Advertisement

Electric, plug-in hybrid, hybrid, or conventional? Polish consumers’ preferences for electric vehicles

  • Milan Ščasný
  • Iva Zvěřinová
  • Mikołaj Czajkowski
Original Article
  • 72 Downloads
Part of the following topical collections:
  1. Energy and Climate Economic Modelling

Abstract

Poland aims at stimulating the market to reach a target of 50,000 plug-in and battery electric vehicles by 2020. However, as in other Eastern European countries, the market penetration stays very low. In Poland, there were only 475 battery electric vehicles and 514 plug-in electric vehicles registered in 2017. To identify effective support measures, this paper examines the preferences of Polish consumers for three types of electric vehicles: battery, hybrid, and plug-in hybrid vehicles. We use a discrete choice experiment to estimate the willingness to pay of a representative sample of consumers intending to buy a car in Poland. We find that electric vehicles are significantly less preferred than conventional cars, even under a public programme that would enable slow-mode charging in places where respondents usually park. We quantify the marginal willingness to pay for increasing the driving range, reductions in charging time, the availability of fast-mode charging stations, and the provision of policy incentives. The novelty of the paper lies in presenting a scenario with the slow-mode and availability of several levels of fast-mode charging stations and examination of the extent to which the heterogeneity of consumer preferences is driven by place of residence (urban, suburban, rural), intention to buy a new versus a used car, and the annual mileage. This is also the first discrete choice experiment on electric vehicles conducted in Eastern Europe. To stimulate the electric vehicle market, we recommend a pricing policy that affects the operating costs and other incentives along with an effective up-front price incentive scheme.

Keywords

Battery electric vehicles Hybrid vehicles Discrete choice experiments Willingness to pay Driving range Fast-mode charging infrastructure Recharging time Incentives 

Abbreviations

BEV

Battery electric vehicle, a vehicle set in motion by an electric motor. Powered by electricity, it has a battery which can be recharged from a regular electric socket.

PHEV

Plug-in hybrid vehicle, a vehicle with an internal combustion engine (petrol or diesel) and batteries that can also be charged from a regular electric socket. The car can drive several tens of kilometres solely on electricity. When the batteries are empty, the car will automatically switch to the internal combustion engine.

HEV

Hybrid vehicle, a vehicle with batteries but without a plug. It has both an internal combustion engine and an electric engine. The combination allows the electric motor and batteries to help the conventional engine operate more efficiently, reducing fuel use. Switching between the two engines occurs automatically without the driver’s intervention. The battery is charged from the energy produced by the combustion engine during driving or while braking. A hybrid car drives several kilometres solely on electricity.

EV

Electric vehicle, includes BEV, PHEV, and HEV

CV

Conventional vehicle, drives on an internal combustion engine that can be fuelled by petrol, diesel, or oil derivatives such as LPG.

Introduction

Electromobility is part of the European Union strategy to reduce energy dependence, improve air quality, reduce noise in urban/suburban agglomerations, and contribute to CO2 emissions reduction (Directive 2014/94/EU; EC 2014a). Accordingly, electric vehicles should be integrated into a smart grid to contribute to the stability of the electricity grid by recharging batteries in periods of low demand. In the more distant future, these vehicles could even feed power from their batteries back into the grid in periods of high demand (The 2016 European Strategy for Low-Emission Mobility). In addition, the transition to low-carbon transport is supported by EU Regulation 443/2009, amended by EC (2014b) which has established emission performance standards for new passenger cars at 95 g of CO2 per km as of 2020.

To fulfil these targets, Directive 2014/94/EU requires each EU member state to adopt a national policy framework for developing the alternative fuel market and related infrastructure, including policy measures to support the construction of infrastructure for alternative fuel vehicles (AFVs). In order to prepare a national policy framework and effective support measures, an understanding of consumer preferences for AFVs and their characteristics is important. Although consumer demand for specific vehicle technologies can be modelled using market penetration data or consumer decisions observed in the actual markets (revealed preferences), in the case of constrained supply of certain goods or negligible market penetration of certain technologies, these approaches are less useful. The alternative is to apply stated preference (SP) methods, such as the discrete choice experiment (DCE) presented in this paper (Carson and Czajkowski 2014).

A number of studies conducted in the 1990s applied SP methods to analyse consumer demand for various characteristics of passenger vehicles (Bunch et al. 1993; Kurani et al. 1996; Golob et al. 1997; Brownstone and Train 1998). In recent years, SP methods have also been used to study preferences for AFV, particularly in the USA, Canada, Asia, and Western Europe (see Liao et al. 2017 for a comprehensive review).

However, there are currently no studies available for Eastern Europe, which differs from the West with respect to the development of the alternative fuel vehicles market, economic conditions, and culture. Understanding the barriers to and factors supporting the adoption of AFVs in this region is of particular importance, as consumers in most EU member states in Eastern Europe, and partly in Southern Europe, have been purchasing far fewer electric vehicles than their counterparts in western and northern member states. According to the European Alternative Fuel Observatory (EAFO 2017), battery electric passenger vehicles represented 1–2% of new registrations in Austria, France, the Netherlands, and Sweden (21% in Norway), with the EU average of about 0.64% in 2017. In contrast, the share of battery electric vehicles on new registrations was only about 0.1–0.2% in the Czech Republic, Estonia, Italy, Lithuania, Poland, Romania, and Slovakia, with even less in Bulgaria and Greece. Similarly, the share of plug-in hybrid vehicles has been much lower in these countries (about 0.09–0.2%, with only 0.11% in Poland) than in western and northern European countries (0.5% in Austria to 4.2% in Sweden, and 19% in Norway in 2017).

This study aims to shed more light on consumer preferences for electric (BEV), plug-in hybrid (PHEV), and hybrid vehicles (HEV) relative to conventional vehicles (CV) in Poland. We select Poland for our study due to the low deployment of these specific technologies in the country. Although the number of BEVs and PHEVs has been growing, there were only 475 BEVs and 514 PHEVs registered in 2017 (compared to 138 and 132 in 2016), resulting in a very low market share of 0.10% and 0.11%, respectively (EAFO 2018).

Our study also contributes to the ongoing policy debate concerning clean mobility in Poland. According to the 2016 Polish National Policy Framework, the government intends to stimulate the AFV market with the aim of increasing the number of BEV and PHEV to 50,000 by 2020, and to 1 million by 2050. The recent Act on Electromobility and Alternative Fuels signed by the President of Poland on 5 February 2018 provides the framework for the construction of public vehicle-charging infrastructure by 2020 and introduces new incentives such as parking fee exemptions, excise tax exemptions for electric cars, and clean transport zones.

To investigate the factors that drive vehicle purchase decisions, including the effectiveness of selected policy incentives, we use the discrete choice experiments (DCE) and elicit Polish consumers’ preferences for the three alternative fuel technologies relative to conventional vehicles. The attributes include vehicle purchase price, expected operation and maintenance cost, driving range, refuelling/recharging time, and availability of fast-mode recharging infrastructure. We also examine the effect of two policy incentives to support the purchase of alternative-fuelled vehicles: free parking and free use of public transport. We elicit preferences from a representative sample of consumers who intend to buy a passenger car. Given the importance of the second-hand vehicle market in Poland, we elicit preferences from both segments of consumers: those intending to buy a new car and those planning to buy a used car.

The preferences were elicited under a hypothetical scenario in which a public programme would be introduced allowing for slow-mode charging of BEVs or PHEVs in places where respondents usually park, even if they do not own a garage. We find that even under these favourable conditions, BEVs are still less preferred, on average, than conventional vehicles. In line with other studies, we find that the technical attributes of EVs matter. In particular, the driving range is an important attribute of passenger cars for Polish consumers. They are willing to pay approximately EUR 257 more for a conventional vehicle with an additional driving range of 100 km. This value is two times larger for PHEVs (EUR 506) and even four times larger for BEVs (EUR 1067). Recharging time and the availability of fast-mode charging stations are other influential barriers to increasing the market share of electric vehicles. On average, the willingness to pay (WTP) for a one-hour reduction in charging time was EUR 311, and marginal WTP for the “high” availability of fast-mode charging infrastructure was EUR 2700. Providing other benefits increases the probability of buying BEVs, but their effect is not sufficient to outweigh the importance of technical attributes. The heterogeneity in preferences is considerable and can be partly attributed to consumers’ place of residence (urban, suburban, or rural area), expected mileage, and the intention to buy a used versus a new car.

The remainder of this paper is structured as follows. “Literature review” provides a review of relevant studies. “Methods and data” describes the methods of our stated preference study and characterises the sample. “Estimation results” presents the results—the main tendencies and the analysis of preference heterogeneity. “Conclusions and policy recommendations” concludes with a summary of findings and provides policy recommendations.

Literature review

The first discrete choice experiments to elicit the stated preferences of consumers for clean-fuel vehicles were undertaken in the early 1990s (Bunch et al. 1993; Kurani et al. 1996; Golob et al. 1997; Brownstone and Train 1998).1 Although the number of these studies has grown rapidly over the last ten years, most have been conducted primarily in North America and Western Europe, with several studies in China, Japan, and South Korea (see a review by Liao et al. 2017).

In the literature, discrete choice experiments usually consist of interviewing recent or potential car buyers. Hoen and Koetse (2012) interview the members of households who drove a car most frequently, while Dagsvik et al. (2002) and Lebeau et al. (2012) target the general public. Golob et al. (1997) and Chorus et al. (2013) elicit preferences for company cars. Hoen and Koetse (2014) and Jensen et al. (2013) also interviewed people who chose used cars. Since second-hand cars dominate the Polish market (PZPM 2017), we paid special attention also to sampling people who intend to buy a used car and elicited preferences only from those who intend to buy a car in the near future, regardless of whether they owned or could use any car during the time of the survey.

In this review, we focus on electric passenger vehicles for which preferences have been elicited in previous studies along with preferences for conventional technology represented by a vehicle fuelled by diesel and/or petrol. These vehicles are described by several attributes, including monetary, technical, infrastructure, and policy (incentive) attributes.

All previous studies described each alternative by the purchase price that was often defined by a pivoted design, or they simply used the respondent-specific mark-up price of the intended car. Each technology was also described by operating and maintenance costs related to either per 100 km or gallon, a certain time period, distance travelled, or the costs of a conventional car. We also follow this strategy in our study and define both monetary attributes (i.e. purchase price and operating/maintenance costs) as respondent-specific to ensure our contingent scenario is credible and realistic. Both of these attributes negatively affect the decision to purchase a car, and this has been mainly explored as a linear relationship (with the exception of Ziegler 2012). People with higher incomes typically assign lower importance to fuel cost (Liao et al. 2017), but Helveston et al. (2015) find that more affluent Chinese are more sensitive to high fuel costs.

