Can Green Car Taxes Restore Efficiency? Evidence from the Japanese New Car Market

We investigate the efficiency of vehicle taxation in second-best settings. A random-coefficients logit model is estimated for quarterly automobile sales data between 2004 and 2012 from the Japanese new car market. The quasi-experimental nature of the data is exploited in two ways. First, we construct the location of product-specific tax rates in the characteristics space as a set of instruments to control for endogeneity of observed car prices. Second, the large and persistent variation in effective vehicle prices, caused due to Japan's green car tax policy since 2009, are used to obtain consistent estimates of the own- and cross-price elasticities. Our results indicate evidence for substantial scale and composition effects: Though the policy successfully reduced sales-weighted average emissions, it also increased total sales substantially. Consequently, the policy-induced reduction in annual vehicle CO2 emissions was small. In contrast, a modified version of the emissions-based vehicle tax" la Fullterton and West (2002), based on the fuel efficiencies of car models, could have reduced annual vehicle CO2 emissions substantially more while increasing total economic surplus relative to the no policy counterfactual.