مالیات بهینه از مالکیت اتومبیل، استفاده از خودرو و حمل و نقل عمومی: بینش حاصل از انتخاب مدل بهینه سازی عددی گسسته

We formulated and numerically evaluated a model of car ownership, car use and public transport use for peak and off-peak hours of the day. The model was used to study the optimal tax structure for passenger transport in Belgium, with special emphasis on the optimal tax treatment of diesel versus gasoline cars. We obtained a number of interesting results. First, if the government can set all fixed and variable transport taxes optimally, the higher marginal external cost of diesel use implies that the optimal tax per kilometre for the use of a diesel car is higher than for the use of a gasoline car. Moreover, high congestion implies that the taxes on car use in the peak period are more than twice their current levels. However, the optimal tax on ownership of a diesel car is some 200€ below its current level. Second, if the government uses kilometre taxes that do not differentiate between fuel types, the optimal ownership tax on a diesel car is twice as high as the tax on a gasoline car. Furthermore, if political constraints restrict user taxes to their current levels, we find that optimal ownership taxes on diesel cars double, whereas those on gasoline cars rise by 30%. Finally, subsidies to public transport are found to be optimal as long as variable car taxes are not differentiated between periods.

The social costs associated with transport externalities such as congestion, pollution and accidents are known to be substantial. Recent estimates suggest that, in highly congested European cities such as Amsterdam and Brussels, the marginal external cost of car use amounts to more than 1€/km (De Borger and Proost, 2001). Consequently, policies to reduce time losses and other external transport costs are high on the political agenda in many countries. Economists have strongly emphasized the need to use pricing instruments, such as road pricing, to correct transport externalities. Not surprisingly, a large theoretical and empirical literature on the optimal tax treatment of transport externalities has developed over the past decades.1 With very few exceptions, the available studies ignore the two-part tax structure on vehicles that is typical of real-world tax systems. It is well known that most European governments tax car ownership (e.g., through annual vehicle taxes) and car use (e.g., through fuel taxes, road tolls, etc.) separately. The purpose of the current paper is, therefore, to explicitly consider the optimal tax treatment of car ownership, car use and public transport use in the presence of externalities. The analysis was motivated by two observations. First, policy-makers face restrictions on the variable pricing instruments they can use to correct external costs of congestion and pollution; this implies a potential role for fixed ownership taxes to cope with externalities (see e.g., Chia et al., 2001; De Borger, 2001; Fullerton and West, 2000). To appreciate this, assume that it would be feasible to implement ‘Pigovian’ taxes that perfectly tax users of transport services at the relevant marginal external cost. The optimal tax literature then convincingly suggests that there would be no explicit role for annual ownership taxes in correcting externalities.2 Unfortunately, application of Pigovian taxes would require measuring the external costs of congestion, emissions and noise of various car types, at each point in time and space, and to charge users accordingly. Such a sophisticated system of full electronic road pricing on the complete transport network is not feasible in the short-run. Moreover, it is unclear whether road pricing will be politically and socially acceptable (Schade and Schlag, 2003). Therefore, imperfect variable tax instruments, such as fuel taxes or kilometre taxes, will have to be used for the foreseeable future. The important point is that, as long as perfect road pricing is not in place, fixed annual taxes on car ownership may play a useful role in correcting externalities. For example, when kilometre taxes are used, fixed ownership taxes may be useful in correcting externality differences between diesel and gasoline cars. This brings us to the second motivation for this paper, viz., that explicit consideration of ownership taxes is necessary to empirically evaluate the tax treatment of cars that use different types of fuel. This issue is currently being widely debated in a number of European countries. Consider the situation in Belgium as an example. External cost estimates for 2000 (see De Nocker et al., 2001) suggest that the marginal air pollution cost of an average diesel car was, per kilometre, about 2.8 times higher than that of a gasoline car. The current tax structure implies that fuel taxes are substantially higher on gasoline than on diesel, but ownership taxes on diesel cars by far exceed those on cars using gasoline.3 This raises a number of relevant questions. How does the observed tax structure compare with an optimal tax system? Can the current tax structure be justified if the government is restricted to the use of fuel taxes as the main variable tax instrument? When does it make sense to differentiate fixed ownership taxes to correct externality differences between diesel and gasoline cars? Is such a differentiation appropriate if fuel taxes are used, or if kilometre taxes are used? How do car taxes interact with public transport fares for various policy instruments? The contribution of this paper is twofold. First, we developed a detailed numerical optimization model that incorporates consumers’ car ownership decisions and their demands for kilometres by car and by public transport in different periods of the day. The model assumes that consumers have a choice between different car types; alternatively, they can decide not to own a car. All consumers have access to public transport. When designing optimal taxes, the government cares about consumer welfare, external transport costs and tax revenues. Externalities taken into account included congestion, the air pollution associated with 11 pollutants, and external accident costs for three types of accidents. Then, using the best available evidence on marginal external costs, we analysed the optimal tax structure in Belgium with special emphasis on the optimal treatment on the ownership and use of cars that operate on diesel and gasoline. Introducing various sets of restrictions on the transport tax instruments allowed us to mimic optimal kilometre taxes, optimal fuel taxes, the optimal tax structure for the current set of tax instruments, and so on. We found a number of interesting results. First, assuming the government has perfect instruments, the current differences in tax treatment between diesel and gasoline cars cannot be justified by considerations of external costs and budgetary constraints. Owing to higher pollution costs for diesel, the user tax for a diesel car actually exceeds that for a gasoline car. Optimal ownership taxes are found to be substantially below current levels, and they hardly differ between the two car types. To cope with congestion, the optimal tax per kilometre, for both fuel types, would be more than twice their current levels in the peak period. Second, if the government uses optimal kilometre taxes that do not differentiate between fuel types, the higher external cost of diesel use results in an optimal ownership tax on a diesel car that is approximately twice as large as that on a gasoline car. Third, if political constraints restrict user taxes to their current levels, we find that optimal ownership taxes on diesel cars double, whereas the tax on gasoline cars rises by some 30%. We also found that public transport fares vary widely across scenarios: the current high level of subsidies generally declines, but subsidies do remain as long as variable car taxes are not time-differentiated. Finally, consistent with earlier work, we found that the largest welfare effects can be obtained by time-differentiated variable taxes that allow adjusting taxes to variations in congestion. A few recent papers have studied issues related to those analysed in this paper. First, Chia et al. (2001) also studied the relative merits of fixed and variable transport taxes. However, they only considered one car type and they did not study restrictions on tax instruments. Moreover, unlike our model, they made the assumption, inappropriate to the European setting, that car owners travel exclusively by car. Second, Mayeres and Proost (2001) numerically investigated the implications of changing the relative ownership taxes on gasoline and diesel cars. However, we focused on optimal tax structures and looked at a richer set of transport tax instruments. Finally, Verhoef and Rouwendal (2004) did explicitly incorporate car ownership decisions in a model of pricing and investment on simple road networks. However, they did not deal with the tax treatment of different car types and the interaction with public transport availability. The structure of the paper is as follows. In Section 2, we specify the discrete/continuous choice model and summarize the optimal tax rules derived from this model under highly stylized assumptions. In Section 3, we discuss the data underlying the numerical analysis and describe in detail the characteristics of the reference situation. In Section 4, we discuss optimal tax results for nine different scenarios that reflect various constraints on the pricing instruments. Finally, Section 5 concludes.

