01 Working-Remotely-Selection-Treatment-Market-for-Remote-Work
The paper investigates how remote work affects worker productivity, career development, and labor market outcomes, distinguishing between the causal treatment effect of working remotely and the selection effect of workers self‑selecting into remote jobs. Using daily call‑center logs, personnel records, and survey data from a large U.S. Fortune‑500 firm (Jan 2019–Oct 2021), the authors exploit the exogenous office closures during COVID‑19 as a natural experiment. Difference‑in‑differences regressions (Equation 3) compare workers who were initially on‑site and forced to work remotely (treatment) with those who were already remote (selection). Key findings: (1) Before COVID, remote hires answered ~12 % fewer calls per hour, indicating negative selection; (2) When on‑site workers were forced remote, their call‑answer productivity fell ~4 % (≈1/3 of the pre‑existing gap is a treatment effect, 2/3 is selection); (3) Remote work increased hold times and callback rates, especially for less experienced workers; (4) Remote workers received fewer manager one‑on‑one meetings and less new‑skill training, and their promotion rates fell by up to 55 % relative to on‑site peers; (5) Heterogeneous effects show women experience larger declines in call volume and larger increases in hold time than men. Additional tables examine full‑remote periods after office closures, showing persistent productivity losses (≈5‑10 % fewer calls per hour) and modest changes in customer satisfaction. The authors develop a two‑period theoretical model of the remote‑work market, showing that adverse selection (high‑productivity workers prefer on‑site jobs because remote work obscures performance) creates a wedge between marginal and average costs, leading to under‑provision of remote jobs. The model predicts that improving firms’ ability to evaluate remote workers (raising po‑pr) or increasing taste heterogeneity (higher s) can move the market toward the efficient equilibrium. The paper concludes with policy implications: firms should consider the hidden career costs of remote work, improve remote performance monitoring, and weigh the trade‑off between short‑run productivity gains and long‑run talent allocation.
Topics
Empirical Strategy to Distinguish Treatment and Selection Effects of Remote Work
The study uses the exogenous office closures during COVID‑19 as a natural experiment, applying difference‑in‑differences regressions that compare workers forced to work remotely with those already remote. This design isolates the causal treatment effect of remote work from the pre‑existing selection differences among workers who self‑select into remote jobs.