Critical Analysis of Scholarly article

The Pipeline to the Top: Women and Men in the Top Executive Ranks of U.S. Corporations by Constance E. Helfat, Dawn Harris, and Paul J. Wolfson

Executive Overview People often ask about the pipeline of women in line for the top position in major U.S. corporations. Despite persistent interest in this issue, we do not yet have good answers to the question of how long it will take until more than a token number of women hold the CEO position. This study provides numerical estimates that help to answer this question, and also provides new information regarding the job responsibilities and positions in the executive hierarchy of women and men below the rank of CEO. This article presents the results of an extensive data collection effort that has yielded a comprehensive census of top executives in U.S. Fortune 1000 firms as of the year 2000. With regard to the pipeline to the CEO position, our data suggest that we should expect to see a slow increase in the percentage of CEOs that are women in the next five to ten years. Nevertheless, the percentage of CEOs that are women is likely to remain relatively low. As a result, our estimates suggest that if current trends continue, perhaps 6 percent of CEOs in the Fortune 1000 will be women by 2016. We also document the little known fact that almost 50 percent of the firms in the Fortune 1000 had no women as top executives as recently as the year 2000. Moreover, even firms with women executives generally had only 1 or 2 per firm.

T he business press frequently bemoans the dearth of women at the top ranks of business. Academic research has documented this scar-

city as well. And virtually everyone wants to know what the pipeline of women in line for CEO positions looks like. We continually hear people ask a question that so far has lacked an answer: How long will it be until we see more than a token number of women as CEOs in major U.S. corpo- rations? In answer to this question, we provide numerical estimates of the percentage of CEOs that are likely to be women in 2010 and 2016. Based on comprehensive new data, we also ana- lyze how women below the rank of CEO compare with male top executives in terms of functional area job responsibility, position in the executive hierarchy, age, company tenure, and tenure in current position. This analysis provides the basis for an assessment of the future prospects for the representation of women in top management, and for recommendations for improvement.

To date, statistical analyses have contained rel- atively little direct comparison of the characteris- tics of women and men at the executive level (with notable exceptions such as Cappelli and Hamori 2004). But without a baseline for com- parison–namely, male executives who comprise the overwhelming majority of top management–it is difficult to fully assess the position of women in the top executive ranks. In order to provide a comprehensive picture of the status of and pros- pects for women in top executive ranks, we re- quire a large, well designed data set that will enable us to draw reasonably precise, robust con- clusions. This article presents the results of an extensive data collection effort that has yielded a comprehensive census of top executives in U.S. Fortune 1000 firms as of the year 2000. Based on this data set of nearly 10,000 individuals, we have performed a detailed analysis of the characteristics of women and men of executive rank in the U.S.

* Constance E. Helfat ([email protected]) is the J. Brian Quinn Professor of Technology and Strategy at the Tuck School of Business at Dartmouth. Dawn Harris ([email protected]) is an Associate Professor in the Graduate School of Business at Loyola University, Chicago. Paul J. Wolfson ([email protected]) is a Statistical Research Associate at the Tuck School of Business at Dartmouth.

42 NovemberAcademy of Management Perspectives

These data provide a benchmark from which to gauge current and future progress.

Our analysis contains three main findings. First, we find that almost 50 percent of the firms in our sample had no women as top executives. Second, in firms that had women among their executives, we find evidence consistent with “to- ken” status of women on the top executive team. Third, partly as a consequence of the first two findings, the pipeline of women in line for the CEO position, although growing, remains small. Despite these sobering findings, we also have pos- itive data to report. Contrary to popular percep- tion, the women in our sample are not underrep- resented in certain important functional areas such as accounting and legal, relative to the over- all percentage of women in the top executive ranks. In addition, the women do not cluster at the very lowest levels of the hierarchy, but instead tend to hold positions one or two levels below the second-in-command executive. Finally, our results suggest that many of the firms that have women in top management have actively recruited and pro- moted them to executive rank.

In what follows, we first elaborate on the mo- tivation for our study. Then we explain our data collection and coding. We present the results of the study and provide an analysis of the pipeline to the CEO position. We also analyze factors that differentiate firms in terms of the representation of women among top executives.

Why Study Women Executives?

T here is a broad literature on the career ad- vancement of women in business. Topics of research have included gender differences in

career success (e.g., Kirchmeyer 1998) and career mobility (e.g., Valcour & Tolbert 2003). Research also has focused specifically on the “glass ceiling” (for a review, see Powell 1999). This term, coined by Hymowitz and Schelhardt of the Wall Street Journal in 1986, denotes an invisible barrier to the upward movement of women and minorities in management. Morrison and Von Glinow (1990) pointed to systematic barriers to advancement of women in management, including lack of oppor- tunities, power, mentors, and role models. Good- man, Fields, and Blum (2003) investigated

whether greater career opportunities for women within organizations mitigated the glass ceiling effect in a sample of medium-to-large size work establishments in the state of Georgia in the early 1990s. They found that greater opportunity for women due to higher management turnover and emphasis on internal promotion and develop- ment, as well as a larger pool of female non- management employees (as a percentage of all such employees), was associated with a greater likelihood of women in top management.

Prior research has advanced three general types of explanations for generally low percentages of women in management: person-centered; situa- tion-centered; and social-system-centered (Powell 1999). Person-centered explanations refer to indi- vidual-level factors that cause women and men to make different career decisions and to perform job tasks differently. Situation-centered explanations refer to group and organizational-level factors that affect the differential hiring and promotion of women and men. Social-system-centered explana- tions refer to factors in society (political, social, governmental, and economic) that affect the dif- ferential hiring and promotion of women and men.

Our study focuses on the situation-centered, or organizational, level of analysis. Within that level of analysis, we focus on top management. The leader of an organization plays an important role in directing strategy and operations, and as a sym- bol to the outside world (Hambrick & Mason 1984; Hayward, Rindova, & Pollack 2004). Re- search has demonstrated the impact of CEO hu- man capital on firm performance (e.g., Bailey & Helfat, 2003, 2005; Harris & Helfat 1997, 1998; Hitt, et al. 2001). For these reasons, both the business press and organizations dedicated to in- creasing the overall representation of women in business have focused special attention on the CEO position.

Yet in order to understand decision making at the top of the organization, we must also look beyond the CEO. What has come to be called the “top management team” (TMT) is the set of in- dividuals at the top of the organization responsible for the strategic and organizational decisions that affect the direction, operations, and performance

2006 43Helfat, Harris, and Wolfson

of the company as a whole. Studies have shown that TMTs, as well as CEOs, have an important influence on firm strategy and performance (for a review, see Finkelstein & Hambrick 1996).

A key issue regarding the effectiveness of deci- sion making in teams, including TMTs, has to do with the diversity of the team. Diversity within organizational teams leads to greater search for information, range of perspectives, and generation of alternative solutions (e.g., Dutton & Duncan 1987; Watson, Kumar, & Michaelsen 1993). Greater heterogeneity, however, may also lead to offsetting negative effects such as greater conflict and communication difficulties among top team members (Miller, Burke, & Glick 1998). Based on a review of the evidence, Finkelstein and Ham- brick (1996) conclude that on balance, the ben- efits of diversity in experience and background among top managers outweigh any negative im- pact.

Should we include gender when considering the benefits of diversity in top management? Some prominent business people have answered strongly in the affirmative. For example, the former CEO of Newell-Rubbermaid, a major con- sumer products company, has argued that the company must promote managers who can best understand its customers—including women and minorities.1 James Preston, CEO of Avon, and Larry Johnston, CEO of Albertson’s grocery chain, have made the same argument with regard to including women on boards of directors (Daily, et al. 1999). Women control 88 percent of all purchases in the U.S. (Kanner 2004). If women in management have additional insight (beyond that of men) into the purchasing decisions of other women, then companies should benefit from in- cluding women in the top management team.2

In addition to the diversity argument, compa- nies routinely state how important it is to obtain the best possible managers. Logically, if talent is at

a premium, then firms should benefit from having as large a pool of potentially qualified individuals to draw upon as possible. Today women make up over half the managerial and professional work- force, including much lower levels of management than those examined here (Bureau of Labor Sta- tistics 2003). Unless women inherently have less top management potential than men, the sheer increase in the size of the labor pool that comes from including women should benefit companies.

