Our understanding of the performance of nascent businesses is greatly hampered by data limitations. In this research we exploit a unique dataset coming from customer records from Barclay’s Bank. The data source has several unique qualities which have been previously unavailable to academic researchers. First, it accurately identifies the month and year at which a business actually started trading, defined as the time at which the business generated its first sales. Second, all financial data are then available. Third, there is a commercial incentive to ensure the information is correct.

We investigate the characteristics and determinants of firm growth and survival, although our main thrust is to investigate the determinants of start-up size. It has been suggested in theoretical work that prior entrepreneurial experience should lead to superior performance. On the one hand, the outcomes of new businesses are notoriously unpredictable and have been likened to a random process (which precludes learning). On the other hand, some evidence (but not all) finds that individuals with prior business experience have higher survival rates. These two could be reconciled if experienced entrepreneurs have not learned to play the game better, but increase their chances of survival by having a larger start-up size (i.e. arriving at the ‘gambling table’ with more ‘poker chips’; or playing the lottery with more lottery tickets to start with). Instrumental-variable quantile treatment effect estimates (instrumented by parent’s business experience) suggests that prior business experience does lead to higher start-up size. Other factors are also observed to affect start-up size.