Skills and experience are most important (63.2%).

• • Targets are not necessary or advisable/Targets are arbitrary (10.3%).

• • We are too small for a diversity policy or to consider targets (3.5%).

• • Targets are not an appropriate method to increase diversity (2.6%).

• • Targets do not result in identifying or selecting the best candidate (2.5%).

• • Infrequent turnover makes targets impractical (1.0%).

Figure 1 shows the proportions of firms that fall into one of the three compliance categories. The solid line illustrates the full sample, with the long-dashed line representing firms with female directors in 2013, and the short-dashed line representing firms with all-male boards in 2013. Graph A shows the proportion of firms with a gender diversity policy (item 2). Graph B indicates a significant surge in the proportion of firms classified as Partial Compliers around the amendment, from 19% in 2013 to 74% in 2014. Firms with all-male boards in 2013 are more likely than firms with a female director in 2013 to fall into the Partial Complier category after the OSC amendment. However, the percentage of Partial Compliers diminishes for both groups after 2014 as firms advance to full compliance status. Indeed, over the sample period, Panel C shows a notable increase in Full Compliers: with just 6% in 2014, this figure rises to 30% by 2017 and then to 41% in 2018. Notably, the proportion of Full Compliers increases more substantially for firms that had all-male boards in 2013, exhibiting a ninefold increase from 2014 to 2018, compared to a twofold increase for firms that had at least one female director in 2013.

FIGURE 1Gender Diversity Policy in Canada Following OSC’s Amendment

Figure 1 presents data on Gender Diversity Policy for Canadian firms included in the S&P TSX Composite Index, with directorship data in BoardEx and financial data in Compustat, over the 2011–2018 period. The sample is split by whether the firm had female directors in 2013. Graph A plots the average fraction of firms with a gender diversity policy each year (complying with item 2 in OSC’s regulation). Graph B plots the average fraction of Partial Compliers—firms that consider gender diversity in their director nomination process but do not disclose a targeted number of women directors (complying with items 2 and 3 but not 5 in OSC’s regulation). Graph C plots the average fraction of Full Compliers—firms that consider gender diversity in their director nomination process and disclose a targeted number of women directors each year (complying with items 2, 3, and 5 of OSC’s regulation). Graph D plots the average fraction of Explainers—firms that comply with 2 by disclosing a gender diversity policy but do not comply with 3 and 5. The solid lines represent the full sample, the short dashed line represents firms with all-male boards in 2013, and the long dashed lines represent firms with at least one female director in 2013.

FIGURE 1 Long description

A four-panel line graph set. Each graph shares a horizontal x-axis representing years from 2010 to 2018. The vertical y-axis represents percentages. Three data series are present in each: a solid line with squares for the Full Sample, a short dashed line with circles for All-Male Board in 2013, and a long dashed line with triangles for firms with greater than 0 Female Directors in 2013.

* Graph A, Percent with a Gender Diversity Policy (Item 2): The y-axis ranges from 0 to 100 percent. All three lines remain below 20 percent until 2013, then show a sharp vertical increase to nearly 100 percent by 2014, plateauing through 2018.

* Graph B, Percent with Partial Compliance (Items 2 and 3): The y-axis ranges from 0 to 100 percent. Lines show a sharp spike from under 20 percent in 2013 to approximately 70 to 80 percent in 2014, followed by a steady decline toward 50 to 60 percent by 2018.

* Graph C, Percent with Full Compliance (Items 2, 3, and 5): The y-axis ranges from 0 to 50 percent. All lines are near 0 until 2013. From 2014 to 2018, there is a steady linear increase. The triangles (firms with female directors) lead, reaching nearly 50 percent, while circles (all-male boards) lag, reaching approximately 30 percent.

* Graph D, Percent with Explanation (Item 2, but not 3 or 5): The y-axis ranges from 0 to 50 percent. Trends remain low, peaking around 2016. The all-male board group (circles) shows the highest rate of explanation at approximately 15 percent, while the group with female directors (triangles) remains the lowest, near 5 percent.

Graph D shows that the proportion of Explainers, firms with only a diversity policy (item 2), increases from 6% in 2014 to 8% between 2015 and 2017, then declines to 7% by 2018. After adopting a diversity policy, many firms begin to add items 3 and 5 (considering women in the director identification or selection process and setting targets for female board representation), moving from Explainers to Partial or Full Compliers. This progression underscores a shift toward increased adoption of structured gender diversity policies in board governance over time, even among firms that initially had no women directors. We find similar trends toward more supportive language regarding board gender diversity used in the proxy statements post-amendment (see Supplementary Material Table C.7 and Supplementary Material Figure C.4).

1. Multivariate Analyses of Determinants of Compliance

We next study cross-sectional differences that explain varying levels of compliance with the OSC amendment. We posit that firms facing greater frictions in identifying and appointing female directors are less likely to adopt more female-friendly diversity policies. We measure these frictions using two proxies capturing a firm’s access to female directors: i) the number of female directors on corporate boards for all firms within 100 km of the firm’s headquarters that are from the same 1-digit SIC industry, and ii) the average number of interlocks between the board’s directors and female directors on other firms’ boards. We expect higher values for both proxies to predict a greater likelihood of adopting more friendly gender diversity policies post-amendment.

Building on the idea that a firm’s culture can be influenced by its connections with other firms, we posit that firms with directors who are interlocked with directors of other firms that have already implemented board diversity practices will be more likely to adopt such practices themselves. Specifically, we use the following network variables to measure the influence of director networks on board diversity policy: i) the number of board interlocks with directors at firms that consider gender diversity in board nominations, and ii) the number of board interlocks with directors at firms with female director targets.

Finally, we posit that firms whose management is insulated from market pressures are more likely to eschew board diversity practices, opting to explain rather than comply. To capture whether management is shielded from market pressures, we consider whether the firm is a controlled corporation, measured by dual-class voting shares or closely held shares exceeding 30% of shares outstanding, and we also control for board independence. In the same vein, we posit that firms with greater media coverage are more likely to comply, as they are likely subject to more external scrutiny. We measure media coverage by the number of times a firm is cited in the Financial Times, Financial Post, and Toronto Star during the year. Additionally, we control for board characteristics such as size; average age and tenure of directors; average degree centrality of a board’s directors based on the number of other directorships ever held; and firm characteristics such as total assets, market-to-book assets, return-on-assets, institutional ownership, and debt/assets. All specifications include year and 1-digit SIC industry fixed effects. When constructing industry fixed effects, we use a 1-digit SIC classification rather than a finer 2-digit classification because, although Canada has the same industries as the U.S., finer categories often contain few firms, leading to sparse counts.