The importance of drawing from the largest possible talent pool, and the potential benefits to TMT decision making, constitute the crux of the “business case” for including women in the top executive ranks of corporations. Yet we still have an incomplete picture of the population of women in top management. In addition, we know rela- tively little about how this population of women compares with the population of men in top man- agement, particularly below the level of the CEO. In what follows, we elaborate on the need for additional data, before explaining our data collec- tion process and coding.

Women in Top Management: What We Know Thus Far

Perhaps the most widely quoted source of data on women in executive rank is the Catalyst bi-annual Census of Women Corporate Officers and Top Earners. Catalyst collects data on women who are corporate officers in the Fortune 500 companies from publicly available company reports to the Securities and Exchange Commission (SEC) and to stockholders, and then asks each company to verify these data. As part of the verification pro- cess, companies can add female corporate officers that do not appear in official company filings or annual reports to stockholders.3 Catalyst also asks the companies to verify the functional area re- sponsibilities of the executives. Based on these data, Catalyst reports the number and percentage of officers who are women.4

1 In February 2004, Joe Galli, CEO of Newell-Rubbermaid at the time, made this statement in a speech at the Tuck School of Business at Dartmouth.

2 A study by Richard, et al. (2004) suggests that under some conditions gender diversity within the top team is positively associated with firm productivity, but the study did not investigate the impact of gender diver- sity on the decision making process of TMTs.

3 This information is based on a conversation between one of the authors of this article and the person at Catalyst responsible for directing the research for the Catalyst Census of Women Corporate Officers and Top Earners in 2000.

4 It is difficult to know what the percentages represent, since it is unclear if the companies also add the names of male executives to their lists

44 NovemberAcademy of Management Perspectives

Academic studies also have examined the rep- resentation of women in top management, often focusing on Fortune 500 companies (for a review, see Powell 1999). The data in the Catalyst reports and other studies show a growing proportion of women in top management in recent years. Most estimates of women in top management in the 1970s through 1990 ranged from zero to 3 percent (Powell 1999). The percentages reported by Cat- alyst for the Fortune 500 companies in the second half of the 1990s were much higher: 8.8 percent in 1995, 11.2 percent in 1998, 12.5 percent in 2000, and 15.7 percent in 2002 (Catalyst 1998, 2002). Hillman, et al. (2005) also found that 7.34 per- cent of top executives in the largest 1,000 com- panies during the period 1990-2003 were women. Finally, Cappelli and Hamori (2004) found that 11 percent of top executives in the Fortune 100 in 2001 were women.

Despite this upward trend, the data show less representation of women among executives most directly in line to be CEO. Catalyst reports that in 2002 women held 9.9 percent of line (profit-and- loss responsibility) officer positions and comprised 5.2 percent of the top five most highly compen- sated corporate officials, up from 7.3 percent and 4.1 percent respectively in 2000. Bertrand and Hallock (2001) found that previously, during the period 1992-1997, 2.4 percent of the top five highest paid executives per firm in the Standard and Poor’s (S&P) ExecuComp data base (cover- ing firms in the S&P 500, S&P Midcap 400, and S&P SmallCap 600) were women. Moreover, Dai- ley, et al. (1999) found that in the Fortune 500 only eight inside directors in 1996 were women (down from 11 in 1987), amounting to 0.006 of the total number of inside directors in the sample. Since these are the women most directly in line to become CEO (Zelechowski & Bilimoria 2003), this very low figure essentially predicts that the percent of CEOs who are women would be small in the near to medium term. Current data bear

this out. As of March 2005, 1.8 percent of CEOs (9 individuals) in the Fortune 500 were women. This is a relatively small change from the 1.2 percent of CEOs (6 individuals) in the Fortune 500 that were women in 2002.

Our Data

The foregoing studies provide a useful starting point for assessing the representation of women in top management. In order to provide a fuller pic- ture, we collected data with several goals in mind. First, in order to gain information about a larger segment of the U.S. economy, we included all companies in the Fortune 1000, rather than only the Fortune 500 companies examined in most prior studies. Second, we coded data for both men and women. This enabled us to make explicit comparisons between them. Third, in order to prevent potential over-reporting by firms of the number of women executives, we included only executives that companies listed in their official filings and reports. Although companies differed in the number of executives they reported, we have accounted for these differences in the statis- tical analysis. We did not restrict our analysis to a subset of the executives listed per firm, such as the five most highly compensated executive officers that all companies must report. This approach would have excluded a large proportion of top executives that are women.

Our data collection began with the list of For- tune 1000 companies from the year 2000.5 We coded information from the short biographies of every individual in the List of Executive Officers reported by each company in their annual 10-K reports to the Securities and Exchange Commis- sion or in proxy statements. We developed a de- tailed coding protocol, which we relied on to pre-code every biography by hand. Research assis- tants then entered the data into an Excel spread- sheet. Other research assistants proofread the spreadsheet entries. The final database consisted of 942 firms and 9,950 individuals. A number of firms on the Fortune 1000 list did not file 10-K reports in 2000 due to mergers, acquisitions, bank-

for Catalyst. The Catalyst reports also include the percentage of women who are among the top five most highly compensated individuals of their companies, are “line” officers, and have selected job titles below the rank of CEO. In addition, the reports contain a breakdown of the number and percentage of female officers by industry. Finally, the reports list the titles held by female officers in each company, but do not list the titles held by male officers for purposes of comparison.

5 We started collecting data in the summer of 2000, and used the most recent Fortune 1000 list available.

2006 45Helfat, Harris, and Wolfson

ruptcies, or private ownership, which reduced the sample size. In addition, some companies did not list information for all of the data items that we coded. We therefore have small amounts of miss- ing data for most of the variables in our data base (other than gender).

In order to compare the jobs and career pat- terns of women versus men, we coded several variables. First, we coded more fine-grained dis- tinctions between functional area responsibilities than in prior studies of women in top manage- ment. Although prior studies have distinguished between line (profit-and-loss responsibility) and staff (non-line) positions (see e.g., Catalyst 2000, 2002), they have not examined gender differences in staff positions. Since some types of staff posi- tions (e.g., finance) are often perceived as more influential than others, we sought to understand whether women and men differed in their staff as well as their line job responsibilities. The Appen- dix explains how we coded the functional area job classifications.

In order to further gauge the relative authority and status of women within the top management of each firm, we coded the relative rank of each executive, female and male, within each top exec- utive hierarchy. Although studies have examined the representation of women among executives holding particular executive titles (e.g., Catalyst 2000, 2002; Powell 1999), prior research has not analyzed within-company executive hierarchies. Corporate titles mean different things in different companies. For example, some companies use the Senior Executive Vice President title and others do not. The title that an individual holds does not necessarily tell us where that person falls within the top management hierarchy of each company. The Appendix explains the algorithm that we used to assign a within-company hierarchical rank to each executive.

We also coded the number of years that each executive had held his or her current position, and the number of years that the executive had worked for the company. These data can help us to understand whether firms used early promotion and outside hiring more often for women than for men. Currently, we do not know whether firms use these approaches to increase the representa-

tion of women in top management. We recorded several other data items as well: age of the exec- utive, industry, number of corporate officers re- ported by each company, and total revenues (a measure of firm size) and profits of each compa- ny.6

Our data base comprises what to our knowledge is the most comprehensive set of information on the characteristics of women and men of execu- tive rank in the United States. What emerged after a time-consuming data collection effort is a baseline from the year 2000 that can be used to assess the progress of women in the recent past. These data also can inform us about the future. For example, our data include women at lower levels of the executive hierarchy not directly in line for promotion to CEO in 2000. Since it can take many years to be promoted through the ranks of the executive hierarchy, our data can help to predict the extent to which we should expect to see a significant number of women who are CEOs in the next 5 to 10 years.

In what follows, we compare women and men at the executive level using the data just de- scribed. Then we provide estimates of the percent of women in the Fortune 1000 who are likely to reach the CEO position in 2010 and 2016. Fi- nally, we report the results of statistical analyses that assess differences between companies in the representation of women among top executives.

