Air Quality, High-Skilled Worker Productivity and Adaptation: Evidence from Github

Abstract

Highly skilled knowledge workers are important drivers of innovation and long-run growth. We study how air quality affects productivity and work patterns among these workers, using data from GitHub, the world's largest coding platform. We combine panel data on daily output, working hours, and task choices for a sample of 25,000 software developers across four continents during the period 2014-2019 with information on concentrations of fine particulate matter (PM2.5). An increase in air pollution reduces output, measured by the number of total actions performed on GitHub per day, and induces developers to adapt by working on easier tasks and by ending work activity earlier. To compensate, they work more on weekends following high-pollution days, which suggests adverse impacts on their work-life-balance. The decline in output arises even at concentrations in line with current regulatory standards in the EU and US and is driven by a reduction in individual coding activity, while interactive activities are unaffected. Exposure to PM2.5 levels above the city-specific 75th percentile reduces daily output quantity by 4%, which translates into a loss in output value by approximately $11 per developer.

Keywords: Air pollution, productivity, high-skilled workers, adaptation, GitHub

JEL Classification: D24, J22, J24, L86, Q52, Q53