We report the results of linear probability regressions in Table 7.Footnote ²¹ Model 1 examines the determinants of full compliance for the entire sample, with the dependent variable indicating whether the firm is a Full Complier (i.e., complies with items 2, 3, and 5). Thus, model 1 compares Full Compliers with the remainder of the sample (Partial Compliers and Explainers). The results show that firms headquartered in regions with a higher concentration of local female directors within the same 1-digit SIC industry are significantly more likely to fully comply.Footnote ²² Similarly, firms with more board interlocks to companies that have female directors or to Full Compliers are also more likely to fully comply. In addition, greater media coverage is associated with a higher likelihood of full compliance, while firms with concentrated ownership structures (i.e., controlled corporations) are significantly less likely to fully comply.

TABLE 7Regression Analysis of Gender Diversity Policies

TABLE 7 Long description

A regression table with four columns representing different models.

Header Information:

* Column 1: Dependent Variable is Full Compliance; Sample is Full Compliers vs. Partial Compliers and Explainers.

* Column 2: Dependent Variable is Full Compliance; Sample is Full Compliers vs. Partial Compliers.

* Column 3: Dependent Variable is Partial Compliance; Sample is Partial Compliers vs. Explainers.

* Column 4: Dependent Variable is Target Magnitude percentage; Sample is Full Compliance.

Key Regression Coefficients (Standard Errors in parentheses):

* No. of Local Female Industry Directors: 0.003* (0.001) in Model 1; 0.002* (0.001) in Model 2.

* Interlock with Female Directors: 0.039** (0.019) in Model 1; 0.046** (0.022) in Model 2.

* Interlock with Full Compliance: 0.606*** (0.113) in Model 1; 0.645*** (0.115) in Model 2.

* Controlled Corporation: -0.118*** (0.041) in Model 1; -0.112** (0.043) in Model 2.

* Institutional Ownership: 0.0006** (0.0003) in Model 4.

* Media Coverage: 0.062*** (0.020) in Model 1; 0.061*** (0.020) in Model 2.

Summary Statistics:

* Other firm controls, Year fixed effects, and Industry fixed effects are included (Yes) for all models.

* Number of observations: 1,269 (Model 1), 1,151 (Model 2), 959 (Model 3), 309 (Model 4).

* Adjusted R-squared: 0.309 (Model 1), 0.310 (Model 2), 0.074 (Model 3), 0.317 (Model 4).

The economic significance of these effects is substantial. A 1-standard-deviation increase in the number of female directors in the same industry and geographic area raises the probability of full compliance by 5.9 percentage points. Given that only 25% of firms in the sample are Full Compliers, this translates to a 23.4% increase relative to the baseline compliance rate, highlighting the importance of local access to qualified female directors. Similarly, a 1-standard-deviation increase in board interlocks with female directors is associated with a 5.9 percentage point increase in the likelihood of full compliance, a 23.4% relative increase. Connections appear especially influential: firms with more interlocks to other fully compliant firms see an 11.3 percentage point higher probability of full compliance, a 45% increase over the baseline. In contrast, controlled corporations are 11.8 percentage points less likely to fully comply. This suggests that agency concerns or resistance to external pressure may play a role in discouraging adoption of comprehensive diversity practices.

Model 2 examines the determinants of full compliance by comparing Full Compliers with Partial Compliers, excluding Explainers. The results closely mirror those in model 1, which compares Full Compliers to both Partial Compliers and Explainers, suggesting that Partial Compliers and Explainers behave similarly along observable dimensions. To further explore this, model 3 focuses specifically on distinguishing Partial Compliers from Explainers—the two groups that do not adopt female director targets. Model 3 shows limited explanatory power (Adjusted R ² = 0.074), and none of the predictors are statistically significant. This likely reflects the narrow behavioral distinction between these two groups. Variables that effectively explain full compliance—such as board interlocks, proximity to potential female directors, connections to full compliers, or media attention—do not meaningfully differentiate firms that opt out of targets. The choice to partially comply versus explain may instead be shaped by unobserved or qualitative factors, such as disclosure preferences or organizational culture. Alternatively, this is consistent with the notion that partial compliance is essentially similar to explaining from a cost perspective: access to female director proxies or governance mechanisms does not predict whether a firm partially complies versus provides an explanation for non-compliance. In other words, partial compliance does not appear to be systematically driven by these observable determinants, in sharp contrast to full compliance.

More broadly, the results suggest that firms’ responses to the OSC amendment are heterogeneous rather than uniformly substantive. While some firms appear to embrace the spirit of the amendment through more comprehensive diversity disclosures and policies, some firms may instead respond in a more symbolic manner by limiting compliance to disclosures that avoid measurable commitments. The observational similarity between Partial Compliers and Explainers is consistent with this interpretation.

Model 4 examines the determinants of the magnitude of self-adopted female director targets, restricting the sample to firms that fully comply (i.e., those that set a target). In this specification, the dependent variable is continuous, capturing cross-sectional variation in the level of the target rather than the decision to adopt one. The results indicate that the observable factors that strongly predict the adoption of targets in models 1 and 2—such as director supply, board interlocks, and exposure to other compliant firms—do not explain variation in the size of the targets once a firm chooses to set one. Most coefficients are small and statistically insignificant, with the exception of institutional ownership, which is positive and significant, suggesting that firms with greater institutional ownership adopt marginally higher targets. Overall, the findings imply that while network connections, local supply of female directors, and external visibility are important for the extensive margin (whether to adopt a target), they do not meaningfully shape the intensive margin (how ambitious the target is). This pattern suggests that the level of self-adopted targets is likely influenced by pressure from institutional investors and other unobserved factors, rather than the observable governance and network characteristics considered here.