A Comparison of Women and Men in Top Management

O f the 942 firms in our sample, 8.25 percent (821 executives) of the total of 9,950 execu- tives in those firms were women. As reported

in Table 1, almost one-half of the companies in our sample (48 percent) had no women as exec- utives. Twenty-nine percent of the companies had just one woman of executive rank and only 23 percent had more than one woman. Thus, the overall figure of approximately 8 percent women

6 We also recorded the educational background of each executive where available, but since the majority of the biographies did not provide this information, we did not analyze these data. We had originally hoped to analyze the job histories of the executives as well, but this information was spotty and often not comparable between companies.

46 NovemberAcademy of Management Perspectives

in top management masks the fact that almost half of the companies had no executives who were women.

Table 1 indicates that the Fortune 1000 com- panies had between zero and 8 women of execu- tive rank per firm. Few firms, however, had more than 3 women at the top level. The variation across firms was much greater for the percentage of executives per firm that were women, ranging from a low of zero to a high of 60 percent. To quantify the dispersion across firms in the percent- age of executives that were women, we can use the coefficient of variation (the ratio of the standard deviation to the mean). This ratio is 1.20, com- pared with a ratio of 0.10 for the percentage of executives per firm who were men. These figures tell us that there were far greater differences be- tween firms in the percentage of executives who were women than in the percentage who were men.

In order to obtain additional information about

the representation of women in top management, we next compared the ages of female and male executives. We also examined how long the women and men had worked for their current employers, and how long they had held their current positions. Table 2 reports these data.

With regard to age, we found that on average women were approximately 5 1/2 years younger than the men (statistically significant at the 0.0001 level).7 Women had an average age of 46.7, whereas men had an average age of 51.1. Cappelli and Hamori (2004) found almost identi- cal results for executives in the Fortune 100 in 2001: an average age of 47 for women and 52 for men. Figure 1 provides further information regard- ing the age distribution of women and men at executive levels. As the figure shows, close to half the women— 42 percent—were age 45 or less, compared with only 24 percent of the men. Clearly, the women were substantially younger than the men.

We also found that on average women had worked for their current employers for approxi- mately 2.5 years less than the men: 8.1 years for women versus 10.7 years for men, a statistically significant difference at the 0.0001 level. Figure 2 compares the distribution of company tenure for

7 For all t-tests reported here, we tested for equality of the variances in the two sub-samples. If the variances were unequal, we performed a t-test under the assumption of unequal variances and report that result here.

Table 1 Number of Executives per Firm

Number of Women Per Firm

Number of Firms

Percent of Firms in the Sample

0 450 47.77 1 276 29.30 2 148 15.71 3 44 4.67 4 12 1.27 5 6 0.64 6 4 0.42 7 1 0.11 8 1 0.11

Minimum Value

Maximum Value

Number of Female Executives Per Firm

0 8

Number of Male Executives Per Firm

2 50

Total Number of Executives Per Firm

3 54

Lowest Level in the Hierarchy Reported

3 7

Table 2 Executive Age, Company Tenure, and Tenure in Current Position

Women Mean Standard Deviation

Minimum Value

Maximum Value

Age 46.70 6.11 29.00 78.00 Years of Company

Tenure 8.08 7.13 0 46.00

Years in Current Position

2.62 2.55 0 17.00

Men Mean Standard Deviation

Minimum Value

Maximum Value

Age 51.07 7.46 28.00 91.00 Years of Company

Tenure 10.70 9.60 0 72.00

Years in Current Position

3.45 3.98 0 53.00

2006 47Helfat, Harris, and Wolfson

women and men. This figure shows that 47 per- cent of the women had worked for their compa- nies for 5 years or less, compared with 38 percent of the men. Companies appear to have been more willing to hire high-level women than men from outside the firm, perhaps in order to seed the pool of women in top management.

Additionally, women on average had approxi- mately one less year of tenure in their current executive positions than men: 2.6 years for women compared with 3.5 years for men, a statis- tically significant difference at the 0.0001 level. Figure 3 compares the distribution of tenure in current position (defined as the period for which an individual had no change in job titles) for women and men. Sixty-four percent of the women had held their current position for two years or less, versus 56 percent of the men. This disparity further suggests noticeable efforts by some firms to promote women to executive rank.

In sum, these data indicate that whereas nearly half of the Fortune 1000 firms had no women executives, the other half of the firms were ac-

tively recruiting and promoting women to the top executive ranks. Relative to the men, the women were younger, and had worked for their companies and held their current positions for shorter periods of time.

Job Positions of Women and Men at the Executive Level

Although the analysis thus far suggests that a subset of the Fortune 1000 companies were ac- tively promoting women to executive rank, women in management are sometimes perceived to hold less influential positions than men. We investigated three ways in which women might or might not hold less powerful positions at the ex- ecutive level. First, we analyzed where women ranked in the top executive hierarchy relative to men. Second, we analyzed the job responsibilities of women and men in order to ascertain whether women were more likely than men to hold non- operational “staff” positions. Executives who hold staff positions generally have less say in decision making than those that hold “line” responsibility

Figure 1 Age of Executives

48 NovemberAcademy of Management Perspectives

for business operations. Third, within staff posi- tions, some are perceived as more influential than others. For example, finance is often viewed as a relatively important “staff” position. Indeed, the position of chief financial officer has become a more common route to the CEO position than in the past. We therefore analyzed the representation of women versus men in different functional “staff” areas.

Table 3 reports the total number of executives and the number of women and men at each level in the executive hierarchy. The highest possible within-company level in the hierarchy was a rank of 1 and the lowest reported level was a rank of 7. Few executives had a rank of 6 or 7, reflecting the fact that most firms reported only very high level executives. At Level 1, the very top of the hier- archy, only 0.62 percent (7 out of 1,119) of exec- utives were women. (A number of companies had more than one executive at Level 1 in the hier- archy.) At Level 2 (second-in-command execu- tives), only 1.7 percent of executives (7 out of 424) were women. The representation of women

increased sharply below Level 2, rising to 6.4 percent at Level 3, 10.4 percent at Level 4, and 12.8 percent at Levels 5, 6, and 7 combined.

A majority of the women held positions just below the top two rungs of the executive hierar- chy. Table 3 indicates that 66 percent of the women had attained either Level 3 or Level 4 in the hierarchy. Figure 4 and table 3 show how each gender was distributed across different levels in the hierarchy. Levels 1 and 2 combined contained only 1.7 percent of the total number of women, compared with 16.7 percent of men. The disparity largely disappears in Levels 3 and 4, which to- gether contained 66 percent of the women and 63.5 percent of the men. Although a smaller per- centage of the women (23 percent) were in Level 3 than the men (30.3 percent), relative to their overall numbers, women were well-represented in the two combined levels just below the second- in-command position.

We next investigated differences between women and men in line versus staff positions and in functional area responsibilities. By definition,

Figure 2 Company Tenure

2006 49Helfat, Harris, and Wolfson

the top two levels in the hierarchy are line posi- tions, with direct profit-and-loss responsibility. Positions such as CEO, President, and Chief Op- erating Officer have direct responsibility for the operations of the entire company. Therefore, we coded a position as having “line” responsibility if: an individual was in Level 1 or 2 in the hierarchy; the title indicated that the individual was head of an operational subsidiary; or at least one of the individual’s functional area responsibilities in- cluded operations, marketing, or sales. We in-

cluded marketing and sales as line positions since these positions often come with profit-and-loss responsibility in many companies, such as those in consumer products, retailing, and financial ser- vices.

