|1Canadian Interlocking Directorates: Gender, Industry and City Differences
Milford B. Green
Department of Geography
Social Science Centre
University of Western Ontario
London, Ontario, CANADA N6A 5C2
Department of Geography
100 College Drive, Box 5002
North Bay, Ontario, Canada P1B 8L7
This paper examines the corporate interlocking directorates for Canada for the year 2000 by gender. Basic descriptive statistics are provided. Based on data from the Financial Post, differences in the number of interlocks are detected by both a firm’s industry and headquarters location. Females are under-represented in all cases. Regression equations identify the more unusual portions of the interlock network. A network analysis of the interlocks finds the economic core of Canada and provides measures of importance for major urban areas to that core. An examination of the ages of the directors show that females are on average five years younger than males.
Key words: interlocking directorates, gender, economic core, Canada
Similar to a number of fields, concern with women’s roles in the corporate boardroom has increased substantially in the recent past. In particular, research organizations (Catalyst, 1998, 1999; Spencer Stuart, 1998, 1999a, 2000, 2001) have emphasized annual surveys to examine female participation and recruitment in the United States and Canada. The results reveal an increasing involvement of females sitting on corporate boards. However, their presence still remains minimal when compared to their male counterparts.
While our understanding of female participation in the Canadian corporate boardroom has increased, the authors of this paper argue that additional consideration of these women is necessary. For example, women composed a hefty 50% of business and financial professionals in 2001. Nevertheless, only 35% of managers, 23% of senior managers (Statistics Canada, 2001b, pg. 8), and a nominal 7.1% (this study) of directors were women. Further research is necessary to understand this conflicting connection between corporate governance rank and female participation.
One avenue to explore this relationship is via interlocking directorates. This occurs when a director of one company sits on the board of directors of another company. Termed an interlocking directorate, this method is a way of determining the upper echelon of Canadian women directors. These women must not only meet the demands of being a director of a single company, but are also sought after by additional ones. An examination of interlocking directorates in Canada necessitates a detailed study of whether women based interlocks are different from previous studies on female participation in corporate governance and if so how. The purpose of this paper is to provide a comparative look at the interlocking directorates by gender with a spatial perspective.
Women Directors Studies
Since female interlocking directorates can only occur if women are selected for board membership, it is instructive to summarize why they may succeed or fail to be chosen. Hughes (2000) and Marzolini (2002, pg 19) among others provide a comprehensive review of female directorship literature that need not be repeated here. Their studies succinctly list facilitating and inhibiting factors for female corporate board membership. These include:
There is a growing pool of qualified women who possess the necessary experience and expertise (Burke and Kurucz, 1998; Business Week, 2001).
The emergence of more objective and professional corporate board practices generated by calls for better corporate governance (Spencer Stuart, 2001; Brown and Brown, 1998, Leighton and Thain, 1997, Millstein et al., 1998). More objective criteria would favor the inclusion of women.
Increased pressure for greater board diversity, by consumers and interest groups.
A smaller pool of qualified men as candidates as corporations undertake more rigorous evaluation methods and expect more from board members.
Some recruiters and corporations believe there is still a shortage of qualified women. This may be due to women who are not interested because of work or family commitments, or are unwilling to serve as a token appointment (Quacquarelli, 2002; Banks and North, 1996, National Post, 2000)
Embedded gender biases in the traditional profile and competencies required for a director.
There is lack of mentoring and sponsorship for female candidates.
Men are uncomfortable in dealing with women at such a professional level. This leads to a lack of acceptance into personal networks.
Lack of previous women directors that have opened the way.
Many firms still rely on interpersonal networks and contacts for recruiting, and are unwilling to appoint what they perceive to be untested women (Leighton, 1999; Catalyst, 1995). This is mirrored in a lack of commitment by corporations to recruitment.
Facilitating and inhibiting factors can be anticipated to vary by both geographic location and industrial class. Consequently, it has been verified that female participation rates vary industrially (Chemical and Engineering News, 2001), internationally (CWDI, 2003), and regionally (CWDI, 2002; Wiles, 2003; BDN, 2002). Such variations imply that both industry and geographic location may play a role in the participation by women in corporate interlocks.
Models of Interlock Formation
As the business environment has changed over the last 100 years, so too have different paradigms as chief explanations of interlock formation. In general, the rationale has moved from interlocks as a controlling mechanism to interlocks as a transfer mechanism. Disagreement exists, however, over the nature and importance of those functions. The models used by researchers can be roughly divided into five categories: (l) management control, (2) class hegemony, (3) resource dependency, (4) bank or financial control and (5) knowledge networks.
