Measuring Knowledge Work [From Home] Using Big Data
Assistant Professor of Finance
HKU Business School , The University of Hong Kong
This paper introduces a measure of work-from-home activity using a novel dataset of firm-mapped internet content consumption covering tens of thousands of business and consumer interest websites. I build a classifier of incoming traffic into residential, business, VPN and mobile networks. Validating this classification, I find remote IPs spike in traffic in the middle of March, the share of remote IP traffic in a county exhibit a high, 0.765 covariance with mobile phone data on workplace visits, and is higher in firms who ex-ante had better remote IT and in knowledge-intensive industries. Studying organizational factors affecting WFH, I find a strong role of social interactions. I document a lower shift toward remote work in response to local social distancing or Covid-cases in organizations with stronger social connections and less generalist workers, in firms with stronger culture (especially teamwork culture), and in areas with more social capital. If local cases or social distancing comprise exogenous variation in WFH, I find firms spend more time on business relevant reading, but less on researching new topics. These findings provide suggestive evidence for the ostensible tradeoff WFH poses between productivity and creativity.