Abstract
Challenging social capital research, we argue that network top management teams (s) established to support entrepreneurship in strategic multipartner networks should be careful in developing ties to outside organizations and networks. We suggest that such ties limit the network 's ability to engage in the strategy‐making processes needed to facilitate entrepreneurial orientation on a network level. Based on five‐year panel data from 53 formalized networks of small and medium‐sized enterprises, we demonstrate that homogenous and highly educated network s can compensate the negative effects of ties to other organizations, but not the negative effects of interlocking directorates.
Notes
18. We also considered an alternative operationalization where we looked at the proportion of TMT members with ties to external organizations and interlocking directorates. While doing so, as previously, we simultaneously controlled for network size. The results were consistent with the ones reported in Table in terms of coefficient signs. Significance level was generally somewhat lower with the alternative operationalization of these constructs: the direct effect of ties to external organizations attained significance of p < .10 (p < .01 for the operationalization reported in Table ); the direct effect of interlocking directorates previously significant at p < .10 has lost its significance although retained the sign; the interaction of homogeneity and ties to external organizations has actually gained in significance (p < .01 as opposed to p < .05); and the interaction of education and ties to external organizations has dropped its significance somewhat (p < .10 as opposed to p < .001).
19. As a robustness check, we retested our models using alternative estimation techniques including ordinary least squares (OLS) regression, OLS regression with standard errors corrected for cluster effects at the network level; feasible generalized least squares estimation, and the regression with Driscoll and Kraay's (Citation1998) heteroscedasticity‐consistent standard errors that are robust to cross‐sectional and temporal dependence as introduced in Hoechle (Citation2007). The results remain consistent across different models. Results are available from the authors upon request.
Additional information
Notes on contributors
Joakim Wincent
Joakim Wincent is professor of Entrepreneurship & Innovation at Luleå University of Technology and professor in Entrepreneurship at Hanken School of Economics.
Sara Thorgren
Sara Thorgren is assistant professor of Entrepreneurship & Innovation at Luleå University of Technology.
Sergey Anokhin
Sergey Anokhin is associate professor of Marketing and Entrepreneurship at Kent State University.