A relatively short driving range and long battery charging time are considered some of the most significant limitations of electric vehicles hampering their market adoption (Dimitropoulos et al. 2013). In particular, Dagsvik et al. (2002) argue that electric vehicles will not become fully competitive unless the limited driving range increases substantially. A meta-analysis by Dimitropoulos et al. (2013) found the distribution of WTP estimates for a 1-mile increase is positively skewed with a mean of $67 and a median of $55 (2005 US$). Studies assuming logarithmic utility function yield lower estimates than the studies assuming a linear specification, supporting concave WTP function increasing with driving range. WTP for an increase in range is larger for BEVs relative to PHEVs. Larger preference heterogeneity is found, especially for smaller ranges, up to 100 km (Rasouli and Timmermans 2016), whereas preference heterogeneity seems to vanish with higher range values, particularly if ranges reach 400 km. Although BEVs with relatively smaller ranges have recently dominated in the Polish market, we also include higher ranges to elicit preferences for more advanced technologies that are likely to enter this market.

In principle, two recharging modes can be distinguished: slow charging, usually at home or at work (6–8 h), and fast recharging that typically takes up to an hour. However, only Golob et al. (1997) specifically distinguish these two recharging modes in their choice experiments. Some studies make this attribute specific to PHEVs and BEVs (Hoen and Koetse 2014), as we have also done in our study but only addressing fast recharging. Battery charging time is found to be twice as important in the case of BEVs compared to PHEVs (Hackbarth and Madlener 2013), and larger heterogeneity is typical for shorter recharging time levels (Rasouli and Timmermans 2016).

To some extent, the limited driving range of electric vehicles can be counterbalanced by a sufficiently widespread charging infrastructure. As charging infrastructure is undoubtedly important when attempting to enhance the uptake of electric vehicles, many studies investigate the effect of this attribute, although relying on different specifications. Most studies use the percentage share of all service stations equipped with charging stations or relate their number to gasoline stations. Others rely on the distance from home to the closest charging station (Kim et al. 2014; Valeri and Cherchi 2016), the detour to purchase alternative fuel (Caulfield et al. 2010; Chorus et al. 2013), or the presence of a charging station in different areas, e.g. at home, at work, or in shopping malls, or they simply define charging availability in general as low, medium, or high (see Liao et al. 2017). In most studies, a higher availability of charging infrastructure has a significant positive effect on the choice of electric cars. However, Valeri and Danielis (2015) find the refuelling distance attribute insignificant, and others find a non-linear relationship indicating diminishing marginal utility (Achtnicht 2012; Bunch et al. 1993; Hackbarth and Madlener 2016). A limitation of the reviewed studies may lie in the fact that most do not differentiate between slow-charging posts and fast-charging stations, which are used for different purposes. In particular, Hackbarth and Madlener (2016) note that individuals would accept considerable mark-ups on the electricity price for large-scale fast-charging infrastructure. In this study, we therefore introduce a fast-mode charging infrastructure with three levels defined by different coverage of the charging stations at fuelling stations and at various public locations.

Various policy instruments have been proposed to promote higher market adoption of alternative fuel vehicles; hence, the acceptability of most of them has been examined in choice experiments. Incentive attributes include various forms of purchase price and usage cost reductions, charging station incentives, and non-financial incentives such as park-and-ride subscriptions, free public transport, free parking in cities, access to priority/bus lanes, or free (highway) tolls (Liao et al. 2017). Reduction or exemption of vehicle purchase/registration tax is found to be significant in all studies, whereas direct purchase price reduction is significant only in some (Hess et al. 2012). Policies reducing usage costs are all found to be significant, regardless of whether they favour “cleaner” vehicles or whether they place the burden on conventional vehicles.

Nevertheless, some studies point to the overall inefficiency of pricing strategies in significantly altering the relative market power of conventional and alternatively fuelled vehicles in favour of alternative technologies (Valeri and Danielis 2015). Axsen et al. (2009) suggest that using non-financial incentives may result in more efficient outcomes than financial strategies. Based on a systematic review of the literature, Hardman et al. (2017) concluded that the effective purchase incentives should support BEVs and PHEVs with high electric ranges and that VAT and purchase tax exemptions should be the most effective. All incentives need to be designed with longevity in mind. Incentives should not be provided for high-end BEVs, and education and awareness campaigns should advertise incentives to consumers.

It is important to examine the effect of introducing policies aimed at encouraging both public transport and the spread of electric vehicles, as synergies in such interconnected systems have been shown. Therefore, apart from commonly used non-financial incentives, such as free parking in cities, we also explore preferences for free public transport tickets for all family members.

However, Bahamonde-Birke and Hanappi (2016) find that free public transportation does not increase the probability of purchasing electric vehicles. Similarly, reducing tolls (Hess et al. 2012) or toll exemptions (Mersky et al. 2016) were insignificant to predict BEVs purchase in California or Norway, respectively. Free access to bus lanes also does not seem to motivate purchases of BEVs at all locales (Chorus et al. 2013; Hess et al. 2012; Hoen and Koetse 2014; Mersky et al. 2016; Potoglou and Kanaroglou 2007; Qian and Soopramanien 2011).

Many reasons may explain the conflicting findings and insignificance of non-financial incentives. In particular, the location of the respondent’s residence appears to be important. People living in areas without congested traffic may not value access to express lanes, while those living in less populated areas with less developed public transportation may not demand public transport even if it is for free; for some, this mode is essentially not available. For these reasons, we analyse preferences for the incentives among different consumer segments according to the location of residence in this paper.

Doubts about the general effectiveness of policy incentives to promote the purchase of electric vehicles have been raised by Hackbarth and Madlener (2016) and Ewing and Sarigöllü (2000). Both of these studies find evidence that an increase in the fully electric driving range to a level comparable with all other car alternatives has the same impact as a multiple-measure policy intervention package. These incentives may, however, also involve an adverse effect that may result in a reluctance of decision-makers to implement them. For instance, exemptions from tolls will reduce revenues meant to finance road infrastructure, which is needed by all types of vehicles; EVs occupy parking spaces just like any other vehicle; and use of transit lanes will involve congestion and consequently bring additional travel costs for public transport users (see Aasness and Odeck 2015).

Previous research on examining the real—as opposed to contingent or simulated—impact of policy incentives has mainly addressed HEVs (see, e.g. Heffner et al. 2005; Gallagher and Muehlegger 2011; Zhang et al. 2013). However, given the short presence of EVs on the market and the resulting limited availability of empirical data, literature evaluating the impact of existing incentives is very limited. A very recent EEA Report by German et al. (2018) examines evidence of the impact of taxation and other incentives on low-carbon vehicles, including electric vehicles, in seven European countries. In line with the results from the SP studies, this review supports the use of financial incentives. The authors find that consumer adoption of lower-CO2-emitting vehicles increases where taxes and incentives are sufficiently large and targeted (France, Ireland, the Netherlands, and Norway), whereas the rate of emission reductions is lower in Germany and Poland.

Among the EU Member States, recent incentives have drastically fostered the adoption of electric vehicles in the Netherlands, resulting in BEV + PHEVs market share of 6.0% in 2016 and 2.2% in 2017 (EAFO 2018). In addition to financial incentives, the Netherlands offered other non-financial benefits such as free parking space or circumvention of a waiting list for parking permits in some regions (Amsterdam, Rotterdam), which stimulated wider adoption of EVs.

In line with these findings, Bjerkan et al. (2016), surveying nearly 3400 BEV owners in Norway, find that an up-front price reduction was the most powerful incentive in promoting BEV adoption. More than 80% of their respondents considered exemptions from purchase tax and VAT critical incentives. Nevertheless, a substantial number of BEV owners considered exemptions from road tolls and bus lane access as the only decisive factor. Figenbaum (2017:32) attributes the significant success of BEVs in Norway to “the result of a long chain of events leading to opportunities that could be exploited more efficiently over time”. In addition to purchase incentives large enough to make BEVs a price competitive alternative, their sales were also spurred by increased model selection, improved technology, and extensive marketing (ibid.).

Quantitative analysis of the real effect of incentives on the adoption of BEVs and PHEVs is still very limited, mainly relying on potential obstacles perceived among potential consumers or adoption comparison of different regions (Egbue and Long 2012). Li et al. (2017) use data from the USA and find that a 10% increase in the supply of charging stations increases EV sales by 8%. To the best of our knowledge, Fluchs and Kasperk (2018) is the only empirical ex post study that analysed the BEV market reaction to various policy incentives over time. In particular, they find that direct rebates increase BEV adoption in China, while the market share of PHEVs increases more compared to the share of BEVs if both vehicle types receive comparable incentives. However, even this very recent study leaves the investigation of the real effect of local non-financial incentives for future research.

Methods and data

Stated preference choice setting

In our discrete choice experiment, respondents were given a hypothetical scenario and asked to choose the best of the presented alternatives. In our case, respondents were asked to think about the situation in which they would be making their next car purchase (our sample consisted of respondents who said they intend to buy a car). The alternatives presented referred to the four vehicle types—BEV, HEV, PHEV, and CV—which were presented to different respondents in random order. To better distinguish the four technologies, we used four different colours attributed to each vehicle type. These colours were attributed to a specific vehicle type again at random, but keeping the same colour for the same respondent.

The alternatives were described using a predefined set of attributes, which could take the levels that were experimentally varied around the level expected by the respondent (see Table 1 for a summary of attributes and attribute levels). There were two monetary, three non-monetary, and two policy attributes describing each vehicle type.
Table 1

Design of the discrete choice experiment

Attribute

Technology-specific attribute

No. of levels

Levels

Purchase price [PRICE]

CV

HEV, PHEV

BEV

1

7

7

PRICE (CV) as said by a respondent in the survey

{80%,90%,100%,110%,120%,130%,140%}*PRICE (CV)

{80%,90%,100%,110%,125%,133%,145%}* PRICE (CV)

Operational and maintenance costs per 100 km [COST = FC + MC/(KM/100)]

per month

CV

HEV

PHEV

BEV

4

2*4

3*4

3*4

FC (CV) = {25,30,40,50} PLN & MC = 4000 PLN

FC (HEV) = {90%,100%}*FC(CV) & MC = 5000 PLN

FC (PHEV) = {70%,90%,100%}* FC(CV) & MC = 5000 PLN

FC (BEV) = {25%,40%,75%}* FC (CV) & MC = 2000 PLN

Driving range [RANGE]

CV, HEV, PHEV

BEV

3

4

500, 700, 900 km

150, 250, 350, 500 km

Charging time [TIME]

PHEV

BEV

3

3

30 min, 1 h, 3 h

2 h, 4 h, 7 h

Availability of fast-mode recharging [FASTMODE]

PHEV, BEV

3

Low (20% of fuel stations + at several public places), medium (60% of fuel stations + at half of public places),high (90% of fuel stations + at almost all public places)

Free parking [PARKING]

HEV, PHEV, BEV

2

Yes

No

Free public transport [PUBLTRAN]

HEV, PHEV, BEV

2

Yes

No

Note: FC denotes fuel costs per 100 km; MC is the annual technology-specific costs of maintenance and repairs, tyres, technical checks, and car insurance expressed per year; KM is a respondent-specific annual mileage. Operational and maintenance costs per month are computed from COST per 100 km considering the respondent’s annual mileage

PRICE represented all one-time expenses associated with a car purchase (including the price and taxes). The purchase price of a CV was always the same for a respondent across all eight choices. This price was based on a price that the respondent expected to pay.2 Although the current prices of the three electric vehicle types are higher, it was explained that these technologies may be supported by a new public policy to motivate customers to buy them due to environmental objectives. As a result, the price of an HEV, a PHEV, and a BEV was lower (10% or 20%), the same, or higher (up to + 45%) than the price of a conventional alternative.

