Can we hope for a collective shift in electric vehicle adoption? Testing salience and norm-based interventions in South Tyrol, Italy Nives DellaValle

Energy Research & Social Science 55 (2019) 46–61 Contents lists available at ScienceDirect Energy Research & Social Science journal homepage: www.elsevier.com/locate/erss Original research article Can we hope for a collective shift in electric vehicle adoption? Testing salience and norm-based interventions in South Tyrol, Italy T Nives DellaValle , Alyona Zubaryeva ⁎ Eurac Research – Institute for Renewable Energy, Via A. Volta 13/A, 39100 Bolzano, Italy A R TICL E INFO A BSTR A CT Keywords: Electric vehicles Mobility Behavioral economics Survey experiment Electric-drive vehicles (EVs) are a promising technology for containing environmental problems of the transport sector. However, to be effective on the environment, these need to be purchased by consumers in large quantities. Not only material barriers, but also cognitive biases prevent their diffusion. Particularly, EVs fail to capture a significant passenger car market share, even in regions that can overcome most material barriers. We report the findings from an online survey experiment administered to respondents of a region identified as a potential EV lead market, but which still fails to reach a significant EV market share. We test the effect of two behaviorally informed strategies on EV preferences: a norm-based and a salience intervention. To better identify treatment effect, we control for pro-environment self-identity, heterogeneity in key economic preferences, and size. Results show that making future cost savings salient significantly increases the likelihood of choosing EVs. However, the effect is limited to individuals displaying preferences for big vehicles and for valuing future benefits, and high values in the pro-environment self-identity measure. In addition, results show that EV choices are unaffected by the descriptive norm embedded in the norm-based intervention. 1. Introduction In recent decades, road transport has largely relied on oil-based fuels, reaching the level of 96% of energy consumed by vehicles in 2015 [1]. This and other factors, such as a growing trend of green-house gas emissions and urban air pollution, have led to steady and continuous efforts to introduce alternative fuelled vehicles, in particular electricdrive vehicles (EV – comprising plug-in hybrid electric vehicles (PHEV), battery electric vehicles (BEV) and fuel cell vehicles (FCV)) in the European market. These vehicles are seen as one of the most relevant innovations that can make the transport sector more sustainable and address environmental problems caused by conventionally fuelled vehicles [2,3]. However, to be fully beneficial to the environment, these alternative vehicles need to be purchased by consumers in large quantities. In Europe, the transport sector is still the second largest source of greenhouse gas (GHG) emissions, accounting for about a quarter of total emissions [4]. The European Union set the target of reducing CO2 emissions in all sectors by 80% by 2050. Also, the 2011 EC White Paper on Transport set a target of reducing road transport emissions by 60% of 1990 levels by 2050, and within this to “halve the use of ‘conventionally fueled’ cars in urban transport by 2030 and phase them out in cities by 2050” [5]. Both short- and long-term estimates of EV market penetration vary greatly [6]. However, despite the rapidly growing sales,1 EVs have failed to capture a significant passenger car market share and continue to be dependent on support measures, such as financial incentives [7]. There are various factors that may promote the market uptake of EVs in Europe. Among these, a high GDP/capita, fuel cost savings from gasoline and petrol, a high share of renewable energy share (RES) in the energy mix and contribution to energy security are some crucial features of potential lead markets for EVs [8]. Italy is one of the main countries displaying such a potential (Fig. 1) [9], and South Tyrol (Fig. 2) is a particularly interesting case study. Located in the Italian Alps, this region exhibits most of the favorable conditions for the diffusion of EV and hydrogen fuel cell vehicles [8]. Corresponding author. E-mail addresses: nives.dellavalle@eurac.edu (N. DellaValle), alyona.zybaryeva@eurac.edu (A. Zubaryeva). 1 Source: European Automobiles Manufacturares Association Statistics. ⁎ https://doi.org/10.1016/j.erss.2019.05.005 Received 9 November 2018; Received in revised form 28 April 2019; Accepted 2 May 2019 2214-6296/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Fig. 1. Lead market forecast for BEV potential of NUTS2. Fig. 2. Study area showing the province examined in this study. Moreover, it has a population with high environmental2 and autonomy concerns.3 Firstly, these high concerns for the environment [10] and autonomy [11] may substantially contribute to promotion and adoption of renewable energy systems. Secondly, the high share of electricity produced in the region from hydro-power (91.6%) might unlock environmental benefits of EVs.4 Moreover, many economic barriers might be overcome thanks to the mix of high GDP/capita5 and the regional government's commitment to reduce CO2 emissions largely from the transport sector down to 1.5 tons per year per capita by 2050, this being the major contributor to CO2 emissions in the area [12]. In particular, to reach the goal of 60% of vehicle-km travelled represented by zero-emission vehicles by 2050 [13], the regional government has incentivized EV purchases through direct incentives; moreover, it has committed to improving the charging infrastructure,6 and to offering test drives and tax reductions to the citizens and companies.7 2 In 2016 the population voted against the opening of an airport via referendum – Source: Comune di Bolzano, Alto Adige. 