In a recent study, we examine whether a director’s “diversity” explains two employment outcomes: being retained on the board and being promoted to board leadership positions. We define diversity as the director’s dissimilarity to the other members of the board (s)he serves on, along four key dimensions: age, gender, ethnicity, and skills. A director’s diversity along any of these dimensions is her distance from the board’s average on the dimension. Consider, for instance, age. A director is more “age-diverse” if the director’s age is very different from the average age of the remaining directors. Thus, an old director is diverse when the rest of the board is young, and an old director is not diverse when the rest of the board is also old, on average.
We find that diversity, i.e., how dissimilar a director is relative to the remaining directors, explains employment outcomes across a range of specifications. On the dimension of gender, women directors are diverse given the dominant presence of men on U.S. boards. Women are more likely than men to be retained. However, women are less likely to be promoted to board leadership positions. On the dimensions of age and ethnicity, we find distinctive anti-diversity effects. Directors who are more diverse in age or ethnicity relative to the board are less likely to be retained and are also less likely to be promoted to board leadership positions. In contrast, we find consistent pro-diversity results on the dimension of skills. Skill-diverse directors, i.e. those who have unique skills compared with the other directors, are more likely to be retained and promoted to leadership positions.
Our sample comprises firms at the intersection of the BoardEx and the CRSP-COMPUSTAT databases. Between fiscal 2004 and 2014, we have 5,032 U.S. listed companies, and 34,455 non-executive directors. The sample is larger than those used in many prior studies and is updated to the post Sarbanes-Oxley Act period. A notable feature of our empirical specification is the use of high dimensional firm-year interactive fixed effects. Using such effects lets us localize comparisons to those within boards. We thus compare a director on a board with another director on the same board in the same year. The key explanatory variables are our diversity constructs but we include other director-level controls. For example, we identify roles of directors from BoardEx and standardize the titles to generate indicators for key leadership roles such as board chair, chair of the compensation committee, or lead director.
We replicate many of the key findings in the literature concerning director turnover. Director turnover is greater in firms with poor accounting or stock performance and in years when there is CEO turnover. For example, the director turnover rate is 8.14 percent when the CEO stays versus 13.48 percent when there is a CEO departure. Key directors have lower baseline turnover rates (6.27 percent) than do non-key directors (10.69 percent). This result underlines the need to control for other director-level variables such as role, and we do so in our empirical specifications.
Turning to details, we measure director diversity using identity distances along the axes of age, ethnicity, gender, and skills. Director age and gender are reported in BoardEx. Age diversity is the absolute difference between a director’s age and the average age of all directors on the board. As boards are predominantly male, gender diversity occurs essentially when a director is a woman. Ethnicity is derived by applying machine learning algorithms to “bi-chars” or two-character sequences of first and last names. Skills are based on a set of 18 skill categories, the presence or absence of which is inferred from biographical data in BoardEx. On both dimensions, a director’s diversity is measured as a difference from the average of the rest of the board.
Our diversity measures explain director retention and promotion, but of particular interest is the asymmetry between gender and non-gender dimensions of sociodemographic identity. On the dimension of gender, women directors are more likely to remain on boards but age- and ethnically-diverse directors are less likely to be retained. The results on promotion are slightly different. Gender matters, but curiously, the results are the opposite of what we see in turnover. Women directors are less likely to be promoted to key director positions. The greater retention of women in boards is apparently accompanied by a reluctance to promote them to key positions. Likewise, age- and ethnically-diverse directors are less likely to be promoted to key positions. Skill diversity also matters. Directors with diverse skills relative to the rest of the board are more likely to be retained and promoted. However, skill diversity does not subsume identity diversity, and that finding remains significant even after controlling for skill diversity.
We further probe and unpack the economics of our findings. If non-key directors can be replaced more easily because they are in greater supply, then identity variables should matter more for non-key directors and be less relevant for key directors. This is exactly what we find. Across a range of specifications and samples, identity is not significant in explaining key director turnover but remains a significant determinant of outcomes for non-key directors. Related tests focus on samples with and without CEO turnover. Identity is significant in determining unemployment outcomes in multiple turnovers and in samples with and without CEO turnover.
One way of thinking about the results is to view them through the lens of the exogenous variation in the types of diversity pressures that firms face. On the dimensions more subject to external pressure, such as gender and its effect on turnover, we observe a pro-diversity pattern. Women are less likely to leave boards. However, on the “silent” dimensions that are less subject to outside pressure, we do not see pro-diversity results. Age-diverse and ethnicity-diverse directors are more likely to leave the board and diverse directors, on all three dimensions, are less likely to be promoted to key directorships.
Our results speak to emerging questions raised by institutional investors. The 2018 diversity report of the prominent proxy advisory firm Institutional Shareholder Services (ISS) states that, while gender dominates board conversations, there has been less conversation on other types of diversity such as ethnicity, skills, and age. Moreover, the assignment of directors to leadership positions within boards remains an open question. We agree with the 2018 ISS diversity report that conversations about diversity in academic and policy circles should be broader. In particular, the debate can perhaps be enriched by focusing on multiple dimensions of identity beyond gender. The conversations should also focus on multiple employment outcomes such as the roles assigned within organizations rather than the mere presence on boards of minorities, women, and older people. Our study takes a step in this direction.
This post comes to us from professors N.K. Chidambaran at Fordham University’s Gabelli School of Business, Yun Liu at Claremont Colleges’ Keck Graduate Institute, and Nagpurnanand Prabhala at The Johns Hopkins Carey Business School. It is based on their recent article, “Director Diversity, Turnover, and Promotion,” available here.