The second monetary attribute described operational and maintenance cost (COST), which represents the average costs of driving a car, including all expenditures, such as the costs of fuel, maintenance and repairs, tyres, technical checks, car insurance, and others. COST consisted of two components in the design, following the design by Hoen and Koetse (2012): fuel cost per 100 km (FC) and annual costs on maintenance, repairs, tyres, technical checks, and car insurance (MC). FC of CVs ranged from PLN 25 to 50 (EUR 5.8–11.6), covering the recent and expected price range. FC of hybrid technologies were the same or 10–30% lower, and fuel costs of BEVs amounted to 25 to 75% of the FC of CVs. This variation may be a consequence of oil price changes in the future, increased fuel taxation, or a subsidy provided for electricity for recharging batteries. Considering current costs, the MC of CVs was set at PLN 4000 (EUR 930), while maintenance of hybrids was assumed to be slightly more expensive due to dual powertrains and BEVs to be cheaper to maintain. This design implied total COST per 100 km for CVs in the range of EUR 8–43 (mean = 19), EUR 8–51 (mean = 21) for HEV, EUR 7–50 (mean = 20) for PHEV, and EUR 2.6–24 (mean = 9) for BEV. The cost attribute was presented per 100 km, and we provided its monthly equivalent using information about annual mileage as stated by a respondent in the survey.

RANGE represents the maximum distance that can be covered by a car after it is fully fuelled or its batteries are fully recharged. The driving range of BEVs described the current state as well as technological progress, with a maximum set at 500 km for BEVs. The driving range interval of the remaining three technologies was the same, from 500 to 900 km.

Refuelling of a CV, HEV, and PHEV always took 2 min. Recharging the batteries of a BEV and PHEV (TIME) took between 2 and 7 h, and between 30 min and 3 h, respectively. Recharging thus described the normal rather than the fast mode. The hypothetical scenario also discussed the possibility of rapid-charging technology, which would make it possible to recharge PHEVs and BEVs much faster, i.e. in a few dozen minutes instead of several hours. The availability of such a fast-mode recharging infrastructure was the next attribute (FASTMODE).

Similarly to Achtnicht (2012) and Hackbarth and Madlener (2013), we linked the availability of such stations to existing petrol station locations, and following Jensen et al. (2013), we also included their availability in other frequently visited locations.3

The last two attributes conveyed the presence of additional benefits associated with purchasing a specific alternative technology. Those who will drive an electric vehicle might park their car in any public parking place throughout Poland for free (PARKING). Second, all family members of a person who owns a BEV or PHEV may use the public transportation system, including railway or buses, and the park-and-ride system fully for free (PUBLTRAN).

While two monetary attributes and driving range described each vehicle type, two policy incentives were only assumed for the three alternative technologies, and two attributes describing recharging batteries were only relevant for PHEVs and BEVs. An example of a choice situation presented to respondents is shown in Fig. 1.
Fig. 1

Example of a choice card. Note: The choice card as shown in the online survey is displayed in the online SI document. We asked the first question and then showed the second question, and then asked for the last third choice

Experimental design

In our experiment, each respondent participated in 8 choice tasks, each consisting of the 4 alternatives presented above. The 8 choice tasks were randomly selected from a set of 40 choice sets. The order of choice sets, the order of technologies (alternatives), and the colours used to highlight technologies were all randomised across respondents.

As our respondents intended to buy very different cars, we pivoted the levels of PRICE and COST according to the respondent’s intended purchase (Rose et al. 2008). Specifically, using information stated earlier in the survey by a respondent about intended purchase price and expected mileage, we derived the respondent-specific PRICE and COST of the reference option (i.e. CV) to make all alternatives, and hence all choices, more realistic.

The experimental design (combinations of attribute levels put together in choice sets) was optimised for D-efficiency (Sándor and Wedel 2001; Ferrini and Scarpa 2007) of the multinomial logit model using Bayesian priors (Scarpa and Rose 2008). The efficiency was evaluated by simulation; we used a median of 1000 Sobol draws as an indicator of the central tendency (Bliemer et al. 2008).4

Econometric framework

Respondents’ choices reveal their preferences and can be modelled using the random utility framework (McFadden 1974). Respondent i’s indirect utility associated with choosing alternative j of the J available alternatives in choice task t can be expressed as:
$$ {U}_{ijt}={\mathbf{X}}_{ijt}{\mathbf{b}}_i+\left({I}_i- PRIC{E}_{ijt}\right){a}_i+{\varepsilon}_{ijt} $$
(1)
where X represents a vector of alternative-specific attributes (COST, RANGE, TIME, FASTMODE, PARKING, PUBLTRAN), including alternative-specific constants that capture all other technology-related characteristics. PRICE is an additively separable purchasing price of a vehicle, I is the respondent’s income, and vector b and a are coefficients. Note that the coefficients are indexed by individuals, allowing for preference heterogeneity. Instead of separately estimating the parameters for each individual, we follow a common practice and assume that the parameters follow specific distributions, which leads to the mixed (random parameters) multinomial logit (MXL) model (Revelt and Train 1998).
To facilitate interpretation of our results, we estimate the model in WTP-space (Train and Weeks 2005), which rescales the utility function to be money-metric:
$$ {U}_{ijt}={\alpha}_i\left({\mathbf{X}}_{ijt}\frac{{\mathbf{b}}_i}{\alpha_i}+\left({I}_i- PRIC{E}_{ijt}\right)\right)+{\varepsilon}_{ijt}={\alpha}_i\left({\mathbf{X}}_{ijt}{\boldsymbol{\upbeta}}_i+\left({I}_i- PRIC{E}_{ijt}\right)\right)+{\varepsilon}_{ijt} $$
(2)

n this specification, the vector of parameters βn = bn/αn can be directly interpreted as a vector of implicit prices (marginal WTPs) for the non-monetary attributes Xijt. In addition to facilitating interpretation of the results, an advantage of this formulation is the possibility to specify a particular distribution of WTP in the population, rather than the distribution of the underlying utility parameters, thus avoiding implausible WTP values.5 In our case, we assume that all WTP distributions are normal and the distribution of the preference-space (negative of) PRICE parameter is lognormal. The model allows for full correlation of random parameters.

The model is estimated using maximum likelihood techniques. An individual will choose alternative j if Uijt > Uikt, for all k ≠ j, and the probability that alternative j is chosen from a set of C alternatives is given by
$$ P\left(j|C\right)=\frac{\exp \left({\alpha}_i\left({\mathbf{X}}_{ijt}{\boldsymbol{\upbeta}}_i+\left({I}_i- PRIC{E}_{ijt}\right)\right)\right)}{\sum_{k=1}^C\exp \left({\alpha}_i\left({\mathbf{X}}_{ikt}{\boldsymbol{\upbeta}}_i+\left({I}_i- PRIC{E}_{ikt}\right)\right)\right)} $$
(3)
There exists no closed form expression of (3) when applying a random parameter logit model, but it can be simulated by averaging over D draws from the assumed distributions. We follow Czajkowski and Budziński (2017) and use 10,000 scrambled Sobol draws for simulations. The simulated log-likelihood (3) function becomes:
$$ \log L=\sum \limits_{i=1}^N{w}_i\log \frac{1}{D}\sum \limits_{d=1}^D\prod \limits_{t=1}^{T_i}\sum \limits_{k=1}^C{y}_{ikt}\frac{\exp \left({\alpha}_i\left({\mathbf{X}}_{ijt}{\boldsymbol{\upbeta}}_i+\left({I}_i- PRIC{E}_{ijt}\right)\right)\right)}{\sum_{k=1}^C\exp \left({\alpha}_i\left({\mathbf{X}}_{ikt}{\boldsymbol{\upbeta}}_i+\left({I}_i- PRIC{E}_{ikt}\right)\right)\right)} $$
(4)
where yikt is a dummy taking the value 1 if an alternative is chosen in choice situation t, and zero otherwise, and wi is the weight corresponding to respondent i’s contribution to the quota-controlled sample. Maximising the log-likelihood function in (4) gives estimates for the parameters.6

Questionnaire and survey

The final version of the questionnaire was prepared according to the usual guidelines (e.g. Champ et al. 2003; Bateman et al. 2004) and based on a pre-survey and pilot testing. The questionnaire structure follows a common ordering (e.g. Bateman et al. 2004), with several socio-demographic questions placed at the beginning to monitor quota attainment and the screening question. The key valuation part was placed around the middle of the instrument, after questions on driving habits, current car(s), and decisive factors to buy a car, and before questions on motivations, home and travel habits, car-sharing participation, and socio-demographics.

The data was collected by a professional public opinion surveying company, Millward Brown, in compliance with ICC/ESOMAR Code on Market and Social Research, using computer-assisted web-based self-interviewing (CASI). The data were collected during January 2015. In our sampling strategy, we originally targeted two specific populations. Sample A consists of respondents who intend to buy a passenger car within the next three years, as confirmed in a screening question. Since the majority of Polish consumers buy second-hand cars (PZPM 2017), we set the share of new car buyers arbitrarily at half to have sufficiently large sub-samples (i.e. new car buyers and used car buyers). Sample B is representative of the general population of Poland. Quota sampling was based on age, gender, region, and the size of the place of residence. In total, merging both samples, the entire dataset includes the data from 2613 respondents, including 407 from the pilot. After excluding speeders, there are 2475 valid observations.7

Socio-demographic characteristics of the sample

For the purposes of this analysis, we merge the data from two samples and consider only those who intend to buy a car and who responded to the choice questions. This strategy gives us in total 17,248 responses to the choice experiments received from 2156 respondents. We apply weights based on the four quota variables (age, education, region, and the size of residence) to improve the representativity of this new sample.8
Table 2

Socio-demographic variables used for deriving the weights, N = 2156

Quota characteristic

Relative frequency

Relative frequency after weighting

Quota for sample B (general population)

Gender

 Male

49.2%

49.2%

50.0%

Age

 18–29

26.9%

31.7%

24.0%

 30–49

46.8%

39.0%

40.0%

 50+

26.4%

29.6%

36.0%

Regions

 Centralny

19.7%

19.2%

21.0%

 Południowy

22.7%

21.7%

21.0%

 Wschodni

17.7%

16.2%

17.0%

 Północno-zachodni

15.5%

18.3%

16.0%

 Południowo-zachodni

10.5%

8.5%

10.0%

 Północny

14.0%

16.1%

15.0%

Education

 Primary and vocational

14.1%

44.4%

46.0%

 Secondary

43.4%

42.0%

35.0%

 Higher

42.6%

17.1%

19.0%

Size of residence

 Up to 20,000

45.3%

45.8%

52.0%

 20,000–200,000

28.3%

29.9%

27.0%

 200,000 and more

26.4%

24.3%

21.0%

Table 3

Characteristics of a car that respondents plan to buy, after weighing, N = 2156

Are you going to buy a new or used car?

