A longitudinal study (1995-2007) explaining
the developments of the Corporate Elite in the Netherlands
Master Thesis March 2009 University of Groningen
MSc International Business and Management First Supervisor: Dr. K. van Veen
Second Supervisor: J. Kratzer
Floris van der Lee – 1335405
The Evolution of the
ABSTRACT
This longitudinal study (1995–2007) analyzed networks of Dutch firms and directors and found three ways in thrying to explain the developments concerning the corporate elite in the Netherlands.
The findings were the following. First, by examining the changes in the firms’ network over time, broad developments became apparent, that laid the basis for a more in-depth analysis. It showed the big linkers disappeared, causing the firms’ network to diminish in cohesiveness and density. Second, the stability of the links between two firms were analyzed, and only 7 out of 503 linkages between firms had a reconstituted character. Moreover, a marginal part of the linkages (31 out of 503 edges) were maintained for 10 years of longer. Finally, in trying to identify whether a career was identifiable for the directors in this study, for those who had positions for ten or more consecutive years, a supervisory career after reaching the top of the corporate executive ladder became clear; none of the directors leave their board(s) an executive director.
Then, this study proposes that several developments, mainly the restrictions stipulated in the Tabaksblat Code (2004), but also the internationalization, may have recently changed the labor market for directors, and has given the opportunity for a new group of directors to become ‘bigger’ linkers, explained by the labor queue theory. By expressing the supervisory positions in so-called Tabaksblat-points, which gives a indicator for the ‘preferredness’ of the directors present on each firm’s board, this study found that while (the difference in) the average points per firm per index remained practically the same from 1995 – 2004, significant changes occurred afterwards. In 2007, apparently the ‘preferredness’ of AEX directors decreased substantially, while that of the smaller indices increased quite well relatively. Even though internationalization has an influence, this has not been relevant only since 2004, but already before that. These findings suggest that bigger linkers were morally obliged to give up some of their positions at firms – mainly the largest, which were taken by directors, predominantly representing the smaller listed firms.
TABLE OF CONTENTS
ABBREVIATIONS LIST 7
INTRODUCTION 9
Sub questions 10
SOCIAL NETWORK STUDIES 12
THEORETICAL FRAMEWORK 14
Corporate arguments for interlocking 15
Individual argument for interlocking 17
Stability of networks 19
Institutional and environmental factors influencing interlocks 21
THE CORPORATE ELITE IN THE NETHERLANDS 23
The ‘rise’ of the Old Boys Network 23
The Old Boys Network falling apart 24
The rise of the new elite 24
Importance of informal ties 25
Governmental influence 25
METHOD 27
Data collection 27
Dataset 27
Sample size over the period 28
Developments of the Dutch IDN and corporate elite over time 29
Stability of edges 30
Individual career 31
RESULTS 33
Developments of the Dutch IDN 33
Stability of linkages between firms 36
Developments of the individual’s network and career 40
Individual characteristics 40
Careerpath characteristics 40
Explaining the developments of the networks 44
CONCLUSION AND DISCUSSION 49
Limitations 51
GLOSSARY 53
ENDNOTES 55
ABBREVIATIONS LIST
AEX Amsterdam Exchange Index AMX Amsterdam Midkap Index AR Annual report
AScX Amsterdam Smallcap Index CG Corporate Governance
CME Coordinated Market Economy IDN Interlocking directorates network LME Liberal Market Economy
OBN Old Boys Network
OECD Organization for Economic Cooperation and Development
For a list of the abbreviations of the firms, used in the network visualizations, see table 15 (appendix).
INTRODUCTION
When an individual has two or more simultaneous board positions in different firms, that individual is an interlocking director and those firms are interlocked. Subsequently, not only do those interlocked firms create a network around the director, but when several interlocking directors hold positions simultaneously on the board of one firm, a network of interlocked firms arises, also known as an interlocking directorates network (IDN). Such a network, especially when it consists of the larger firms with billions of Euros in revenue and thousands of employees, is capable of exerting substantial influence on society, since those firms hold key positions in the economy.
Concerning IDNs, Mizruchi (1996: 277) strikingly argues that “interlocks occur between organizations, but (…) are created by individuals”; these specific individuals carry the expertise and knowledge that adds value to firms. In terms of the firm’s network, these individual’s play a crucial role as an information intermediary, since the director controls the exchange of relevant information between firms. Several interlocking directors regularly meet each other on board meetings, but also in informal circumstances, and form a clique that is able to influence the information streams between these firms: the so-called corporate elite.
Taking this into account, it is surprising to see that within empirical research concerning the IDNs, which have been quite abundant, the focus lies mainly on interlocking from a corporate perspective in stead of an individual (centered) perspective. From the corporate perspective, research dominantly focuses on the density and centrality of these networks (e.g. Davis and Mizruchi, 1999; Mariolis and Jones, 1982), or on transnational networks (e.g. Carroll and Fennema, 2002; Kentor and Jang, 2004; Van Veen and Kratzer, 2008). Recently, a new stream emerged, linking corporate governance (practices) to corporate networks (e.g. Bezemer, Maassen, Van den Bosch, and Volberda, 2008). These studies prove to be insufficient when it comes to providing more information on the individual networks that corporate networks inevitably underlie.
Empirical studies from an individual perspective are few (e.g. Heemskerk, 2007). Moreover, during recent studies, authors have recognized that the allotment of board positions among individuals, and therefore networks, can be based on informal connections, and are influenced by for example status and typical characteristics of a member (e.g. Bezemer et al., 2008; Galaskiewicz, Wasserman, Rauschenbach, Bielefeld, and Mullaney, 1985). From this perspective, Westphal and Stern (2006b, 2007) argue that ingratiatory behavior can lead to obtaining board positions. Since these positions generally concern well paid and longer term functions, it implies that by behaving in this desired way, a individual’s career in supervisory board(s) is possible and its career path identifiable over time, theorized by Zajac (1988) in his personal advancement model.
never properly tested. Since social networks are identified in the Netherlands (see Helmers, Mokken, Plijter, and Stokman, 1975) and have been studied extensively for more than three decades, it is chosen as the focus of this research. Moreover, the corporate elite in the Netherlands, known as the Old Boys Network (OBN), has been discussed in great detail. For example, Heemskerk (2007) describes the decline of the Dutch business elite over 1976, 1996, and 2001. Although longitudinal, this recent study, as many in this stream, lacks an up-to-date dataset. By looking at the board members of Dutch listed firms, which are predominantly multinational corporations, over the period 1995-2007, not only a recent, compared with other studies (e.g. Heemskerk, 2007; Stokman and Wasseur, 1985), but also a comprehensive dataset of thirteen consecutive years is available to thoroughly analyze networks on firm and individual level in the Netherlands.