3 In South Tyrol, there are three different language subgroups (Italian, German and Ladin speakers) with special cultural differences. Therefore, the Italian government guarantees South Tyrolean legislative and administrative autonomy greater than the competencies of a region governed under normal statute – Source: Provincia autonoma di Bolzano, Alto Adige. 4 Source: Istituto provinciale di statistica - Provincia autonoma di Bolzano, Alto Adige. 5 Source: Istituto provinciale di statistica - Provincia autonoma di Bolzano, Alto Adige. 6 Source: h2-suedtirol, Alperia Group. 7 Source: Green Mobility Alto Adige. 47 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Fig. 3. New BEV/PHEV registered in South Tyrol. Source: Ministero delle Infrastrutture e dei Trasporti. However, despite this fertile ground, since 2010 only 0.1% of total new vehicle registrations are represented by EVs in South Tyrol (Fig. 3).8 Why is EV uptake still low in South Tyrol? The reason might lie in the fact that material barriers are only some of many factors preventing EV adoption [2]. As an example, individuals might fail to adopt EVs because they tend to overvalue the high purchase price and undervalue the low lifecycle operating costs [14]. Such a pattern of behavior can be well described by the field of behavioral economics. This field has been influential in disrupting policy-making that has traditionally been based on a rational choice model of human behavior. By providing evidence that individuals exhibit systematic and predictable patterns of decision-making that depart from this model's assumptions (i.e. cognitive biases), behavioral economics has provided policy-makers not only with a richer model of human behavior, but also with a richer policy toolbox, which usually comprises taxes, incentives and regulations. One concrete illustration is choice architecture: by intervening in the decision problem, governments can redirect behavior without forbidding any option or changing economic incentives [15]. As an example, governments might implement strategies that simplify the decision problem and rely on the power of social influence or cognitive biases [16]. These strategies might help governments design interventions that are more effective at tackling several policy challenges [4]. In this study, we investigate how the challenge of low EV uptake can benefit from insights from behavioral economics. In particular, we test the effect of two behaviorally informed strategies to promote preferences for EVs: one that relies on social influence (i.e., norm-based intervention [17]), and another that simplifies the problem of deciding to purchase a vehicle (i.e., salience intervention [18]). Firstly, informing individuals that similar peers have already chosen to purchase EVs might be effective at shifting preferences for EVs. Secondly, making EV future cost savings salient might prove to be an effective strategy for helping individuals to appreciate future benefits of an EV. At the same time, heterogeneity in behaviour might also affect several individual choices [19]. As an example, those who display higher capacity to value 8 future consumption, i.e., higher willingness to delay benefits, are more likely to invest in energy-efficient appliances [20]. Therefore, to isolate the effect of our treatments on preferences for EVs, we control for major sources of heterogeneity that might correlate with preferences for EVs. The paper is divided into five parts. Section 2 explores the conceptual background and introduces the research questions. Section 3 presents the method and the sample. In Section 4, we perform the statistical analysis, of which the results are discussed in Section 5. Finally, Section 6 concludes. 2. Conceptual background and research questions To encourage diffusion of EVs, current governments, including South Tyrol, have promoted several incentivizing policies by providing financial subsidies, driving privileges, or tax rebates [21]. These policies have been proven to positively influence consumers’ EV adoption intentions [22], however the magnitude of their efficacy is often lower than expected [23]. One reason might be that these policies do not take into account consumer heterogeneity [24]. As these policies may turn out to be costly and inefficient, governments might look for alternative cost-effective strategies aimed at promoting large-scale EV adoption. As an example, they might implement strategies that act on the problem of deciding to redirect behavior without forbidding any option or changing economic incentives [15]. In particular, they might implement strategies that simplify the decision-making problem and rely on the power of social influence [16]. 2.1. Norm-based intervention EVs need to be adopted by a large mass of individuals to release their full potential on the environment. This implies that a collective behavioral change is required. A norm-based intervention, i.e., an intervention relying on social influence, can be a candidate tool to shift preferences in favor of EVs. However, compared to other interventions, it is crucial to take a more cautious approach before implementing a norm-based one, given that individuals might draw different inferences on the type of information provided. Therefore, we first need to diagnose the nature of behavior to change [25]. Source: Ministero delle Infrastrutture e dei Trasporti. 