New

17%

Used

61%

I do not know yet

22%

What kind of fuel or alternative technology should the car you plan to buy use?

Gasoline

58%

Diesel

36%

Natural gas (CNG)

8%

With LPG gas fittings/I am going to install fittings after purchase

34%

Biofuels

1%

Electricity (electric or hybrid car)

3%

What alternative fuel vehicle do you plan to buy? (percentage from the entire sample)

Battery electric car

0.3%

Hybrid car

1.7%

Plug-in hybrid car

1.4%

Average purchasing price of a car

New car

Used car

Do not know

EUR 15,660

EUR 4537

EUR 8048

Table 4

Description of driving habits, car ownership, purchase intentions, and household income of different consumer segments (frequencies of responses), after weighting

 

All

Urban

Sub-urban

Rural

New car

Used car

Plan to buy a CV

Plan to buy an EV

Number of respondents (not weighted)

2156

839

672

645

465

1691

2017

85

How many km a year do you or others expect to drive this car?

11,861

11,748

12,046

11,822

14,071

11,397

11,865

11,719

How often did you drive your car for a trip during the last year?

 Short-distance trips (up to 100 km one way)

79.63

73.45

73.45

90.36

67.75

82.12

79.95

70.29

 Medium–long trips (up to 500 km one way)

9.25

7.8

11.73

8.67

10.62

8.96

9.34

6.54

 Long-distance trips (more than 500 km one way)

4.11

1.78

6.27

4.87

1.86

4.58

4.21

1.36

Will the car you intend to buy

 Serve as an additional one?

15.1%

13.9%

14.7%

17.0%

13.7%

15.4%

15.0%

19.6%

 Replace the cars you already have?

64.3%

62.2%

66.6%

64.5%

74.7%

62.1%

64.2%

64.9%

What will you do with the vehicle you already have?

 We will keep it

16.1%

15.1%

14.6%

18.7%

11.3%

17.2%

16.2%

15.1%

 We will re-sell it

57.2%

56.1%

59.7%

56.3%

59.4%

56.7%

57.3%

55.2%

 We will give it away

7.6%

9.1%

8.6%

5.1%

15.1%

5.9%

7.7%

5.9%

 We do not know

19.0%

19.7%

17.2%

19.9%

14.2%

20.1%

18.9%

23.8%

Number of vehicles you have/can use?

 How many…

1.58

1.4

1.62

1.76

1.84

1.53

1.57

1.86

 We have no car.

7.6%

11.4%

7.3%

3.3%

2.2%

8.7%

7.6%

7.4%

Household net monthly income

 EUR a month

971

971

1019

926

1277

904

963

1216

Residence

 Urban

38.1%

   

37.0%

38.3%

37.7%

48.5%

 Suburban

29.7%

   

39.0%

27.7%

29.7%

28.8%

 Rural

32.2%

   

24.0%

33.9%

32.5%

22.7%

Table 5

Estimation results of the MXL model—consumers’ WTP for selected attributes of a passenger car (in EUR 1000 of the purchase price)

 

Mean(s.e.)

Standard deviation(s.e.)

Hybrid (vs. conventional)

− 2.7675***

(0.4106)

6.9174***

(0.6491)

Plug-in hybrid (vs. conventional)

− 4.5227***

(0.4798)

8.2629***

(0.6331)

Battery electric (vs. conventional)

− 6.9803***

(0.5608)

11.8791***

(0.6877)

Range (100 km)—conventional

0.2567***

(0.0364)

0.5158***

(0.0315)

Range (100 km)—hybrid

0.2097***

(0.0539)

0.9121***

(0.0400)

Range (100 km)—plug-in hybrid

0.5060***

(0.0455)

0.4566***

(0.0328)

Range (100 km)—electric

1.0668***

(0.0894)

1.1095***

(0.0779)

Charging time (h)—electric

− 0.3113***

(0.0554)

0.7751***

(0.0462)

Charging time (h)—plug-in hybrid

0.0168

(0.0810)

1.1971***

(0.0874)

Fast-charge stations—medium (vs. low)

2.0802***

(0.1918)

2.9389***

(0.1444)

Fast-charge stations—high (vs. low)

2.7246***

(0.1989)

3.5535***

(0.2021)

Free parking

0.7743***

(0.1177)

1.4858***

(0.0953)

Free public transport

0.4310***

(0.1195)

1.1536***

(0.0949)

 Operating costs (EUR/10 km)

3.1717***

(0.2677)

5.0106***

(0.2802)

 Purchase price (EUR 1000)

− 0.8994***

(0.0465)

1.1999***

(0.0575)

Model diagnostics

 LL at convergence

− 17,780.75

 

 LL at constant(s) only

− 23,151.04

 

 McFadden’s pseudo-R2

0.2320

 

 Ben-Akiva-Lerman’s pseudo-R2

0.4741

 

 AIC/n

2.0774

 

 BIC/n

2.1381

 

n (observations)

17,248

 

r (respondents)

2156

 

k (parameters)

135

 

Notes: *, **, and *** represent coefficients significantly different from zero at 0.1, 0.05, and 0.01 significance level, respectively. Standard errors are provided in parentheses. Log-likelihood function is weighted to improve the representativity of the sample. All WTP distributions are normal except the parameters for the (negative of) purchase price, which are log-normally distributed (the coefficients of the underlying normal reported). All parameters are freely correlated

Table 6

Estimation results of the MXL model—consumers’ WTP for selected attributes for different segments (in EUR 1000 of the purchase price)

 

Main effects

Interactions

Mean

Standard deviation

Rural (vs. urban)

Suburban (vs. urban)

New car (vs. used)

Mileage(normalised)

Hybrid (vs. conventional)

− 3.3799***

(0.5901)

6.4513***

(0.4916)

0.1697

(0.8434)

1.7538**

(0.7847)

0.9908

(0.8645)

− 0.3506

(0.5808)

Plug-in hybrid (vs. conventional)

− 5.3989***

(0.6801)

7.6837***

(0.5447)

− 0.9349

(0.8243)

0.2026

(0.8205)

3.3693***

(0.8820)

− 1.2863**

(0.5869)

Battery electric (vs. conventional)

− 7.4959***

(0.7743)

12.0713***

(0.6020)

0.1089

(0.9524)

1.4969

(0.9614)

− 2.0602*

(1.1409)

− 3.9525***

(0.6786)

Range (100 km) —conventional

0.2976***

(0.0554)

0.4596***

(0.0272)

0.0228

(0.0692)

0.1007

(0.0724)

− 0.0497

(0.0798)

− 0.0467

(0.0558)

Range (100 km) —hybrid

0.4029***

(0.0681)

0.7923***

(0.0320)

− 0.1641*

(0.0868)

− 0.2053**

(0.0875)

− 0.0641

(0.1084)

0.0167

(0.0555)

Range (100 km) —plug-in hybrid

0.6738***

(0.0706)

0.4772***

(0.0310)

− 0.1203

(0.0813)

− 0.0776

(0.0853)

− 0.2028**

(0.0931)

0.0722

(0.0538)

Range (100 km) —electric

1.1392***

(0.1155)

1.0145***

(0.0669)

0.2171

(0.1367)

− 0.2662*

(0.1402)

− 0.0952

(0.1643)

0.3439***

(0.0968)

Charging time (h) —electric

− 0.2499***

(0.0725)

0.4972***

(0.0440)

− 0.1711*

(0.0930)

− 0.0057

(0.0897)

0.1738*

(0.0963)

− 0.0324

(0.0631)

Charging time (h) —plug-in hybrid

0.1442

(0.1269)

1.1246***

(0.0689)

0.4717***

(0.1568)

− 0.1831

(0.1533)

− 0.6241***

(0.1616)

0.0478

(0.0921)

Fast-charge stations —medium (vs. low)

2.3704***

(0.2552)

2.9490***

(0.1485)

− 0.2944

(0.3107)

0.1904

(0.3052)

− 0.7464**

(0.3152)

0.1669

(0.1993)

Fast-charge stations —high (vs. low)

3.0233***

(0.2666)

3.4002***

(0.1608)

− 0.3922

(0.3308)

0.4616

(0.3049)

− 0.8452***

(0.3216)

0.2191

(0.1853)

Free parking

0.9422***

(0.1381)

1.3187***

(0.0846)

− 0.4802***

(0.1679)

− 0.4419**

(0.1739)

0.3603*

(0.1898)

0.2597**

(0.1022)

Free public transport

0.3548**

(0.1468)

1.0800***

(0.0918)

− 0.1158

(0.1746)

0.2796

(0.1743)

0.0442

(0.1955)

− 0.1815*

(0.1056)

 Operating costs (EUR/10 km)

3.7095***

(0.3548)

4.4739***

(0.1627)

0.8328**

(0.3551)

0.3257

(0.3350)

1.2289***

(0.4056)

3.7184***

(0.3075)

 Purchase price (EUR 1000)

− 0.8320***

(0.0644)

1.2546***

(0.0604)

− 0.0243

(0.0831)

0.0114

(0.0853)

− 0.1158

(0.0932)

− 0.1764***

(0.0521)

Model diagnostics

 LL at convergence

− 17,667.67

     

 LL at constant(s) only

− 23,151.04

     

 McFadden’s pseudo-R2

0.2369

     

 Ben-Akiva-Lerman’s pseudo-R2

0.4753

     

 AIC/n

2.0713

     

 BIC/n

2.1590

     

n (observations)

17,248

     

r (respondents)

2156

     

k (parameters)

195

     

Notes: *, **, and *** represent coefficients significantly different from zero at 0.1, 0.05, and 0.01 significance level, respectively. Standard errors are provided in parentheses. Log-likelihood function is weighted. All WTP distributions are normal except the parameters for the (negative of) purchase price, which are log-normally distributed (the coefficients of the underlying normal reported). All parameters are freely correlated. Mileage is normalised to the mean value of expected kilometres (i.e. 12,320 km a year), and the scale is expressed in 10,000 km. The value equal to 1.0 corresponds to 22,320 km a year, i.e. 10,000 km above the mean level

The resulting dataset after weighting is representative of the population of Poles who intend to buy a car, using information from Sample B. In this sample, there is an even number of males and females (49.2% males) with an average age of 39.2 years. There are on average 3.5 persons living in a family, and 55% are childless. About 64% of respondents are employed full time or part time, and 10% are self-employed. About 14% of respondents are retired persons, and 9% are recently unemployed. There are 7.6% of respondents without any own income, and the median personal net income ranges between EUR 419 and 535 per month (PLN 1800-2299). Median household income ranges between EUR 814 and 1046 per month (PLN 3500–4500); the mean is EUR 971 (PLN 4178) a month. There are about 10% of respondents who preferred not to answer the income questions. About 16% of respondents live in the city/town centre, and another 22% live outside the centre; these two categories constitute the sub-sample URBAN. Further, 30% of respondents live in the suburbs of a city or town (SUBURB). The remaining 32% of respondents live in a village or small town or a remote village or house (RURAL); all presented statistics here are after weighting.