In order to compare the representation of men and women in line positions, we compared the percentage of women who held line positions with the percentage of men who held line positions. The percentage of women in line positions (25.3 percent) was approximately half that of men (52.5

Figure 3 Tenure in Current Position

Table 3 Level in the Executive Hierarchy

Level in the Hierarchy

Total Number of Executives

Per Level

Number of Men

Per Level

Number of Women

Per Level

Women as percent of

Total Executives Per Level

Women as percent of

# Total Female

Executives in Full

Sample

Men as percent of Total # Male Executives in Full Sample

1 1119 1112 7 0.63 0.85 12.18 2 424 417 7 1.65 0.85 4.57 3 2955 2766 189 6.40 23.02 30.30 4 3385 3032 353 10.43 43.00 33.21 5 1719 1499 220 12.80 26.80 16.42 6 326 287 39 11.96 4.75 3.14 7 22 16 6 27.27 0.73 0.18

50 NovemberAcademy of Management Perspectives

percent). Since the remainder of the executives held staff positions of some type, these data indi- cate a corresponding overrepresentation of women in staff positions. To further investigate this finding, we excluded executives in Levels 1 and 2 of the hierarchy from the analysis. Because very few women held these high-level line posi- tions, including the Level 1 and 2 positions could overstate line/staff differences between men and women. When we excluded Levels 1 and 2, the percentage of men in line positions dropped by 10 percentage points to 42 percent, while the per- centage of women in line positions dropped only slightly to 24 percent. The gap between women and men, however, remained substantial.

Next we compared the representation of women and men in each of the individual func- tional areas. Since many executives had more than one functional area responsibility, we as- signed each job responsibility (rather than each individual) in the database to a functional area. Hence, the number of job responsibilities exceeds the number of executives. Then for each func- tional area, we computed the ratio of the number

of job responsibilities held by women in that func- tional area relative to the total number of job responsibilities held by women in all functional areas together, and converted this to a percentage. We performed the corresponding calculations for men. Although these data depend on the number of job responsibilities and titles (which often re- flect job responsibilities) reported per person, they provide a sense of whether or not women and men clustered in different functional areas.

Figure 5 enables us to discern functional areas in which women and men were under- or over- represented relative to their overall representation among top executives. The data reveal large dis- parities in the following areas: operations, finance, accounting, secretary, legal, public relations, and human relations. Men were much more heavily represented in the first two areas, especially oper- ations, and women were much more heavily rep- resented in the latter areas. Some of these findings accord with popular perception, particularly with regard to operations, finance, and public and hu- man relations. Other findings, however, do not necessarily fit popular perceptions. Women were

Figure 4 Level in the Executive Hierarchy

2006 51Helfat, Harris, and Wolfson

more heavily represented in accounting and legal positions than men. Furthermore, we do not see wide disparities in areas such as strategy, opera- tions support, and information technology, which might generally be perceived as positions held by men rather than women.

Women and men had fewer disparities in job responsibilities than a popular reading of the situation might suggest. Although underrepre- sented relative to their total numbers in line and finance positions, women were overrepre- sented in some functional areas, notably ac- counting and legal, and held their own in in- formation technology and strategy. Women were underrepresented at the very highest levels of the hierarchy, but two-thirds of them at- tained ranks in the hierarchy just below the top two rungs. This suggests that eventually we may see greater numbers of women at the very top of the hierarchy. We next use our data to provide some rough estimates of what we might expect to see and how soon.

The Pipeline Question

I n order to estimate how many women may move up in the executive hierarchy and how soon, we need to account for several factors. First, we

must account for the fact that not all executives at each level of the hierarchy are promoted to the next higher level in the hierarchy. Second, we must account for the speed at which individuals who are promoted move up in the hierarchy. Third, because line positions, and more recently the CFO position, are often a route to the top, we need to know the representation of women in line plus CFO positions in relevant levels of the hier- archy. To assess the pipeline to the CEO position, we start with a base case scenario that accounts for the foregoing factors, and then suggest some alter- natives to the base case estimates. Table 4 reports the range of estimates discussed below.

For the base case, we assume that average CEO tenure is 5 years. This is probably on the low side, but it ensures that our estimates of the rate at which women may become CEO will not suffer

Figure 5 Functional Areas of Responsibility

52 NovemberAcademy of Management Perspectives

from undue pessimism. In accordance with this assumption about average CEO tenure, we assume that other promotions from one level to the next within the executive hierarchy occur on average every five years as well. Additionally, we assume that women will be promoted to the next higher level in the hierarchy in accordance with their representation at their current level of the hierar- chy. Thus, if women were to comprise 10 percent of all Level 4 executives in 2005, we would expect that after promotion, on average they would com- prise the same 10 percent of all Level 3 executives in 2010.

Using these (perhaps optimistic) base case assumptions regarding the speed and rate of promotion, our data enable us to make rough predictions regarding the pipeline to the CEO position in the near to medium term, meaning in the next five to ten years. Since our data come from the year 2000, an average CEO ten- ure of five years implies that by 2005 the aver- age CEO position would have turned over. Of-

ten, new CEOs come from the ranks of second- in-command executives at the same or another firm. Thus, in 2005, the executives most likely to reach the top two rungs by 2010 and 2015 would have been at Levels 3 and 4, respectively, of the executive hierarchy in the year 2000.

To begin, we take the most optimistic scenario and assume that all executives, not just those in line or CFO positions, are potential candidates for CEO. In the year 2000, 6.4 percent of all execu- tives at Level 3 in the hierarchy were women. This would imply that by 2010, with an average five-year tenure in each level of the hierarchy, this same 6.4 percent of CEOs would be women. Sim- ilar reasoning suggests that since 10.4 percent of all executives in Level 4 in 2000 were women, by 2015 we might expect to see this same percentage of executives reach Level 1 in the hierarchy. We emphasize that these are only estimates, and that we certainly cannot forecast actions by companies that might alter these estimates. Furthermore, our estimates are based on averages, and things may

Table 4 Pipeline Estimates

Average Years of Tenure Per Level in the Executive Hierarchy Year

Estimated Percent of CEOs That May be Women

Base case: 5 Years (All Executives Eligible for CEO Position)

2010 6.4

5 Years (Only Line and CFO Executives Eligible)

2010 4.9

8 Years (All Executives Eligible)

2010 (reach CEO in 2008)

1.7*

8 Years (Only Line and CFO Executives Eligible)

2010 (reach CEO in 2008)

1.7*

4 Years (All Executives Eligible)

2010 (reach CEO in 2008)

6.4

Base case: 5 Years (All Executives Eligible)

2015 (and 2016) 10.4

5 Years (Only Line and CFO Executives Eligible)

2015 (and 2016) 6.2

8 Years (All Executives Eligible) 2016 6.4 8 Years

(Only Line and CFO Executives Eligible) 2016 4.9

4 Years (All Executives Eligible) 2016 12.8 *These estimates are based on the percentage of women in Level 2 in 2000, which by definition is a line position. Hence, the two

estimates with asterisks are the same.

2006 53Helfat, Harris, and Wolfson

move more slowly or more quickly than averages would predict.

The base case does not account for the fact that executives in line and CFO positions generally have a greater likelihood of promotion to the top position. We next refine the base case estimates under the assumption that only CFOs or execu- tives in line positions are eligible for promotion to higher levels in the hierarchy. This approach yields the following train of logic. In 2000, 4.9 percent of all executives in Level 3 in line or CFO positions were women. An average five-year ten- ure in each level of the hierarchy implies that by 2010 this same 4.9 percent of CEOs or chairmen of the board would be women. Similar logic sug- gests that since 6.2 percent of all executives in Level 4 in line or CFO positions in 2000 were women, by 2015 we might expect to see this same percent of executives reach Level 1 in the hierar- chy.8

These estimates are sensitive to assumptions about the number of years of tenure in each level of the hierarchy. If executives have a longer av- erage tenure of eight years in each level in the hierarchy for both women and men, executives in Level 2 (a line position) in 2000 would be pro- moted to the CEO position in 2008, and execu- tives in Level 3 in the hierarchy in 2000 would be in line for the CEO position in 2016. In this case, 1.7 percent of all executives would be women in 2008 (and therefore in 2010). Between 4.9 and 6.4 percent of CEOs would be women in 2016. If we assume a shorter average tenure of four years in each level in the hierarchy, the base case esti- mates given earlier for 2010 do not change, al- though women would reach these percentages two years earlier in 2008. The estimates for one year beyond 2015, however, do change from the base case. Executives at Level 5 in the hierarchy in 2000 would reach the CEO position in 2016. In this case, up to 12.8 percent (the proportion of all

executives at Level 5 in 2000) of CEOs would be women in 2016.9

The foregoing estimates suggest that between 1.7 percent (based only on line and CFO execu- tives in 2000) and 6.4 percent (including all ex- ecutive positions) of CEOs and Chairmen of the Board may be women by 2010. Between 4.9 and 12.8 percent may be women by 2016. This broader range in the more distant future reflects the fact that there were more women lower down in the hierarchy in 2000. Companies therefore will have a greater pool of women from which to draw for the CEO position.

