Dayton Lambert

This paper estimates the relationship between broadband internet speeds and the survival and employment of business establishments. The analysis combines publicly accessible broadband speed availability data from the Federal Communications Commission with microdata on over 10 million U.S. businesses from the National Establishment Time Series database across 2015 to 2020. We use a two-stage Heckman procedure to identify and correct for attrition bias resulting from businesses that exit the dataset over the study period. Probit models estimate business survival in the first stage.