Overall, the results in Table 7 suggest that access-related frictions and external visibility significantly influence firms’ compliance with the OSC’s gender diversity amendment. Firms are more likely to fully comply when they are located near a larger pool of qualified female directors and when their directors are connected to female directors or to peers at other fully compliant firms, highlighting the importance of local access and network ties.Footnote ²³ In addition, firms with greater media coverage are significantly more likely to fully comply, suggesting that public scrutiny may serve as a reputational incentive for adopting stronger gender diversity policies. Conversely, controlled corporations, which typically face weaker market and governance pressures, are less likely to fully comply, indicating that ownership structure may dampen responsiveness to regulatory and social expectations.

A. Changes in Board Composition

During the 3 years prior to the OSC amendment (2011–2013), only 20% of firms appointed a female director. In contrast, more than half of all firms (51%) appointed a female director in the 3 years after the amendment (2015–2018). The response was particularly strong among firms with all-male boards in 2013: ~59% added a female director within 4 years post-amendment, compared to 48% of firms that already had at least one female director in 2013 (see Supplementary Material Figure C.1).

These data demonstrate that Canadian firms increasingly added women to the boards post-amendment. However, it remains unclear whether this increase is directly attributable to the OSC amendment or part of a broader trend. To address this, Figure 2 shows the annual trend in the average proportion of female directors on Canadian boards from 2011 to 2018, compared with various samples of U.S. firms as controls. For Canadian-listed firms, the solid line in Graph A reveals a noticeable kink—a steepening of the upward trend—after 2013, suggesting that the amendment prompted firms to accelerate the appointment of female directors, raising the average rate at which women are added to boards. This increased rate persists for the remainder of our sample period. For comparison, Graph A also tracks U.S. firms in the S&P 500 index at any point between 2010 and 2016. The U.S. firms, which are not subject to any such regulation during our sample period, do not exhibit the same trend after 2014, reassuring us that this phenomenon is unique to Canadian firms and is thus plausibly a consequence of the OSC’s amendment.

FIGURE 2Comparison of Changes in Female Director Ratios in Canada and the U.S.

Figure 2 presents data on female directorships for Canadian firms included in the S&P TSX Composite Index and a comparison group of U.S. firms, with directorship data in BoardEx and financial data in Compustat, over the 2011–2018 period. The graphs plot the average fraction of firms’ boards consisting of female directors each year. Graph A reports data for Canadian firms and U.S. firms that were included in the S&P 500 index. Graph B reports data for Canadian firms and a matched sample of U.S. firms, each selected from within the same 3-digit SIC industry and closest in total assets to the corresponding Canadian firm in 2013 (only Canadian firms with a match available are included). Graph C reports data for Canadian firms that are cross-listed on a U.S. stock exchange and a matched sample of U.S. firms, each selected from within the same 3-digit SIC industry and closest in total assets to the corresponding Canadian firm in 2013 (only Canadian firms with a match available are included).

FIGURE 2 Long description

A multi-panel figure containing three line graphs labeled A, B, and C. All graphs share a common Y-axis titled Average Female Director Ratio ranging from 0 percent to 30 percent in 5 percent increments, and an X-axis titled Year ranging from 2011 to 2018. A legend below each graph identifies a solid line for Canada and a dashed line for the U.S.

* Graph A. Comparison of Canadian S P T S X Firms and U.S. S P 500. The U.S. dashed line starts at approximately 16 percent in 2011 and rises steadily to 25 percent in 2018. The Canadian solid line starts lower at 9 percent, remains relatively flat until 2013, then rises sharply to 22 percent by 2018, narrowing the gap with the U.S.

* Graph B. Comparison of Canadian S P T S X and U.S. Matched Firms. The Canadian solid line is identical to Graph A, starting at 9 percent and rising to 22 percent. The U.S. dashed line for matched firms starts lower at 7 percent and rises more slowly to 14 percent in 2018, showing Canada significantly outperforming the matched U.S. sample.

* Graph C. Comparison of Canadian S P T S X A D Rs and U.S. Matched Firms. The Canadian solid line represents cross-listed firms, starting at 10 percent in 2011 and rising to 21 percent in 2017. The U.S. dashed line starts at 6 percent and rises to 13 percent in 2017. Similar to Graph B, the Canadian ratio remains consistently higher than the U.S. matched sample throughout the period.

In Graph B, we compare Canadian firms to U.S. firms matched on total assets and 3-digit SIC industry.Footnote ²⁵ U.S. firms are geographically proximate and more closely linked to Canadian firms in terms of board connections, culture, and business relationships relative to firms in other countries and thus present a plausible control group. Prior to the amendment, the average female director ratio moves in parallel fashion for both groups, but after the OSC amendment, an upward kink appears for the Canadian sample. In Graph C, we compare Canadian firms cross-listed on a U.S. stock exchange (and thus subject to both Canadian and U.S. exchange regulations) to matched U.S. firms (subject only to U.S. regulations). Despite being governed by U.S. SEC regulations, cross-listed Canadian firms display an upward kink and an increased trajectory post-amendment, a pattern not observed in their U.S.-based counterparts. These findings further support the conclusion that the kink in the female director ratio on Canadian boards is driven by the OSC amendment.

Next, we compare pre-treatment firm characteristics of the treated and control firms used in the analyses in Graph B (full sample) and C (Canadian cross-listed firms) to make sure that the two groups are similar prior to OSC’s rule. As shown in Supplementary Material Table C.9, the Canadian treated and matched U.S. control groups are similar, as we find only a couple of statistically significant differences between the means of observable characteristics of the two groups (return on assets (ROA) and institutional ownership are higher for the matched U.S. control group). Prior literature suggests that higher institutional ownership may lead to increases in board gender diversity, but we observe higher institutional ownership in the U.S. matched control firms, suggesting that differences in institutional ownership are unlikely to explain our findings. Importantly, the two groups are similar in female director ratio and the rate of change in this ratio, suggesting parallel trends in the outcome variable prior to the amendment. The overall similarity in observable firm characteristics is reassuring because it makes it less likely that unobserved differences between the groups are driving our results.

The regressions in Table 8 confirm the statistical significance of the patterns shown in Figure 2. The unit of observation is firm-year, with the dependent variable being the proportion of a firm’s board consisting of female directors. The sample period is 2011–2018. Control variables include the level of compliance in the prior year, institutional ownership, media coverage, total assets, market-to-book assets, return-on-assets, and debt/assets. All specifications include firm fixed effects and focus exclusively on Canadian-listed firms. In model 1, the coefficient on the post-2014 indicator is significantly positive, indicating that the representation of female directors increased by about 7.4 percentage points, on average, following the 2014 OSC amendment.