Whether one sees the pipeline picture as rosy or dreary depends on which estimates seem most realistic. We favor the estimates where only line and CFO executives are candidates for promotion to the top. Our base case estimates for this pool suggest that 4.9 percent of CEOs could be women by 2010 and 6.2 percent of CEOs could be women by 2015 (and 2016). These estimates may slightly underestimate the full pool of CEO candidates, since firms occasionally promote executives in other functional areas such as legal or audit to the CEO position. The speed at which women reach the top could also be somewhat faster than we have estimated, particularly if firms promote large numbers of women from Level 3 positions directly to the CEO position or if firms promote women through the executive ranks more quickly than men.10

Our analysis nevertheless strongly suggests that while we should expect to see more women at the top of Fortune 1000 firms, progress will be slow. The facts simply do not support statements such as that by James Preston, former CEO of Avon, that “women are in the pipeline in droves” (Himel- stein & Forest 1997:64). Indeed, it is difficult to have a full pipeline of women in line for promo- tion to CEO when as recently as the year 2000,

8 Having a line or CFO position is probably particularly important for promotion from Level 3 to Level 2, since the second-in-command position entails line responsibility. Unless women in Level 4 who lack line or CFO positions obtain them when they move to Level 3, the percentage estimates given in the text for female candidates for Level 1 positions would not change.

9 Since Level 5 is relatively low in the hierarchy of top executives, and since there are fewer CFO positions at this level, we do not attempt to refine the estimates to reflect only line and CFO women at Level 5 in 2000.

10 The pipeline estimates become complicated if men and women move up at a different pace. Differential promotion rates imply that during the periods of time when women have been promoted before the men catch up, either the total number of executives at each level must change or the number (and percentage) of men must decrease to keep the total number of executives at each level the same.

54 NovemberAcademy of Management Perspectives

almost half of the companies in the Fortune 1000 had no women as top executives at all.

What (If Anything) Explains Differences Between Firms?

The fact that almost half of the companies had no women as top executives raises the question: what, if anything, explains differences between firms in the representation of women? We considered three possible explanations that our data enabled us to examine. First, we assessed whether firms of different sizes had systematically different repre- sentation of women. Some prior reports and re- search have analyzed larger firms that comprise a subset of our sample, such as the Fortune 100 (Cappelli & Hamori 2004) or the Fortune 500 companies (e.g., the Catalyst reports). We com- pared these sets of companies with the remainder of the firms in our larger sample. Second, we examined differences between industries. Prior re- search has found that companies in services in- dustries tend to have greater representation of women in top management (Cappelli & Hamori 2004; Hillman, et al. 2005; Goodman, Fields, & Blum 2003). In addition, some CEOs have stated that companies need women executives in order to gain the best possible understanding of their customers. We therefore investigated whether consumer products companies had greater repre- sentation of women than other firms. Finally, we examined whether the total number of executives reported per firm had a bearing on the reported representation of women in top management. Companies differed in the total number of exec- utives that they reported in the List of Executive Officers in their 10-K reports and proxy state- ments. Generally, companies that reported a greater number of executives also reported more executives lower down in the executive hierarchy. Since women tended to cluster at somewhat lower levels in the executive hierarchy than men, we investigated whether companies that reported larger numbers of corporate officers in total also reported greater representation of women.

With regard to differences in firm size, we found almost no disparity in the representation of women between the top and the lower portions of the Fortune 1000 companies. Women comprised

7.8 percent of executives in the Fortune 100, 8.3 percent in the Fortune 500, and 8.2 percent in the Fortune 501-1000 companies. These percentages did not differ significantly statistically between the companies in the Fortune 500 and those in the Fortune 501-1000, or between the Fortune 100 and the Fortune 501-1000.

Unlike the results for firm size, we found large differences between industries (as classified by Fortune) in the representation of women. To mea- sure the extent of these differences, for each in- dustry we computed the average percentage of executives per firm that were women and the average percentage of men. We then calculated the coefficient of variation (the ratio of the stan- dard deviation to the mean) across industries for each of these percentage measures. That coeffi- cient of variation is 0.46 for women and 0.04 for men. The difference in these ratios underscores the large differences between industries in the representation of women, compared with the min- imal differences between industries in the repre- sentation of men. Table 5 reports the ten indus- tries with the highest mean percentages of women executives per firm and the ten industries with the lowest mean percentages. Some of these industries might accord with general preconceptions of where women are well or poorly represented, such as a low representation of women in trucking and a relatively high representation in soaps and cos- metics. But we find a few surprises as well. Com- puter software and transportation equipment are in the top two industries and furniture is in the bottom ten.

We also examined whether the representation of women differed between service industries, con- sumer products industries, and the remainder of the industries in the Fortune 1000. The percentage of executives in each sector who were women was: 9.6 percent in service companies, 8.2 percent in consumer products companies, and 7.2 percent in the other industries. An F-test revealed that these percentages differed significantly statistically at the 0.001 level. Subsequent t-tests revealed that the difference between service and all other com- panies drove these results (significant at the 0.001 level); consumer products companies did not dif- fer significantly statistically from either the service

2006 55Helfat, Harris, and Wolfson

or the other industry companies. These results add further confirmation to the findings of Cappelli and Hamori (2004), Hillman, et al. (2005), and Goodman, Fields, and Blum (2003), who also found greater representation of women in service industries. Goodman, et al. (2003) proposed that this finding may reflect the greater value placed on women’s interpersonal skills in non-manufac- turing industries.

Next we examined whether companies that reported having more executives in total also re- ported greater representation of women in their executive ranks. Figure 6 displays a plot of the

number of women executives per company against the total number of executives per company. The number of women executives per firm appears to increase modestly as the number of executives per firm increases. In particular, the figure suggests that as the number of executives reported per firm increases, so does the likelihood that the total includes at least one woman.

Figure 7 displays a plot of the percentage of executives per firm that were women against the number of total executives per firm. Notably, the percentage of women executives decreases as the number of executives per firm increases. The fact that the proportion of women falls as the total number of executives rises suggests the possibility of “tokenism” with regard to the representation of women. We know from the raw data that 29 percent of the firms had one woman of executive rank and only 23 percent of the firms had more than one woman. Of the latter set, few firms had more than two women. As Figure 7 demonstrates, even in the firms with women at the executive level, the number of women does not increase in proportion to the total number of executives per firm. This evidence of tokenism of women in top management echoes that of Farrell and Hersch (2005), who found strong evidence of tokenism on boards of directors. They found that boards added women when they had low or zero repre- sentation of women, and sought to attract a new female board member when a woman left the board.

Thus far, we have seen that the percentage of executives who were women differed by industry of operation and that firms that reported more women also reported more executives in total. But how important were these factors relative to one another, and relative to other factors that might differentiate firms with regard to the representa- tion of women? To answer this question, we next turn to regression analysis, which enables us to incorporate multiple potential explanatory factors simultaneously.

Regression Analysis

To measure the representation of women among top executives, we used two different dependent variables: the number of executives per firm that

Table 5 Industry Percentages of Executives Who Are Women

10 Industries with the Highest Percentage of Executives Who Are Women

Industry Percentage of

Women Publishing, Printing 15.8 Transportation Equipment 15.7 Securities 14.8 Health Care 14.6 Temporary Help 14.5 Airlines 13.8 Food Services 13.6 Computer Software 13.4 Soaps & Cosmetics 13.1 Pharmaceuticals 12.5

10 Industries with the Lowest Percentage of Executives Who Are Women

Industry Percentage of

Women Semiconductors 1.3 Energy 2.8 Waste Management 3.6 Trucking 3.8 Aerospace 3.8 Mail, Package, & Freight Delivery

3.8

Pipelines 3.9 Motor Vehicles & Parts 3.9 Furniture 4.2 Electronics, Electrical

Equipment 4.3

56 NovemberAcademy of Management Perspectives

were women and the percentage of executives per firm that were women. The implication contained in Figure 7, namely, that the number of women may be an artifact of company reporting policies, makes it especially important that we examine the percentage of executives per firm that were women. If one of the key ways that firms show that they have more women among their execu- tives is by reporting a larger executive team, ana- lyzing the number rather than the percentage of executives that are women may be misleading.

