Several recent papers have studied the optimal taxation of car ownership and car use. These models (see, e.g., De Borger (RSUE 2001), Chia, Tsui and Whalley (NBER 2001), Fullerton and West (JEEM 2002), De Borger and Mayeres (EER, 2007)) are typically based on the assumption that households either prefer to have no car or, in case of ownership, that at most one car is owned by each household.

The purpose of the current paper is to use a recently developed model of car ownership and car use (Rouwendal 2008), allowing for multiple car ownership, to study the optimal tax treatment of car ownership and car use. The model assumes households have the choice between owning no car, owning one car or owning two cars (generalization to more cars is conceptually straightforward). For each car there is an annual cost of ownership and, in case of car use, a variable cost per kilometer. Cars differ in terms of fixed and variable cost: for example, the first car households own is typically of higher quality than possible second cars.  

Conditional on ownership, the model implies linear demand equations for kilometers per car and allows for substitution effects. If the two cars use different fuel types, an increase in the price of one fuel type will lead to substitution towards the other car. If two cars use the same fuel type but differ in efficiency, the substitution will be towards the more fuel efficient car. Moreover, two-car households may decide to abandon one car when the fuel price increases, and a single-car household may decide to switch to a more fuel efficient type, or to become no-car households. The model is therefore able to cover a rich set of behavioral responses to fuel price changes on ownership (number and type of car) as well as use of cars. A change in car ownership will usually involve a discontinuity in household demand for car kilometers.

In this paper we study the effect of car taxes. The government levies both variable taxes (e.g., fuel or kilometer taxes) and fixed ownership taxes. Within this setting, the model first studies how the different taxes affect the decision to own zero, one or two cars, how it affects the demand for kilometers driven using the individual cars, and how it affects car quality. A crucial ingredient of the model is that multiple car ownership introduces nonlinearities in the demand for total household kilometers. Next, the model develops welfare optimal tax rules for car ownership and car use to study some specific policy questions. For example, are there conditions such that the first-best tax structure involve fixed ownership taxes? In a second best setting, will it be welfare optimal to tax ownership of first and second cars differentially for pure efficiency reasons, i.e., even in the absence of distributive concerns? What role do discontinuities in demand for kilometers play in the optimal tax structure?