Tsvetovat, MaksimBerea, Anamaria2012-10-05NO_RESTRIC2012-10-052012-10-05https://hdl.handle.net/1920/7950Evidence shows that only a small number of entrepreneurial endeavors - high-impact companies - create most of the new employment in the US. This research is looking for the causes of emergence and uses computational methods to analyze speci c aspects of these companies. First, this research proves or disproves some "popular conjectures" regarding the age, location, industry and entrepreneurial character of high-impact companies. Secondly, the agent-based model shows how a company grows in employment and revenue based on two layers of organization: 1) one is the heterogeneous team formation and interaction at the mezzo level of a company organization and 2) another one is the heterogeneous employees skills and interaction at the individual level of a company organization. The model advances and tests the hypothesis that companies that learn the "fastest" from failed projects while retaining access to capital are more likely to become high-impact companies. The experiments replicate the high-impact rate given by the real life data and show that the high-impact phase of a company growth is achieved for speci c learning parameters and failed projects. The purpose is to provide a coherent theory-model-evidence analysis on high-impact entrepreneurship, that adds new insights for researchers, policy makers and business practitioners, in addition to the qualitative information, informal knowledge or hands-on experience that they currently posses.enEntrepreneurshipHigh-impact companyConstructive failureAgent-based modelDissertation