Management control theory (Mace, 1971), the dominant paradigm until the 1970’s, argues that interlocking directorates are relatively unimportant to the corporation. It is assumed that managers appoint directors who offer few disruptions to the power structure of the company. Management control theorists look to the Enron and Bre-X scandals as examples, where directors did not practice due diligence.
Since chief executive officers (CEOs) most often nominate the directors, and are seen as the preferred candidates (Daily and Dalton, 1999), women could expect broader admittance to the boardroom only when senior women executives became more common. This is a corollary of the ‘old boys’ network barrier, which women have yet to overcome on a large scale.
This view implies a random and chaotic network structure and argues the network of interlocking directorates occurs by chance. But Mintz and Schwartz (1985) contend that a highly integrated system of interlocking directorates contradicts the fundamental reasoning behind the management control. Furthermore, this theory is undermined by the number of studies that have found significant relationships between the interlock structure and firm financial characteristics and performance (Green, 1983; Pfeffer, 1972; Burt, Christman, and Kilburn, 1980; Allen, 1974, 1978).
Class hegemony theory (Sonquist and Koenig, 1976) contends that interlocks emphasize upper-class participation in business. The model argues that an upper class elite exists that has cohesiveness, self-consciousness, and a consensus on social issues. The unity of this class is promoted through common life experiences, of which membership to a corporate board is an example. Since the size of this elite class is small, a number of members must fill multiple positions, hence the interlocking directorate (Domhoff, 1967; Sonquist and Koenig, 1976; Stanworth and Giddens, 1975; Blumberg, 1975; Useem, 1980).
Similar to the management control theory, interlocks are viewed as an end in themselves (a controlling mechanism) rather than a means to an end. If elite individuals are always appointed to the board of directors, they will continually control corporate power. A dearth of female directors would be expected under such a model because women have traditionally found it difficult to gain acceptance into these inner circles. Historically, such elites tend to be conservative in outlook. Women would be seen as homemakers rather than executives. Such a belief would be difficult to overcome.
Today, research generally acknowledges that resource dependency (Pfeffer and Salancik, 1978) is the best explanation for interlocking directorates. This paradigm contends that interlocks are established to reduce uncertainty. Through an interlock, a firm creates a relationship to ensure access to a resource not produced internally. Hence, interlocks are considered a transfer mechanism.
It could be argued that the fourth paradigm, financial control theory, although earlier in time, (Mintz and Schwartz, 1985), is simply a branch of the resource dependency model. Resource dependency fails to highlight the fact that financial institutions play a central role in the interlocking network. Resource dependency asserts that interlocks reduce uncertainty while financial control theorists argue that access to liquid currency raises the greatest concern. In a rapidly changing financial environment, a firm’s capital needs cannot always be met by recourse to bond or stock markets. Firms thus need ready access to the financial resources of banks, insurance companies, and other financial corporations. Access to these resources does not come without a cost. The bank can use a firm’s financial needs to forge a long-term borrowing and financial services arrangement and may come to influence or even control corporations through that dependency. The interlocking directorates derive from the practice of the bank requiring that one of its representatives serve on the dependent firm’s board of directors. Results (Kotz, 1979; Mariolis, 1975, 1976; Dooley, 1969; Mintz and Schwartz, 1981 ; Levine, 1972, 1976) suggest that financial institutions play a pivotal role in the interlock network.
While these paradigms warrant recognition, the latest branch of research (Useem, 1984; Lorsch and MacIver, 1989; Haunschild and Beckman, 1998; Carpenter and Westphal, 2001) argues that interlocks are an avenue for leaders to exchange knowledge and strategy between firms. This concept is a second derivative of resource dependency theory (Pfeffer and Salancik, 1978); viewing interlocks as a mechanism for reducing uncertainty by increasing the knowledge of top management decision-makers. Applied to the context of interlocking directorates, resource dependency views board directors as “important boundary spanners that link with the environment and extract resources for successful operations, and predicts that in an uncertain environment, firms will use interlocking directorates to achieve better coordination with other organizations and reduce uncertainty” (Au, Peng, and Wang, 2000, p. 31). The notion that board members and interlocking directorates provide useful information has been empirically verified by a number of researchers (Pennings 1980; Provan, 1980; Lorsch and MacIver, 1989; Boyd 1990; Useem, 1984).
