Private household demand for vehicles on alternative fuels and drive trains: a review

Attitude theory [30] is a central concept in environmental psychology and typically relates attitudes to behaviour through an intermediary intention construct. One of the most applied theories to study behaviours of environmental relevance is the theory of planned behaviour (TPB) [5]. This theory stipulates that behavioural intention is shaped by attitudes towards the specific behaviour (overall evaluation of its possible consequences), subjective norms (the perception of a person about the normative expectations of others such as close friends or family) and perceived behavioural control (PBC) (personal feeling that one could easily change its behaviour and that one has the possibilities to do this). TPB pretends that if attitudes and subjective norms are favourable, the PBC will be larger and the intention for behavioural change will be stronger [4,5]. In this context, the adoption of an AFV will be immediately influenced by a person’s intention to adopt the technology. Other conceptual frameworks in the field of environmental psychology include the norm-activation model (NAM) (by [77,78]), which focuses on moral values and personal norms to explain (altruistic) behaviour, and the value-belief-norm (VBN) theory (by [83,84]) to explain commitments to protect the environment.

Methodological approaches to measure the correlation between attitudes and behaviour in environmental psychology often apply quantitative methods, qualitative methods (e.g., semi-structured interviews, focus groups, group discussions etc.) or a combination of both.

Although the TPB and other theories are well tested, they rely on complicated links between consumer values, knowledge, beliefs, attitudes, intentions and behaviour. A large size of factors is often difficult to quantify, predict and manage and even if they are known, there is a loose causal link between intentions and actions. The explanatory value of attitudes on behaviour is thus limited, but might enlarge when behaviour measures are self-reported, than when observed [8,75]. In addition, the choice of survey questions might be subjected to potential biases (e.g., framing, sample selection bias etc.). Quantitative surveys might for example provide less opportunity to let people frame their own public perceptions and attitudes than qualitative surveys, but the latter encounters the limitation of being representative for wider populations [74].

Table 3 shows an increasing emergence of attitudinal surveys within the last 7 years, mainly executed in Europe. Except for Ricci et al. [74] and Schulte et al. [76], quantitative surveys were applied to reveal the consumer preferences for AFVs. Within the surveys, there is a predominance of measuring the attitudes towards hydrogen [2,66,74,86] and biofuel vehicles [69,89,91]. Overall, the general concern about environmental issues is found to be high (general positive environmental attitudes), but very often, this is not translated into changes in purchasing behaviour [2,66,74,94]. This discrepancy between environmental attitudes and ecological behaviour is known as the attitude-action gap [56]. Most of the reviewed studies have addressed this attitude-action gap and have consistently reported two findings.

First, attitudes and behaviour towards AFVs are not merely determined by environmental considerations, but are the outcome of a complex trade-off involving economic [69,74,89], performance [66,74,86] and psychological factors [67] including less conscious determinants such as status [52] and symbolic motives [67]. Moreover, individuals will rather value the environmental benefits of AFVs in terms of potential monetary savings, than out of environmental concern [69,89].

Second, limited knowledge levels currently prevent to build up awareness of AFVs, which is the key to their adoption [66,74,86]. Prior levels of AFV knowledge and hence awareness are found to be associated with socio-demographic (e.g., gender, age) and personal characteristics (e.g., education) [66,86], environmental knowledge [35,66,86,94], pro-environmental attitudes [35,86], cultural dispositions [2], product involvement [66,94], direct experience (e.g., in terms of practical experiences and demonstrations) and familiarity with AFVs [74,94].

Besides attitudes, an increased belief of the ability to influence the environment positively (PBC) is found to affect behavioural change in the same way [67,69,91]. To strengthen the PBC, information is required that focuses on the opportunities and possible solutions (e.g., range of actions that can be taken within car purchase to reduce energy problems), instead of messages that contain negative information and strengthen the seriousness of the problem and its detrimental consequences [66,69,91].

The need for information is a recurrent finding throughout most of the reviewed studies [2,35,66,74,85,91]. Audiences are found to behave differently and require information to be differentiated according to the interests and knowledge levels, socio-demographic characteristics, attitudes and cultural dispositions of the specific target audience [2,74,91,94]. Performance attitudes are found to be positively related to information search [94], whereas people with environmental attitudes are less affected by information provision [91,94]. In this respect, tailored information which focuses on the new technical developments that improve environmental performance while maintaining car performance might for example help to enhance the support for AFVs [94].