We report two sets of regressions, one set per dependent variable.11 The explanatory variables include firm-level characteristics as well as vari-

ables that reflect characteristics of the executives in each firm. Since the regressions use the firm as the unit of observation, the variables that reflect executive characteristics are firm-level averages. The Appendix describes the explanatory variables and the rationale for their inclusion in the regres- sions. As a caveat, we do not attribute causation to the explanatory variables, since our cross-sec- tional data do not permit us to preclude potential endogeneity of some of the variables.

Table 6 reports the regression results. The re- gressions show that an important factor associated with differences between firms in both the num- ber and percentage of women is industry of oper- ation. The industry dummy variables are signifi- cant as a group: some industries, and therefore the firms in them, have greater representation of women than others. This result is consistent with our earlier analysis that showed substantial differ- ences in the representation of women across in- dustries. In the regression for the percentage of executives that were women, none of the variables other than industry of operation were significant.

11 For the regression for the number of women per firm, we use Poisson maximum likelihood estimation. Poisson estimation is appropriate when the dependent variable consists of integers (termed “count data”), often of low values (including zeros). We also tested for overdispersion of the variance relative to the mean, a potential problem in Poisson regression, but found no evidence of such in the regressions reported here. For the regression for the percentage of women per firm, we use Tobit maximum likelihood estimation. Tobit estimation is appropriate when a variable is censored at a lower threshold such as zero, as well as when a variable also has an upper threshold, such as for a percentage variable. We used STATA to estimate all of the regressions.

Figure 6 Number of Women vs. Total Number of Executives

2006 57Helfat, Harris, and Wolfson

In the regression for the number of executives who were women, only three variables were significant at the .10 level or less (two-tailed test): number of (male) executives reported, lowest level in the executive hierarchy reported, and profitability.

As a sensitivity analysis, we assessed whether firms that had a woman in the top position were more likely to have greater representation of women in the remainder of the top executive team. To conduct this analysis, we excluded women in Level 1 from the dependent variables and added an explanatory dummy variable indi- cating whether or not the firm had any women in Level 1. The dummy variable for a woman in the top position was not significant and the other results did not change.

The regression analysis confirms the results of the two scatter plots presented earlier in Figures 6 and 7. In the regressions, a greater number of male executives per firm is associated with an increased likelihood that we observe a greater number but not a greater percentage of executives per firm who

were women. Consistent with this result regarding company reporting policies, the regressions also show that firms that reported executives lower down in the hierarchy reported a greater number but not a greater percentage of executives who were women. The fact that the total number of reported executives and the lowest reported level in the hierarchy is positively associated with a greater number but not a greater percentage of women executives per firm again suggests a degree of “tokenism.” Furthermore, although we find that greater profitability is associated with a greater number of women per firm,12 this result does not hold for the percentage of women per firm. Over- all, the regressions indicate that when we analyze the percentage rather than the number of execu- tives per firm who are women, the significance of

12 Catalyst and Adler (1999) found similar results for the relationship between profitability and the number of women executives per firm using simple correlations, as did Hillman, et al. (2005) using regression analysis. These authors did not analyze the percentage of women executives per firm.

Figure 7 Percent of Executives Who Are Women vs. Total Number of Executives

58 NovemberAcademy of Management Perspectives

variables other than industry of operation disap- pears. For this reason, we recommend that future research on the representation of women pay par- ticular attention to the percentage rather than the number of executives per firm who are women.

The regressions further show that characteristics that differentiated women from men in the Fortune 1000 companies – age, company tenure, and tenure in current position – did not differentiate firms in terms of either the number or percentage of execu- tives who were women. Although the women were younger and had less company tenure and tenure in their current positions than the men, this did not occur because firms with more women had younger executives, relied more heavily on outside hiring, or promoted executives more quickly. Instead, the rep- resentation of women in top management appears to reflect aggressive promotion and hiring of women specifically.

Discussion and Conclusion

W e began this study by asking about the pipe- line of women to the CEO position. Based on an extensive data collection effort, we

have provided quantitative estimates of the per-

centage of CEOs that may be women in 2010 and 2016. We estimate that in 2016, between 4.9 and 12.8 percent of CEOs may be women. Within this range, we believe that a particularly likely esti- mate is 6.2 percent, because it includes only line and CFO executives as candidates for the CEO position. Although low, this estimate is substan- tially greater than the 1.8 percent of CEOs in the Fortune 500 who were women in 2005.

We found other promising signs as well. In companies with women executives, the women were younger, had less company tenure, and less tenure in their current positions than the men. These factors suggest that many companies were aggressively hiring and promoting women into the top executive ranks. Moreover, although women clustered lower down in the executive hierarchy than men, two-thirds of the women executives held positions in the two levels just below the second-in-command. Once women are in the ex- ecutive hierarchy, they do well in terms of rank.

In order for women to continue to make and accelerate the sort of progress that our data indi- cate, they need to reach executive rank in the first place. Therefore, getting more qualified women

Table 6 Firm-Level Regressions

Dependent Variable Number of Women Per Firma Percentage of Women Per Firmb

Constant Term �1.055 (0.717) �0.107 (11.007) Average Age of Male Executives �0.015 (0.013) �0.242 (0.207) CEO Age 0.006 (0.006) 0.032 (0.092) Average Years of Company

Tenure of Male Executives �0.009 (0.008) �0.077 (0.116)

Average Years in Current Position of Male Executives

�0.025 (0.015) �0.221 (0.248)

Number of Male Executives 0.048 (0.070)*** �0.070 (0.134) Lowest Level in Executive Hierarchy of

Male Executives 0.093 (0.048)* 0.694 (0.72)

Ratio of Line to Total Positions of Male Executives

0.056 (0.223) 0.843 (3.326)

Profitability (Return on Sales) 0.007 (0.003)** 0.065 (0.059) Revenues (logarithm) 0.018 (0.046) 0.862 (0.678) Industry Dummy Variables Included Included

N�900 Standard errors are in parentheses. Two-tailed significance levels: *** � � 0.001, ** � � 0.05, * � � 0.10 aPoisson estimation bTobit maximum likelihood estimation

2006 59Helfat, Harris, and Wolfson

into the executive hierarchy is critical. This first and foremost requires a change in the almost 50 percent of firms in the Fortune 1000 that had no women at the executive level. These companies can draw lessons from those companies that have made advances in this area. Our findings suggest that companies achieved greater representation of women in the top executive ranks through aggres- sive promotion and hiring, policies that compa- nies lacking women executives could emulate. Moreover, even companies that had women ex- ecutives could benefit from additional efforts of this sort, in light of our findings suggestive of “tokenism.”

Aggressive promotion and hiring of women into top management requires a pool of available talent. Companies cannot simply recruit talented women from other firms, since eventually this approach will leave some firms short of talented women. Companies must further develop and in- crease the overall pool of talent from which to draw female executives. This has particular import for the still low proportions of women in line positions, which are an important route to the top of the executive hierarchy. Unless firms find ways to move women into line positions and retain them, the route to the top will remain much more difficult for women than for men.

The extant literature contains many useful suggestions for developing the pool of women with top management potential. For example, providing developmental experience for lower- level female managers can increase the propor- tion of women in top management (Powell 1999). Specific approaches that companies can take to develop the careers of women include: mentoring (formal and informal); developing and utilizing women’s networks inside and out- side of the organization; and creating and im- plementing leadership development programs for women (Society for Human Resource Man- agement 2004). A good deal of research has documented the importance of mentoring and networking to the career success of women (e.g., Metz & Tharenou 2001), as well as differences in networking success for men and women (e.g., Lyness & Thompson 2000; van Emmerik, Eu- wema, Geschiere, & Schouten 2006).