Johnson, Daily, and Ellstrand (1996, pg. 425) summarize a number of studies which prove knowledge of corporate strategy transferred via interlocking directorates can lead to a firm making better business decisions than they would have otherwise. Davis (1991), for example, found that poison pill anti-takeover provisions diffused through the interlocking directorate. Mizruchi (1989) demonstrated that proximity in the interlocking directorate accounted for similarities in political contributions by major corporations to political action committees. The diffusion of more fundamental managerial innovations and norms, such as adoption of multidivisional corporate structures (Palmer, Jennings, and Zhou, 1993) or general acquisition strategies (Haunschild, 1993), also provides evidence that directors not only advise, but may also initiate important strategic changes in corporate strategy.
Although resource dependency is now the dominant paradigm it is not without its flaws. O'Hagan and Green (2002a) argue that a shortcoming of Pfeffer and Salancik’s resource dependency model is the heavy emphasis placed on the individual firm. Not only can knowledge be acquired from other firms, but by interacting with a variety of organizations over space, a firm exists in much broader environment.
All of the proposed explanations of interlocking have merit. It seems unnecessary to choose any one explanation as the sole one. In fact, it is quite likely that all of the explanations are true for specific firms and circumstances. It is more fruitful to adopt a more eclectic view. Surely the main point is that interlocking directorates are not random undirected phenomena, but rather they serve a variety of needs of firms and their corporate boards. To fully understand the patterns of interlocking directorates, industry-to-industry and city-to-city knowledge linkages should be examined.
Geographical Research on Interlocking Directorates
The paradigms presented in the previous section attempt to determine why interlocks occur. But geographers argue this answers only one component of interlocking research. By examining the problem spatially, geographers offer a different direction by attempting to determine where and why they occur. Green (1981, 1983) and Green and Semple (1981) were the first to examine the geography of interlocking directorates. They hypothesized that the US manufacturing belt was dominated by intraregional interlocking. This was advantageous when firms of the manufacturing belt were on the cutting edge of technological development. But over time other regions in the United States became increasingly important. By not interlocking with firms in these regions, the manufacturing belt was denied critical information.
Rice and Semple (1993) examined interlocks spatially in support of a different purpose. Using interlocks to examine the distribution of corporate control in Canada, they found an increasing concentration of power in Toronto. On the other hand, Montreal decreased to become a regional center in Eastern Canada, and Calgary emerged to become a regional center in Western Canada. Furthermore, it was found that regional centers were important in certain industries.
Some sociologists appreciate the spatial implications of interlocking directorates. Kono et al. (1998) suggest that determining why interlocks occur is more powerful when differentiating between local and non-local links. For example, when local and non-local interlocks were aggregated into the same sample, the independent variable “presence of exclusive upper-class social clubs in a corporation's headquarters city” did not influence corporate interlocking. However, an interesting result occurred when the sample was separated into local and non-local links. It was found that when corporations were headquartered in cities with exclusive upper class social clubs, they were more likely to maintain local interlocks. On the other hand, corporations not headquartered in cities with exclusive upper class social clubs were less likely to maintain local interlocks. Using a geographic approach, the authors were able to prove that these finding supported the class hegemony model.
In the most recent geographical studies, O'Hagan and Green (2002a; 2002b) explored the gap between previous resource dependency-knowledge transfer research and geographical findings. Three notable findings are worth mentioning here. First, a spatial component was added to the resource dependency paradigm. Second, components of a city that initiate and attract interlocking, and thus knowledge transfer, were identified. It was found that geography was more important in the American network than in the Canadian network. Third, it was found that the American knowledge network is increasingly dispersing while the Canadian network is converging into a few cities.
The purpose of this paper is to build upon past research by extending analyses to question the gender of interlocking directorates. Specifically, this paper addresses whether gender differences exist. It is hypothesized that industry class and geographical location of a firm creates these gender differences. Such characteristics may also explain age differences in interlock selection.
The electronic version of the Financial Post’s Directory of Directors, 2000 provides the data used in this paper. It is arguably the most comprehensive source on corporate directors for Canada. The Directory provides coverage for over 1800 Canadian firms and over 5,400 directors. Information is provided on names of the firms’ corporate boards on which the directors serve, the directors’ name, gender, date of birth, and job titles. The firms are classified into one of 31 industry classes and their headquarters locations are provided. Using this information as a base, analysis of gender related differences among Canadian corporate directors can be pursued.
Plan of Analysis
The data available from the Financial Post dictates an analysis that is organized along the three classificatory variables provided. These occur at two levels, the firm and the individual director. Analysis proceeds for both of these levels. For the firms, the industrial class and headquarters location is provided. For the individual director gender and age is available. The gender variable is addressed concurrently within the industry and geographic analyses. Age is analyzed in a small final section.