Therefore, this research makes the following contributions. First, using a thirteen year period will give thorough understanding of the developments of the Dutch IDN and its corporate elite. Second, an overview is given to which extent linkages between firms within the IDN are stable over time. Finally, the individual network over time is examined, which will give a better image on the potential existence of a director’s career. In this way, the developments are examined from both an individual and corporate perspective, and so a unique light is shed on IDNs. Following the aforementioned, the following main question is formulated: how can developments of the network of Dutch firms and directors over time be explained by the individual and corporate argument?
Sub questions
First, it is relevant to analyze in what way the Dutch IDN has developed; to comprehend the outcomes on a corporate level will help to thoroughly interpret the individual-level results. By isolating the interlocking directors and observing the year on year changes of (the density of) this network (i.e. the network of individuals representing the same firm simultaneously) over 1995 – 2007 will give an overall frame of reference for the more detailed analyses described below. Thus, sub question (1) will be: how has the IDN (and corporate elite) developed in the Netherlands since 1995?
the argument behind the linkage. Therefore, the sub question (2) will be: to what extent are linkages between firms stable over time?
Third, it is relevant to know whether the individual member’s network over time is evolutionary of nature, i.e. a career is identifiable. Previous empirical research predominantly compares changes in the network as a whole on a corporate level (e.g. Davis and Mizruchi, 1999). But when the individual network, meaning the firms that are interlocked by one specific director, expands over the years, this may be a quantitative indication of the existence of a corporate career. Also is it important to provide a better understanding of the individual network because, as said before, it serves as the input for the corporate network. But in the end, this would give profound support for the individual argument for interlocking, since the director is personally driven to expand its ‘firm-portfolio’. Therefore the following sub question (3) is formulated: to what extent is a career visible based on the individual’s network?
Actor
Dyad / Edge
Relation
Note: In this figure, de actors are the firms, and the relation (line) is represented by an interlocking
Clique
Every actor (firm) has a direct link with every other actor in this (sub)group, through one or more SOCIAL NETWORK STUDIES
Social network studies involve the research on the social relationships in a group. It tends to distinguish itself because it does not see individuals as discrete units of analysis; the focus lies on how the structure of ties affects individuals in the group and their relationships.
Network analysis emerged as a specific technique from traditional social studies, like anthropology and sociology. Related concepts were used for more than a century already, but Barnes started to use the term systematically in 1954 (Freeman, 2006). Since then, many scholars, such as Borgatti, Everett, Faust, Freeman, Granovetter, and Wasserman, have expanded the use of network analysis and developed specific theories, methods, and software. This allowed the analysis of complex social groups in an organized and quantitative way.
Network analysis has a specific jargon that not only helps in understanding the basic characteristics of the network, but also the underlying social implications. They will be discussed briefly, which descriptions draw heavily on Wasserman and Faust (1994) – WF – and the appendix of Emirbayer and Goodwin (1994: 1447-1449) – EG.
An actor is a discrete individual, corporate, or collective social unit (WF: 17), which can be a firm in a network visualization as presented in figure 3 (page 37). This research focuses on a collection of actors that are of the same type, which is called an one-mode network (ibid). The simplest level of a linkage (or relationship) is a tie between two actors, called a dyad. In this research such a dyad is referred to as an edge, specifically meaning the linkage between two firms. A subset of three actors and the (possible) tie(s) among them is called a triad, and a subgroup is defined as any subset of actors and their ties (ibid: 19). When actors are directly linked to all the others in a (sub)group, this is labeled a clique (EG). Following, the collection of ties between actors in a group is called a relation. Then, a social network consists of a finite set or sets of actors and the relation(s) defined on them (WF: 20).
FIGURE 1.
The network itself can have several structures, such as a star, pyramid, circle, or reciprocal clique (Nollert, 2005), representing the different ways individuals or firms can be linked. Each network can have a (combination of) different structure(s), and will therefore vary concerning density, i.e. the ratio of the number of links to the maximum possible that could arise, and centrality, which represents the number of links a given actor has.
In the remainder of this research, the interlocked firms (as in figure 3) are the actors of the network, with the interlocking directors providing the relationships (linkages) between them, by creating edges.
THEORETICAL FRAMEWORK
Concerning firm interdependency, Pennings (1980) identified three types, knowingly horizontal – between competing firms, vertical – between firms that either supplies to or buys from another firm, and symbiotic – between complementary firms. Then, several authors have proposed theories on motives why firms should interlock those interdependencies, labeled the interorganizational perspective (Allen, 1974; Dooley, 1969; Pennings, 1980; Pfeffer, 1972; Pfeffer and Salancik, 1978), and later developed by authors such as Fennema and Schijf (1979), Koenig, Gogel, and Sonquist (1979), and Useem (1980).
Apart from the theories, studies that examine why two specific firms should interlock one another are not extensive, especially not recently. Burt, Christman, and Kilburn (1980) found that when market constraints are present, firm try to coopt this by installing interlocking directorates, which occur through ownership, direct interlocking, and/or indirect financial interlocking. Ornstein (1984) refers to Berkowitz, Kotowitz, and Waverman (1976) who found evidence for a relationship between interfirm ownership and interlocks. Other studies found no consistent relationship, although Kono et al. (1998) do suggest the importance of physical closeness of firms.
Nevertheless, authors generally agree that reasons for interlocking can be categorized in corporate and individual arguments1 (Heracleous and Murray, 2001; Mizruchi, 1996; Nollert, 2005; Stokman, 1986). Corporate arguments, explained by Koenig et al. (1979) as the reciprocity model, are collusion, cooptation and monitoring – the means of dealing with uncertainty, and the aim for legitimacy2 (Mizruchi, 1996; Nollert, 2005). Then, individual arguments are career advancement and social cohesion3 (Mizruchi, 1996; Nollert, 2005). Although not the focus of this research, other authors also add the concentration of the industry as an antecedent (Palmer, Friedland, and Singh, 1986). This idea stipulates that for example the higher the concentration of an industry, the less competition a firm faces within that industry and the better firms in that industry can bargain with buying or supplying firms from another industry. Interlocks might therefore occur between firms of the same industry to limit competition or between firms of a different industry to limit any constraint that might occur (ibid).
Institutional and environmental factors/requirements Corporate arguments 1. Collusion 2. Cooptation and Monitoring 3. Legitimacy
Note: the dashed circular lines indicate the dynamic character of the network over time. Moreover, industry
concentration is displayed in grey, since it is not part of this research.
Source: Heracleous and Murray (2001), Mizruchi (1996), Nollert (2005), Palmer, Friedland, and Singh (1986).
Reasons to interlock Industry concentration Board positions of individuals (per year) Firm Network Individual arguments 1. Career advancement 2. Social Cohesion Stability Stability Individual Network
Then, no matter what the argument behind the interlock, it is subjected to certain macro-level factors that will influence the character of the interlock. For example, recently many codes for good corporate governance (CG) were drafted, that contained guidelines on the maximum number of positions to limit the influence some individuals could exert.