A structured hiring and promotion process that holds decision makers accountable also has less room for personal biases to affect hiring and pro- motion decisions (Powell & Butterfield 1994). This in turn can help to increase the percentage of top executives who are women (Powell 1999). Companies therefore can benefit by reviewing hu- man resources policies and practices to insure that they are fair and inclusive (Society for Human Resource Management 2004).

In addition to the indirect steps just men- tioned, companies can seek to directly increase the proportion of women who are candidates for top (and other) management jobs, which is par- ticularly important when most job incumbents and applicants are men (Powell 1999). Specific hiring and promotion policies and processes in- clude: incorporating the advancement of women into performance goals for line management; training line management to raise awareness and understanding of barriers to the advancement of women; identifying best practices that support the advancement of women; tracking the advance- ment of women in the organization; and develop- ing a list of women for succession planning (So- ciety for Human Resource Management 2004).

Representation of women in top manage- ment is also sensitive to the interest of women in holding these jobs and in remaining in the organization if faced with limited career oppor- tunities (Powell 1999). Support for women with families, such as flexible work schedules, child and elder care assistance, and temporary leaves of absence for family reasons, can increase the interest of women in holding top management jobs and in remaining with the organization (Powell 1999). Having more women in top management positions also may lead to less turnover of women at lower levels of the orga- nization, in part by influencing the perceptions of women that the organization provides oppor- tunities for career advancement (Powell 1999; Cohen & Elvira 1997).

The observation that having women in top management may lead to more women with potential for promotion to top management brings us back to a key conclusion of our anal- ysis: getting more qualified women into the

60 NovemberAcademy of Management Perspectives

executive hierarchy remains a critical priority. Achieving this goal requires leadership and commitment from the most senior executives of business organizations.

APPENDIX Functional Area and Executive Rank in Hierarchy Variables

Functional area job responsibilities: One author of the study assigned each of the job titles and job responsibilities in the data base to a narrow func- tional area classification. Many of the executive biographies contained short descriptions of job responsibilities, which aided in this coding. A second author checked these functional area as- signments for accuracy and differences were re- solved through discussion. Some executives had more than one title or job responsibility. Each title and job responsibility was assigned a func- tional area, and some executives had responsibil- ity for more than one functional area within the firm. In all, we assigned 572 different titles plus thousands of job descriptions in the database to 18 functional area classifications. In making these assignments, we relied on our own knowledge gained from prior research on top executives (Har- ris & Helfat 1997, 1998; Bailey & Helfat 2003, 2005). Additionally, we consulted several experts in the various functional areas and in the study of top management teams to ensure the correct as- signments of titles and job responsibilities to func- tional areas. The functional area classifications that we used are: operations; marketing; sales; information technology; research and develop- ment (including new product development); op- erations support (including engineering and qual- ity control); legal (including regulatory and government affairs); secretary; finance; account- ing; miscellaneous staff (including corporate and shared services); administration; real estate; sup- ply chain; customer service; public relations; hu- man resources; and strategy.

Rank within the Executive Hierarchy per Firm: After the raw data from the biographies were entered into the database, we wrote a computer algorithm to assign ranks within the top exec- utive hierarchy for each company. We first

ranked all of the 572 distinct titles in the data base as though every company had every title. A title of CEO or Chairman was assigned a hier- archical Level of 1 for the highest ranking of- ficers. A title of President or COO (Chief Op- erating Officer) was assigned a Level 2. President and COO in particular are considered second-in-command titles (Hambrick & Can- nella 2004). We consulted with preeminent scholars of top management teams in assigning these and other titles to ranks within the exec- utive hierarchy. A full description of the rank assigned to each of the titles in the data base is available from the authors on request.

Next we assigned a preliminary hierarchical level to each individual in the database, based on the full ranking of titles. If an individual had more than one title, the person was assigned the level for their highest ranking title. We then sorted the preliminary hierarchical levels by company. For the individuals that had prelimi- nary Levels of 1 or 2, in the final within-com- pany ranking the highest ranking individual(s) in the company received a Level 1, even if they did not hold the title of CEO or Chairman, since not all companies used these titles for their top ranked executives. (All companies in our sample had at least one executive with a title of CEO, Chair, or President, however.) Some firms did not have a second-in-command executive. Notably, most of the executives at Level 1 held multiple titles, often including a second-in-command title of President or COO. Some firms leave the second-in-command posi- tion unfilled and give the title to the top rank- ing executive, particularly during the early years of a CEO’s tenure. Later in the top executive’s tenure, these companies may transfer the sec- ond-in-command title to a potential successor to the CEO, who then holds the number 2 position in the company. Because firms some- times purposely leave the second-in-command position unfilled, in the final within-company rankings, we retained the Level 2, even though some companies had no executives in this po- sition. Then, beginning with the preliminary Levels of 3 and greater (meaning lower levels in the executive hierarchy), we closed up any gaps

2006 61Helfat, Harris, and Wolfson

in the levels within companies, and re-ranked titles within companies from Level 3 up to and including the lowest within-company hierarchy level reported, which was a Level 7.

Explanatory Variables in the Regressions

The firm-level explanatory variables are: aver- age age of all male executives, average years of company tenure of male executives, average years of tenure in their current positions of male exec- utives, the ratio of the number of line positions to the total number of positions held by male exec- utives, the total number of male executives, the lowest level in the executive hierarchy held by men, and dummy variables reflecting industry of operation (based on a total of 61 Fortune industry classifications). The rationale for including each of the explanatory variables is as follows.

Earlier we identified some key differences be- tween men and women at the executive level. On average, the women were younger than the men, had fewer years of tenure within the company and within their current positions, held lower posi- tions in the executive hierarchy, and were more likely to hold staff than line positions. We inves- tigated whether differences between companies on these dimensions were correlated with differ- ences in the representation of women per com- pany. For example, were women at the executive level younger than their male counterparts be- cause companies with greater representation of women also had younger executives overall? Since we also found that companies with more women reported a larger number of executives, and since the representation of women differed substantially by industry, we included these factors in our anal- ysis as well.

Because our dependent variables measure the representation of women executives at each firm, we avoided using explanatory variables that would be influenced by the characteristics of the women in the executive ranks of each company. This is particularly important because some of the vari- ables differed substantially by gender. For exam- ple, as reported earlier, the ratio of line to total positions for men is twice that for women. Includ- ing women in the ratio of line to total positions per firm causes the ratio to drop substantially

relative to a variable that only includes the men. If we were to include the women in this explan- atory variable, we would be regressing the repre- sentation of women on a variable that reflects the representation of women– creating endogeneity of the explanatory variables. To avoid creating this sort of problem, we expressed the explanatory variables that reflected the characteristics of ex- ecutives only in terms of the characteristics of male executives.13 Since the executive teams in most companies consist largely of men, these vari- ables reflect the most common executive charac- teristics per firm.

In addition to the foregoing variables, we in- cluded revenues as a control for size of company. Larger companies may have a larger pool of em- ployees to draw from, which in turn could influ- ence the representation of women in top execu- tive ranks. We included profitability as well, measured as return on sales (calculated as profits divided by revenues, using data reported in the Fortune 2000 survey). Prior research has found that profitability is positively correlated with the number of executives who are women (e.g., Adler 1999), although the reasons for this prior finding and the direction of causation are unclear. Finally, to account for the possibility that the CEO may play a key role in determining the representation of women in his or her top executive team, we included a variable associated with individual CEOs. We used the age of the CEO on the theory that younger CEOs might be more used to work- ing with women and perhaps more likely to pro- mote women to the top executive ranks.14 Table A1 provides descriptive statistics for all of the variables in the regressions. Since we have missing observations for some variables, the means re- ported in Table A1 for the number and percentage of women per firm differ slightly from those re- ported earlier for the full sample.

13 A sensitivity analysis that included both female and male executives in these variables revealed noticeable changes in some of the regression coefficients, in a direction consistent with endogeneity. This further con- firmed our decision to exclude female executives from the explanatory variables in the regressions.