TABLE 8Regression Analysis of Changes in Female Directorships in Canada

TABLE 8 Long description

The table presents the Female Director Ratio as the dependent variable across three O L S regression models.

Independent Variables and Coefficients:

* Post-2014: Model 1 is 0.074 triple asterisk (0.004); Model 2 is 0.047 triple asterisk (0.007).

* Partial Compliance sub t minus 1: Model 2 is 0.028 triple asterisk (0.008); Model 3 is 0.020 double asterisk (0.008).

* Full Compliance sub t minus 1: Model 2 is 0.061 triple asterisk (0.012); Model 3 is 0.038 triple asterisk (0.013).

* Institutional Ownership: Model 2 is 0.021 (0.023); Model 3 is 0.030 (0.021).

* Media Coverage: Model 2 is minus 0.001 (0.022); Model 3 is minus 0.002 (0.041).

* Log Assets: Model 1 is 0.017 double asterisk (0.007); Model 2 is 0.014 double asterisk (0.007); Model 3 is 0.0001 (0.007).

* Market-to-Book Assets: Model 1 is 0.0005 (0.002); Model 2 is 0.0001 (0.0001); Model 3 is 0.0002 (0.0005).

* Return-on-Assets: Model 1 is 0.009 (0.006); Model 2 is 0.009 (0.006); Model 3 is 0.006 (0.004).

* Debt/Assets: Model 1 is 0.034 double asterisk (0.012); Model 2 is 0.033 double asterisk (0.013); Model 3 is 0.012 (0.015).

* Constant: Model 1 is minus 0.042 (0.055); Model 2 is minus 0.030 (0.58); Model 3 is minus 0.008 (0.062).

Fixed Effects and Statistics:

* Year fixed effects: No for Models 1 and 2; Yes for Model 3.

* Firm fixed effects: Yes for all models.

* Number of observations: Model 1 is 2,254; Models 2 and 3 are 2,030.

* Adjusted R-squared: Model 1 is 0.747; Model 2 is 0.767; Model 3 is 0.798.

Statistical significance is indicated by triple asterisk for 1 percent, double asterisk for 5 percent, and single asterisk for 10 percent levels.

In model 2, we include independent variables for partial and full compliance and find that the post-2014 indicator remains statistically significant, though the magnitude of the effect is slightly reduced compared to model 1. Model 2 shows that partial compliance is associated with a 2.8 percentage point increase in the female director ratio the following year,Footnote ²⁶ while full compliance, which includes disclosing a director gender diversity target, is associated with a 6.1 percentage point increase. Given that female directors accounted for an average of 10% of directors in Canadian firms before the amendment, these results imply that partial compliance leads to a 28% increase in female director representation, with full compliance leading to a 61% increase. In model 3, these findings remain statistically significant when we replace the post-2014 indicator variable with year dummies.

1. DiD with U.S. Control Sample

The increase in board gender diversity observed in the previous section may reflect a broader positive trend in female representation rather than a direct response to the OSC amendment. To better isolate the effect of the OSC amendment, we employ a DiD approach in Table 9.Footnote ²⁷ In model 1, we compare Canadian firms with all U.S. listed firms in the S&P 500 index (similar to Graph A of Figure 2). The coefficient on “Canadian Firm×Post-2014” in model 1 suggests that following the amendment, Canadian firms increased female director representation by about 3.3 percentage points more, on average, than U.S. firms during the same period. In model 2, we match U.S. firms to Canadian firms based on total assets and 3-digit SIC industry, and in model 3, we match Canadian firms cross-listed on a U.S. exchange with firms listed only in the U.S. In both cases, we find a similar relative increase in female director representation for Canadian firms (4.3 and 3.8 percentage points, respectively). Since female directors accounted for an average of 10% of directors in Canadian firms before 2014, the coefficients imply that the OSC amendment is associated with a 38%–43% increase in female director representation.

TABLE 9Difference-in Differences Analysis of Changes in Female Directorships

TABLE 9 Long description

The table presents O L S regression results for the Female Director Ratio across three sample models.

Header Information:

* Model 1: Canada and U. S. S & P 500 Index.

* Model 2: Canada and U. S. Matched.

* Model 3: Canada A D R and U. S. Matched.

Regression Coefficients (Standard Errors in parentheses):

* Canadian Firm times Post-2014: Model 1 is 0.033*** (0.006); Model 2 is 0.043*** (0.008); Model 3 is 0.038*** (0.011).

* Institutional Ownership: Model 1 is 0.031** (0.013); Model 2 is 0.048*** (0.016); Model 3 is 0.026 (0.022).

* Log Assets: Model 1 is 0.002 (0.005); Model 2 is 0.006 (0.002); Model 3 is 0.008 (0.008).

* Market-to-Book Assets: Model 1 is 0.001 (0.002); Model 2 is 0.003** (0.001); Model 3 is 0.002 (0.001).

* Return-on-Assets: Model 1 is 0.001** (0.001); Model 2 is 0.000 (0.001); Model 3 is 0.002 (0.004).

* Debt/Assets: Model 1 is 0.002 (0.011); Model 2 is 0.034** (0.016); Model 3 is 0.097*** (0.031).

* Constant: Model 1 is 0.033*** (0.006); Model 2 is 0.020 (0.050); Model 3 is 0.013 (0.067).

Model Statistics:

* Year and Firm fixed effects are included (Yes) for all models.

* Number of observations: Model 1 is 5,819; Model 2 is 2,787; Model 3 is 1,071.

* Adjusted R-squared: Model 1 is 0.756; Model 2 is 0.802; Model 3 is 0.828.

Statistical significance levels: *** 1 percent, ** 5 percent, * 10 percent.

