Vehicle Type Choice and Differentiated Registration Taxes

Vehicle Type Choice and Differentiated Registration Taxes


S Mabit, DTU Transport, DK


We analyse the tax reform of the vehicle registration tax in Denmark in 2007 to see to what extent the reform may explain the changes in purchase behaviour, e.g. the increase in fuel efficiency of new cars.


In 2007 a tax reform of vehicle registration in Denmark was introduced. During the year the fuel efficiency of new cars registered rose and there was an increase in the diesel share from 0.24 to 0.44. The average fuel efficiency increased from 15.9 km/l in the first trimester to 17.7 km/l in the third trimester. For petrol cars fuel efficiency rose from 15.1 km/l to 15.9 km/l and for diesel cars it rose from 18.4 km/l to 19.9 km/l. Statistics from previous years show that the increase in 2007 is remarkably large and not a result of seasonal variation.

In May 2007, the registration tax was differentiated based on the fuel efficiency of a new car. Petrol cars with fuel efficiency above 16 km/l became less expensive while those with fuel efficiency below 16 km/l became more expensive. For diesel cars the threshold was at 18 km/l. In this paper we develop a discrete choice model to capture vehicle purchase behaviour. Using our model we analyse the 2007 vehicle tax reform to see to what extent the reform may explain the changes in purchase behaviour.

We use a disaggregate model based on random utility maximisation to elicit preferences. In this way we follow the general framework used in, e.g. Manski and Sherman (1980), Berry, Levinsohn, and Pakes (2004) (BLP), and Train and Winston (2007).

We use a random sample with 10,000 individuals who registered a new vehicle in 2007. Each purchase is matched to one of 420 alternatives defined by make/model/fuel/class. We estimate an MNL model using all alternatives. We calibrate the alternative-specific constants using the contraction mapping introduced by BLP. As fuel prices rose through 2007 this could explain some of the changes. Therefore we control for the changes in fuel prices through 2007. While an MNL model may seem simple it is a general result in the literature on vehicle type demand that heterogeneous preferences are difficult to capture with cross-section and even panel data. We plan to expand the investigation using information on second choices. This has been found in the literature to be necessary to identify unobserved preference heterogeneity.

Investigation of the tax reform using the model indicates that the tax differentiation by fuel efficiency may only explain a change from 16.8 km/l to 17.0 km/l in fuel efficiency and an increase in the diesel share from 0.34 to 0.35 from the first trimester to the rest of the year. We are lead to conclude that while the tax reform appeared in the same year as a large increase in fuel efficiency it may only explain a small part of the shift in fuel efficiency that happened.

The contribution of the paper is that it applies innovative methodology to capture heterogeneity in preferences related to vehicle choice. Furthermore the application is relevant since fuel efficiency measures play an important role in many countries as a way to make the vehicle fleet more environmentally friendly.

Berry, S., Levinsohn, J., Pakes, A. (2004) Differentiated Products Demand Systems from a Combination of Micro and Macro Data: The New Car Market, Journal of Political Economy 112, pp. 68-105.

Manski, C., Sherman, L. (1980) An empirical analysis of household choice among motor vehicles, Transportation Research A 14, pp. 349-366.

Train, K., Winston, C. (2007) Vehicle Choice Behavior and the Declining Market Share of U.S. Automakers, International Economic Review 48(4), pp. 1469-1496.


Association for European Transport