Fuel taxes versus car taxes to reduce fuel consumption

Laura Grigolon, Mathias Reynaert, Frank Verboven 10 January 2015



To reduce CO2 emissions governments often opt for car taxes that are based on fuel efficiency, measured in liters per 100km (or miles per gallon). Proponents of an upfront car tax argue that it would be more effective than a fuel tax paid at the pump because of an investment inefficiency – consumers may be short sighted, failing to take future fuel taxes into account when choosing a new vehicle. Put differently, an upfront car tax can immediately influence the decision to purchase a car with high fuel efficiency, whereas a fuel tax requires that consumers take into account future fuel costs.

There is a long awareness of this problem, at least since the work of Hausman (1979) who found that consumers undervalue the future energy costs for air conditioners. For the US car market, Busse et al. (2012) find more or less correct valuation, but Allcott and Wozny (2014) find moderate undervaluation of future fuel cost savings. Based on this evidence there is a risk that fuel taxes only have a small impact on the purchase decision. At the same time, fuel taxes have a limited impact on the car usage decision; see for example Hughes et al. (2014) for evidence that mileage is quite inelastic.

In a recent paper we contribute to the debate on the effectiveness of car taxes versus fuel taxes (Grigolon et al. 2014). We stress the crucial importance of accounting for consumer heterogeneity in annual mileage. Mileage heterogeneity implies an important advantage of fuel taxes over car taxes: fuel taxes better target the right consumers, those with a high mileage, to purchase the most fuel efficient cars. This targeting effect makes fuel taxes more effective in reducing total fuel consumption than car taxes, even when consumers are moderately short sighted and mileage is perfectly inelastic.

We will explain our contribution to this debate in two steps:

  • We first ask whether there is an investment inefficiency – are consumers short sighted or do they correctly value their discounted future fuel costs when purchasing a new car?
  • We then ask what this implies for the relative effectiveness of fuel taxes versus car taxes, based on the cars’ fuel economy. We define the effectiveness of both taxes in terms of their reduction in total fuel usage.

Consumer valuation for future fuel costs

Using a rich dataset on seven European countries, we estimate consumer sensitivity to a car’s purchase price and the fuel costs. We can identify sensitivity to fuel costs thanks to variation in fuel prices across countries and over time, and observe further variation by two common types of fuel in Europe – gasoline and diesel. Table 1 shows that cars with a diesel engine are on average 30% more expensive than cars with a gasoline engine. But they also need about 20% less fuel to travel the same distance, and they drive on diesel fuel, which is on average 20% cheaper than gasoline fuel.1 Consumers thus face an investment trade-off between a higher initial purchase price and future fuel cost savings. This trade-off varies across vehicles, time, and countries, and to the extent that consumers are forward looking this will result in different market shares for diesel cars.

Table 1. Variation between diesel and gasoline engines

Using this identification approach, we find that car buyers are slightly less responsive to future fuel costs than to the initial purchase price of a car. For a one euro saving in the discounted future fuel costs, consumers are willing to pay € 0.87 in the form of a higher initial purchase price. European consumers do not take full account their future fuel costs.

Effectiveness of fuel and car taxes

Does this finding of moderate consumer myopia imply that an upfront car tax is more effective than a fuel tax in reducing fuel consumption? To address this question, we compare a fuel tax increase of € 0.5 per liter with a revenue-equivalent car tax, set at € 767 for each liter/100 km.2

We find the following effect on the sales-weighted average fuel efficiency of the vehicle fleet:

  • An increase in the fuel tax of € 0.5 per liter increases average fuel efficiency by 1.3%.
  • A revenue-equivalent car tax per liter/100km increases average fuel efficiency by 1.6%.

A fuel tax is therefore 20% less effective than a car tax in increasing the sales-weighted average fuel efficiency of the vehicle fleet. Fuel taxes are less effective in increasing fuel efficiency because consumers are not fully responsive to the increase in future fuel taxes.

However, it is more relevant to look at the change in total fuel usage (fuel efficiency times miles driven), since this determines the size of the externalities from CO2 emissions. We then obtain the opposite conclusion:

  • An increase in the fuel tax of € 0.5 per liter decreases total fuel usage by 1.8%
  • A revenue-equivalent car tax per liter/100km decreases total fuel usage by 1.6%.