14 If a firm had more than one executive in Level 1, we used the average age of the Level 1 executives in the firm.

62 NovemberAcademy of Management Perspectives

Acknowledgements We received very helpful comments on an earlier draft of this paper from Peter Cappelli and an anonymous reviewer. We are also grateful to our many research assistants: Ashley Nowygrod, Igor Fuks, Marcella Gift, Patrick Jou, Jesse Kiefer, Raina Kim, Stanley Kim, Jennifer Lee, Seungyeon Lee, Veronica Mendez, Pierre Nguyen, Mark Permann, and Marcia Sajewicz.

References Adler, R. D. 1999. Women in the Executive Suite Correlate

to High Profits. Pepperdine, CA: Evergreen Project on Equal Pay.

Bailey, E. E. & Helfat, C. E. 2003. External management succession, human capital, and firm performance: An integrative analysis. Managerial and Decision Economics, 24: 347–369.

Bailey, E. E. & Helfat, C. E. 2005. External succession and disruptive change: Heirs-apparent, forced turnover and firm performance. Strategic Organization, 3(1): 47– 83.

Bertrand. M. & Hallock, K. F. 2001. The gender gap in top corporate jobs. Industrial and Labor Relations Review, 55(1): 3–12.

Bureau of Labor Statistics. 2003. Current Population Sur- vey. Washington, DC: U.S. Government Printing Of- fice.

Cappelli, P. & Hamori, M. 2004. The path to the top: Changes in the attributes and careers of corporate exec- utives, 1980-2001. Working Paper 10507: National Bu- reau of Economic Research.

Catalyst. 1998. Catalyst census of women corporate officers and top earners of the Fortune 500. New York: Catalyst.

Catalyst. 2000. Catalyst census of women corporate officers and top earners of the Fortune 500. New York: Catalyst.

Catalyst. 2002. Catalyst census of women corporate officers and top earners of the Fortune 500. New York: Catalyst.

Cohen, L. & Elvira, M. 1997. The effects of organizational sex composition on the turnover of men and women: Is leaving just the same? Presented at the annual meeting of the Academy of Management, Boston.

Daily, C. M., Certo, S. T., & Dalton, D. D. 1999. A decade of corporate women: Some progress in the boardroom, none in the executive suite. Strategic Management Jour- nal, 20(1): 93–99.

Dutton, J. E. & Duncan, R. B. 1987. The creation of momentum for change through the process of strategic issue diagnosis. Strategic Management Journal, 8: 270 – 295.

Farrell, K. A. & Hersch, P. A. 2005. Additions to corporate boards: The effect of gender. Journal of Corporate Fi- nance, 11: 85–106.

Finkelstein S. & Hambrick, D. C. 1996. Strategic leadership: Top executives and their effects on organi- zations. Minneapolis/St. Paul: West Publishing Com- pany.

Goodman, J. S., Fields, D. I, & Blum, T. C. 2003. Cracks in the glass ceiling: In what kinds of organizations do women make it to the top? Group & Organization Man- agement, 28: 475–501.

Hambrick, D. C. & Cannella, A. A. 2004. CEOs who have COOs: Contingency analysis of an unexplored structural form. Strategic Management Journal, 25(10): 959 –979.

Table A1 Descriptive Statistics for Regression Variables

Variable Name Mean Standard Deviation Minimum Maximum

Number of Female Executives Per Firm 0.872 1.093 0.0 7.0 Percentage of Female Executives Per Firm 7.927 9.570 0.0 60.0 Average Age of Male Executives Per Firm 50.972 3.702 37.25 65.333 CEO Age 55.648 7.359 34.0 85.0 Average Years of Company

Tenure of Male Executives 10.871 5.959 1.0 41.0

Average Years in Current Position of Male Executives

3.692 2.746 0.0 25.0

Number of Male Executives Per Firm

9.696 4.790 2.0 50.0

Lowest Level in Executive Hierarchy of Male Executives

4.394 0.860 2.0 7.0

Ratio of Line to Total Positions of Male Executives

0.492 0.177 0.0 1.0

Firm Profitability (Return on Sales) 4.870 10.449 �93.230 131.020 Firm Revenues (logarithm) 8.252 0.965 7.058 12.150

Number of observations � 900

2006 63Helfat, Harris, and Wolfson

Hambrick, D. C. & Mason, P. A. 1984. Upper Echelons: The Organization as a Reflection of Its Top Managers. Academy of Management Review, 9: 193–206.

Harris, D. & Helfat C. E. 1997. Specificity of CEO human capital and compensation. Strategic Management Journal, 18: 895–920.

Harris, D. & Helfat, C. E. 1998. CEO duality, succession, capabilities and agency theory: Commentary and re- search agenda. Strategic Management Journal, 19: 901– 904.

Hayward, M. L. A., Rindova, V., & Pollack, T. G. 2004. Believing one’s own press: The causes and consequences of CEO celebrity. Strategic Management Journal, 25(7): 637– 653.

Hillman, A. J., Shropshire, C., & Cannella, A. A. 2005. Organizational predictors of women in top management teams and corporate boardrooms. Working Paper, Ari- zona State University.

Himelstein, L. & Forest, S. A. Breaking through. Business Week. 3 February 1997, 64.

Hitt, M. A., Bierman, L., Shimizu, K., & Kochhar, R. 2001. Direct and moderating effects of human capital on strat- egy and performance in professional service firms: A resource-based perspective. Academy of Management Journal, 44(1): 13–28.

Hymowitz, C. & Schelhardt, T. D. The glass ceiling. Wall Street Journal Special Report on Corporate Women. 24 March 1986.

Kanner, B. 2004. Pocketbook power. New York: McGraw- Hill.

Kirchmeyer, C. 1998. Determinants of managerial career success: Evidence and explanation of male/female differ- ences. Journal of Management, 24(6): 673– 692.

Lyness, K. S. & Thompson, D. E. 2000. Climbing the corporate ladder: Do female and male executives follow the same route? Journal of Applied Psychology, 85(1): 86 –101.

Metz, I. & Tharenou, P. 2001. Women’s career advancement: The relative contribution of human and social capital. Group & Organization Management, 26(3): 312–342.

Miller, C. C., Burke L. M., & Glick, W. H. 1998. Cognitive diversity among upper-echelon executives: Implications for strategic decisions. Strategic Management Journal, 19(1): 39 –58.

Morrison, A. & Von Glinow, M. A. 1990. Women and minorities in management. American Psychologist, 45: 200 –208.

Powell, G. N. 1999. Reflections on the glass ceiling: Recent trends and future prospects. In G. N. Powell (Ed.), Hand- book of Gender & Work (pp. 325–345). Thousand Oaks, CA: Sage Publications.

Powell, G. N. & Butterfield, D. 1994. Investigating the “glass ceiling” phenomenon: An empirical study of ac- tual promotions to top management. Academy of Man- agement Journal, 37: 68 – 86.

Richard, O. C., Barnett, T., Dwyer, S., & Chadwick, K. 2004. Cultural diversity in management, firm perfor- mance, and the moderating role of entrepreneurial ori- entation dimensions. Academy of Management Journal, 47(2): 255–266.

Society for Human Resource Management, 2004. The glass ceiling: Domestic and international perspectives. HR Magazine, 2004 Research Quarterly, 49:2–10.

Valcour, P. M. & Tolbert, P. S. 2003. Gender, family and career in the era of boundarylessness: Determinants and effects of intra- and inter-organizational mobility. Journal of Human Resource Management, 14(5): 768 –787.

Van Emmerik, I. J. H., Euwema, M. C., Geschiere, M., & Schouten, M. F. A. G. 2006. Networking your way through the organization: Gender differences in the re- lationship between network participation and career sat- isfaction. Women in Management Review, 21(1): 54 – 66.

Watson, W. E., Kumar, K., & Michaelsen, I. K. 1993. Cultural diversity’s impact on interaction process and performance: Comparing heterogeneous and diverse task groups. Academy of Management Journal, 36(3): 590 – 602.

Zelechowski, D. D. & Bilimoria, D. 2003. The experience of women corporate inside directors on the boards of For- tune 1,000 firms. Women in Management Review, 18(7): 376 –381.

64 NovemberAcademy of Management Perspectives