Studies that map interlocks and their networks are abundant. Characteristics concerning the density of the network and direction of linkages, the visualization, and the (dominant) presence of firms and individuals, of which some longitudinal, have been described for the Netherlands (Heemskerk, 2007; Heemskerk, Mokken, and Fennema, 2007; Helmers et al., 1975; Stokman, Van der Knoop, and Wasseur, 1988), the US (Allen, 1974; Davis, Yoo, and Baker, 2003), Europe (Van Veen and Kratzer, 2008), countries comparatively (Conyon and Muldoon, 2006; Fennema, 1982; Stokman and Wasseur, 1985), and globally (Carroll and Fennema, 2002; Kentor and Jang, 2004).
All the above, including the outcomes of the reasons to interlock, is visualized in the conceptual model (figure 2), which will be further elaborated on in this chapter.
FIGURE 2.
Conceptual model concerning this research
Corporate arguments for interlocking
line with the resource dependency lies the financial theory (e.g. Dooley 1969; Allen 1974), which stipulates a link between a non-financial (production) firm and a financial firm, where the first is dependent on financial capital provided by the latter. In this way, banks and insurances cannot only monitor their capital investments (Koenig et al., 1979), but also “occupy a hegemonic position in domestic economies” (Nollert, 2005: 292).
Second, interlocks are used to increase a firm’s environmental legitimacy through influential connections (Mizruchi, 1996). When directors on a firm’s board have important linkages with other firms, this may be another argument to be associated with such a legitimate firm. This is an argument maybe the least discussed in the literature.
Finally, collusion occurs when interlocks are installed purposely to restrict competition (Heracleous and Murray, 2001). Not only have several countries, already in the early twentieth century, tried to restrict interlocks between competing firms to limit collusion (Mizruchi, 1996; Nollert, 2005), evidence of price-fixing scandals show that interlocks are not an essential factor in causing collusive behavior of firms (Mizruchi, 1996). Moreover, the actual influence of ‘collusive interlocks’ is both hard to identify and measure as shown by a study of Pennings (1980).
Several firm characteristics have proven to be related with the number of interlocks, such as the size of the firm – larger firms tend to be more powerful, prestigious, and have more interactions, therefore have more interlocks (Dooley, 1969; Pfeffer, 1972; Pennings, 1980; Peng, Au, and Wang, 2001), if a firm grants financial capital (Allen, 1974; Dooley, 1969; Helmers et al. 1975; Ornstein, 1984), if a firm is foreign-controlled, or foreign owned, the industry it operates in – firms from similar industries show less interlocks (Helmers et al. 1975), or even its location (Kono et al., 1998).
Kono et al. (1998: 865) accentuated the relevance of distance concerning interlock; firms that are headquartered in a location where elite clubs are also present, are more likely to maintain local industrial interlocks. Nevertheless, as they already indicate themselves (their data sample is from 1964), increasing mobility has “progressively reduced the importance of space”. As many authors have shown (e.g. Staples, 2007, 2008; Van Veen and Marsman, 2008; Van Veen and Kratzer, 2008), increasing mobility, globalization, and cross-border mergers and acquisitions have increasingly internationalized boards over the last decades.
Individual argument for interlocking
Firms also choose their directors as individuals, where personal traits are sought after by the firm. In this way, the selection of a specific individual can benefit both the firm as director (Zajac, 1988). Schoorman et al. (1981: 244) propose four benefits of recruiting interlocking directors for a board position, i.e. horizontal and vertical coordination, the personal expertise a director can provide the firm, and the image the firm creates by having a certain composition of directors on its boards. Also, firms want directors to add prestige, their expertise, their reputation (ibid), accessing an individual’s network and multidisciplinary knowledge (Fennema and Heemskerk, 2008), and the influence a name represents (Van Hezewijk, 1986). Two theories are derived from this idea.
The first individual argument for interlocking is the personal advancement model, proposed by Zajac (1988), where directors aspire a career as a board member, where many board positions are sought after, selfishly driven by remuneration, prestige, contracts and future opportunities; arguments which were already introduced by Mills (1956). Stokman (1986) argues an executive function offers easy access to the corporate elite and supervisory positions, which are maintained and even expanded after the executive career. Moreover, the share of directors that access this elite without an executive career is marginal. Lester (2003) studied the career of directors in US firms, and found for example that executive directors with demonstrated accomplishments and those from larger firms are more likely to obtain positions as an outside director. Moreover, whether or not executives are retired does not affect the career as an outside director. Executives dismissed from their position seek rebuilding their human and social capital through a position as an outside director. Then, Westphal and Stern (2005) argue that behaving in an ingratiatory way directed at influential people, like a CEO, can compensate for a lack in elite social and educational credentials for either a ethnic or gender minority (i.e. women), and can offer an alternative route in obtaining board positions.
Second, this practically implies that interlocking directors “represent social ties among the (…) elite”, explained by the social cohesion theory, meaning that future directors are most likely to be selected from a small number of potential individuals (Mizruchi, 1996: 279). As stipulated in the class hegemony model, an elite is existent which is characterized by commonalities, meaning they share similar social believes, life experiences, and ethics, creating a secluded clique of individuals, acting as a social network, with “unwritten rules of acceptable code of conduct” and “economic power and social prestige” (Koenig et al., 1979: 177).
of social ties, which are suggested to serve as a vehicle for facilitating the elite (Domhoff (1970, 1974, 1979) in Useem (1982) and Ornstein (1984); Heemskerk, 2007; Mace, 1971). Kono, Palmer, Friedland, and Zafonte (1998: 891) confirm that elite clubs may facilitate friendships, which subsequently offer easy access to board positions. As a result, networks may represent “local concentrations of interlocks” (Ornstein, 1984: 212), especially when highly integrated. This ‘limited recruitment process’ can create a climate facilitating conflicts of interest and impeding opinionating critique and pro-active intervention (Fennema and Heemskerk, 2008; Stokman, 1986), since directors will tend to be personally biased in their business function due to unavoidable familiarity within the elite. Within this corporate elite, the network of big linkers, or inner circle (Useem, 1984), is crucial, since they not only “promote new ideas on political and economic issues” and act as “spokesmen of the corporate elite” (Heemskerk, 2007: 95), but also “facilitate political cohesion within the business community” (Burris, 2005: 249).
These two individual arguments seem are extensively intertwined; pursuing a personal career and a position in the elite, offers opportunities to strengthen an individual’s position – once part of the elite, by easier access to future directorate positions, which subsequently boosts the individual’s career. Stokman (1986) argues the ‘pool of potential directors’ is not only defined by the firm, but can also be influenced by an individual’s efforts (see Westphal and Stern, 2005), implying obtaining board positions can be done either passively (being invited by a firm) or actively (by lobbying for them).