In experimental and quasi-experimental studies, individuals are put in a natural or an artificial setting to observe their behaviour towards a group of individuals not exposed to the experimental treatment. Experimental studies often make use of vehicle trials or clinics, activity-based approaches, gaming simulations or design spaces [11,55]. Activity analysis includes the use of household travel diaries, activity location maps, videos and other informational material to observe daily travel patterns and understand consumers’ needs with respect to AFVs [55]. Gaming simulations are experimental contexts in which respondents seek solutions to a particular problem or issue (e.g., range limit) within their activity space [40,55]. Design spaces are used to elicit the consumers’ design priorities and preferences of AFVs, which is consistent with theories of constructed preferences that view consumer preferences as outcomes (and not as inputs) of decision contexts and processes [11,15].

Although experimental designs rely on virtual or real-world experiences of the technology and are more realistic than other survey techniques, they have their limitations. Trials may for example evoke the “Hawthorne” effect, indicating that people will produce upward biased estimates of interest in AFVs since they receive special attention. As a result, fewer people than those who expressed a purchase intention are likely to purchase such a car. Additionally, there may arise several measurement problems related to the duration and length of the trial. Trials may also provide reactions to a specific category or product because of the opportunity that the participants have to experience competing technologies (e.g., conventional gasoline cars) [39].

All reviewed experimental studies, displayed in Table 4, were performed in California (US). Many of them were carried out in the 90s to understand how consumers could address EV limitations. In line with technological developments, recent experimental studies rather concentrate on HFCVs and PHEVs. All vehicle technologies included in the experimental designs face driving range and infrastructure challenges and rely on an electric motor powered by a unique fuel source [55,61].

Most experimental studies shared the opinion that there is a strong concern for the environment, and a strong belief that lifestyle changes are required to solve environmental problems [38,39,54,55,90]. However, they also discovered that the environment was the lowest rate issue when purchasing a vehicle (see for example [55,90]). Although environmental awareness may not lead to the purchase of an AFV, it might encourage households to seek out and evaluate AFVs for purchase considerations [55].

Short-term exposure improves the respondents’ overall impressions of AFVs, especially with respect to their environmental benefits [39,61]. But, as they gain experience with AFVs, those perceived environmental benefits become a lower priority as other vehicle features will enter the decision set [39]. On the one hand, comparisons will be made based on range, purchase price and convenience of use [38]. Some of the reviewed studies reveal that respondents did not change their perception about the desired range [38,39,61]. Despite the demonstrated utility of AFVs, respondents still desire ranges to be similar to that of a conventionally fuelled car, even when travel diaries showed that they were usually travelling less on a daily basis [38,61]. Acceptable driving ranges are found to be 160 km for EVs [38,54] up to 480 km for HFCVs [61]. On the other hand, consumer preferences might emerge with respect to the new qualities of AFVs (e.g., quiet ride, low maintenance costs) [38,54,55]. In this respect, Kurani et al. [55] argue that multi-vehicle households will combine AFVs and conventional vehicles in their stock to achieve the advantages of the different propulsion systems. These “hybrid households” will not be discouraged by the limited driving range of AFVs as they allocate household travel according to the different operational characteristics of the vehicles (i.e., conventional vehicles for longer trips, AFVs for shorter trips). The attractiveness of AFVs will not lie in their competitive, but in their complementary relation to conventionally fuelled vehicles. In addition, Kurani and his colleagues assert that any disutility of reduced range can be more than offset by the value of home recharging [9,11,55]. They stipulate that a succesful market launch of AFVs will not depend on the duplication of the performance attributes of conventionally fuelled vehicles, but on the recognition that there is a potential market for less ambitious AFV designs with shorter driving ranges. According to these authors, the commercialisation of more ambitious AFVs should be inevitably accompanied with financial incentives, large-scale vehicle demonstrations and persuasive information campaigns to overcome the financial barriers and the lack of inherent interest.

Preference valuation studies are another technique to analyse the potential demand for environmental goods or services which are usually not traded within the market mechanism. Consumer preferences are usually inferred by stated or revealed preference techniques. The stated preference (SP) technique is a survey-based technique that allows researchers to uncover how people value different product/service attributes. The most common SP techniques used in transport studies are the choice modeling (CM) method and the contingent valuation method (CV). CM originates from conjoint analysis, information integration theory in psychology [6] and discrete choice theory in economics/econometrics [58,59]. It applies a choice experiment approach using a variety of instruments (e.g., pencil and paper, computer aided personal survey instrument (CAPI), internet-based survey) to indirectly elicit attribute values based on either ranking or rating of products described by a number of attributes in several labelled or unlabelled choice sets [18,41,47]. Subsequently, via statistical techniques, the analysis will derive a value for each of these attributes and thus express the relative preferences among vehicle attributes [29,79]. Conventional discrete choice (DC) models analyse situations in which respondents are asked to choose one alternative from a set of mutually exclusive hypothetical alternatives [17,50]. Recently, multiple discrete-continuous extreme value (MDCEV) models have been introduced that deal with the existence of multiple-vehicle households, where households own and use multiple vehicles for satisfying their travel needs [16,17]. Other recent progresses in CM aim to improve realism, by for instance adding a no-choice option [31], or by customising attribute levels based on respondent’s current vehicle choices [10,31,62].

