In 2008, the French government introduced a policy taxing cars with high carbon emissions and rebating low carbon emission cars, better known as a feebate policy or bonus-malus écologique. This type of policy is appealing for two reasons: first, because it provides incentives to purchase less polluting cars, and secondly, because it can be designed to be revenue neutral since the revenue collected through the taxes subsidies the rebates.
In a recent paper, I conduct a quantitative evaluation of this policy, with a particular focus on its distributional effects: it is particularly relevant in this case to identify the winners and losers of the policy. I also analyse the effect of this policy, which is based on carbon emissions, on other local pollutants like particulate matter or nitrogen oxide. By nature, a policy that targets carbon emissions favours diesel cars, which consume higher levels of emissions of nitrogen oxide and particulate matter than petrol cars. Particulate matter and nitrogen oxide are known to have a direct impact on air quality and hazardous effects on health. While carbon emissions have a global impact, these local pollutants’ emissions raise the question of distributional impacts of the feebate policy in terms of health effects.
To measure these effects, I build a structural model of market equilibrium for the automobile industry. This implies estimating the supply and demand for the different car models using data on car characteristics and sales, which can then be used to simulate what the market would have looked like had there been no feebate policy in 2008. Comparing the observed market equilibrium with the counterfactual one, I can thus deduce the policy’s causal effect. Relying on a structural model is especially useful because some outcomes of interest cannot be observed directly, but can be expressed in terms of the model parameters. This is the case for car manufacturers’ profits and consumer surplus.
A notable challenge in modelling this market and being able to distinguish the winners and losers of this policy is to incorporate a large dimension of heterogeneity in individuals’ preferences for cars and their attributes. I make the assumption that individual heterogeneity in preferences is related to observable demographic characteristics, and leverage the correlation between composition of car sales and demographic characteristics at the municipality level. For instance, observing that the cars purchased in rural areas tend to be more fuel efficient than in urban areas reveals that individuals in rural areas tend to drive more, and are thus likely to be more sensitive to fuel costs than those living in urban areas. I also find a positive correlation between horsepower and income, which can be observed from the sales in wealthier municipalities.
On the supply side, I model the competition between car manufacturers and their pricing strategies, with and without the feebate policy. I do not model the choice of car characteristics and consider they are identical regardless of the regulatory environment. The marginal cost of each car model is estimated under the assumption that the car prices in 2008 are the optimal prices under the feebate policy. In the simulation of the market equilibrium absent the feebate policy, I predict prices and sales for each car model since both are jointly determined by demand and supply.
What is important here is that when setting its prices, the firm anticipates that consumers get a rebate or pay a tax and take up a part of the rebate or the tax. How much is left to the consumer depends on the competition of the market and the market power of car manufacturers.
In the end, the feebate policy improved consumer surplus and firms’ profits, surpassing the 223 million euros it cost in 2008. I find that the feebate caused a decrease in average carbon emissions of 1.56%, while average emissions of local pollutants – carbon monoxide, hydrocarbon, NOx, and PM – all increased. Emissions of local pollutants and carbon dioxide, however, increased once converted into annual tons. The increase in annual carbon emissions can be explained not only by the higher share of diesel cars, which implies more kilometers are driven, but also by the increase in the number of cars purchased. Indeed, the cars with low carbon emissions, which are already cheap cars, become even cheaper because of the rebates. This means that individuals who were not initially buying a car do buy a car, at least in my model. Nonetheless, including the cost of carbon and local pollutant emissions using standard levels still implies that the policy is globally welfare improving, with an estimated net benefit of 124 million euros.
Shifting the focus to the impact on income distribution, the main insight is that the feebate favoured the middle-income category at the expense of low and high-income classes. Moreover, given that the policy was not revenue neutral and contributed to a net deficit, the feebate could have been made redistributive if it were to be compensated by a proportional to income tax.
Clear winners and losers also appear among the car manufacturers. Car manufacturers are typically very specialised in different car segments: French manufacturers specialize in small, fuel-efficient cars, whereas bigger cars are the mainstay of the German car manufacturers. It comes as no surprise that the model points towards PSA and Renault, the two French manufacturers, as the winners of the feebate policy. The feebate policy increased their profits by 3.4% and 4% respectively, a considerably higher gain compared to increase in profits of the total industry (2.1%). Fiat group, the Italian manufacturer, increased its profits by 6.2% while Volkswagen, a German manufacturer very active on the compact car segment, only increased its profits by 0.3%. The other German manufacturers such as Porsche, BMW, and Mercedes-Daimler, were all severely hurt by this policy.
Finally, looking at the heterogeneity of the policy effects in terms of emissions of local pollutants, I find that average emissions increased the most in low emission municipalities. The policy generated a decrease in average emissions of local pollutant in some areas, but a high degree of heterogeneity can be observed across the country.
The analysis is concluded by an evaluation of the feebate in terms of redistribution and limitation of local pollutant emissions. The idea is to ask whether it would have been possible to improve consumer surplus, achieve more distribution across individuals, or limit the increase in emissions of local pollutants with the same budget and the same effect on average carbon emissions. In this exercise, I restrict the set of alternative policies to be simple linear feebates with different slopes for rebates and taxes. Interestingly, I find that average consumer surplus cannot be further improved, while there are large potential gains in terms of profits. Alternative feebate schemes could limit the rise in emissions of local pollutants, but the gains are not very large, and the best outcomes for the different pollutants cannot be achieved with a single feebate scheme: this reveals that there is an arbitrage to be made between the various pollutants.
by Isis Durrmeyer