【正文】
etail gasoline prices, and social preferences on hybrid sales. We first estimate and pare the effects of government incentives designed to stimulate purchases and market incentives created by changed gasoline prices. We evaluate a number of testable hypotheses related to relative sales of higheconomy and loweconomy hybrids. In particular, we would expect rising gasoline prices to have an especially large effect on the sales of higher efficiency hybrid models, models for which the relative fuel economy improvement over similar cars is substantial. In addition, we expect that the fuel efficiency and lower emissions of higher efficiency models would be particularly attractive to individuals with a strong preference for the environment or energy security and individuals with higher travel intensity. We then study how incentive efficacy relates to the structure of the tax incentive provided. We pare the effect of sales tax waivers and ine tax credits and find, consistent with the literature on tax salience, that sales tax waivers are associated with significantly greater sales than ine tax incentives. Finally, we test whether proxies for environmental preferences or preferences for energy security are correlated with sales of high economy hybrids, for which the environmental and energy security benefits are the greatest. Several sources of potential bias exist arising from the endogeneity of state incentives and gasoline prices. States may choose which incentive to offer based on the relative efficacy of the different incentives in a particular state. For example, California and Virginia may choose to allow hybrid vehicles to access the HOV lanes because travel/traffic costs are significant in each of these markets. More rural locations where travel/traffic costs are substantially lower may choose to use tax incentives instead. Endogeneity of this form lead our point estimates to be upper bounds on the efficacy of government incentives. State gasoline prices, on the other hand, are plausibly exogenous to hybrid sales. Although higheconomy hybrids are substantially more fuel efficient than parable nonhybrid vehicles, hybrid market peration is fairly low during the study period. Thus, hybrid vehicles account for a small share of total gasoline consumption and are unlikely to affect state gasoline prices. In several quarters in 2020 and 2020, production limitations constrained sales of the Toyota Prius and Honda Civic hybrid. If production constraints affected all states equally, our use of time*model fixed effects would control for stateinvariant scarcity. If automakers allocated a greater proportion of production during periods of scarcity to states with more generous incentives, though, we may inappropriately attribute the effect of production constraints to the government incentives. We believe that this concern does not substantively bias our results. Our conversations with Toyota indicate that the firm allocated scarce production so as to equalize the delivery delays across the different markets. To the extent that the sales of hybrid vehicles during these periods was proportional to existing demand, model*time fixed effects will capture the effect of production constraints. We econometrically confirm this by test the robustness of our estimates to the exclusion of periods in which model production exceeded production limits. We find the exclu