In CV, value elicitation is whole-product based by asking respondents to express their maximum willingness to pay (WTP) for a given improvement of a public good provision level (e.g., cleaning up a lake) or for public goods aspects of a market good (e.g., eco-labeled goods) [44,63]. In the dichotomous CV design (yes/no answers), respondents accept or refuse a payment for a change in the quality or the quantity of a good at a given cost, while open-ended questions (such as payment cards and bidding games) provide a way to elicit the respondent’s maximum WTP [63,64]. CV and CM offer rather different merits and their use entirely depends on the purpose of the study under consideration. CM is particularly suited to measure the marginal value of changes in various characteristics of environmental goods and allows a deeper understanding of the trade-offs between attributes, whereas CV is a better technique than CM when the main objective of the study is to value an overall policy package and for assisting in policy evaluations [25,43,44,50].

Economists have been sceptical towards the use of SP data. One possible problem with hypothetical choices is that it may not reflect the real purchase intentions of the respondents (i.e., hypothetical bias) [20,47]. Another criticism is that SP surveys view consumer preferences as inputs to decision contexts and processes and assume that consumers have preferences for attributes that are unfamiliar to them. Consequently, these surveys might not capture the complexity of vehicle purchase behaviour [9,15,39,55]. Moreover, respondents tend to give socially desirable responses, such as “feel good” responses for environmental benefits or they may provide in contrast anti-environmental survey responses [55]. As a result, they may signal their preference for provision of less pollution, although in reality they would not spend any extra money on purchasing an environmentally friendlier car. Ewing and Sarigöllü [34] also point out that people who are highly concerned about the environment may have a higher motivation to return the surveys. Finally, Kurani et al. [55] state that surveys usually question one person from a household, while vehicle purchases are often made jointly by the whole household. In the literature on CV, an extensive overview of potential sources of bias is given by Mitchell and Carson [63], Carson [25], Bateman et al. [12] and Venkatachalam [95], which can also offer guidance on how to cope with potential bias in hypothetical choice experiments [47].

In contrast to the SP technique, the revealed preference (RP) technique uses real market data from observations on actual choices in order to measure the consumer preferences. So RP data or market data do not have the possibility for confusion or unstated assumptions. But the main problem in predicting a market for AFVs by using RP data is the absence of actual choice observations since only a small market share of AFVs is currently available [70,71]. Furthermore, using RP data makes it difficult to observe the effect of large variations in the variables of interest. Finally, RP data may often produce strong correlations between the variables (multicollinearity) and may evoke difficulties in measuring the vehicle attributes [1]. Because SP and RP have complementary strengths, a growing body of literature applies joint SP-RP modeling techniques (e.g., [10,22]).

Except for Brownstone et al. [22] and Axsen et al. [10], all studies, listed in Table 5, applied SP surveys with a predominance of CM, used for the identification of unique attributes of AFVs and their effects on vehicle purchase. All reviewed studies from the 80s and 90s were performed in America and focused on battery EVs or on a mix of AFVs including either EVs, LPG, CNG or methanol. Later research also included HEVs, PHEVs and HFCVs in the vehicle choice experiments. These reviewed CM studies reveal that the most critical factors for the adoption of AFVs are price characteristics (e.g., purchase price, fuel costs) [3,26,29,33,72,79], followed by performance and convenience attributes (e.g., driving range, recharging times, fuel availability) [13,14,21,24,27,29,33,42,72,79]. Although people express a willingness to pay for reduced emission levels [21,23,87], environmental benefits are consistently found to be of minor importance compared to these attributes [26,27,34,72].

Recent studies also apply CM to capture the dynamics in consumer preferences for new technologies [10,31,62]. These studies reveal that conventional vehicles become less desirable with increases in AFV adoption, given equal monetary costs (i.e., neighbour effect) [10,31,62]. However, these dynamics in consumer preferences seem to depend on the type of new technology. Consumers will more likely switch to HEVs than to HFCVs [62] or to EVs [31]. This can be attributed to the fact that HEVs exhibit low switching costs (e.g., no reliance on the availability of charging stations), whereas HFCVs and EVs possess attributes that are unfamiliar to consumers (e.g., driving range) and require technical infrastructure (e.g., to recharge the car at home) [31,62].