It terms of the network, linkages with an individual argument tend to have an incidental character. In general, the director can decide which boards he likes to represent, and can do so independently, speaking on behave of himself only. In this way, provided no restrictions are stipulated, there is no limit to the number of positions a director can hold, apart from their physical constraints and the limitations they would determine for themselves. Therefore, it is no surprise big linkers occur regularly; Heemskerk (2007) found 68 big linkers in 1976, 47 in 1996, and still 32 in 2001. This would imply that the more big linkers are present in the network, the higher the density of the network tends to be.
The existence of big linkers also emphasizes an important implication concerning the selection of directors. Certain directors become big linkers for a reason. Apparently, they possess those traits, knowledge, contacts, and expertise that firms want. Every firm that has the ability to attract and recruit such a director will try to do so, since it will most likely add to the firm, and therefore those big(ger) linkers will be preferred over other directors also suitable for a board position. From this idea two theories are deducted.
generalizing approach to human resource practices. Four assumptions are crucial in personnel economics, namely that 1) firms and employees are rational maximizing agents, seeking utility and profits, 2) labor and product markets must reach some price-quantity equilibrium, 3) efficiency is crucial, and 4) underlying causal relationships can be identified using econometrics and experimental design (Lazear and Shaw, 2007). These assumptions imply that ‘top directors’ will rationally choose that firm that, in return of a certain amount of tasks, will give the director the (relative) highest utility and profit, meaning the highest remuneration, responsibilities, status, etcetera.
Second, discussed by Reskin (1990) as the concept of labor queue, it assumes that firms prefer certain demographic groups over others when selecting employees, and that these potential employees are ordered by their position in a certain demographic (or social) group, choosing potentials from the most preferred group first. Moreover, the selection of directors would not only be the search for a fit between the firm and the (skills of the) individual, where extrinsic and intrinsic rewards are crucial for the latter, but also be influenced by discriminatory beliefs concerning the demographic and social group they belong to. Actually, Conyon and Muldoon (2006) studied board structure and connectedness in the US, UK, and Germany and found a positive degree correlation, meaning directors with many positions, serve on boards with directors representing many boards as well. Also, directors preferred by firms share boards with similar preferred directors. Therefore, the implication is that directors from within the corporate elite would be such a (most) preferred group to select from, especially those from the inner circle, compared with non-elite members; the largest firms will tend to recruit the biggest linkers first, and after that the bigger linkers, etcetera.
Stability of networks
implying the underlying motive for interlocking is “a consciously formulated strategy”. Moreover, individual ties that are the most embedded in firm linkages are expected to be reconstituted (ibid). Nevertheless, since interlocking directors are installed for different motives and a firms’ environment is “constantly in flux”, the argument to maintain the interlock is similarly subject to change (Mizruchi and Stearns, 1988: 196). For example, reconstitution of links that have served their purpose, are not likely to occur (ibid), for example when an interlock exists for obtaining information (Palmer et al., 1986; Useem, 1984). But an interlock with a control or monitoring function, is likely to be reinstalled, to influence the “decision making apparatus” (Stearns and Mizruchi, 1986: 525). Nevertheless, whether or not interlocks serving a cooptive and coordination purpose will be reconstituted is difficult to forecast and depends on several situational factors, such as levels of uncertainty and dependency (Stearns and Mizruchi, 1993).
Moreover, initiated linkages are clearly different from reconstituted linkages, meaning the argument and rationale behind the two is different (Mizruchi and Stearns, 1988). An initiated linkage, as discussed above, may be representing an answer to changes in the firm’s dynamic environment, whereas a reconstituted linkage represents a “maintenance of the existing tie” (Mizruchi and Stearns, 1988: 196).
Several theories have been brought forward that expect networks to be stable over time, whatever reasons exist for the linkages. Mariolis and Jones (1982) mention firms central in the network are better able to coordinate and influence any commodities flowing through the network, suggesting they strengthen corporate cohesion. From a different perspective, Granovetter (1985) takes the costs of dismantling a linkage into account, which is further elaborated by Kim et al. (2006: 704), who go as far as suggesting networks are reluctant to change, introduced the concept of network inertia, which is “a persistent organizational resistance to changing interorganizational dyadic ties or difficulties that an organization faces when it attempts to dissolve old relationships and form new network ties”, implying embedded networks may be ‘stable’ up to a rigid point. Thus, even when the interdependency has disappeared, the linkage between firms will remain to exist, simply because maintaining it may relatively have the lowest ‘costs’.
Van der Knoop et al. (1984), Ornstein (1982) and Palmer (1983) showed higher probability of reconstitution when two firms are linked by two or more directors compared with only one director. Also, positions held by ‘primary interlocks’ – having an executive function with one or more supervisory function, showed to be more stable (Van der Knoop et al., 1984).
Institutional and environmental factors influencing interlocks
Two factors specifically facilitate and influence the characteristics of interlocks, the market model (Hall and Soskice, 2001) and corporate governance practices.
Hall and Soskice (2001) emphasized the distinction between the coordinated and liberal market economy, and argued that the characteristics of the coordinated market economy (CME) facilitates firm networking, whereas the liberal market economy (LME) stimulates competition, where firm networks are less prevalent. Further elaborated by Heemskerk (2007: 17), who compares the CME with the LME, and argues the CME typically has concentrated ownership, a relatively underdeveloped stock market, a two-tier board structure, stakeholder oriented, and a lack of arm’s length agreements. A difference in market model may influence the network of firms (and individuals); LME countries, such as the US and UK, show less dense networks compared with other European (CME) countries, like Germany, France, and the Netherlands (e.g. Stokman and Wasseur, 1985). The Netherlands are discussed separately (see The Corporate Elite of the Netherlands, page 23).
function when it concerns (good) corporate governance practices, many integrated and applied those CG principles and disclosed and openly discussed their compliance for example in their annual report.
THE CORPORATE ELITE IN THE NETHERLANDS
In this chapter, the relevant developments concerning the elite in the Netherlands will be discussed in order to better understand the background of the elite. Several books (Helmers et al., 1975; Fennema, 1982; Van Hezewijk, 1986, 1988, 2003; Heemskerk, 2007; Fennema and Heemskerk, 2008) have presented an in-depth overview of the history, early networks, and factors that were crucial in shaping the landscape and networks in the Netherlands today. The authors refer generally to the ‘rise’, which refers mainly to the introduction of the concept, and ‘fall’ of the Old Boys Network and some to the rise of the new elite.
