Forecasting Adoption of Ultra-Low-Emission Vehicles Using the GHK Simulator and Bayes Estimates of a Multinomial Probit Model

In the European Union, transport is the largest consumer of oil products and second largest emitter of carbon dioxide (CO2); within the sector, road transport dominates in both regards. Consumer shift to ultra-low-emission vehicles has been regarded as a way to promote sustainable personal transportation. Whereas new low-emission technologies - including battery electric vehicles - have clear benefi ts such as efficiency gains and emission reductions, there are several barriers preventing broad adoption. On the one hand, electric vehicles are much more expensive than standard gas vehicles with a similar build. On the other hand, consumers face reliability issues, namely limited and variable driving range, and lack of refueling stations.

Using stated-preference data on vehicle choice from a Germany-wide survey of potential light-duty-vehicle buyers using computer-assisted personal interviewing, in this paper we analyze market shares of di fferent automotive technologies produced by a discrete choice model with flexible substitution among di fferent fuel types. Effectively, we propose a methodology to use the estimates of a probit model to produce both market-share forecasts as well as Bayesian con fidence intervals for the forecasted shares. These forecasts are simulated from the posterior distribution of a Bayesian model and account for uncertainty. Having better tools to address uncertainty is particularly relevant in the context of modeling consumer response to emerging energy-efficient technologies.

We define a base scenario of vehicle attributes that aims at representing an average of the current vehicle choice situation in Germany. Consumer response to qualitative changes in the base scenario is subsequently studied. Because limited fuel availability is a major obstacle to consumer adoption of low-emission vehicles, we analyze the specifi c e ffect of increasing the density of the network of service stations for charging electric vehicles as well as for refueling hydrogen-fueled vehicles. Our results indicate that if availability of charging is increased to its maximum, electric vehicles would experience a greater than three-fold increase in market penetration.

Daziano, Ricardo A. and Martin Achtnicht (2012), Forecasting Adoption of Ultra-Low-Emission Vehicles Using the GHK Simulator and Bayes Estimates of a Multinomial Probit Model, ZEW Discussion Paper No. 12-017, Mannheim.