A minority of the reviewed SP studies applied the CV method. In recent years, this method was especially applied to unveil the WTP for biofuels [68,80,81], HFCVs [65] and HEVs [32]. Similar to other CV literature, the WTP for AFVs is found to be positively influenced by income [32,80] and environmental concern [32,51,65,80,81]. Despite these levels of environmental concern and supportive attitudes towards AFVs, the WTP is mainly determined by financial considerations. The high initial costs of HEVs currently prevent them from gaining a market share (e.g., in Turkey) [32], the WTP for HFCVs is driven mostly by expectations of personal financial savings (e.g., reduced running costs) [65] and there only exists a WTP for methanol if the cost burden is shared [81].

To widen the acceptance of AFVs, these studies suggest that policymakers should act on the environmental concerns by issuing educational campaigns to raise awareness about the features and benefits of AFVs [31,32,65]. In this respect, consumers seem to react more to emission information (eco-labelling) at the vehicle level than at the class level [60]. In addition, tax incentives (e.g., surcharge of dirtier fuels, differentiation of vehicle taxes based on fuel economy or CO₂ emissions), subsidies or private privileges (e.g., free parking) should make these vehicles more financially competitive as compared to conventionally fuelled vehicles [32,65]. Finally, the availability of a refuelling or recharging network could be ensured by the deployment of public charging stations, which is likely to be facilitated by the involvement and cooperation between all major stakeholders [31].

A fundamentally different approach is symbolism, which does not focus on monetary costs and functional attributes to predict the adoption of AFVs, but on the symbolic meanings associated with them. Studies of symbolism either rely on conceptual frameworks from psychology (see [57,82]) or on ethnographic interviews which originate from anthropology (i.e., semiotics) (see [45,46]). The former studies used social-scientific research techniques to illustrate that the attractiveness of car use not only depends on instrumental-reasoned factors (e.g., travel costs, safety), but also on symbolic-affective motivations (e.g., status and social comparison, feelings of self-expression, feelings of sensation). The latter studies by Heffner et al. [45,46] assume that vehicles are important carriers of symbolic meanings that are used to maintain (product as self-expression) and create (product as self-creation) self-identity. By selecting a particular vehicle, people communicate their interests, beliefs, values and social status.

Semiotic theory often uses qualitative surveys as they allow participants to use their own terminology and value frameworks. In addition, it might overcome some of the challenges associated with the examination of symbolic meanings, such as the tendency to understate the impact of symbolic meanings in vehicle purchases [45,46]. On the other hand, this approach relies on the existence of AFVs on the market and associated symbolic meanings, which take time to appear and to communicate [75].

The reviewed studies, summarised in Table 6, mostly relied on semiotic theory to unveil the symbolic meanings associated to AFVs in California (US). They all agree on the fact that households are not purchasing AFVs for their functional or economic benefits, but to gain access to symbols that are used to define and express who they are.

Symbolism is found to be particularly strong in vehicles that use new types of technologies. They were important to early buyers of EVs in Norway [36] and HEVs in the US [45,46]. Heffner et al. [45] used ethnographic interviews with 25 early HEV buyers to explore how widely recognised social meanings (denotations) are connected to more personal meanings (connotations) and how they affect vehicle purchase. They revealed that households buy HEVs for the meanings they symbolise (such as “environmental preservation”, “financial responsibility”, “independence from oil producers”, “embracing new technology” and “opposition to war”) as well as the connotations linked to these ideas that are relevant to self-identity (such as “concern about others”, “intelligence”, “independence”, “uniqueness”, “ethics”, etc.). While denotations are generally socially-shared, connotations vary from person to person. For example, two households may view their HEV as a symbol for “preserving the environment”, but one household may emphasise the “ethics” connotation, whereas for another household their “concern about others” is an important value to communicate to society [45,88].

Heffner et al. [46] recommend that, just as early buyers of HEVs looked for meanings that were unavailable in other types of vehicles; early buyers of HFCVs will also look for new symbolic meanings in their vehicles that are not offered by competing alternatives. In this respect, HFCVs should improve on existing symbolic meanings (e.g., by providing potential buyers a more authentic access to the ideas of environmental preservation or advanced technology). In addition, they should offer entirely new symbolic meanings as well such as “home refuelling”, which can refer to aspects of independency. Another possibility is to explore the idea of “extended personal territory”, as HFCVs have the potential to produce clean electrical power for purposes other than propulsion. As new meanings will continue to emerge, understanding the meanings, as well as their construction and communication is essential to promote the AFV market.