In this paper we presented a discrete/continuous choice model to study the optimal tax treatment of car use, car ownership and public transport services in the presence of externalities. The model was applied to Belgian data for analysing optimal taxes on cars using different fuel types, viz. diesel and gasoline. Of course, perfect taxation of congestion and pollution requires user tax instruments that allow differentiation according to traffic levels and emissions. As such instruments may not be available for the years to come, a large number of scenarios were analysed that imposed particular restrictions on the available tax instruments; these included the use of fuel taxes or kilometre taxes as the main variable tax instrument. When such restrictions exist, we analysed whether there is a role for optimally differentiated annual ownership taxes on cars in correcting for externalities. We obtained a number of interesting empirical results.19 First, we found that the current tax structure for diesel and gasoline cars, i.e., lower variable taxes on diesel use combined with higher annual fixed taxes on diesel car ownership, cannot be justified on the basis of external cost considerations. Assuming that taxes on car use can be optimally differentiated between fuel types and between peak and off-peak periods, we find that the tax on car use in the peak period would steeply rise as compared to the current levels. This, of course, serves to capture high congestion costs. Interestingly, the user tax per kilometre on diesel should exceed the tax on gasoline; the opposite is true in the current tax structure. Moreover, annual taxes on car ownership are too high: optimal taxes are lower than current levels and, again unlike the current tax structure, they hardly differ between car types. The optimal tax structure further suggests that public transport operating subsidies should be eliminated and replaced by taxes. Second, we find that differentiation in annual fixed ownership taxes on cars may be useful in capturing externality differences between car types if, for political or technical reasons, variable user taxes cannot be optimally adjusted. This will be the case, for example, if policy makers are restricted to the use of fuel taxes as the main variable tax instrument. If the government uses the set of instruments it currently employs (fuel taxes, ownership taxes) but chooses these tax levels in an optimal way, then the current fuel tax on gasoline is approximately optimal, but the tax on diesel should increase substantially. Ownership taxes on gasoline cars rise by some 200€, whereas the tax on diesel cars is close to optimal. Public transport would be subsidized, although at lower rates than in the reference situation. As another example, if the government is unwilling to adjust variable car taxes at all, then very large increases in annual ownership taxes are required to cope with congestion and pollution externalities: the tax on diesel cars approximately doubles and that on gasoline cars rises by some 30%. Importantly, we found that annual ownership taxes also play a useful role if the government uses a system of kilometre taxes as the main variable tax instrument; these do not allow differentiation according to fuel type. Whereas perfect variable tax instruments (and hence differentiation between fuel types) yielded fixed taxes that hardly differed between fuel types, this is not the case if time-differentiated kilometre taxes are used. The inability to differentiate variable taxes between diesel and gasoline implies that, although both fixed taxes decline relative to the reference case, diesel taxes are approximately twice those on gasoline cars to bring the share of diesel cars closer to the optimal one. In other words, the use of kilometre taxes instead of full blown road pricing has large implications for the level of optimal fixed taxes, an observation not previously made in the literature. A third finding of this paper is that, although we showed the usefulness of optimal differentiation in taxes on car ownership to complement imperfect user tax instruments such as kilometre taxes or fuel taxes, the welfare gains of these second-best policies are rather limited: they attain at most some 20% of the welfare gain in the first-best. Consistent with earlier literature, we found that larger welfare improvements can only be obtained when the government is willing to differentiate variable car taxes between periods so as to capture the large differences in congestion between peak and off-peak traffic levels.