Because these books are particularly rich in information concerning the history of the elite, this thesis will theoretically draw on them in this chapter. Even though they will serve as an input for the theoretical background, some remarks need to be made concerning their empirical results. Helmers et al. (1975) and Fennema (1982) have shaped the research concerning interlocks. After that, recent literature seems to have failed to research social networks in the Netherlands from a new perspective. Despite Van Hezewijk’s books are superior when it comes to detailed information concerning specific firms and individuals, they lack the main picture of developments of the networks over time. Then, Heemskerk’s (2007) research is valuable concerning its qualitative research, and it does describe the developments of the network over time, but the intervals of his years of analysis are irregular, making interpretation difficult. Moreover, his analysis of the data lacks depth. Finally, Fennema and Heemskerk’s (2008) book mainly proposes theory and comes short of a valuable analysis. This thesis will provide empirical proof of not only the fall of the OBN, like the disappearance of the stronghold some Old Boys had, like A.A. Loudon, H. de Ruiter, and F.H. Fentener van Vlissingen, but also the rise of the new elite, such as R. Zwartendijk, A.G. Jacobs, and J. Aalberts. Additionally, it will also take into account the developments of the firms and individual’s network over time.
The ‘rise’ of the Old Boys Network
When after World War II the economy was rebuilt, it was Mertens, president of the Dutch Catholic Union (‘Nederlands Katholiek Vakbond’), who caused the corporate elite to receive mass attention and scrutiny in 1968, by stating ‘the entire Dutch economy is controlled by roughly 200 individuals, who know each other well and meet frequently. The group is as competent and financial powerful, as it is frightening.’
directors had more than three positions in 1976, accumulating to 344 positions – contributing 67,5 percent to all interlocking positions.
This elite is formed dominantly of people from the Dutch nobility and the ‘patricians’, i.e. powerful businessmen that have played a prominent role in Dutch economy and society (Van Hezewijk, 1986; Helmers et al., 1975). These groups obtained their dominant position in the Netherlands already in the preceding centuries, and maintained it by for example marriages between the two groups, and finding successors for top positions within their own family (Van Hezewijk, 1986). Since they live mainly in the West of the Netherlands, (potential) members of the elite already met and strengthened ties at some typical ‘elite-universities’ and sorority associations (ibid).
The Old Boys Network falling apart
The OBN fell apart due to several developments, knowingly the internationalization, the growing importance of the financial markets, the democratization of the managing elite (Fennema and Heemskerk, 2008), and the Tabaksblat Code. First, the internationalization was a world wide trend, which did not miss the Netherlands, which was already internationally oriented compared with other European countries (Van Veen and Marsman, 2008). Since 1995, investments coming in and going out of the Netherlands increased significantly, resulting in e.g. more cross-border M&A. This had its impact on the composition of the boards, where Dutch directors had to make place for foreign directors from the overtaking firm. Second, the ongoing liberalization of the stock exchange market has made trade in stocks more attractive. Also, the public trade of pieces of a firm’s ownership changed some criteria for the management board of a firm, meaning that creating shareholder’s wealth has become increasingly important. Especially since shareholders can buy large numbers of stock and therefore exert significant pressure on e.g. the composition of the board. Third, positions in top management became more accessible for non-OBN members; a member’s family background proved to be less important. Also due to retirement- and investment plans for employees the relationship between employer and employee changes; both faces started to point more in the same direction (Fennema and Heemskerk, 2008). Finally, the CG reforms in the Netherlands in 2004 (discussed under Institutional and environmental factors influencing interlocks, page 21) directly impeded the stronghold of the OBN, by setting restrictions on the maximum number of positions. This opened up the boards of many firms and allowed the entrance of ‘new’ directors on the boards.
The rise of the new elite
(Fennema and Heemskerk, 2008). Moreover, the Netherlands recently seems to have shown a gradual change to a more shareholder capitalism (Heemskerk, 2007). Aspiring and obtaining a top position was easier to accomplish for non-elite members, who had increasingly good academic records; many individuals from the new elite have obtained their position, using an ‘unconventional path’, i.e. not relying on the OBN. This meant that the new elite was created and existed alongside the OBN rather than that it formed an ‘integrated elite’ with the OBN (Fennema and Heemskerk, 2008; Van Hezewijk, 2003). According to Van Hezewijk (2006), by 2005 the new elite had dispelled the Old Boys; only 21 percent have origins in the nobility, compared with 83 percent in 1986.
Nevertheless, individuals from the new elite were regularly involved in scandals. Van Hezewijk (2003) discusses several of those scandals, some concerning trading shares with foreknowledge, like WorldOnline, involving Nina Brink, and Maurice de Hond’s Neweconomy.
Importance of informal ties
Besides their primary – executive and / or supervisory – functions, elite members exert influence with their additional jobs, such as advisory positions for firms and government and positions at political and educational institutions (Van Hezewijk, 1986). The numbers are quite impressive: in 1996 directors held 8 additional jobs (among the top 200 directors), which grew to 10 in 2006 (Van Hezewijk, 2006). Moreover, their reach is not limited to formal links in the Dutch economy; many individuals also meet each other in the informal circuit, since they are highly involved in clubs and associations serving a cultural, charitable, and sociable purpose. Over the last decades, these informal ties also cross borders and have created an international elite.
Governmental influence
The influence of the government also deserves notable attention, since politics played a role in shaping and maintaining the corporate elite (Helmers et al., 1975; Van Hezewijk, 1986). In the 1960s, firms had on average 8 linkages with the government, with logically state-owned firms, but also the core of the network consisting of 17 firms, the best linked. Still, 80 out of 86 had at least one direct link with the Dutch government (Helmers et al., 1975). Even nowadays, quite some directors have had a history in politics and the majority is member of a political party (Van Hezewijk, 2006), such as Hans Wijers – former Minister of Economic Affairs, now CEO of Akzo Nobel and Wim Kok – former Prime Minister, now supervisory director at ING, Shell, and TNT. Many female directors also have their credentials at governmental positions, like Neelie Kroes and Trude Maas-de Boer.
OBN is a tightly knit network, it can be expected that Old Boys tend to have a dominant presence on the boards of several firms. Also, because they tend to have an above average number of positions (see Helmers et al., 1975; Heemskerk, 2007) the network of firms will have a higher density (more on this in the Method, page 27). Given that the OBN will fall apart, these two expectations will tend to lessen similarly, i.e. the dominance of a typical group of directors will decrease, as the density of the network.
METHOD Data collection
For this research the focus lies on Dutch listed firms. Therefore, the dataset consists of firms that have been listed on the three largest indices – either AEX, AMX, or AScX – at least for one year. Due to lack of availability of data concerning firms on the local index, this group was excluded. Since the aim of this longitudinal study is to identify the developments of the yearly firm and individual networks, if a firm has been listed on one of the aforementioned three indices for one year, all other years listed at any other index during the period will be included in the dataset as well, even when it concerns a listing on the local index. To obtain a valuable and reliable dataset, via several business websites the year on year changes of the market’s composition are included4. Situations, where predominantly larger firms that were initially listed on the AEX, but during the period got listed on the AMX, AScX, and Local index, will have an influence on the interpretation of the data, since it is likely directors will remain board members of the firm and therefore sustain their interlocking positions. For example, if two AEX firms are linked in one year, but the following year one of those firms is listed on the AMX, the link will classified as a AEX-AMX link. Although the index a firm is listed on says something about the size of the firm, care should be taken with the interpretation, as this example may be (wrongly) interpreted that in the second year AEX firms are better linked with AMX firms. Therefore, having a period of 13 years may neutralize the longer term effect of these index changes. Applying this reasoning and method would not only give better insight in the development of the networks on firm and individual level, but also to ensure reliable and consistent results.
