Geographical differences in U.S. Clean Air Act requirements are often used to identify environmental regulatory impacts. The standard approach abstracts from aspects of the law affecting which areas are regulated, how strictly they are regulated, and when regulatory changes occur. We find that omitting these factors can bias results by contaminating the control group, leading to underestimation of historical employment impacts and overestimation of projected impacts from tightening regulations. Results indicate that 1990 changes to ozone nonattainment provisions reduced power plant employment without significantly affecting generation, suggesting that installation of pollution controls contributed to labor-saving technical change at affected sources.