Start-ups' success factors - Evidence from Balkans

Abstract

Venture capitalists are considered as one of magical forces driving the creation of clusters like Silicon Valley. Yet questions like "is funding from a venture capitalist a sufficient condition to predict the success of a start-up?"and "what other factors than venture capitalists can be used to predict the success of start-ups?"are far from being answered in full, especially when considering the Balkans region. In this paper we are trying to answer these questions using start-up company data from Crunchbase.com. The sample includes data from eleven countries: Turkey, Romania, Greece, Bulgaria Croatia, Serbia, Montenegro, Bosnia and Herzegovina, Albania, North Macedonia, Kosovo. We apply Logistic regression models for the Boolean dependent variable of "Has Exit"(either IPO or M&A) as a proxy for start-up success. The model includes 9 independent variables such as "Number of funding rounds", "Parents granted", "Number of Events", "Trademarks registered", "Has Lead Investor"and others. Due to largely imbalanced dataset, we have also applied Synthetic Minority Oversampling Technique (SMOTE), leading to overall prediction accuracy of 67%.