Local Origins of Business Formation

Using comprehensive administrative data on business applications, we find that startups per capita exhibit enormous variation across counties and tracts in the United States. We decompose this spatial variation into two components: variation in business ideas per capita and in their rate of transition to startups. Both components matter for the variation in startups per capita. Furthermore, local demographic, economic, financial, and business conditions account for a significant fraction of the variation in startups per capita and in its components. In particular, income, education, age, and foreign-born share are strongly and positively associated with idea generation and transition rate. The relationship between local conditions and ideas generally differs in magnitude from the relationship with the rate at which these ideas transition into employer businesses. Interestingly, certain conditions are positively associated with ideas but negatively with transition rates. The predicted rank of locations based only on observable local conditions closely relates to the actual ranking of locations in terms of startups per capita, making it possible to characterize high-startup locations using observable local conditions alone.