The primary source were the annual reports of the listed firms, which were quite readily available on the firm’s website or other business websites5. Annual reports are sufficient as a source, because listed firms perform a role-model function when it concerns (good) corporate governance practices, such as disclosure on board members and their remuneration. Moreover, if necessary, firms were contacted to retrieve the necessary reports and relevant information. To supplement the information on board members presented in the annual reports, apart from the firm’s website, various sites were used, like Google Finance (finance.google.com), the member’s personal website (if available), and others6.
Dataset
To properly answer the three sub questions (page 10), the initial dataset is used – discussed below, but also two datasets that are transformed from the initial set. They are discussed here.
the exit year was logically the last year (s)he was a board member. Several variables concerning the board members are relevant, such as name, date of birth, nationality, current function and entry and exit date concerning the firm and board. This resulted in a dataset covering 13 years, 96 firms over four indices, of which 88 were linked to another firm at least once in those 13 years. Moreover, the dataset for the whole period comprised in total 1590 directors creating a total of 10173 board positions, of which 3844 were executive and 6200 supervisory positions (see Table 1 for an overview). This specific dataset will be used for identifying an individual career (sub question 2) and analyzing the stability of edges (sub question 3).
Then, to correctly visualize and analyze the yearly networks, the initial dataset needs to be transformed by consistently taking one moment per year to map the yearly composition of boards7. This moment is January 1, which is taken as the representation of that year. As can be seen in the row of Yearly Networks (table 1), still 9099 positions (3420 executive and 5562 supervisory), 95 firms, and 1504 directors are included. This dataset serves as the basis to isolate the interlocks and visualize the IDN.
Finally, the interlocks need to be isolated. When this group is isolated – every director holding two or more simultaneous positions in one year, 218 directors (14 percent of the initial dataset) remain, interlocking 88 firms (92 percent), and thereby holding 2675 positions (26 percent of total positions). This selection is used to analyze the IDN’s developments (sub question 1). Logically, almost ninety percent of these interlocking positions involve supervisory functions, which require e.g. less time and tasks. Therefore, directors can hold more positions simultaneously.
Table 1 provides the numbers on the relevant datasets. TABLE 1.
Datasets (1995-2007) used in this thesis
Total positions
Executive Positions
Non-Executive
Positions Firms Directors
Dataset used to answer Initial dataset 10173 3844 (38%) 6200 (61%) 96 1590 Sub quest. 2/3
Yearly Networks 9099 3420 (38%) 5562 (61%) 951 1504 (Sub quest. 1)
Interlocks 2675 278 (10%) 2385 (89%) 88 218 Sub quest. 1 1) Super de Boer is excluded from the ‘yearly networks’ dataset (this only concerns the years 2006 and 2007), because its management is similar to its mother-firm Laurus, which is already part of the dataset. This is done to omit an unrealistic amount of double interlocks between these two firms.
Note: missing values on whether a positions is executive or non-executive are not included in this table,
since they contribute only a minor share (respectively 129 positions (1,3%) for the initial dataset; 117 positions (1,3 %) for yearly networks; and 12 positions (0,4%) for the interlocks). However, logically they are included in the analyses.
Sample size over the period
another firm, or simply bankrupted. The acquired firms are mainly done so by the largest Dutch or foreign firms, such as Polygram by Seagram Company and Gist Brocades by DSM in 1998, Baan by SSA Global and Gucci Group by French PPR in 2003, P&O Nedlloyd by the Danish firm Maersk in 2006, and Numico by French Danone in 2007; all examples of Dutch AEX-firms. Particularly, it was problematic to retrieve annual reports before the respective delisting, since contacting the firm was difficult or no such annual report was available anymore. As can be seen from table 2 (appendix), up to 2000 a quarter of the annual reports could not be retrieved. After this year, the number of missing ARs quickly fell. When presenting the analyses for these years in upcoming chapters, this will be taken into account and referred to accordingly.
Not including the local firms may have an influence on the results presented in this research, but its importance should not be overestimated. As Heemskerk’s (2007) findings show, the number of isolated firms have more than doubled from 54 (1996) to 113 (2001). On the contrary, in this study the share of firms that were interlocked remained almost similar: 87 percent in 1996 and 86 percent in 2001 (table 4, page 32). Then, because the substantial increase in isolated firms from 1996 to 2001 (Heemskerk, 2007) is not identifiable in this study and the dataset of this study consist primarily of the larger firms, it implies it were predominantly local firms – not included in this study – being cut off from the network. Moreover, with the recent developments clarified for the Netherlands (page 22), it would seem illogical to assume the local firms would play a progressively important role for a director.
Developments of the Dutch IDN and corporate elite over time
Calculating the centrality of a network can give various insights to the developments of the IDN. In this study, three measures of centrality are relevant for the analysis: degree centrality, the betweenness, and the density of the network. The added value lies in applying these measures simultaneously, offering different perspectives on the network.
First, the degree centrality is important because actors (firms) central in the network should be most active, meaning they would have the most linkages with other firms (Wasserman and Faust, 1994). It is defined as the number of links incident upon an actor (node). More precisely, the degree of node “i” is given by where aij is the number of links
from i to j.
Second, the density is calculated, since it offers an indication of the group cohesion and is recommended and widely used (Wasserman and Faust, 1994). The density is described as the ratio of actual relations or ties among a set of actors in a network and the maximum possible number of ties. The resulting network-coefficient of DM is then the
sum of all links zij divided by the maximum possible number of links n(n-1).
The density of each network of interlocking directorates is calculated using UCINET VI (Borgatti et al., 2002).
Third, the betweenness is included in the analysis, because it incorporates the paths between two firms, meaning that firms positioned between two firms has the potential to influence interactions between those firms (Wasserman and Faust, 1994). The betweenness of a network is defined as the frequency with which an actor is positioned on the shortest path between other actors in the network. Whereas the degree of an actor is a rather simple measure of centrality, the betweenness of an actor “is a crude measure of the control [an actor] exerts over the flow of (…) any (…) commodity between others”, and therefore takes into account the position it has in the network and “gives quite informative answers” (Newman, p. 6).
Furthermore, several other measures are executed on the networks, like the average distance between two nodes, the compactness, the number of cliques and the multiplicity of edges, to give support to the centrality measures discussed above.
Before centrality measures can be calculated, individual network need to be transformed to dyadic firm linkages8. For example, when a director holds 5 simultaneous positions, its personal network will create 10 dyadic linkages (or edges), according to the formula ‘n(n-1)/2’. In the table below the respective number of dyadic ties are presented for interlocking directors.
TABLE 3.