3. Changes Within the Canadian Sample

To ensure that our results are not an artifact of our choice to use U.S. firms as control groups, we next examine changes in the female director ratio within Canada alone. Specifically, we focus on the trajectory of female director ratios of Canadian firms as shown in Figure 2, where there is a visible kink in 2014 when the new OSC rules were passed, indicating an increased rate of change in female director ratios post-regulation. To verify the statistical significance of this kink, we examine the year-on-year change in the female director ratios of Canadian firms over the 2011–2018 sample period by estimating the following regression specification:

$$ \Delta {Female\ Director\ Ratio}_{i,t}=\alpha +\sum \limits_{j=2012}^{2018}{\theta}_j\bullet \times \unicode{x1D7D9}\left[ Year=j\right]+{\varphi}_{i,t}\bullet Controls+{\gamma}_i+{\varepsilon}_{i,t}. $$

Observations are at the firm-year level, with i indexing firms and t indexing calendar years. The dependent variable is the percentage point change in the proportion of a firm’s board consisting of female directors between the current year, t, and the previous year, t-1. Control variables include Log(Assets), Market-to-Book Assets, ROA, and Debt/Assets. Firm fixed effects ( $ {\gamma}_i\Big) $ are included, and $ {\varepsilon}_{i,t} $ is an error term. The variables of interest are year indicator variables, with the omitted year being 2011 (the benchmark year). Results are presented in Graph A of Figure 3, which plots the coefficient estimates, $ {\theta}_j $ , of the percentage point change in the female director ratio in each year, relative to 2011, with error bars that indicate 95% confidence intervals for the coefficient estimates.

FIGURE 3Dynamic Model of Year-on-Year Changes in the Female Director Ratios of Canadian Firms

Figure 3 plots coefficient estimates and the 95% confidence intervals from the following regression specifications:

$$ \mathrm{Graph}\;\mathrm{A}:\Delta {Female\ Director\ Ratio}_{i,t}=\alpha +\sum \limits_{j=2012}^{2018}{\theta}_j\bullet \times \unicode{x1D7D9}\left[ Year=j\right]+{\varphi}_{i,t}\bullet Controls+{\gamma}_i+{\varepsilon}_{i,t.} $$

$$ {\displaystyle \begin{array}{c}\mathrm{Graph}\hskip0.32em \mathrm{B}:\Delta Female\;{Director\ Ratio}_{i,t}=\alpha +\sum \limits_{j=2012}^{2018}{\beta}_j\bullet All\;{Male\ Board}_{i,2013}\times \unicode{x1D7D9}\left[ Year=j\right]\\ {}+\sum \limits_{j=2012}^{2018}{\theta}_j\bullet \times \unicode{x1D7D9}\left[ Year=j\right]+{\varphi}_{i,t}\bullet Controls+{\gamma}_i+{\varepsilon}_{i,t}.\end{array}} $$

The sample consists of firms that are included in the S&P TSX Composite Index, with directorship data in BoardEx and financial data in Compustat, over the 2011–2018 period. The observations are at the firm-year level, with i indexing firms and t indexing calendar years. The dependent variable is the percentage point change in the fraction of a firm’s board consisting of female directors between the current year, t, and the previous year, t–1. “All-Male Board₂₀₁₃” is an indicator equal to 1 if a firm has no female directors in 2013. Control Variables consist of Log(Assets), Market-to-Book Assets, ROA, and Debt/Assets. All variables are defined in the Appendix. Firm fixed effects ( $ {\gamma}_i) $ are included. The omitted year in the specifications, and thus the benchmark year, is 2011. The graph in Graph A displays the coefficient estimates $ {\theta}_j $ of the percentage point change in the female director ratio in each year, relative to 2011. The graph in Panel B displays the coefficient estimates $ {\beta}_j $ of the difference between the change in the female director ratio for firms with all-male boards and firms with female directors, relative to 2011. The post-regulation period is years 2015 onward. The error bars indicate 95% confidence intervals for the coefficient estimates.

FIGURE 3 Long description

The figure contains two panels.

Graph A: Year-on-Year Change in Female Director Ratio.

The x-axis represents Year from 2011 to 2018. The y-axis represents Percentage Point Change in Female Director Ratio, ranging from -1.00 percent to 3.50 percent. A horizontal dashed line marks the 0.00 percent baseline. The graph is divided into Pre-Regulation (2011 to 2014) and Post-Regulation (2014 to 2018).

* From 2011 to 2013, the change is near zero.

* In 2014, there is a sharp increase to approximately 2.1 percent.

* From 2015 to 2018, the change stabilizes between 1.5 percent and 2.0 percent.

Vertical error bars indicate 95 percent confidence intervals, which are wider in the post-regulation period.

Panel B: Difference in Year-on-Year Change in Female Director Ratio between Firms with All-Male Boards and Firms with Female Directors.

The x-axis is Year (2011 to 2018) and the y-axis is Percentage Point Change ranging from -3.00 percent to 6.00 percent.

* In the Pre-Regulation period, the difference dips below zero, reaching approximately -1.2 percent in 2013.

* In 2014, it rises sharply to 1.5 percent.

* In the Post-Regulation period, the difference fluctuates but shows an upward trend, peaking near 2.8 percent in 2018.

Error bars are significantly larger than in Graph A, particularly from 2014 onward.

Graph A of Figure 3 shows that in the pre-regulation period (2012 and 2013), changes in Canadian firms’ female director ratios are not statistically different from the benchmark year, 2011. However, in 2014, there is a distinct and statistically significant increase of about 2 percentage points in the annual change in female director ratios relative to 2011. The elevated rate of change persists at a similar magnitude of between 1.5 and 2 percentage points and remains statistically significant through 2018. These findings confirm a statistically significant increase in the rate of change in female director ratios for Canadian firms following the OSC amendment, validating the kink observed in Figure 2.

