A new EBRD Working Paper (number 269)
Can policymakers improve macroeconomic performance by encouraging the entry of high-performance startups? To answer this question, we construct a novel and comprehensive data set on 1.3 million startups in ten European countries. We apply cluster analysis to identify distinct startup types and trace their development over time. Three stylized facts transpire. First, we uncover five well-separated startup types that are consistently present across countries, industries, and cohorts. We label these Basic, Large, Capital-intensive, Cash-intensive, and High-leverage. Second, the initial differences between these startup types are persistent. Third, each startup type displays a characteristic life cycle in terms of productivity, employment generation, and exit rates. We feed these empirical results into an agnostic firm dynamics model to quantify how much structural policy could improve macroeconomic performance by shifting the composition of startups. We find that substantial gains in aggregate employment and productivity can be made through policies that benefit high-performance startups (such as large and capital-intensive ones) while discouraging the entry of underperforming firms (such as highly leveraged ones).