Number of dyadic ties (edges) for interlocking directors (n(n-1)/2)
Number of simultaneous positions 2 3 4 5 6 7 Number of dyadic ties between firms 1 3 6 10 15 21
Stability of edges
directors. Also, specific attention is paid to the reconstitution of the links, i.e. whether directors maintaining the same edge literally replace the other’s seat or also share the boards simultaneously (creating edge multiplicity). Then, the edges that were maintained for 10 or more years are visualized. To not disregard the possibility of the linkages being present already before 1995, those links, that were maintained for 9 years during the period and were already instituted in 1995, are included in the visualization as well, since the assumption that these linkages were also present in 1994 are quite likely. The same counts for linkages maintained for 9 years and that were still present in 2007.
Individual career
Finally, by researching the existence of an individual career, this study pays specific attention to the directors. By looking specifically at the individual network, a better understanding can be obtained concerning both the individual and corporate argument of interlocking. First, if there is a career identifiable for the directors, it implies their individual (career) choices define how the corporate network is structured and develops over several years. Second, whether or not an individual career can be defined, depends on the extent to which interlocks are installed because of a corporate argument; if evidence of a career is hard to find, this supports the corporate argument for interlocking.
TABLE 4. DESCRIPTIVES OF INTERLOCKS 1995 – 2007 2 3 4 5 6 7 Interlocking Individuals Interlocked Firms Interlocking Positions (IP) IPs / Indiv. Edges Edges / Position Degree Between-ness Distance Compact-ness4 Density 1995 45 14 7 4 2 - 72 (15%1) 48 (91%2) 192 (33%3) 2.7 174 0.91 8.292 25.375 2.176 0.486 0.1764 1996 44 14 8 3 2 - 71 (15%) 48 (87%) 189 (32%) 2.7 169 0.89 8.083 30.625 2.303 0.507 0.1720 1997 47 20 4 3 1 1 76 (16%) 53 (95%) 198 (33%) 2.6 176 0.89 7.434 38.774 2.491 0.475 0.1430 1998 50 17 7 3 1 1 79 (15%) 57 (95%) 207 (32%) 2.6 180 0.87 7.333 44.807 2.600 0.455 0.1310 1999 52 21 7 3 1 2 86 (16%) 58 (94%) 230 (34%) 2.7 207 0.90 8.414 41.879 2.469 0.476 0.1476 2000 50 21 7 2 2 1 83 (14%) 62 (89%) 220 (30%) 2.7 195 0.89 7.290 54.677 2.793 0.433 0.1195 2001 45 17 9 3 - 1 75 (12%) 63 (86%) 199 (27%) 2.7 180 0.90 6.381 57.794 2.864 0.417 0.1029 2002 49 12 12 3 1 - 77 (13%) 65 (89%) 203 (28%) 2.6 183 0.90 6.215 60.400 2.888 0.412 0.0971 2003 53 19 9 4 2 - 87 (14%) 73 (91%) 230 (30%) 2.7 221 0.90 6.384 61.658 2.811 0.398 0.0887 2004 54 16 10 3 1 - 84 (14%) 69 (87%) 217 (29%) 2.6 192 0.83 5.971 56.290 2.866 0.371 0.0878 2005 60 17 5 1 1 - 84 (13%) 66 (80%) 202 (27%) 2.4 156 0.72 5.030 62.333 3.040 0.368 0.0774 2006 63 16 2 2 - - 83 (13%) 66 (80%) 192 (25%) 2.3 132 0.64 4.333 56.424 3.170 0.305 0.0667 2007 73 16 1 - - - 90 (13%) 69 (83%) 198 (25%) 2.2 122 0.59 3.681 72.710 3.410 0.311 0.0541
1) The percentage indicates the contribution to the total amount of individuals in that year (also including non-interlocks) 2) The percentage indicates the contribution to the total amount of firms in that year
3) The percentage indicates the contribution to the total amount of positions in that year 4) Ranges from 0 to 1; larger values indicate greater cohesiveness
RESULTS Developments of the Dutch IDN
Table 4 (previous page) offers a comprehensive overview of data concerning the interlocks, which provide some interesting findings. First, the number of interlocking directors varies between 71 and 90, and is responsible for a corporate elite that consists of 12 – 15 percent of all directors. Second, despite a small peak in 1998 and 2003, over the whole period the number of interlocking positions per director, as well as the share of interlocking director and positions and the interlocked firms per year decreased. This is mainly caused by a diminishing number of directors holding more simultaneous positions; the big linkers vanished nearly over those 13 years (see table 16 for all big linkers in this dataset). As a result, the density, compactness, and average degree of the firm’s network decreased, whereas the average distance increased. Nevertheless, the average betweenness increased.
Importantly, two periods can be identified, knowingly the period up to 2000, and the period starting 2000 through 2007, which is well represented by the number of interlocking positions the directors hold. In the first 5 years of the period (through 1999) a gradual absolute increase in interlocking directors, interlocked firms, and interlocking positions – especially big linkers – is visible. The years after show a decline – with a minor peak in 2003 – of those variables. Moreover, as discussed before concerning data availability and therefore the number of firms in the sample (see table 2 and page 28), it would be sensible to analyze the period 1995 – 1999 separately from the last seven years concerning the two findings mentioned before.
Moreover, TNT was involved in 10 cliques, of which 6 were added in 1999 (i.e. the size of the clique is 3); Laurus contributed to 9 cliques, of which 2 were added the same year. On an individual level, TNT’s board consisted of six and Laurus’ board of seven interlocking directors in 19999. As becomes clear from this (extreme) example, adding a firm may have a profound impact on the characteristics of the network as a whole and taking into account the number of missing firms in this period, conclusions should be drawn with care.
Then, the last years of the 1990s are characterized by the dominance of the AEX index. It becomes clear that just over half of all edges were internal to the AEX index10; apparently the biggest firms were mainly linked to each other (table 6, Appendix). This becomes even more obvious when looking at table 7 (Appendix), which list the contribution of an index to all edges. In 1995, 88 percent of all edges involved an AEX firm, and that number slightly declined to 84 percent in 1999. The involvement of local firms grew noticeably from 23 (1995) to 34 percent (1999), because mainly the linkages between AEX and local firms increased. Also, these years showed an average degree between 7.3 of 8.4, which is remarkably higher than the period after 1999. The AEX-dominance is confirmed by observing the firms with the highest degree in those years (1995-1999); of the top 10 only in 1995 and 1997 not all the firms were from the AEX, i.e. it contained one firm from the AMX. During 1995 – 1999 the betweenness of the network increased from 25 to 42, indicating despite the network getting less dense, the firms in the network got better positioned within the network on average. This implies whereas at first firm within the AEX were linked well and AMX and local firms are predominantly linked with AEX firms. Then, local firms got slightly better linked with each other (up to 6.1 percent in 1999). This apparently resulted in a higher average distance between two random chosen firms (up to 2.47 edges in 1999).
