Does Telework Widen the Abstract-Task Wage Premium? Evidence from Cross-Occupation Variation in Telework Feasibility 

Did the post-2020 telework expansion amplify or compress the wage premium for abstract-task intensive occupations? I provide a causal estimate of how telework access mediates the abstract-task wage premium, distinct from existing telework-wage studies that do not condition on task content, using a difference-indifferences design that exploits within-abstract-task variation in telework feasibility. Comparing high-abstract occupations that are teleworkable (management analysts, software developers) with highabstract occupations requiring physical presence (surgeons, dentists) in CPS-ORG data from 2017Q1 to 2023Q1, I find that the preferred specification yields a 2.3 percent wage premium for telework-accessible abstract-task occupations, stable at 1.8-2.3 percent across seven specifications. Coefficient stability analysis yields δ = 49.43, indicating that selection on unobservables is unlikely to explain the result. An event study confirms parallel pre-trends (p = 0.886) and gradual post-treatment divergence. The effect concentrates in metropolitan areas (2.7 percent) with a null in non-metropolitan labor markets, consistent with a monopsony-reduction mechanism. Sensitivity analysis following Rambachan and Roth (2023) confirms robustness to moderate violations of parallel trends.