Hence, even though a fuel tax is 20% less effective in increasing the average fuel efficiency of the fleet, it is 12% more effective in reducing total fuel usage. This is because fuel taxes better target the right consumers (those with a high mileage) to encourage the purchase of the most fuel efficient cars. This targeting effect makes fuel taxes more effective in reducing total fuel usage.

In sum, because of limited consumer myopia and substantial mileage heterogeneity, fuel taxes are more effective in reducing total fuel usage than car taxes. This shows that it is important to take heterogeneity in usage into account when designing tax policies to reduce CO2 emissions from automobiles. When proposing car tax or subsidy programs, policymakers appear to be overly concerned with improving the average fuel efficiency of the vehicle fleet rather than reducing total CO2 emissions.

Discriminatory taxes: Taxing only a single fuel type

To further explore the effectiveness of fuel taxes under mileage heterogeneity, we also consider the impact of discriminatory fuel taxes, i.e., separately raising the price of only diesel or gasoline fuel (rather than simultaneously raising both as we did before). This is of broad interest as governments often choose to subsidize specific vehicles with a certain fuel or engine type such as methanol, hybrid, or electric vehicles.

We simulate the effects of an increase in the fuel tax on gasoline, while the tax on diesel remains constant:

  • An increase in the gasoline fuel tax by € 0.5 per liter decreases total fuel usage by 3.1%.

This is surprising: although the gasoline tax applies to only about half of the cars that are sold in the market, the effect is almost twice as large as a tax on both gasoline and diesel fuel (1.8%, as seen before). The reason for this is twofold. First, the gasoline-only tax policy encourages consumers to buy the more fuel efficient diesel engines. The market share of diesel cars increases by 19.3 percentage points.  Second, the effects are again larger for consumers that have a higher mileage.

In sum, discriminatory taxes (or subsidies) towards certain fuel types result in large substitution effects, better achieving policy goals. In the case of gasoline and diesel fuel, European governments have indeed followed a discriminatory policy in favor of diesel cars. This helped to reduce CO2 emissions, though at the expense of other pollutants such as small particles and nitrogen oxide. With the emergence of alternative fuels, policymakers can keep in mind these lessons and design suitable discriminatory taxes towards high polluters.

Concluding remarks: car taxes or fuel taxes?

Our results show that EU consumers moderately undervalue future fuel costs. But this undervaluation is not enough to justify a strong emphasis on upfront car taxes. The car tax will result in a more fuel-efficient vehicle fleet than a fuel tax, but fails to target the high mileage consumers to drive with those more fuel efficient cars. Once we take this targeting effect into account, fuel taxes turn out to be a more effective instrument to reduce fuel usage (even if we assume that car usage is perfectly inelastic). When advising policymakers on the instruments to reduce externalities one should ask who is affected by the measure – will the heavy polluter change her choice?


Allcott, H and N Wozny (forthcoming), “Gasoline prices, fuel economy and the energy paradox”, Review of Economics and Statistics.

Busse, M, C Knittel, and F Zettelmeyer (2013), “Are consumers myopic? Evidence from new and used car purchases”, American Economic Review.

 Grigolon, L, M Reynaert, and F Verboven (2015), “Consumer valuation of fuel costs and the effectiveness of tax policy: Evidence from the European car market”, CEPR Discussion Paper 10301

Hausman, J (1979), “Individual discount rates and the purchase and utilization of energy-using durables, Bell Journal of Economics.

Hughes, J, C Knittel, and D Sperling (2008), “Evidence of a shift in the short-run price elasticity of gasoline demand”, The Energy Journal.


1 This is a European average that masks interesting differences. For example, in the UK diesel and gasoline fuel have similar prices, while the gap is large in Belgium, France and the Netherlands.

2 For example, this would amount to an upfront car tax of € 3835 for a car that uses 5 liters/100 km, and a car tax of € 7670 for a car that uses 10 liters/100km



Topics:  Energy Environment

Tags:  fuel efficiency, emissions, pollution, subsidies, fossil fuels, energy subsidy, carbon

Assistant Professor of Economics, McMaster University

PhD candidate, University of Leuven and University of Antwerp

Professor of Economics at the Catholic University of Leuven and CEPR Research Fellow


CEPR Policy Research