The most prominent Canadian firms in terms of number of firms with which interlocks have been forged is displayed in Table 1. All five of the largest banks are found within the top six firms, headed by Canadian Imperial Bank of Commerce. BCE Inc. (Bell Canada Enterprises), the communications giant, is the only nonbank in the top five. In fact, financial services is the only industrial class with substantial representation with eleven of the top 25 firms. Such a distribution is consistent with previous studies emphasizing the importance of financial institutions in interlock behavior. Admittedly, Canada with its highly concentrated banking sector is an extreme case.
Toronto is the most common headquarters city with sixteen of the top 25 firms as major interlockers. Montreal follows with six of the 25 and Calgary has four of the 25. If the list is expanded to include more firms the general pattern of dominance by Toronto, Montreal and Calgary is sustained for the top 45 firms. The next center to enter is Halifax, home of Aliant Incorporated, a telephone utility, which enters at number 46. In spite of its large population, Vancouver only enters at number 81.1 As the reader will discover later, this is consistent with these cities overall importance in the interlock networks.
The overall distribution of the number of interlocks held by an individual director is summarized in Table 2. Over 80% of directors serve on only one board and thus do not interlock. About nine percent serve on two boards thereby generating a single interlock. Very few directors (2.5%) serve on more than four boards. When taking gender into account, females are somewhat less likely to serve in an interlocking capacity. This true for all categories, except for four through six interlocks, where females have a slightly higher percentage. The likely explanation for this higher percentage is the relative scarcity of experienced women directors. They are highly sought after and with a small pool they will tend to serve on more boards on average (National Post, 2000).
It is also possible to examine interlocking behavior by industry class. Each firm in the interlocks data is divided into one of 31 industry classes defined by the Financial Post. Table 3 shows the number of firms in each class for the corporate core of Canada. As noted later, different industries exhibit differing propensities to interlock, creating effects on the spatial structure of the national network.
Toronto is the most important city for half of the companies in the sample and is overwhelmingly dominant in Financial Services, Consumer Products and Computer Services. Montreal is dominant in seven classes, with Crown Corporations and Entertainment being the largest. As expected, Calgary dominates Oil and Gas and Pipelines. The Gold and Silver, Metals and Minerals classes are most often found in Vancouver.
Table 4 shows the number of firms within each class in the firm sample as well as the average number of interlocks per firm in each industry. The telephone utility class has the highest average number of interlocks with 7.21 per firm. This is the product of having only 14 firms in that industry. One of the 14, BCE has connections with 50 other firms, and 2 of BCE’s major subsidiaries together have 57 connections (see Table 1). Firms such as BCE are the exception however, not the rule.
The industrial class with the second highest average number of interlocks is media. This class has 10 of its 38 firms with more than 20 interlock connections to other firms. The largest is Rogers Communications Inc. with 39 linkages followed by Hollinger Inc with 35 and Southam Inc with 34. The linkages exist between a wider variety of other industrial classes. Rogers for example is connected to firms in 13 other industries. Hollinger shows a similar pattern with 17 and Southam has 15. None of these firms is directly linked to each other.
The third highest average number of interlocks is for the pipelines category with six per firm. This is a situation similar to that of the telephone utility class. Three firms, with above average number of interlocks greatly influence the average. These firms; TransCanada PipeLines Limited with 29 linkages, Trans Mountain Pipe Line Company Ltd with 25 linkages and Westcoast Energy Inc with 21 linkages. None of these firms is directly linked to each other.
The financial services sector has 4.5 interlocks per firm. Financial control theory and information transfer theory would suggest an important role for this sector. Canadian Imperial Bank of Commerce, a member of this sector, has the largest number of interlock connections with other firms at 68. Bank of Nova Scotia and Toronto Dominion Bank are second and third. In fact, as revealed in Table 1, nine of the top 20 firms with other firm connections are in this sector. This provides support for both theories.
At the other end of the spectrum professional services exhibit only .15 connections and advertising only .50 compared to the overall mean of 2.98. In the advertising class only 2 of the 10 firms in the sample exhibited any interlocks at all. The same is true for professional services with only 2 of the 39 firms in the sample having interlock connections. This unimportance may exist because firms in these industries tend to have small boards compared to other firms in the sample (based on revenues or assets). It is widely known in interlock research that the number of interlocks generally declines with firm size (Burt et al, 1980). The reasons for this are many but include; less visibility so they are less prestigious; they have smaller capital needs so they have less need for financial connections; their smaller board sizes means fewer candidates; and smaller firms possess less knowledge when compared to their large counterparts (Haunschild and Beckman, 1998).
When comparing the average interlock rates by gender the patterns are very similar, although the female rate is less than a tenth of the male rate. In total, there are only 355 female interlocks compared to 5062 male ones. The similarity of the pattern of the rates is borne out by a simple correlation coefficient of 0.81 between male and female averages.
