Glass networks : the network structure of women directors on corporate boards : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Business and Administration at Massey University, Palmerston North, New Zealand

Abstract

Boards of directors are overwhelmingly dominated by men. Globally, only 5-20% of directors of substantial corporations are women. Despite three decades of research and intervention this gender ratio has remained static except where affirmative action has forced change. Using Carlile and Christensen’s (2005) management theory building methodology, I develop a new theoretical approach to the study of women on corporate boards of directors called Glass Network Theory. Using concepts from small-world and scale-free network theory, I suggest that semi-permeable invisible barriers or glass nets, rather than glass ceilings, permit limited directors through to become connector directors with multiple board seats. Resistance to change in gendered director networks is framed as a structural and normative characteristic of dynamic networks. Like other natural and social scale-free networks, gendered director networks display an emergent and self-perpetuating order. Low levels of diversity are incorporated as adaptive features of stable self-similar director networks responding to environmental pressures and economic fluctuations. Nested director networks at global, national or local levels show self-similarity at all scales, particularly where ‘shoulder tapping’ or preferential attachment underpins the network formation. Probing gendered director networks with social network analysis tools shows that women directors in global and national networks are more likely to be found in the largest connected component than in the unconnected network components. Both male and female director networks show the characteristics of small-worlds, that is, high clustering coefficients and short path lengths. Both male and female directors also show ‘positive assortativity’ where directors of high degree associate with other directors of high degree. As the distribution of multiple directorships follows a power law, the model of the expected seat spreads is a useful tool to track the effectiveness of governance or affirmative action interventions such as quotas. I conclude that female director networks are not inherently different from male director networks except in size.