48 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva information provision making salient future cost savings, can be a candidate tool to circumvent discounting of future EV cost savings and promote preferences for EVs. Several studies have experimentally validated the positive effect of making future cost savings salient on uptake of energy efficient technologies, such as efficient TVs [41], energy efficiency measures [42], and energy efficiency appliances [43]. On the other hand, for what concerns eco-friendly vehicles, it is still unclear whether making future cost savings salient positively affects individuals’ choices in favor of eco-friendly vehicles. While [44] find that providing information in the framework of the European Energy Label rating scheme increases purchase decisions for energy-efficient vehicles, [45] note that this effect is smaller than that found in a study using a similar design on household appliances [46]. Similarly, [47] find that making running costs salient positively affects eco-friendly car choices, while [48] find that providing information about five-year cost savings does not change stated preferences for purchasing a gasoline, hybrid, plug-in hybrid or battery electric vehicle. There might be different reasons why there is still no agreement on the efficacy of this salience intervention on preferences for eco-vehicles. One reason may be that previous studies testing for the effect of salience of future cost savings did not control for major sources of unobserved heterogeneity that might affect the inference of preferences for EVs. In particular, previous studies did not account for factors that might contribute to the decision to adopt EVs. In this study, we add to previous studies testing for the effect of salience of future cost savings by examining sources of heterogeneity. Following [26]'s definition of collective behaviors, individuals’ choice over a particular type of vehicle might look like a custom – i.e. independently motivated choices of many individuals that happen to be similar to each other. Instead, it is a type of interdependent collective behavior depending on empirical expectations (i.e. descriptive norm) [26], since individuals prefer a type of vehicle depending, among other factors, on what they think (or see) most individuals in their reference group do. Therefore, to shift collective preferences towards EVs, it is crucial to change the empirical expectations about others’ preferences for EVs. A norm-based intervention providing information about what most individuals do can be a useful strategy for changing empirical expectations [17]. In particular, this intervention might be effective at inducing individuals to believe that there exists a descriptive norm about the target behavior (i.e. collective preferences for EVs). However, to be effective, the norm-based intervention must specify the reference group [17]. In particular, mere membership in a social group (such as gender, ethnicity, nationality) is one of the strongest sources for influencing behavior [27]. Therefore, when it is salient, it motivates individuals to conform to the behavior of others relevant to them [27]. Several studies have assessed the positive effect of norm-based interventions on pro-environment intentions or actual behavior. As an example, informing individuals about what others do has been proven to significantly improve household electricity use [28,29], intentions [30] and actual recycling behavior [31], towel reuse [32], and household water use [33]. On the other hand, no previous study has yet tested the role of a norm-based intervention in shifting preferences for EVs. Only [34] surveyed individuals on what individuals think others do about EVs, but it did not test for the effect of providing information about what relevant others do about EVs on preferences to adopt EVs. Norm-based interventions can be candidate tools for changing empirical expectations required to shift collective behaviors, such as preferences for EVs; however, their efficacy is always a matter of testing. As an example, [35] showed that simply providing information about others’ energy consumption might have a differential effect on individuals, depending on whether their behavior is below or above the suggested norm (i.e., boomerang effect). Therefore, in this study, we add to previous research on EVs by testing whether providing normative information is effective at shifting preferences for EVs. As suggested by [17], to increase the efficacy of the intervention, we decrease ambiguity of the reference group, by providing information about EV registrations made by members of an existing natural group (namely, South Tyroleans). 2.3. Some sources of heterogeneity To isolate treatment effects, it is crucial to isolate the role of different sources of hereogeneity that might interact with the choice to adopt EVs. In particular, the choice to adopt EVs can be influenced by certain factors underlying the decision making process, which might differ across individuals even after random assignment to different groups. As an example, the fact that EVs contribute to protecting the environment might contribute to drive intentions to adopt eco-friendly vehicles [49]. As anticipated in the introduction, South Tyroleans might appear to be willing to protect the environment, however, their underlying motivations might differ. For some individuals, protecting the environment can be part of their self-identity [50], while for others, protecting the environment underlines an intrinsic motivation to contribute to any public good [51]. Therefore, to better disentangle the effect our treatments on the likelihood to choose EVs, we control not only for pro-environment self-identity, but also for trust, reciprocity and altruism, these being primary predictors