We next examine variations within the sample of Canadian firms to investigate whether the rate of change in female director ratios differs for firms with all-male boards. We augment the regression from Graph A by adding an interaction between the year indicator and an indicator for firms with all-male boards as of 2013 as follows:

$$ {\displaystyle \begin{array}{c}\Delta Female\hskip0.42em {Director\ Ratio}_{i,t}=\alpha +\sum \limits_{j=2012}^{2018}{\beta}_j\bullet All\hskip0.42em {Male\ Board}_{i,2013}\times \unicode{x1D7D9}\left[ Year=j\right]\\ {}+\sum \limits_{j=2012}^{2018}{\theta}_j\times \unicode{x1D7D9}\left[ Year=j\right]+{\varphi}_{i,t}\bullet Controls+{\gamma}_i+{\varepsilon}_{i,t}.\end{array}} $$

Graph B of Figure 3 displays the coefficient estimates, $ {\beta}_j $ , of the difference between the change in the female director ratio for firms with all-male boards and firms with female directors, relative to 2011.Footnote ²⁹ The figure indicates that following the amendment in 2014, firms with all-male boards in 2013 increased their female director ratios by about 1.5 percentage points more than firms with female directors in 2013 (the difference is statistically significant at the 10% level). The increase persists with similar magnitudes (of between 1 and 3 percentage points) through 2018 and is statistically significant (at the 5% level) in all subsequent years except 2016. The results indicate that, between the pre- and post-regulation periods, firms with all-male boards exhibited a larger increase in the rate of change in female director ratios than did firms with female directors. This pattern is consistent with Graph B of Supplementary Material Figure C.1, which showed that firms with all-male boards were more likely to add female directors in the post-regulation period. These results, which are akin to examining how a DiD coefficient changes dynamically over time, help address our concerns that our findings in Table 9 are driven by another Canada-specific event that also affects female ratios. That is, the results shown in Figure 3 make it less likely that the results in Table 9 are an artifact of the choice of using U.S. firms as a control group.

A potential concern is that the increase in female board representation reflects the same women being appointed to multiple boards, increasing the number of busy female directors, rather than an expansion in the pool of female directors. However, we find little evidence consistent with this interpretation. As shown in Supplementary Material Figure C.5, the number of unique female directors in our sample increases substantially following the OSC amendment, rising from 246 in 2013 to 430 in 2018. At the same time, the average number of board seats held by a female director declines from 3.09 in 2013 to 2.78 in 2018, and the percentage of female directors in our sample holding three or more board seats falls from 54.1% to 46.5% over the sample period. Together, these patterns suggest that firms increasingly drew from a broader pool of female directors rather than relying on the same women to fill multiple board positions.

In addition to board representation, the OSC amendment requires listed firms to disclose policies regarding female representation in the top executive team. Although the focus of our study is board representation, we also examine changes in the composition of the top named executives (from annual proxy circulars) in the years surrounding the OSC amendment. In practice, female executive targets are substantially less common than female director targets: during 2014–2018, only 27 firms in our sample adopt female executive targets compared to 111 firms adopting female director targets, and all firms adopting female executive targets also adopt female director targets.

The increase in the annual average proportion of female top executives at Canadian firms between the pre- and post-regulation periods is much smaller than for female directors, increasing from 6% to 8.9% between 2011 and 2018 (vs. 10% to 22% for directors). Moreover, similar analyses to those in Table 9 do not show substantial evidence of an increase in female executive ratios among Canadian firms relative to U.S. firms (see Supplementary Material Table C.10). We suggest the reason the OSC regulation impacted board gender diversity but not executive gender diversity is due to the higher costs associated with replacing top executives. Directors are up for election annually and can be added without displacing an existing board member. Moreover, unlike directorships, executive positions are not easily scalable, as a single executive typically cannot serve comparable executive roles across multiple firms, potentially making supply constraints more binding and slowing the development of a broader pipeline of female executives. Furthermore, institutional investors who acted upon the regulation appear to have been focused on improving female board representation (see Section VII), and media coverage of the regulation’s initial announcement appears to be more focused on board composition rather than that for executives.Footnote ³⁰

A. Institutional Investor Engagement

The CCGG was established in 2003 by institutional investors to promote improved corporate governance practices among Canadian public companies. CCGG currently represents 51 major institutional investors in Canada, accounting for approximately 22% of the total institutional ownership of Canadian firms in our sample, with $5.5 trillion in assets under management. Since 2008, CCGG has engaged directly with Canadian public company boards on various governance issues. These board engagement meetings offer a private forum for dialogue and exchange of views between independent directors and institutional investors.Footnote ³³ The CEO of CCGG stated, “Board engagements are the appropriate forum to discuss a company’s governance practices, including shareholder rights, board composition, executive compensation, and board oversight of corporate strategy and material business risks….”Footnote ³⁴

We obtained proprietary data from CCGG on all its engagement meetings with Canadian companies from 2008 to 2018. Panel A of Table 10 summarizes these engagements. On average, CCGG engages about 40 companies annually. We received a summary report for each engagement, allowing us to categorize what was discussed and, particularly, whether board gender diversity was among the issues raised. Notably, there was no discussion of board gender diversity in any engagement meeting before 2012. While the topic of diversity began to surface in 2012 and 2013, there was a marked increase after 2013 in discussions of board gender diversity. For example, of the 36 engagements in 2014, 27 (75%) included discussions on board gender diversity. Gender diversity continued to be a popular topic, representing about 59% of all engagements in 2015 and about 40% in 2016 and 2017. Interestingly, many of these discussions on board gender diversity were not initiated by CCGG; instead, firms themselves often used these meetings to inform CCGG of their actions regarding board diversity.Footnote ³⁵ Forty-two percent (15 out of 27) of gender diversity discussions were proactively initiated by firms in 2014, with 62% of gender diversity discussions being proactively initiated by firms in 2015. These statistics suggest that OSC amendment’s disclosure requirements may have heightened boards’ focus on diversity, particularly in the years immediately following the amendment. The last 2 columns show the number of engagements in which CCGG commends or questions the firm on the gender diversity of its board. Examples of questions asked by CCGG include whether the board has considered the issue of gender diversity, whether the board considers gender when adding new directors, whether directors believe it makes sense to have one or more females on the board, what the board is doing to identify appropriate female candidates, or asking directors to comment on the lack of gender diversity on their board.

TABLE 10Engagements by Canadian Coalition for Good Governance (CCGG)

TABLE 10 Long description

Panel A. Summary of C C G G Engagements.

This table tracks engagement data across six columns: Year, Total Number of C C G G Engagements on any Subject, Percentage of Engagements with Discussion of Board Gender Diversity, and three sub-columns under Number of Gender Diversity Engagements in Which: Firm Initiates Discussion, C C G G Commends Firm, and C C G G Questions Firm.

- 2008 to 2011: Total engagements grew from 5 to 41, but gender diversity discussion remained at 0 percent.

- 2012: 40 total engagements, 5 percent gender discussion, 1 firm-initiated, 0 commended, 1 questioned.

