International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
French households’ preferences for alternative fuel vehicles: A discrete choice analysis of Stated Preference data
Akli Berri, Stefan Mabit

Last modified: 28 March 2017


The challenge of transforming the structure of the car fleet in favour of energy-efficient andclean vehicles (e.g. electric, hybrid and bio-fuel cars) is not merely a matter of technicalfeasibility. Equally important is the requirement of economic viability of a technology.Despite an increasing number of models put on the market, purchases of electric vehicles(EVs) and hybrid vehicles remain very low in comparison to the overall new car registrationsin most European countries (CCFA, 2016, p. 75).1 In the case of France, even if the salesincreased over the last few years, the market shares attained only 0.9% for EVs and 3.2% forhybrids (among a total of 1,917,226 new cars) in 2015.2 The sales figures for the remainingalternative fuel vehicle (AFV) types are also still negligible: 129 super-ethanol cars, 38 NGVcars and 1467 bi-fuelled cars (i.e. gasoline + LPG or NGV) during the first 9 months of 2015(CCFA, 2015, p. 33).A wide-scale diffusion of AFVs presupposes acceptance and willingness to adopt by asufficiently large proportion of the consumers. New car buyers constitute the main populationsegment through which this diffusion could materialise. Knowledge of their preferences,opinions and attitudes is essential to elucidate the reluctance to switch to AFVs and drawlessons as to possible levers of action to encourage such behavioural change.We investigate the preferences of households (who own the bulk of the car fleet) in Franceregarding the choice of a new car in view of its technical characteristics and performance,purchase and use costs, and practical aspects of use (e.g. availability of refuelling/chargingstations).The data are from a Stated Preference survey carried out in the second half of 2012. Thesample was drawn from the new car registrations database, based on the energy source of thecar. Thus, the sampled individuals still had in memory the purchase context and the importantfactors that they had to consider in determining their choice. Each respondent was presented12 choice situations comparing each time two alternatives. Five vehicle types wereconsidered: conventional, bio-fuel, hydrogen, electric, and hybrid. Each alternative wasdescribed by a set of attributes: purchase price, cost of fuel/energy per 100 km, density ofrefuelling/charging stations, range, engine power, level of CO2 emissions (g/km), and amount1 Two notable exceptions are Norway (17.1% of new car registrations for EVs and 10.4% for hybrids in 2015)and the Netherlands (12.5% for hybrids but only 0.7% for EVs in 2015).2 EV purchases were 184 in 2010, 2630 in 2011, 5663 in 2012, 8779 in 2013, 10561 in 2014 and 17268 in 2015.The figures for hybrids were, respectively, 9655, 13635, 27889, 46745, 43143 and 61619.2of ecological bonus or penalty. For EVs, two additional characteristics were given: durationof complete recharging of battery at home, and whether the car has a range extender(allowing 50 or 100 km). The attribute levels were pivoted around those of the car bought.This introduces more realism by ensuring that the alternatives proposed to the respondent arenot too different from his/her recent purchase experience. Concerning the characteristics notmentioned, the respondent was asked to consider them as identical to those of the car he/sherecently bought. In addition to car preferences, the survey collected information on, amongothers, the respondent and his/her household, the new car bought and the other car consideredbefore final choice (if any), opinions on car technologies, environmental problems in relationto the car, and attitudes towards the environment. Over 4,300 interviews were completed.The stated choices are analysed by means of Multinomial Logit and Mixed Logit models toestimate the effects on the probability of choosing a vehicle type of, notably, its technicalcharacteristics and performance (e.g., engine power, range, CO2 emission), the density of thenetwork of refuelling/charging stations (when relevant), purchase and use costs, fiscalincentives, and socio-demographic factors (e.g., income, household composition, residentiallocation, education level). The aim is to identify favourable factors for and hindrances to theadoption of AFVs, to assess households' willingness to pay for these vehicles and to estimatetheir market potential under different scenarios about attribute levels.The first modelling results show, in particular, the high importance attached to a sufficientrange. This appears through the effects of the range level, of the density of the network offilling/charging stations, and of the availability of a range extender (RE) for EVs. The resultsalso show a strong penalizing impact of high polluting emissions (CO2) on the choice of avehicle. The effects of the ecological bonuses and penalties, linked to the CO2 emissionlevels, are not symmetric: the dissuasive effect of the penalty is stronger than theincentivizing effect of the bonus. Finally, the fact that the car actually bought by therespondent was a non-conventional one (whether LPG, electric, hybrid or super-ethanol E85)increases the probability of choice for all the AFV types considered in the SP games.Focusing attention on electric vehicles, anxiety about range appears as a major hindrance totheir adoption. Apart from a sufficiently high battery performance, this could be alleviated bythe availability of a RE (even more if it allowed an additional distance of 100 km instead of50 km) and a fairly dense network of public charging stations. A further barrier to theadoption of EVs is their relatively high purchase price. Respondents’ actual new car purchasedecisions (made during the few months preceding the survey) show that in most cases wherethey considered purchasing an EV but ultimately abandoned it in favour of another vehicletype (mainly a conventional vehicle), the EV was significantly more expensive than the carthat was finally bought (a price difference of at least 10%). Also hindering is the negativeperception of EVs as to their reliability and/or safety by part of the respondents (among thereasons frequently put forward to motivate systematic rejection). Finally, owning only onecar lowers the chances of a shift to an EV, in view of its (perceived) limited range and thediversity of trips made by car (including long distance trips for holidays). The EV seems tobe perceived rather as a second car. Indeed, all but one of the (very rare) respondents whobought an EV owned at least two cars. Furthermore, when an EV was considered and thenabandoned, the car finally bought was in a sizeable proportion of cases the only car owned.


CCFA (2016). L'industrie automobile française. Analyse et statistiques ─ 2016. Comité desConstructeurs Français d’Automobiles (CCFA), Paris.CCFA (2015). Tableau de bord automobile – 3ème trimestre 2015 (No. 44). Comité desConstructeurs Français d’Automobiles (CCFA), Paris.

Conference registration is required in order to view papers.