of decisions to contribute to a public good [52]. The decision to purchase an EV involves a degree of risk tolerance, since adoption consequences are uncertain [53,54]; moreover, it is an investment decision in which individuals trade-off utility at different points in time (i.e. high purchase price versus high future cost savings) [20,55]. In this study, we thus control for individual time discounting and risk aversion preferences. Finally, to better isolate the effect of our interventions, we also control for size, which has been proven to affect preferences for EVs. As an example, those who are willing to purchase large EVs (i.e., high-end EVs) have been found to be different from those who are not willing to purchase large EVs. While they have been found to be less sensitive to purchase price, their higher interest in cash incentives suggest that this interest is not related to the monetary value per se, but more in a form of acknowledgement of their greener behavior [56,57]. For potential high-end EV buyers, one may also hypothesize a conspicuous consumption effect, i.e. undertaking costly actions to signal their type as green [58], which cannot be disregarded when evaluating the treatment effect. 2.2. Salience intervention When purchasing a vehicle, individuals are usually poorly informed about and cognitively constrained in assessing fuel economy [36]. Moreover, EVs being more expensive than conventional vehicles, individuals might not adopt EVs, because they anchor to the high purchase price. This is due to the fact that individuals tend myopically to overvalue things that are closer in time, and undervalue those that are distant in future [14,37]. At the same time, due to a lack of the basic information needed to make optimal decisions as assumed by rational decision-making [38], individuals might also not adopt EVs because they make errors in estimating future cost savings. As argued by [39], this undervaluation of future consequences might help explain the general individual underinvestment in energy-efficient technologies (i.e. “energy paradox” [40]). This implies that governments might provide individuals with more information about fuel economy to help them make more informed purchasing decisions. However, information alone does not necessarily result in a change in behavior and, thus, in shifting preferences. Instead, information might be more effective at changing behavior when it is framed in a way that acknowledges individuals’ imperfect informationprocessing capacities [16]. A salience intervention [18], i.e., an 49 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva recent years.13 In the salience treatment, compared to the baseline, individuals were also provided information about the vehicles’ life-cycle operating costs (i.e. energy costs, variable costs, fixed costs and charging infrastructure) to make the EVs’ future cost savings salient.14 Before presenting the two hypothetical scenarios, a cheap talk message was sent to reduce the hypothetical bias, based on that of [60].15 In the third stage, respondents were asked to state a ranking over seven incentives (car-sharing, subsidies, free parking, leasing, use of bus lines, free highway, and better charging infrastructure) that might increase their willingness to adopt EVs.16 Then, they were presented a reduced version of the Collective Self-Esteem scale (CSE; [61]) and an item based on [62], to assess normative beliefs on Southern Tyroleans’ green behaviour. For the Norm treatment, these scales allow us to isolate the moderating role that identification with the group (i.e. South Tyroleans) and perceptions that others’ green behaviour is normal (or not) have on the intention to conform to the suggested descriptive norm. In the fourth section, we included eight items from the scale developed by [50], to measure pro-environment self-identity. Then, we adopted the reduced-form module validated by [63] as an instrument able to capture economic preferences (time and risk preferences, altruism, reciprocity and trust) in incentivized experiments.17 In the last section, we added a survey about transport habits, car statistics and a set of socio-demographic factors. 2.4. Research questions The aim of this study is to test the effect of two behaviorally informed interventions on preferences for EVs. In particular, we test whether making EV's future cost savings salient and whether informing about EV choices made by members of the reference group are two effective interventions for promoting preferences for EVs. To improve identification of treatment effect, we control for key factors that might confound preferences for EVs. In particular, we control for key economic preferences, pro-environment self-identity and size. Following [59]'s principles for more rigorous energy social science research, we clearly state the research questions that we aim to address in our study: 1. Is informing about what similar others do an effective intervention for promoting preferences for EVs? 2. Is making future cost savings salient an effective intervention for promoting preferences for EVs? 3. Does controlling for economic preferences, pro-environment selfidentity, and size improve identification of treatment effects on preferences for EVs? 3. Experimental design An online survey experiment was designed and 591 participants from South Tyrol completed the survey. The survey was distributed using newsletters and social media to reach a wider audience and was designed using a web-based survey system.9 Treatments were randomized in the survey redirect page. In particular, participants were assigned in a between-subject fashion to either the baseline, the Salience or the Norm treatments. The three surveys were the same except from the additional treatment-information provided in the hypothetical scenarios stage.10 3.2. Description of the subject pool The sample consists of 591 participants (see Table 1). 