Prospective car buyers

About 71% of respondents from the sample representative of the general population9 said that either they or another member of their family intends to buy a passenger car sometime in the future. Of all who plan to buy a car sometime in the future, 32% intend to buy a car within a year, 50% within 2 to 3 years, 13% within 4 to 5 years, and 5% later or without knowing the exact period of car purchase. Sixteen per cent of respondents who plan to buy a car consider this car as an additional one, while 70% plan to buy a car to replace their current car.

Our survey confirms general knowledge about the car market in Poland, i.e. that the majority of passenger cars have been purchased as used cars. Indeed, 61% of our respondents plan to buy a second-hand car, whereas only 17% stated a strong preference for buying a new car. The share of those who intend to buy a new car is larger among those who live in the suburbs (23%) and who also intend to purchase an EV (37%). Table 3 also reports the shares of the intended technology: the majority prefer a petrol-fuelled car (58%), 36% prefer a diesel car, and 34% prefer LPG gas fitting or prefer to install fitting after purchase (a multiple-choice option used). Only 1.7% prefer a hybrid car and 1.4% a plug-in hybrid. Even less, about 0.3%, prefer a battery electric car (compared with 0.1% of BEVs reported by EAFO in 2017 and 2018).

The mean purchasing price of a new car is about EUR 15,660, and the median is EUR 13,000. Naturally, the intended price of second-hand cars is lower; the mean is EUR 4537, and the median is EUR 2906. The purchasing price of a car for which a respondent did not yet know whether it should be new or used is in between (with the mean at EUR 8048 and the median at EUR 5813).

Regarding the vehicle size category, a small family-size car (C class—for instance, Škoda Octavia, VW Golf, Honda Civic, or Ford Focus) is preferable by most Polish respondents (30%), followed by the small-car category (B class—e.g. Ford Fiesta, Opel Corsa, or Peugeot 208) and large-size car (D class—e.g. Audi A4, Ford Mondeo, VW Passat) with 18% and 15% shares, respectively. Only 3% of respondents plan to buy an executive (E class) or luxury car (F/G class)—the category in which most of the hybrid cars are recorded. About 5% prefer an SUV, 6% prefer a VAN or multi-purpose vehicle, and 14% did not know the size category during the period of the survey. These shares, particularly for hatchbacks and sedans, are in line with the real market situation during 2015–2017, covering the period within which 82% of our respondents intended to buy a car.10

Table 4 reports the descriptive statistics for the vehicle fleet, driving habits, and income. Respondents with the intention to buy a car own on average 1.58 cars, but the majority (50%) has just one car, 29% own two cars, and almost 8% cannot use any car during the time of the survey. Respondents living in a rural area own more cars than people living in urban areas. Respondents who intend to buy a new car and, in particular, who plan to buy a BEV or PHEV, tend to own a higher number of cars. The majority of respondents (64%) who already have a car consider their next car to replace their existing car. This share is larger for the new-car segment (75%) compared to the used-car segment (62%), but it does not differ across residence types. About 57% intends to re-sell the car they currently have and 8% plan to give it to someone else. Only 15% considers the next car to serve as an additional one. However, almost 20% of those who intend to buy an electric car plan to use it as an additional car.

The same table also displays statistics on expected mileage and current trips. Average mileage is slightly lower than 12,000 km. Those who intend to buy a new car also intend to drive more—about 23% more than those who do not intend to buy a new car. There are large differences across the segments in the number of trips. Families living in a rural area drove more than 90 short-distance (up to 100 km) trips a year, while families from urban and suburban areas drove 73 such trips. Families living in a suburban area take medium- and long-distance trips more. The new-car segment has larger mileage with more trips up to 500 km one way, but with less short- and long-distance trips relative to the used-car segment. Interestingly, those who said they could imagine buying an electric car also took all kinds of trips much less, having the same total mileage as the respondents who were not thinking to buy an EV. There are, however, very few respondents with the prior intention to buy an EV, representing 3.4%.

Of the total respondents, 38% live in an urban area and 32% live in a rural area. Respondents living in suburban areas are more likely to buy a new car, while respondents from rural areas are less likely to buy a new car. Those who considered buying an EV are more likely to be living in urban areas and less likely in rural areas (see Table 4).

Average net household income is PLN 4178, which corresponds to EUR 971 a month. Respondents who intend to buy a new car or EV and who live in suburban areas have higher incomes.11

Estimation results

WTP estimates from the basic model

In what follows, we interpret the results of the MXL model (see Table 5).12 The estimated coefficients correspond to the means and standard deviations of the marginal WTP distributions for the attributes of a passenger car that a respondent is planning to buy.

Overall, we find that Polish consumers are reluctant to choose BEVs. On average, they would require an HEV to be approximately EUR 2750 cheaper, a PHEV to be EUR 4500 cheaper, and a BEV to be almost EUR 7000 cheaper than a CV to provide them with the same utility level.13 We note, however, that we observe substantial preference heterogeneity regarding these technologies, as indicated by relatively large and significant standard deviations. We investigate the sources of this heterogeneity in “Drivers of consumers’ preference heterogeneity”.

The driving range of a fully tanked/charged car increases WTP for a car. Each additional 100 km of the range of a CV or an HEV is worth an additional EUR 210–260 in their purchase price, while for a PHEV it corresponds to approximately EUR 500 and for a BEV over EUR 1000. This result is in line with the diminishing marginal utility of range, which is generally lower for BEVs than any other cars.

Longer charging times reduce the demand for BEVs further. Each additional hour of expected charging time to full charge reduces WTP for such vehicles by EUR 300 on average. Interestingly, the charging time of a PHEV was not statistically significant on average. However, again the substantial preference heterogeneity indicates that this would be an attribute of importance for at least some of the respondents. PHEVs are also mainly fuelled by fossil fuels, while electricity is considered as an additional “fuel” that might explain the difference.

The availability of fast-mode recharging infrastructure was found to translate to an increase in WTP for BEV and HEV cars by approximately EUR 2000 and EUR 2700, respectively, for their medium and high levels, as described in the survey. Interestingly, offering free parking or free public transport to all family members of EV owners only increased their WTP by approximately EUR 770 and EUR 430 on average, respectively.

Finally, we found that our respondents were, on average, willing to pay an additional EUR 317 for a car that had EUR 1 lower operating and maintenance costs per 100 km. The coefficient of the last attribute presented in Table 5 does not have a direct interpretation other than that the increase in purchase price reduces respondents’ utility.

It is worth noting that for each of the attributes we observed substantial and significant preference heterogeneity. In what follows, we investigate to what extent this heterogeneity can be explained using respondents’ location, whether respondents intend to buy a new or used car and respondents’ expected mileage.

Drivers of consumers’ preference heterogeneity

Empirical literature suggests several factors may influence a family’s choice of vehicle fuel. These include driving pattern and mileage, the family’s wealth, trust in a certain technology, and environmental concerns. The type of residence affects both charging opportunities as well as the driving pattern (longer vs. shorter, or regular vs. less frequent travels). In this paper, we investigate the effect of residence (living in urban, suburban, or rural areas) and annual mileage as a proxy for various driving and mobility patterns. We also examine whether preferences for EVs vary between new-car buyers and used-car buyers, while simultaneously controlling for the effect of residence and mileage.

Table 6 presents these results of the MXL model with the interactions of mean WTP. The main effects represent the WTP of respondents in the baseline group, which is living in an urban area, with an intention to buy a used car, and with mileage of 12,320 km a year, while the four vectors of the two-way interactions show the deviations from this group for various groups of consumers. The two-way terms describe specifically the interactions between all vehicles’ attributes and living in a rural area (third column) or a suburban area (fourth column), the respondent’s intention to buy a new car (fifth column), and the expected annual mileage of the intended car (sixth column). Assuming the indirect utility function is additive in its attributes, the final WTP values are given as a sum of the respective WTP estimates. For instance, in the case of free parking for a respondent living in a rural area, intending to buy a new car rather than a used car, and with mileage 10,000 km larger than the average sample mileage, the resulting WTP equals EUR 1082 (i.e. EUR 942 − EUR 480 + EUR 360 + EUR 260).

We find that respondents’ location is an important driver of their preferences towards EVs. Consumers living in rural areas appear significantly more sensitive to the charging time of a PHEV and operating cost, possibly because of higher levels of this component for those living in rural areas. At the same time, they are substantially less concerned with the “free parking” policy incentive. Respondents who live in suburban locations are somewhat less negative about HEVs in general and less sensitive to their driving range. Similarly to rural respondents, free parking would be substantially less effective to motivate them to buy these alternative electric technologies. The respondents living in the suburbs also do not need such a long driving range of BEVs as compared to people living in urban or rural areas.

The differences in preferences of consumers who intend to buy a new rather than used car are even larger. This segment appears relatively more likely to choose PHEVs and less likely to buy BEVs compared to people planning to buy a used car. They are also less sensitive to the driving range of PHEVs and more sensitive to how long it will take to charge them. We also find that respondents who consider purchasing a new car value the development of fast-mode charging infrastructure less than other respondents.

