Three Essays on the Survival Time of Firms and Their Growth

The first essay in this dissertation uses US iron and steel shipbuilding data (1825-1914) used by Thompson (2005). It proposes a Two-Stage Discrete Finite Mixture hazard model to account for selection effect associated with a high first-year exit rate, and omitted variable bias associated with missing information such as a shipbuilder's pre-entry experience. In the first stage, the model uses a Probit model to explain the selection effect by employing both a firm's production share and production-selection component at the time of entry as exclusion restrictions. The results identify two latent classes as proxies for pre-entry experience used in the Weibull model by Thompson. The model is useful when important factors for new-firm survival, such as human capital, might be missing from databases.The second essay examines whether starting with different types of bank credit has a heterogeneous effect on the future performance of small business in the Kauffman Firm Survey (2004-2011) (KFS). The analysis uses a structural equation model to account for endogeneity bias, entailing three interrelated equations: one for the firm's survival time, one for its start-up with a type of bank credit, and one for its revenue in 2007. To identify high-quality firms starting with business credit, the model employs three exclusion restrictions: a firm's entry share of tangible assets in its cohort; the entry-selection component of tangible assets, and per-capita state-level secured loans. The results reveal a positive endogenous treatment effect on a firm's future revenue when starting with business credit, and a negative one when starting with personal credit.The third essay evaluates the direct effect of knowledge spillover on small businesses in the KFS. It investigates whether their entrepreneurs could transform abstract knowledge into economic knowledge. Knowledge spillover can take place when a firm directly collaborates with other institutions or firms. To identify the causality effect, the empirical analysis uses collaboration as a proxy for transformed knowledge. To improve identification, it implements a Combined Difference in Difference approach. The results indicate that firms teaming with other institutions outperform those without such treatment. Teaming induces higher growth in real revenue, employment size, and real wages.