- 2013: 48 total engagements, 29 percent gender discussion, 6 firm-initiated, 0 commended, 8 questioned.

- 2014: 36 total engagements, 75 percent gender discussion, 15 firm-initiated, 5 commended, 7 questioned.

- 2015: 41 total engagements, 59 percent gender discussion, 15 firm-initiated, 3 commended, 6 questioned.

- 2016: 45 total engagements, 40 percent gender discussion, 11 firm-initiated, 5 commended, 2 questioned.

- 2017: 44 total engagements, 41 percent gender discussion, 8 firm-initiated, 6 commended, 4 questioned.

- 2018: 36 total engagements, 31 percent gender discussion, 9 firm-initiated, 0 commended, 2 questioned.

Panel B. Regressions of Changes in Female Directorships Following C C G G Engagements.

The dependent variable is Female Director Ratio across four models.

- C C G G Post-Engagement: Coefficients are 0.051 (significant at 1 percent), 0.033 (significant at 10 percent), 0.032 (significant at 10 percent), and 0.025.

- C C G G Engagement Fixed Effect: Coefficients are minus 0.0385 (significant at 10 percent) and minus 0.033 (significant at 5 percent).

- C C G G Ownership: Coefficients are 0.0009 and 0.0007.

- Controls: Models 1 and 3 have no other controls; Models 2 and 4 include other controls.

- Fixed Effects: All models include Year fixed effects. Models 1 and 2 include Industry fixed effects. Models 3 and 4 include Firm fixed effects.

- Observations: Range from 2,155 to 2,239.

- Adjusted R-squared: Ranges from 0.313 to 0.784.

Panel B estimates the impact of CCGG’s activism on female corporate board representation. The dependent variable is the proportion of a firm’s board made up of female directors. The CCGG Engagement Fixed Effect is an indicator equal to 1 if CCGG questioned the firm on director gender diversity in any year in our sample (between 2011 and 2018). CCGG Post-Engagement is an indicator variable that equals 1 if CCGG questioned the firm on director gender diversity in a prior year (i.e., equals 1 in year t and all subsequent years if a firm was engaged in year t–1). All models are linear probability models and include year fixed effects. Models 1 and 2 include 1-digit SIC industry fixed effects, while models 3 and 4 include firm fixed effects. Models 2 and 4 also include observable firm characteristics and governance controls.

The coefficient of our main explanatory variable (CCGG Post-Engagement) indicates the change in the female director ratio following CCGG’s engagement, relative to firms not engaged on gender diversity during this period. The coefficient is positive and statistically significant in three of the four models, demonstrating that CCGG engagements on gender diversity lead to a higher subsequent female director ratio. In terms of economic significance, the coefficient in model 2 suggests an increase of about 3 percentage points in the female director ratio following CCGG’s engagement. The negative and statistically significant coefficient on CCGG Engagement Fixed Effect indicates that firms targeted by CCGG have a female director ratio that is, on average, 3 percentage points lower. This suggests that CCGG targets firms with lower female board representation and that these engagements result in an increase in female board representation. Our findings are consistent with Doidge, Dyck, Mahmudi, and Virani (Reference Doidge, Dyck, Mahmudi and Virani2019), who study CCGG engagements related to the adoption of majority voting, say on pay, and specific compensation policies—but not diversity policies—and find that firms engaged by CCGG are more likely than non-engaged firms to adopt CCGG’s specific governance proposals.

We acknowledge potential caveats to these findings. First, firms engaged by CCGG are not randomly selected, raising concerns about causal interpretation. Engaged firms might have improved board gender diversity even without these engagements. However, consistent with a causal interpretation, CCGG are unlikely to expend scarce resources to engage firms that already plan to make these governance changes. Moreover, we show that CCGG targets firms with lower female board representation (without a gender diversity target), and we find that these engagements are followed by increases in female board representation. Second, it’s possible that an omitted variable (e.g., the same societal trends that spurred the OSC regulation) affected institutional investors’ attitudes toward board diversity around 2014 and led to their engagements. While we acknowledge this possibility, our discussions with CCGG suggest enhanced disclosure of the OSC amendment facilitated their engagement efforts. Catherine McCall, CEO of CCGG, noted that the OSC amendment provided their institutional investor members “meaningful information and disclosure, allowing them to express their disapproval of a lack of diversity by voting against one or more directors.”Footnote ³⁶

B. Voting Outcomes

The OSC stated that its gender disclosure regulation aimed to provide investors and stakeholders with information on a company’s efforts to advance women’s representation on boards and in senior management, suggesting these disclosures might influence investment and voting decisions. Notably, CCGG indicates that its members may withhold votes based on diversity considerations. CCGG’s Gender Diversity Policy notes, “Investors are looking to see the extent of diversity when evaluating a company and boards are increasingly facing investors’ higher expectations in this area,” and also mentions that “several of CCGG’s members have proxy voting guidelines that recommend voting to withhold from some directors in certain circumstances where gender diversity is considered to be inadequate.”

The OSC amendment requires firms to disclose their stance on board diversity, giving investors additional data upon which to act. This increased post-amendment disclosure provides investors with information that may allow them to express dissatisfaction with board diversity. We posit that investors will be more likely to express such dissatisfaction via votes against the chair of the nominating committee post-amendment when firms must disclose their stance on board diversity. To analyze whether the OSC regulation impacted voting decisions in director elections, we obtain shareholder voting data from Institutional Shareholder Services (ISS) and from the Johnston Centre for Corporate Governance Innovation at the University of Toronto.Footnote ³⁷ Since the ISS data for Canadian firms start in 2013, our analysis in Table 11 covers 2013–2018 (2 years pre- and 4 years post-regulation). The dependent variable is the percentage of votes withheld and against, reflecting the lack of support for the chair of the nominating committee. The main explanatory variables are the interaction of female director ratio and a post-regulation dummy (Post-2014 × Female Director Ratio) in models 1 and 3 and the interaction between gender diversity target dummy and post-regulation (Post-2014 × Female Director Target) in models 2 and 4. All specifications are OLS regressions with year fixed effects. Models 2 and 4 also include firm characteristics and governance controls. Models 1–4 include 1-digit SIC industry fixed effects, while models 5 and 6 include firm fixed effects.

