Congestion Pricing and Local PM2.5: Early Evidence from Fixed Monitors in New York City

This paper studies whether implementation of New York City's congestion pricing program is associated with local changes in ambient PM2.5. I assemble a monitor-hour panel of fixed-site pollution readings across New York City and align each observation to New York local time, allowing treatment status and charged hours to match the program's legal geography and toll schedule. The empirical design compares treated and untreated monitors before and after implementation using difference-in-differences and triple-difference specifications with monitor, date, hour, and date-hour fixed effects, hourly weather controls, and small-cluster inference checks. Across specifications, PM2.5 declines more at legally treated monitors than at control monitors after the charge begins. The association is strongest under the baseline legal treatment definition, remains negative under a broader geographic treatment definition, and is more pronounced on weekdays and during charged hours. These patterns are robust to wild-cluster bootstrap inference and are supported by placebo-assignment counterfactuals that randomly reassign treatment status across the monitor network. Event-study diagnostics, fake-date placebo tests, and donut-window checks further support the timing interpretation while also reinforcing the need for caution. The results do not estimate a full citywide causal effect, nor do they observe traffic volumes, route choices, or individual exposure. They provide early fixed-monitor evidence that implementation of the congestion charge is associated with a relative decline in local PM2.5 at monitored sites most directly tied to the policy's geography and timing.