Optimal Policy and Network Effects for the Deployment of Zero Emission Vehicles

Emissions from land transport are a major source of greenhouse gas emissions (approximately 24 % for the EU28). Urban pollution, in particular particulate matter (PM2.5) and ground level ozone, is causing 3 millions of premature deaths yearly. Battery and fuel cell electric vehicles (BEV and FCEV respectively) are thought to be attractive technologies to face these societal challenges. However, with the exception of Norway, the current market shares remain quite low for BEV and anecdotal for FCEV. The common explanations for this slow penetration are: the high price of electric cars, their limited range and the lack of filling infrastructure. The last two issues explains “range anxiety”: the fear of running out of power.
We formalize the interaction between three major factors that drive the deployment of zero emission vehicles: indirect network effects for adopters (i.e. range anxiety), scale effects to reduce the cost at the production stage (learning-by-doing and spillovers), and the degree of competition in the market with its influence on the price of cars. These three factors are embedded into a static partial equilibrium model. Consumers derive utility from transportation and incur a utility loss from filling. Con

sumers pay for the car, the fuel but not for the stations themselves. The benefit derived from the size of the network of stations is unpriced. Each firm production cost depends on the aggregate car production through a scale effect. Car producers compete à la Cournot. Filling stations are price-takers on the fuel retail market, and each has a limited capacity (convex cost). Our analysis explores various stages of deployment: take-off, building-up and expansion stages. From an economic standpoint our three factors may be interpreted as three externalities, or market failures, which may induce a distorsion between the market equilibirum and the social optimum. The relative magnitude of the distorsion will depend on the stage of deployment. At the take-off stage we may have a degenerate market equilibrium with no cars while the social optimum would imply a positive deployment.

At the building-up stage there may be three market equilibria, the equilibrium with the largest deployment generating the highest welfare. The intermediate equilibrium is a tipping point. Indeed if the initial market position lies below the intermediate equilibrium it will converge to the lowest one while if it lies higher it will converge to the preferred equilibrium. At the expansion stage the distorsion is reduced and eventually disappears as the significance of externalities decrease with the size of the market. For each possible stage we investigate the joint optimal subsidies for infrastructure and car adopters (i.e. price rebates) so that the social optimum can be implemented as a market equilibrium, i.e decentralized through market forces. Our results are illustrated with data on hydrogen cars (FCEV) based on Creti et al. (2018).

According to our calibration, a subsidy of approximately 80 % of the fixed capital cost of a hydrogen retailing station and a rebate of approximately 10 % on the listed price of cars would be necessary at the take-off stage. If the market is stagnating with a low deployment, strong public-private initiatives involving temporary demonstration projects, may be needed to pass the tipping point. The total level of subsidies would significantly increase as the deployment builds-up to eventually vanish as market failures disappear. A side result of our analysis shows that if the regulator can only subsidize vehicles or infrastructure, but not both, the return in welfare terms and in the size of the car park is higher with the former policy instrument. Our static model has the advantage of providing analytical solutions and explicit guide- lines for policies. It provides a framework which fits ra

ther well with the observed deployment of electric vehicles in Norway as extensively discussed in (Figenbaum, 2016). However it should be considered as a first step to build more elaborate analytical models, in par- ticular dynamic ones. Indeed a dynamic model would be more appropriate to analyze the efficiency of the many instruments that have been put in place by authorities in different megalopolis in reaction to the growing concern with urban pollution. Ideally the interaction between the various technologies such as BEV and FCEV with these policies should also be introduced. This is the direction followed by Harrison and Thiel (2017). We think that our model provides a useful complement for the interpretation of such large complex models.