Hoen and Koetse (2014) and Jensen et al. (2013) found that purchasing price is more important for people who choose used cars. We do not find this among Poles. Price is more important for those Poles who expect to drive more. Contrary to findings from many other studies, we find that Poles who intend to buy a used car are less sensitive to changes in operating and maintenance costs, although they are less wealthy than those who intend to buy a new car (see Table 4). One possible explanation is that respondents who plan to buy a used car also expect to drive less on average (11,400 km compared to 14,070 km) and make less short-distance trips (up to 100 km) than those who plan to purchase a new car. We found a significant and negative effect of income on WTP for operating costs in our previous research (Ščasný et al. 2015), and this finding is in line with what Helveston et al. (2015) found in China.

Finally, we observe that expected mileage is an important driver of respondents’ preference heterogeneity.14 This interaction is normalised for the mean mileage in the sample and expressed in 10,000 km per year. As a result, respondents whose expected mileage is 10,000 km above the mean level in the sample (i.e. about 22,500 km) are willing to pay nearly EUR 1300 less for a PHEV and almost EUR 4000 less for a BEV. At the same time, their marginal WTP for an additional 100 km range of BEV is approximately EUR 350 larger. Respondents who expect to drive frequently are also more sensitive to the free parking incentive and prefer less the free public transport incentive. As expected, they are also substantially more sensitive to operating and maintenance costs and less concerned with the vehicle purchase price.

The mean willingness to pay for EUR 1 savings of operating and maintenance costs per 100 km (i.e. approx. 5% of current costs) is EUR 371 for used cars, living in an urban area, and driving average mileage. However, it increases to EUR 454 for a rural area, EUR 494 for new cars, and even to EUR 577 for new cars and living in a rural area. Each 1000 km above the average mileage increases these WTP values by EUR 37; see Fig. 2. Assuming a ten-year lifetime (vehicle usage), the implicit discount rate for average mileage is in the range of 27–56%, is larger for used vehicles and the urban/suburban segment, and declines sharply with mileage.
Fig. 2

WTP for EUR 1 savings of operating costs per 100 km (on the left) and implicit discount rate (on the right; assuming a lifetime of 10 years) for different consumer segments. Note: Abbreviation “urb/sub” describes respondents living in the urban or suburban area. USED and NEW are two segments intending to buy a used or a new vehicle, respectively

Conclusions and policy recommendations

In this paper, we provide estimates of the willingness to pay (WTP) of Polish consumers who intend to buy a passenger car for several specific attributes of electric-driven passenger vehicles, specifically hybrid, plug-in hybrid, and battery electric vehicles. We used discrete choice experiments to elicit preferences for electric vehicles. We analyse data from a sample representative of the segment of the Polish adult population that intends to buy a car.

We find that Polish car buyers prefer electric-driven vehicles significantly less than conventional vehicles. However, decreasing the purchase price and operating costs, developing technologies that increase the driving range, and decreasing charging time can all serve to strengthen preferences for electric vehicles. In addition, the deployment of charging infrastructure can encourage the spread of battery electric vehicles in particular.

The driving range is a very important attribute of electric cars. In fact, Poles are on average willing to pay for each additional kilometre of driving range of a PHEV about EUR 5 and of a BEV about EUR 11. People living in suburban areas are willing to pay less (EUR 8.7) for each additional kilometre of driving range of a BEV. Each 10,000 km above the mean mileage increases the WTP for each kilometre of driving range of a BEV by EUR 3.4. We converted these WTP values using purchasing power parity to USD15 in order to compare our results with the mean values reported in a meta-analysis by Dimitropoulos et al. (2013). In this paper, the mean WTP for a 1-mile increase of driving range of a BEV corresponds to $55 for the whole sample, $42 for people from suburban areas and $16 for a 1-mile increase above the mean mileage. These estimates for Poland are similar to the mean values reported by Dimitropoulos et al. (2013), that is $67 for all studies and $44 for European studies, and are close to the range of $62–73 found for model specifications similar to our study.

The estimated average willingness to pay values for driving range were found to be between USD 35 and USD 80 for an extra mile (Golob et al. 1997; Hidrue et al. 2011; Axsen et al. 2015), and the incremental WTP was found to decrease at higher distances in the Netherlands (Koetse and Hoen 2014). WTP for driving range was larger and more important for BEVs—between EUR 16–33 in Germany (Hackbarth and Madlener 2013) and EUR 50 per kilometre increase in Italy (Valeri and Danielis 2015), while the WTP for a 1-km increase as a generic attribute was found to be only EUR 8–17 and EUR 7.5, respectively. The lower wealth may explain slightly smaller WTP estimates for Poland. In line with the above two studies in Italy and Germany, we also find larger preferences for driving range for BEVs, which is two times the WTP value for PHEVs and even four times the WTP value for CVs. Those who expect to drive double the kilometres as the average are willing to pay almost EUR 15 for each kilometre increase in driving range of BEVs.

Recharging time and the availability of charging stations are currently the most influential barriers to increasing the consumption of plug-in electric vehicles. On average, Polish car buyers are willing to pay about EUR 311 for each hour saved for charging a BEV, but they are not sensitive to this attribute in the case of a PHEV, with one exception. Respondents living in urban and suburban areas who intend to buy a PHEV as a new car prefer significantly shorter-charging batteries of PHEVs and are willing to pay EUR 624 for each hour less of charging. Interestingly, this value is much larger than the WTP for charging BEVs for these two segments (EUR 76 and EUR 250, respectively). There is, however, considerably large preference heterogeneity for charging BEVs.

Preferences for electric-driven vehicles increase sharply when the availability of fast-mode recharging improves from a low level (20% of fuel stations + at several public locations) to a medium level (60% of fuel stations + at half of public locations) or even to a high level (90% of fuel stations + at almost all public locations). The medium availability of fast-mode recharging infrastructure increases the willingness to pay for electric-driven vehicles by EUR 2100 and the high availability by EUR 2700.

Providing other benefits, e.g. free parking and public transport for all family members, increases the probability of choosing electric-driven cars, particularly providing free parking for people living in urban areas intending to buy a new car and with larger mileage.

The results of the mixed logit models indicate that consumer preferences for electric-driven vehicles and their characteristics are highly diverse. A model with two-way attribute-income interactions (Ščasný et al. 2015) shows that higher levels of income increase the probability to purchase hybrid cars and plug-in hybrids and, conversely, weaken the effect of the operational cost attribute. The effect of income on other attributes seems to be insignificant.

Based on our findings, we propose several ways to promote greater uptake of electric vehicles in Poland. First, installation of battery-charging infrastructure and increasing the visibility of charging stations need to be supported. This recommendation has already been provided in the recent Act on Electromobility and Alternative Fuels approved in February 2018. However, according to the results of simulations run by van der Vooren et al. (2012), policy makers should allocate substantial financial resources to public support for infrastructure development of infrastructure-dependent vehicle technologies, rather than providing small financial support for the infrastructures of many different vehicle technologies that may decrease the chances of any vehicle technology benefiting from increasing returns to scale.

Second, from the consumer perspective, support for research and development should focus on improving driving range and battery charging. Since our survey was conducted, there has been rapid development in both of these technology components.

Third, alternative mobility options for “long journeys”, e.g. public transport, various forms of car sharing or pooling systems, and deployment of autonomous driving, need to be promoted (Ščasný et al. 2015). Such policy incentives can motivate consumers to buy electric vehicles, but they are less important than the technical attributes and the purchase price. A pricing and tax policy that will lower the purchase price or the operating costs could affect consumer choices much more concerning achieving higher market shares of electric vehicles.

There is a general concern that purchasing an electric car will simply be adding to the family’s vehicle fleet. For instance, in the most developed BEV and PHEV market in the world, Norway, typical owners of an electric car still buy their BEV or PHEV as an addition to their petrol or diesel car (Haugneland et al. 2016). On the positive side, however, the second vehicle running on conventional oil fuels is only used occasionally for longer trips. There is also a substantial share of EV owners in Norway who have been using their EV for longer journeys (ibid.). In Poland, with its less saturated car demand, we find that only 15% of respondents consider their next car to serve as an additional one. While it is true that respondents who considered buying their next car as an EV thought of it more as an additional car, the share of such respondents was still less than 20%.

Another issue to consider in the policy design is how consumers will substitute between vehicles of different fuel types and emission ratings. For instance, Xing et al. (2018) estimate for the US market that BEVs and PHEVs replace vehicles that are relatively fuel efficient. About 79% of BEVs and PHEVs replace gasoline vehicles with an average fuel economy of 27.2 mpg, and 12% of them replace hybrid vehicles. In order to achieve a large environmental effect, a pricing policy that affects the operating costs and other incentives should be considered along with an effective up-front price incentive scheme.

Footnotes

  1. 1.

    Stated preferences make it possible to measure willingness to pay for configurations of goods which do not currently exist, such as new characteristics of existing or new products to be introduced in the market (Hanley and Czajkowski 2017). In addition, the ability to exogenously and systematically vary attributes of alternatives from which the respondent chooses serves the joint purpose of allowing for clean identification (e.g. allaying endogeneity and collinearity concerns associated with market-observed attribute level combinations; Earnhart 2001; Freeman et al. 2014; Phaneuf and Requate 2017) and increasing the efficiency of preference parameter estimation (Scarpa and Rose 2008).

  2. 2.

    To elicit this information, we used 13 price categories for a new car and 16 price categories for a used car (as the used cars are typically cheaper), and we asked each respondent to choose the one (s)he is most likely to expect to pay for their next car. In the design, the price of a CV equalled the midpoint of the selected price category. If a respondent did not know the purchase price, we attributed randomly one of these prices, PLN 45,000; 55,000; 65,000; or 80,000, for a new car or PLN 12,500; 17,500; 25,000; or 35,000 for a used car or for a car if the respondent did not know whether their next car should be new or used (these ranges were based on the Polish market purchases). Of the respondents, 4.7%, 6.1%, and 16.6% did not know the price of a new, a used, or a non-specified car, respectively.

    Hereinafter, to convert PLN to EUR, we use the purchasing power corrected exchange rate of 0.2325 EUR per PLN.

  3. 3.

    Specifically, fast-mode recharging infrastructure was described as “Recently, very fast recharging devices have become available, which make charging faster. Recharging an electric vehicle entirely takes only 10 minutes, compared to 6 to 8 hours if recharged from an AC socket at home. A hybrid vehicle with a plug-in can then be recharged within 5 minutes only. The fast-mode charging stations can be available to users to various degrees. They can be located at some of existing petrol stations, for example, 20%, 60%, or 90% of petrol stations, or other frequently visited places (e.g. supermarkets, cinemas and sport stadiums).”

  4. 4.

    All prior estimates were assumed to be normally distributed, with the exception of the priors for alternative specific constants, which were assumed to be uniformly distributed to represent potentially larger heterogeneity of respondents’ preferences with respect to propulsion technologies. The means of the Bayesian priors were derived from the MNL model estimated on the dataset from the pilot survey, and standard deviations equal to 0.25 of each parameter mean (with some arbitrarily selected minimum level).

  5. 5.

