The positive impact of social capital on startup survival has been well described in the literature,but the impact of business incubators and clusters remains inconclusive in spite of the opportunity for social capital diffusion in these high density sites. We hypothesize that being in an environment in which social capital is readily accessible does not imply that the startup will engage it. We also hypothesize those startups that collaborate with other agents (universities, industries, and government organizations) outperform startups that do not.

To test the above hypotheses, this work develops a model of startup survival that distinguishes between the density of a startup’s ecosystem and the startup’s exploitation of its ecosystem. Using the longitudinal Kauffman Firm Survey of 4928 companies founded in 2004 and the US Census Bureau County Business Patterns, we compute the density of a startup’s ecosystem as the number of companies in the startup’s ZIP code with the startup’s 2-digit NAICS code. This density measure is used as a proxy for the availability of relevant social capital and tacit knowledge to the startup.Life table and parametric survival analysis models are applied to these integrated databases using collaboration and density as time-varying covariates.