Although increasing somewhat until 1999, still characteristics of the network seem to be relatively stable with some slight fluctuations. Contrasting this, remarkable changes occurred starting 2000. Not only did the big linkers disappear, with merely one director holding 4 simultaneous positions in 2007, the interlocking directors were able to make the firm network more evenly connected with less interlocking positions and edges and a lower degree.
person (see table 3), and average degree – of the top 10 firms with the highest degree the average halved from 16,7 (2000) to 8,2 (2007). The AEX firms got slightly worse linked to the firms from the smaller indices – from 43 percent in 2000 to 41 in 2007 (table 6). Firms in general got a better position in the network, indicated by the average betweenness that increased from 55 (2000) to 73 (2007) and the increasing involvement of firms from the smaller indices – from 32 percent (2000) to 39 percent (2007)11. This suggest that although firms have less links between each other (fell from 7.3 in 2000 to 3.7 in 2007), those links that remained in 2007 where spread more evenly between the firms within the network and positioned them better, mainly improving the linkages between the smaller firms with each other.
Concerning the slight peak in 2003, the same explanation holds as with the 1999 peak. Imtech was only included starting 2003, because before that year unfortunately no annual reports could be retrieved. Imtech was well connected, having 13 links with other firms (6th place), being part of 9 cliques (of 4 were added that year), and having 5 interlocking directors on its board. Imtech’s contribution remained quite substantial until 2005, after which it quickly reduced12.
Several conclusions can be drawn from the developments of the Dutch IDN. First and foremost, the corporate elite did surely decline, not in the number of directors, but definitely in the number of positions they hold; the big linkers have nearly disappeared. This conclusion unanimously corresponds with all previous research (e.g. Heemskerk, 2007). More precisely, the size of the corporate elite found in this study – between 11 and 15 percent of the corporate elite – is comparable with previous research during this period, arguing that is 11 – 14 percent (Heemskerk, 2007)13. Second, in the first period (up to 2000) the emphasis lay on the AEX firms interlocking mainly each other (the centre of the network), and firms from the smaller indices – the peripheral firms – were predominantly linked directly with the AEX firms, indicated by a high degree and a relatively low betweenness of the network – mainly due to the dominant centre. In the second period (from 2000), this emphasis shifted to a network were the centre of the network became less dense (dominant) and the peripheral firms got better positioned in the network and therefore also with each other, indicated by a decreasing degree for AEX firms and an increasing betweenness of the firms in the network.
increased from 32 to 39 percent14. The decrease is due to fewer linkages between AEX and AMX firms, linkages internal to the AMX index, but also between AMX and Ascx/Local firms. The increase is mainly caused by more linkages between AEX and Local firms (see table 6). Apparently, the emphasis has moved from the larger to the smaller indices; where directors previously linked AEX and AMX, now the larger firms are more directly linked with the AScX and Local firms. Even though the direction of the link is unknown, it seems logical to assume this development is caused by directors at Local firms obtaining supervisory positions at larger, i.e. AEX, firms, in stead of the other way around.
Stability of linkages between firms
When the focus lies solely on the links between firms (edges), for the whole period, 503 ‘unique’ edges appear (i.e. no double values). To see whether or not the dataset contains support for the various theories written on dependencies between firms, which would imply firms have many arguments to sustain a structural link between two firms, the table below presents which of those edges have a seemingly structural character.
TABLE 9.
Stability and reconstitution of edges (N=503 edges)
Edge maintained by 5 years
or longer 10 years or longer full 13 years 1 director 186 16 0 2 or more directors 62 (14) 20 (7) 5 (3)
Note: the number between brackets indicates the number of links between firms
where two directors were on the same board for maximum 2 years, of which one exited the board, implying one director is intentionally replacing the other. For example, when a director leaves a board in 1997 (and the link between two specific firms ceases to exist), and another member sustains the link between those firms by entering the same board(s) in 1996, 1997, or 1998, it is counted as 1.
implying not only that either one of those directors would suffice to label the link as ‘stable’, but also that is significantly contributes to the edge multiplicity (table 8). In line with the findings put forward in Developments of the Dutch IDN (page 33) – the dominance of AEX firms is confirmed here as well: of the 22 firms that are linked by more than one director for 10 or more years, only 2 firms have not been listed on the AEX index.
The most ‘stable’ links are visualized in figure 3a through 3c (next page). Three visualizations are presented; one with linkages in 1995, one of 2005, which is included mainly because this year in combination with 1995 incorporates all maintained links of more than 10 years listed in table 9, and one of the situation in 2007. These years were also chosen because they plainly visualize the decreasing number of edges that can be interpreted as ‘stable’; clearly, during this period, several edges active in 1995, have been not been reconstituted by 2005 and 2007, since the number of stable edges decreased from 45 (1995) to 29 (2005) and even further to only 17 edges in 2007.
FIGURE 3A.
Stability of edges between Dutch firms (1995, N=45 edges)
Note: The dotted lines indicate links that have been present for 9 years until 2007, suggesting the possibility of being active for 10 years consecutively, i.e. still in 2008 or later, is substantial (4 linkages, of which 0 sustained by 2 directors). The single solid lines indicate linkages active for 10 or more consecutive and sustained by only one director (9 linkages). Finally, the thick solid lines indicated linkages sustained for 10 or more consecutive years by 2 or more directors (16 linkages, of which 7 seem to have a reconstituted character). See appendix for the firm codes.
FIGURE 3B.
Stability of edges between Dutch firms (2005, N=29 edges)
FIGURE 3C.
Stability of edges between Dutch firms (2007, N=17 edges)
Therefore, a corporate argument, based on interdependencies, to sustain the link between two firms seems most unlikely, or at least increasingly unlikely in the current economy, indicated by the sudden drop of stable edges in the last 3 years.
Concerning the reconstitution of the edges, only 14 linkages (out of 62) between firms seem to have a reconstituted character. When the strict interpretation of reconstitution would be applied, i.e. when one director literally replaces the seat of a leaving director and therefore the more lenient criteria mentioned in the notes of table 9 would be ignored, of these 14 links only 7 would have a reconstituted character. Contributing to only 1.4 percent to the total edges, the argument of deliberated reconstitution of edges seems hard to hold for this dataset.
In conclusion, it appears that directors not only hold a board position for quite a while, but that links between firms created by these directors are based more on personal preferences and choices of both the director and firm than deliberate positioning by the interlocked firms.
As indicated before, holding a position up to twelve years is reasonable, as it is stipulated in the Tabaksblat Code (2004). From the perspective of the firm and director it is similarly understandable; when both parties in a collaboration have no reason to end it, there is no incentive not to sustain it for the maximum twelve years. One extreme example is A.J. van Puijenbroek, who has been a supervisory board member of Telegraaf Media Groep since 1975, when he was 28 years old. Due to the Tabaksblat Code he has to resign in 2009.