    There is a direct translation between asymptotic parameters in models estimated in preference space and WTP space (Scarpa et al. 2008), and the two expressions of utility are behaviourally equivalent. Any distribution of parameters in preference space implies some distributions in WTP space, and vice versa. In some cases, however, the resulting distributions can lead to implausible values for WTP or preference parameter estimates (Carson and Czajkowski 2013).

  6. 6.

    The software codes for estimating the MXL model were developed in Matlab and are available at http://github.com/czaj/DCE under Creative Commons BY 4.0 licence.

  7. 7.

    For the identification of speeders, we followed the recommendation of Survey Sampling International (Mitchell 2014) and excluded those who completed the survey in 48% of the median time. Following this strategy, 5.6% of respondents were excluded.

  8. 8.

    These weights should not be based on the quota for the general population nor on the quota for the automotive population, which are generally both available, compared to the population of Poles who intend to buy a car that is unknown. In order to have our dataset representative of the population who intend to buy a car, we utilise the data from sample B (representative of the general population) and derive the relative frequencies for the quota variables using observations from respondents with this intention. Since sample A is representative of the general Polish population and background information about the characteristics of car purchasers was missing, we believe this approach is sufficiently sound to provide information about the shares and hence the weights. The resulting shares after weighting are compared to the quota for the general population of the Poles in Table 2.

  9. 9.

    This subsample provided a basis to derive the weights for the dataset used in this paper. From the general population sample, 9% of respondents do not have a car and also do not intend to buy one in the future, and 5% do not have a car now but would like to have one later.

  10. 10.

    According to the Polish 2017/2018 Automotive Industry Report (PZPM 2017), in 2015–2017, the first registrations of passenger cars by market segment were 17–19% (B class), 29–30% (C class), 10–11% (D class), 1.7–2.6% (E–G class), 24–28% (SUV), and 6–7% (MPV).

  11. 11.

    Average equivalised household income in 2015 was PLN 2223, taking 2.6 persons a household (Eurostat) and assuming 1.8 “equivalised persons” give PLN 4000 a month per household.

  12. 12.

    The supplementary results, such as the estimation results of other models, including Conditional logit and Mixed logit without fully correlated parameters, are provided in the on-line Supplementary Information material. Code is available from http://czaj.org/research/supplementary-materials.

  13. 13.

    Note that the alternative specific constants capture the utilities associated with the baseline levels of all the attributes (e.g. range = 0, charging time = 0, operating cost = 0, low availability of fast-charge stations and no policy incentives) and the otherwise uncontrolled perceived differences between labelled alternatives.

  14. 14.

    Note that this interaction may to some extent control for the differences between urban, suburban, and rural respondents. We find, however, that on average mileage does not differ across these three residence segments, whereas it is significantly larger for the used-car buyers than for the new-car buyers.

  15. 15.

    All USD values reported in this paragraph are PPP-adjusted 2005 USD based on OECD statistics.

Notes

Acknowledgments

This research has been supported by the Czech Science Foundation (GA15-23815S; Ščasný), Charles University (PRIMUS/17/HUM/16; Zvěřinová), and the National Science Centre of Poland (Sonata 10, 2015/19/D/HS4/01972; Czajkowski). Data collection and preliminary analysis were financed by the Polish NCBiR (Centre for Research and Development), within the framework of the project “Development of an Evaluation Framework for the Introduction of Electromobility – DEFINE” provided to the Center for Social and Economic Research (CASE Poland). This article is a part of research presented at the ECOCEP Conference on Economic Modelling for Climate-Energy Policy (FP7-PEOPLE-2013-IRSES, No. 609642) and secondments funded by the H2020-MSCA-RISE under GA 681228. This support is acknowledged. The views expressed here are those of the authors and not necessarily those of our institutions. Responsibility for any errors remains with the authors.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary material

12053_2018_9754_MOESM1_ESM.pdf (716 kb)
ESM 1 (PDF 716 kb)
12053_2018_9754_MOESM2_ESM.pdf (1 mb)
ESM 2 (PDF 1057 kb)