It is also possible to determine the propensity of interlocking by industry class. This propensity is defined as the number of reported interlocks in an industry class divided by the total number of directors in that industry. This can be interpreted as the likelihood of an interlock occurring. Table 5 reveals that Holding Companies are the most likely to generate an interlock with the average board having .64 interlocks. This number needs to be interpreted with caution since many of the interlocks are with firms controlled by the holding company and do not represent real arms lengths connections. The average board sizes are consistent with the reported averages of 13.7 directors per board for Canadian firms with revenues above $5 billion and 11.2 for firms with revenues $1-$5 billion (Spencer Stuart, 2001, pg 15).
The media industry class appears again, with a propensity of 0.46. The financial sector with a value of 0.34 is barely above the overall mean likelihood of 0.31. The financial sectors importance in the interlock network of Canada is a function of both the importance of the top banks (see Table 1), their large board size (Spencer Spencer, 2001, pg 16), as well as the number of firms in the sample (see Tables 3 and 4).
An examination of the differences by gender in propensity to interlock shows general agreement in pattern. The simple correlation coefficient between the male and female propensity rates is r=0.68, lower than that for gender averages (Table 5). To objectively determine where the greatest difference occurs between the male and female patterns, a simple regression was calculated. The female rate was used as the dependent variable and the male rate was the independent. The regression should not be viewed as testing a causal relationship, but merely a convenient way to generate standardized residuals for comparison purposes. A positive standardized residual indicates an industry where the female propensity is under-predicted, while a negative standardized residual indicates an industry where the equation over-predicted the female propensity.
The greatest difference between the two sets of propensities as measured by standardized residuals for a simple regression between the two sets of propensities is for Crown Corporations with a value of 2.15, followed by the Media sector with a value of 1.98. This not unexpected since governmentally controlled bodies are often progressive in their personnel policies.
By contrast the mining sectors (the gold and silver sector [-1.47] as well as the metals and minerals sector [-1.54]), consumer and industrial products [-1.60] have a much lower than expected propensity for women interlocking directorates. This certainly coincides with the popular conception of the mining sectors being male dominated
The predictive ability of the gender and industries of interlocks variables is tested via a multiple regression model. The approach is suggested by an appeal to regression based spatial interaction models where the logarithm of levels of interaction between places is influenced by the size or mass of those places. Here the places are represented by the industries, and the mass or size terms are represented by the number of firms in each of the industry classes. The concept behind the regression is the same as in a spatial interaction model. Larger numbers in an industry class (or a place as in the traditional spatial interaction model) should generate more interlocks ceteris paribus. Table 6 provides the estimated equation as well as all the standardized residuals with values at or above 2.0 and below -2.0.In table 6, Industry 1 and Industry 2 are nondirectional. In other words, the Conglomerate to Media linkages is the same as Media to Conglomerate linkages. Each industry to industry linkage is counted only once. The regression has gender included as well to determine if it is a significant factor in the number of cross industry interlock linkages.
The results are reasonably strong with a multiple correlation coefficient of r=.70 (r2=0.49). The number of firms in the industry at each end of the linkage as well as gender are statistically significant at p=.001. An examination of the residuals from the regression equation reveals that of the twenty-four largest residuals (Table 6), five are between firms within the same industry class (bolded Table 6). In all of these cases, the equation greatly under-predicts (cases with positive residuals) the number of linkages. For the Media, Gas/Electrical, Gold and Silver, Metals and Minerals, and Telephone Utilities sectors there is a need for interconnections far beyond what the number of firms in each class would indicate. That is to say, the number of interlocks reported is substantially larger than one would expect given the number of firms in each industrial class. Several of the under predicted linkages mirror resource complementarities. Gas/Electric Utilities with Pipelines, Media with Paper and Forest Products and Gold and Silver with Metals and Minerals are examples.
For some of the other large residuals cases such as the Advertising and Professional Services sectors (seven of the cases), the inhabitant firms tend to be small and relatively unattractive as candidates for linkage and would tend to generate few interlocks. The model thus over-estimates the number of interlocks. The residuals with Holding Companies can be understood as an administrative convenience for the firms and simply mirror ownership patterns.
The gender variable is an important variable in the regression equation, explaining about seventeen percent of the variance. Only three of the large residuals are associated with female director interlocks. In two of the three cases the residuals are positive indicating under-prediction, both being associated with holding companies and can be discounted. The third link between the Financial Service and Health Services classes is held by a faculty member in international business at the University of Toronto. She maintains a high corporate profile and substantial experience and was no doubt sought out because of it.