Fuel Prices vs. Automobile Fuel Economy Standards in a CO2-Constrained Transport Sector

Abstract

One way to raise the fuel efficiency and reduce CO2 emissions of new cars is through fuel economy (FE) standards; more than 20 countries worldwide currently implement such standards. A second way is to increase fuel taxation in order to induce purchases of more efficient cars and discourage private car travel. Although the adoption of standards has induced FE improvements, there are arguments against standards and in favor of fuel tax increases. The aim of this paper is to analyze the impact of FE standards and fuel prices in new car fuel economy with the aid of cross-section time series analysis of data from 18 countries. Similar work was previously conducted for the U.S. only, and mostly with data up to 1990. We estimated a log-linear equation with new-car FE as the dependent variable and the following explanatory variables: FE standard, real gasoline price (with lags of 0 to –3), and a time trend to capture autonomous technical progress and changing consumer preferences. Data were obtained from official sources such as the U.S. EPA, the IEA and the European Commission, covering the U.S. (cars and trucks), Canada (cars and trucks), Australia, Japan, Switzerland and 13 EU countries, thus building an unbalanced panel of 279 observations. For Japan and some EU countries, we employed Chow tests to test for the existence of a structural break between two periods: one for the years up to 1995 (approximately the time of adoption of the first FE target values in both Japan and the EU), and one for the post-1995 ‘with standards’ period. For all those countries, the hypothesis of no break was clearly rejected. Therefore, we ran separate regressions for the ‘pre-standard’ and the ‘with standards’ sample using the above mentioned variables through pooled least squares with country fixed effects. In both samples, only one price variable was found to be statistically significant, that of lag 1. Estimated coefficients (i.e. ‘elasticities’) for the ‘with standards’ panel were approximately 0.7 for FE standards, -0.1 for price and -0.002 for the time trend and were all significant. Using the ‘pre-standard’ sample of 41 observations with lagged gasoline price and time trend as regressors, we estimated statistically significant coefficients of –0.3 and –0.007 respectively. Then we selected those countries for which both pre- and post-standard observations were available. Running the same regression for these countries and the whole period (pre- and post-standard), the price and time trend coefficients were almost the same as previously (–0.3 and –0.008 respectively). In all estimations, heteroskedasticity and serial correlation consistent standard errors were calculated. The results have significant policy implications: Firstly, they help to assess how much fuel prices should be raised in order to achieve future FE targets without resorting to higher FE standards. Secondly, they provide an indication about how FE might evolve without stricter standards. This is a very relevant issue as several European long-term energy/transport models assume that automobile FE will continue to improve at fast rates even without post-2010 FE regulations. Results show that without stricter FE standards and at fuel prices not higher than $50(in 2004 prices) per barrel, one could expect only minor FE improvements between 2010 and 2020. Still, the cross-section time series analysis shown here cannot help to draw conclusions on the cost-effectiveness and the welfare impact of alternative policy paths.