TABLE 11Gender Diversity and Changes in Chair of Nominating Committee Voting Support

TABLE 11 Long description

A table titled Gender Diversity and Changes in Chair of Nominating Committee Voting Support. The dependent variable is Chair of Nominating Committee Voting: Withheld and Against. The table presents six O L S regression models.

Key independent variables and coefficients:

* Female Director Ratio: Model 1 is -16.580 (significant at 1%), Model 4 is -12.817 (significant at 5%).

* Female Director Target: Model 2 is -1.560, Model 3 is -0.607.

* Post-2014 times Female Director Ratio: Model 1 is -19.420 (significant at 5%), Model 3 is -19.610 (significant at 5%), Model 5 is -30.128 (significant at 1%).

* Post-2014 times Female Director Target: Model 2 is -5.586 (significant at 1%), Model 4 is -4.469 (significant at 5%), Model 6 is -5.163 (significant at 1%).

Control variables include Institutional Ownership, Media Coverage, Controlled Corporation, Independent Board, Log(Assets), Market-to-Book Assets, Return-on-Assets, Debt/Assets, Board Size, Board Degree Centrality, Board Tenure, and Board Age.

Fixed Effects and Statistics:

* Year fixed effects: Yes for all models.

* Industry fixed effects: Yes for models 1 through 4; No for models 5 and 6.

* Firm fixed effects: No for models 1 through 4; Yes for models 5 and 6.

* Number of observations: 1,187 for models 1 and 2; 1,120 for models 3 through 6.

* Adjusted R-squared: Ranges from 0.050 in Model 2 to 0.318 in Model 5.

In Table 11, the coefficient on Post-2014 × Female Director Ratio is negative and statistically significant in models 1 and 3, indicating that after the OSC amendment, firms with lower female director ratios receive less voting support for the chair of their nominating committee than they did before the amendment. Similarly, the negative and statistically significant coefficient on Post-2014 × Female Director Target in models 2 and 4 suggests that after the amendment, not having a director gender target is associated with less voting support for the nominating chair than it was before the amendment.Footnote ³⁸ These results are consistent in models 5 and 6 with firm fixed effects. In terms of economic significance, the coefficient in model 5 implies that after the OSC amendment, a 1-standard-deviation lower female director ratio results in 3.7% lower voting support for the nominating committee chair, while the coefficient in model 6 suggests that not having a gender target results in 5.2% lower voting support relative to the pre-amendment period. In untabulated regressions, we find similar results using the average votes withheld and against the nominating committee as the dependent variable (as opposed to the nominating chair).Footnote ³⁹

Table 11 shows that shareholders were more likely to vote against the nominating committee chair at firms with low female board representation after the OSC amendment, but not before. This suggests that the amendment increased shareholder attention to gender diversity on boards. Importantly, the OSC amendment also required firms to disclose their position on board gender diversity. Accordingly, Table 12 examines voting outcomes for nominating committee chairs based on common justifications provided by explainers and partial compliers for not setting gender diversity targets. Among firms with all-male boards, we find that the chair receives 7% more votes withheld or cast against when the firm claims that targets are unnecessary or inadvisable, and 8.4% more when the firm states that targets are not an appropriate tool to improve diversity. In contrast, we find no similar effect for firms citing merit-based appointments, firm size, or board turnover constraints. These findings suggest that shareholders distinguish between firms perceived as dismissing board diversity initiatives and firms providing more substantive or operational explanations for not adopting targets. More broadly, the evidence indicates that investors respond not only to the presence or absence of diversity disclosures but also to their specific content and framing in proxy statements.

TABLE 12Quality of Explanations and Changes in Chair of Nominating Committee Voting Support

TABLE 12 Long description

The table presents O L S regression results with ‘Chair of Nominating Committee Voting: Withheld and Against’ as the dependent variable.

* Column Headers: Six models numbered 1 through 6.

* All-Male Board: Coefficients range from 3.164 to 4.000, with statistical significance in models 2, 3, 4, and 6.

* Interaction Terms:

- Appointments Based on Merit/Skills cross All-Male Board: 0.883 (Model 1).

- Targets Not Necessary or Advisable cross All-Male Board: 7.030 with 5 percent significance (Model 2).

- Too Small for Target cross All-Male Board: minus 5.174 (Model 3).

- Targets Not Appropriate cross All-Male Board: 8.440 with 5 percent significance (Model 4).

- Targets Do Not Result in Best Candidate cross All-Male Board: 1.438 (Model 5).

- Turnover too Infrequent for Target cross All-Male Board: 4.460 (Model 6).

* Controls and Fixed Effects: All models include Other controls, Year fixed effects, and Industry fixed effects.

* Model Statistics:

- Constant: Values range from minus 3.017 to 0.555.

- Number of observations: 707 for all models.

- Adjusted R-squared: Ranges from 0.092 to 0.095.

Standard errors are provided in parentheses below each coefficient. Asterisks indicate significance levels: three for 1 percent, two for 5 percent, and one for 10 percent.

In Supplementary Material Table C.11, we focus on firms that fully comply and examine whether the adopted female director target is associated with voting support for the chair of the nominating committee. The coefficient on the Female Director Target is negative and statistically significant, indicating that firms adopting higher targets receive lower percentages of withheld or against votes. Using the estimate from model 2, a 1-standard-deviation increase in Female Director Target (0.056) is associated with approximately a 2.4% reduction in withheld or against votes. We do not find evidence that this relation differs for firms with all-male boards. Overall, the evidence suggests that, among Full Compliers, shareholders respond more favorably to firms adopting more ambitious female director targets.

C. Summary of Mechanisms

Although the OSC amendment did not explicitly provide “best practices” guidelines, it was designed to give investors insights into firms’ board diversity practices, potentially influencing investment and voting decisions. Following the regulation, the OSC facilitated access to this information by mandating the disclosure of firms’ board diversity practices. The evidence suggests that the OSC’s reasoning was correct: board diversity increased even without an explicit mandate. This increase appears to be driven at least in part by investor pressure: post-amendment, institutional investor engagements on board diversity increase, and investors are more likely to vote against nominating chairs and committees of firms with non-diverse boards. Moreover, investors respond to diversity disclosures through voting. Overall, our evidence suggests that the disclosure of diversity policies required by the amendment contributed to increased board diversity through investor engagement and voting.