Measuring small business dynamics and employment with private-sector real-time data

Abstract

This paper proposes a novel methodology to distinguish true business openings and closings from sample churn in private-sector data and to evaluate the representativeness of the resulting estimates by leveraging supplementary high-frequency information on individual business activity. The methodology produces both real-time estimates using only concurrent information and retrospective estimates that incorporate additional information as it becomes available, reflecting a fundamental trade-off between timeliness and accuracy. The methodology is applied to a real-time sample of small businesses widely used during the COVID-19 pandemic to demonstrate its usefulness under extreme circumstances. The application highlights the importance of properly accounting for business openings and closings and at the same time yields two important insights about small business dynamics during the pandemic: (i) small business employment in in-person service sectors experienced larger swings at the beginning of the pandemic than employment of larger businesses, primarily due to a spike in temporary closings; (ii) delayed access to loans from the Paycheck Protection Program significantly increased small business closings but had minimal impact on employment of continuing businesses, suggesting the program’s effectiveness operated primarily through preventing closures rather than preserving jobs at operating businesses.