References

  1. Aasness, M. A., & Odeck, J. (2015). The increase of electric vehicle usage in Norway—incentives and adverse effects. European Transport Research Review, 7(4). Google Scholar
  2. Achtnicht, M. (2012). German car buyers’ willingness to pay to reduce CO2 emissions. Climatic Change, 113(3–4), 679–697.CrossRefGoogle Scholar
  3. Axsen, J., Mountain, D. C., & Jaccard, M. (2009). Combining stated and revealed choice research to simulate the neighbor effect: the case of hybrid electric vehicles. Resource and Energy Economics, 31(3), 221–238.CrossRefGoogle Scholar
  4. Axsen, J., Bailey, J., & Castro, M. A. (2015). Preference and lifestyle heterogeneity among potential plug-in electric vehicle buyers. Energy Economics, 50, 190–201.CrossRefGoogle Scholar
  5. Bahamonde-Birke, F. J., & Hanappi, T. (2016). The potential of electromobility in Austria: evidence from hybrid choice models under the presence of unreported information. Transportation Research Part A: Policy and Practice, 83, 30–41.Google Scholar
  6. Bateman, I. J., Carson, R. T., Day, B., & Hanemann, W. M. (2004). Economic valuation with stated preference techniques: a manual. Cheltenham, UK: Edward Elgar ISBN: 1840649194.Google Scholar
  7. Bjerkan, K. Y., Nørbech, T. E., & Nordtømme, M. E. (2016). Incentives for promoting battery electric vehicle (BEV) adoption in Norway. Transportation Research Part D: Transport and Environment, 43, 169–180.CrossRefGoogle Scholar
  8. Bliemer, M. C. J., Rose, J. M., & Hess, S. (2008). Approximation of Bayesian efficiency in experimental choice designs. Journal of Choice Modelling, 1(1), 98–127.CrossRefGoogle Scholar
  9. Brownstone, D., & Train, K. (1998). Forecasting new product penetration with flexible substitution patterns. Journal of Econometrics, 89(1–2), 109–129.zbMATHCrossRefGoogle Scholar
  10. Bunch, D. S., Bradley, M., Golob, T. F., Kitamura, R., & Occhiuzzo, G. P. (1993). Demand for clean-fuel vehicles in California: a discrete-choice stated preference pilot project. Transportation Research Part A: Policy and Practice, 27(3), 237–253.CrossRefGoogle Scholar
  11. Carson, R., & Czajkowski, M. (2013). A new baseline model for estimating willingness to pay from discrete choice models. Sydney: Presented at International Choice Modelling Conference.Google Scholar
  12. Carson, R., & Czajkowski, M. (2014). The discrete choice experiment approach to environmental contingent valuation. In Handbook of choice modelling (p. 202–235).Google Scholar
  13. Caulfield, B., Farrell, S., & McMahon, B. (2010). Examining individuals preferences for hybrid electric and alternatively fuelled vehicles. Transport Policy, 17(6), 381–387.CrossRefGoogle Scholar
  14. Champ, P. A., Boyle, K. J., & Brown, T. C. (ed). (2003). Collecting Survey Data for Nonmarket Valuation. A primer on nonmarket valuation (Vol. 3). Dordrecht: Springer NetherlandsGoogle Scholar
  15. Chorus, C. G., Koetse, M. J., & Hoen, A. (2013). Consumer preferences for alternative fuel vehicles: comparing a utility maximization and a regret minimization model. Energy Policy, 61, 901–908.CrossRefGoogle Scholar
  16. Czajkowski, M., & Budziński, W. (2017). Simulation error in maximum likelihood estimation of discrete choice models protection regulation (GDPR). University of Warsaw Faculty of Economic Sciences Working Papers, 1–47.Google Scholar
  17. Dagsvik, J. K., Wennemo, T., Wetterwald, D. G., & Aaberge, R. (2002). Potential demand for alternative fuel vehicles. Transportation Research Part B: Methodological, 36(4), 361–384.CrossRefGoogle Scholar
  18. van der Vooren, A., Alkemade, F., & Hekkert, M. P. (2012). Effective public resource allocation to escape lock-in: the case of infrastructure-dependent vehicle technologies. Environmental Innovation and Societal Transitions, 2, 98–117.CrossRefGoogle Scholar
  19. Dimitropoulos, A., Rietveld, P., & van Ommeren, J. N. (2013). Consumer valuation of changes in driving range: a meta-analysis. Transportation Research Part A: Policy and Practice, 55, 27–45.Google Scholar
  20. EAFO (2017). The European Commission initiative to provide alternative fuels statistics and information (electricity, hydrogen, natural gas, LPG). European Alternative Fuel Observatory. http://www.eafo.eu. Accessed January 2018.
  21. EAFO (2018). The European Commission initiative to provide alternative fuels statistics and information (electricity, hydrogen, natural gas, LPG). European Alternative Fuel Observatory. www.eafo.eu/content/netherlands. Accessed on 10h July 2018.
  22. Earnhart, D. (2001). Combining revealed and stated preference methods to value environmental amenities at residential locations. Land Economics, 77(1), 12–29.CrossRefGoogle Scholar
  23. European Commission (2014a). Directive 2014/94/EU of the European Parliament and of the Council of 22 October 2014 on the deployment of alternative fuels infrastructure. Text with EEA relevance.Google Scholar
  24. European Commission (2014b). Regulation (EU) No 333/2014 of the European Parliament and of the Council of 11 March 2014 amending Regulation (EC) No 443/2009 to define the modalities for reaching the 2020 target to reduce CO2 emissions from new passenger cars.Google Scholar
  25. Egbue, O., & Long, S. (2012). Barriers to widespread adoption of electric vehicles: an analysis of consumer attitudes and perceptions. Energy Policy, 48, 717–729.CrossRefGoogle Scholar
  26. Ewing, G., & Sarigöllü, E. (2000). Assessing consumer preferences for clean-fuel vehicles: a discrete choice experiment. Journal of Public Policy & Marketing, 19, 106–118.CrossRefGoogle Scholar
  27. Ferrini, S., & Scarpa, R. (2007). Designs with a priori information for nonmarket valuation with choice experiments: a Monte Carlo study. Journal of Environmental Economics and Management, 53(3), 342–363.zbMATHCrossRefGoogle Scholar
  28. Figenbaum, E. (2017). Perspectives on Norway’s supercharged electric vehicle policy. Environmental Innovation and Societal Transitions, 25, 14–34.CrossRefGoogle Scholar
  29. Fluchs, S., & Kasperk, G. (2018). The influence of government incentives on electric vehicle adoption: cross-national comparison (pp. 25–29). Gothenburg Sweden: Paper presented at the 5th World Congress of Environmental and Resource Economists.Google Scholar
  30. Freeman, A. M., Herriges, J. A., & Kling, C. L. (2014). The measurement of environmental and resource values: theory and methods (third ed.). Abingdon, Oxon; New York, NY: RFF Press.Google Scholar
  31. Gallagher, K. S., & Muehlegger, E. (2011). Giving green to get green? Incentives and consumer adoption of hybrid vehicle technology. Journal of Environmental Economics and Management, 61(1), 1–15.CrossRefGoogle Scholar
  32. German, R., Pridmore, A., Ahlgren, C., Williamson, T., & Nijland, H. (2018). Vehicle emissions and impacts of taxes and incentives in the evolution of past emissions: report to EEA. Bilthoven: Eionet Report – ETC/ACM 2018/1. European Topic Centre on Air Pollution and Climate Change Mitigation.Google Scholar
  33. Golob, T. F., Torous, J., Bradley, M., Brownstone, D., Crane, S. S., & Bunch, D. S. (1997). Commercial fleet demand for alternative-fuel vehicles in California. Transportation Research Part A: Policy and Practice, 31(3), 219–233.Google Scholar
  34. Hackbarth, A., & Madlener, R. (2013). Consumer preferences for alternative fuel vehicles: a discrete choice analysis. Transportation Research Part D: Transport and Environment, 25, 5–17.CrossRefGoogle Scholar
  35. Hackbarth, A., & Madlener, R. (2016). Willingness-to-pay for alternative fuel vehicle characteristics: a stated choice study for Germany. Transportation Research Part A: Policy and Practice, 85, 89–111.CrossRefGoogle Scholar
  36. Hanley, N., & Czajkowski, M. (2017). Stated preference valuation methods: an evolving tool for understanding choices and informing policy. University of Warsaw Faculty of Economic Sciences Working Papers, 1–43.Google Scholar
  37. Hardman, S., Chandan, A., Tal, G., & Turrentine, T. (2017). The effectiveness of financial purchase incentives for battery electric vehicles—a review of the evidence. Renewable and Sustainable Energy Reviews, 80, 1100–1111.CrossRefGoogle Scholar
  38. Haugneland, P., Bu, C., & Hauge, E. (2016). The Norwegian EV success continues. Montréal, Quebec, Canada: Presented at EVS29 symposium.Google Scholar
  39. Heffner, R.R., Kurani, K.S., & Turrentine, T.S. (2005). Effects of vehicle image in gasoline-hybrid electric vehicles. UC Davis Institute of Transportation Studies, Davis. European Alternative Fuels Observatory (2018). Netherlands. Retrieved from www.eafo.eu/content/netherlands on 10h July 2018.
  40. Helveston, J. P., Liu, Y., Feit, E. M., Fuchs, E., Klampfl, E., & Michalek, J. J. (2015). Will subsidies drive electric vehicle adoption? Measuring consumer preferences in the U.S. and China. Transportation Research Part A: Policy and Practice, 73, 96–112.CrossRefGoogle Scholar
  41. Hess, S., Fowler, M., Adler, T., & Bahreinian, A. (2012). A joint model for vehicle type and fuel type choice: evidence from a cross-nested logit study. Transportation, 39(3), 593–625.CrossRefGoogle Scholar
  42. Hidrue, M. K., Parsons, G. R., Kempton, W., & Gardner, M. P. (2011). Willingness to pay for electric vehicles and their attributes. Resource and Energy Economics, 33(3), 686–705.CrossRefGoogle Scholar
  43. Hoen, A., & Koetse, M. J. (2012). A choice experiment on AFV preferences of private car owners in the Netherlands (vol. 3). PBL working paper.Google Scholar
  44. Hoen, A., & Koetse, M. J. (2014). A choice experiment on alternative fuel vehicle preferences of private car owners in the Netherlands. Transportation Research Part A: Policy and Practice, 61, 199–215.Google Scholar
  45. Jensen, A. F., Cherchi, E., & Mabit, S. L. (2013). On the stability of preferences and attitudes before and after experiencing an electric vehicle. Transportation Research Part D: Transport and Environment, 25, 24–32.CrossRefGoogle Scholar
  46. Kim, J., Rasouli, S., & Timmermans, H. (2014). Expanding scope of hybrid choice models allowing for mixture of social influences and latent attitudes: application to intended purchase of electric cars. Transportation Research Part A: Policy and Practice, 69, 71–85.Google Scholar
  47. Koetse, M. J., & Hoen, A. (2014). Preferences for alternative fuel vehicles of company car drivers. Resource and Energy Economics, 37, 279–301.CrossRefGoogle Scholar
  48. Kurani, K. S., Turrentine, T., & Sperling, D. (1996). Testing electric vehicle demand in ‘hybrid households’ using a reflexive survey. Transportation Research Part D: Transport and Environment, 1(2), 131–150.CrossRefGoogle Scholar
  49. Lebeau, K., Van Mierlo, J., Lebeau, P., Mairesse, O., & Macharis, C. (2012). The market potential for plug-in hybrid and battery electric vehicles in Flanders: a choice-based conjoint analysis. Transportation Research Part D: Transport and Environment, 17(8), 592–597.CrossRefGoogle Scholar
  50. Li, S., Tiong, L., Xing, J. & Zhou, Y. (2017). The market for electric vehicles: indirect network effects and policy design. Journal of the Association of Environmental and Resource Economists, 4(1), 89–133.CrossRefGoogle Scholar
  51. Liao, F., Molin, E., & van Wee, B. (2017). Consumer preferences for electric vehicles: a literature review. Transport Reviews, 37(3), 252–275.CrossRefGoogle Scholar
  52. McFadden, D. L. (1974). Conditional logit analysis of qualitative choice behavior. In P. Zarembka (Ed.), Frontiers in econometrics. New York: Academic Press.Google Scholar
  53. Mersky, A. C., Sprei, F., Samaras, C., & Qian, Z. S. (2016). Effectiveness of incentives on electric vehicle adoption in Norway. Transportation Research Part D: Transport and Environment, 46, 56–68.CrossRefGoogle Scholar
  54. Mitchell, N. (2014). The changing landscape of technology and its effect on online survey data collection. Survey Sampling International. https://www.surveysampling.com/site/assets/files/1583/the-changing-landscape-of-technology-and-its-effect-on-online-survey-data-collection.pdf. September 2017.
  55. Phaneuf, D. J., & Requate, T. (2017). A course in environmental economics: theory, policy, and practice. Cambridge, United Kingdom; New York, NY, USA: Cambridge University Press.Google Scholar
  56. Potoglou, D., & Kanaroglou, P. S. (2007). Household demand and willingness to pay for clean vehicles. Transportation Research Part D: Transport and Environment, 12(4), 264–274.CrossRefGoogle Scholar
  57. PZPM (2017). Automotive industry report 2017/2018. Polski Zwiazek Przemyslu Motoryzacyjnego. http://www.pzpm.org.pl/en/Automotive-market/Reports/PZPM-Automotive-Industry-Report-2017-2018. September 2017.
  58. Qian, L., & Soopramanien, D. (2011). Heterogeneous consumer preferences for alternative fuel cars in China. Transportation Research Part D: Transport and Environment, 16(8), 607–613.CrossRefGoogle Scholar
  59. Rasouli, S., & Timmermans, H. (2016). Influence of social networks on latent choice of electric cars: a mixed logit specification using experimental design data. Networks and Spatial Economics, 16(1), 99–130.MathSciNetCrossRefGoogle Scholar
  60. Revelt, D., & Train, K. (1998). Mixed logit with repeated choices: households’ choices of appliance efficiency level. Review of Economics and Statistics, 80(4), 647–657.CrossRefGoogle Scholar
  61. Rose, J. M., Bliemer, M. C. J., Hensher, D. A., & Collins, A. T. (2008). Designing efficient stated choice experiments in the presence of reference alternatives. Transportation Research Part B: Methodological, 42(4), 395–406.CrossRefGoogle Scholar
  62. Sándor, Z., & Wedel, M. (2001). Designing conjoint choice experiments using managers’ prior beliefs. Journal of Marketing Research, 38(4), 430–444.CrossRefGoogle Scholar
  63. Scarpa, R., & Rose, J. M. (2008). Design efficiency for non-market valuation with choice modelling: how to measure it, what to report and why. Australian Journal of Agricultural and Resource Economics, 52(3), 253–282.CrossRefGoogle Scholar
  64. Scarpa, R., Thiene, M., & Train, K. (2008). Utility in willingness to pay space: a tool to address confounding random scale effects in destination choice to the Alps. American Journal of Agricultural Economics, 90(4), 994–1010.CrossRefGoogle Scholar
  65. Ščasný, M., Zvěřinová, I., & Czajkowski, M. (2015). Report on determinants and barriers of purchase of low carbon vehicles, including WTP estimates for specific attributes of passenger vehicles in Poland. Deliverable 8.1. of project “Development of an Evaluation Framework for the Introduction of Electromobility” (DEFINE) funded by Era-Net Transport Transnational Call Electromobility+.Google Scholar
  66. Train, K., & Weeks, M. (2005). Discrete choice models in preference space and willingness-to-pay space. In R. Scarpa & A. Alberini (Eds.), Applications of simulation methods in environmental and resource economics (Vol. 6, pp. 1–16). Berlin/Heidelberg: Springer-Verlag.CrossRefGoogle Scholar
  67. Valeri, E., & Cherchi, E. (2016). Does habitual behavior affect the choice of alternative fuel vehicles? International Journal of Sustainable Transportation, 10(9), 825–835.CrossRefGoogle Scholar
  68. Valeri, E., & Danielis, R. (2015). Simulating the market penetration of cars with alternative fuel power train technologies in Italy. Transport Policy, 37, 44–56.CrossRefGoogle Scholar
  69. Xing, J., Leard, B., & Li, S. (2018). What does an electric vehicle replace? Paper presented at the 5th World Congress of Environmental and Resource Economists (WCERE), Gothenburg, June 25–29, 2018. Retrieved from https://www.jianweixing.com/research.html on 5 July 2018.
  70. Zhang, X., Wang, K., Hao, Y., Jing-Li, F., & Wei, Y. M. (2013). The impact of government policy on preference for NEVs: the evidence from China. Energy Policy, 61, 382–393.CrossRefGoogle Scholar
  71. Ziegler, A. (2012). Individual characteristics and stated preferences for alternative energy sources and propulsion technologies in vehicles: a discrete choice analysis for Germany. Transportation Research Part A: Policy and Practice, 46(8), 1372–1385.Google Scholar

Copyright information

© Springer Nature B.V. 2018

Authors and Affiliations

  1. 1.Environment CentreCharles UniversityPrague 6Czech Republic
  2. 2.Department of Sociology, Faculty of ArtsCharles UniversityPragueCzech Republic
  3. 3.Faculty of Economic SciencesUniversity of WarsawWarsawPoland

Personalised recommendations