Quantifying the nonlinear causal impact of commute time on U.S. remote work

Urban commuting is a significant contributor to environmental pollution and emissions, while remote work offers a sustainable alternative. This study investigates the nonlinear causal impact of commute time on remote work decisions using U.S. monthly survey data from May 2020 to March 2025. To address residential self-selection bias, we apply propensity score matching (PSM) and find that commutes exceeding 30 mins increase the odds of higher-category remote work preferences and frequency by 31.1 % and 20.8 %, respectively. We then apply the generalized propensity score (GPS) method and uncover a nonlinear causal effect of commute time on remote work preferences and frequency: the marginal effect is negative within the first 15 mins, turns positive afterward, and plateaus after 50 mins. Heterogeneity analysis reveals the strongest effects among onsite workers, with weaker effects on hybrid and remote workers. These findings offer behavioral insights for remote work policies and sustainable transportation planning.