47% of the participants were female and 47.55% were between 26 and 40 years old. Most participants had a high school diploma (49 %), were German speakers (55%) and came from towns (29%) and small cities (29%). 64 % were employed and 51 % had an income between €12.000–35.000. The majority of respondents had on average 1.79 cars in the household. And these cars are either small (35%) or medium (35%) sized. Among these, only 1 % were EVs. For what concerns transport habits, the average number of flights per year is 2.6, and normally (78 %) green transports (train, foot, bus or bike) are used for the daily activities.18 When asked to evaluate the technical factors when purchasing a vehicle, on average respondents reported price (4.73) as the most important factor, followed by fuel type (3.79) and operating costs (3.72). Finally, when asked to consider which incentive would help increase their willingness to purchase an EV, respondents ranked subsidies (5.51) as the most effective, followed by free parking (4.99) and better charging infrastructures (4.4). 3.1. The survey To start with, respondents were asked to state a ranking over six technical factors (fuel type, operating costs, space, price, emissions, and autonomy) that they might take into account when they purchase a vehicle.11 Then, they were shown two hypothetical purchase scenarios in which information about price and technical characteristics was provided (i.e. speed, fuel type, autonomy, space, and emissions). In particular, they were asked to choose first between a small-sized EV and a small-sized conventional vehicle, and then between a large sport EV and a large sport conventional vehicle, to account for the influence of size.12 In this stage, we evaluate the effect of our treatments on EV choices. In the norm treatment, compared to the baseline, individuals were also provided with a descriptive norm about EV choices made by the members of the reference group (South Tyroleans) in the most 13 The graph showing the registrations made by South Tyroleans is available in the Appendix, and was preceded by the question: “Did you know that South Tyroleans increased in new Electric Vehicle registrations by 178 % in the period 2013–2017?”. 14 The life-cycle operating costs are available in the Appendix and were based on the results of the Total Cost of Ownership (TCO) calculator developed by the Oeko-Institut e.V. 15 “The experience from previous similar surveys is that people often state a higher willingness to pay than what one is actually willing to pay for the good. This difference in percentage is due to the fact that the purchasing questions are hypothetical. To avoid this so-called hypothetical bias, it is important that you make each of the following choices as if you would actually be facing these exact choices in a store.” 16 Ranking was collected on a scale 1–7, with 1 meaning “Not at all important”, and 7 meaning “Very important”. 17 Answers in the third and fourth stages were collected on a Likert scale 1–7, with 1 meaning “I totally disagree”, and 7 meaning “I totally agree”. 18 The extensive use of green transport for daily activities is representative of Southern Tyroleans’ high environmental concerns, as anticipated in the Introduction. 9 We used the web-based survey system Opinio. The survey is available in the Appendix. 11 Ranking was collected on a scale 1–6, with 1 meaning “Not at all important”, and 6 meaning “Very important”. 12 To make the two scenarios plausible, we focused on conventional and electric-fuelled cars that are in the top ten models sold in Italy and are comparable in size and technical features. Information about price, autonomy, space, emissions, etc., was taken from the most known vehicle guide in Italy (alVolante), so as to be accessible to all respondents. Since the aim was to select vehicles comparable in size and technical features, it was not possible to find vehicles with the same price and total costs. In particular, the price differential between the small EV and the conventionally fuelled vehicle is 11.000, and the total cost differential between the small EV and the conventionally fuelled vehicle −7.266. In contrast, the price differential between the big EV and the conventionally fuelled vehicle is 20.000, and the total cost differential between the big EV and the conventionally fuelled vehicle −7.721. 10 50 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Table 1 Description of the subject pool. Table 2 EVs choices. N % Demographics Female Age [18–25] Age [26–40] Age [41–65] Age [ > 65] Language [Italian] Language [German] Language [Ladin] Language [Other] City size [ < 2.000] City size [2.000–10.000] City size [10.001–100.000] City size [ > 100.000] Job [Unemployed] Job [Employed] Job [Self-employed] Job [Student] Education [Primary school] Education [High school] Education [Graduate] Education [Post graduate] Income [ < 12.000] Income [12.001–35.000] Income [35.001–60.000] Has children N house inhabitants 584 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 591 47%(0.5) 21.66%(0.41) 47.55%(0.5) 29.44%(0.46) 1.35%(0.12) 41%(0.41) 55%(0.5) 2%(0.15) 2%(0.14) 15%(0.35) 29%(0.46) 29%(0.46) 27%(0.44) 2.5%(0.16) 64%(0.48) 14%(0.34) 15%(0.36) 0.6%(0.082) 49%(0.5) 11%(0.31) 34%(0.48) 13.6%(0.34) 51%(0.5) 23%(0.42) 36%(0.48) 2.86(1.25) Car and transport statistics N cars Has small car Has medium car Has big car Already uses EV Green transport habits N flights last year 591 591 591 591 591 591 591 1.79(0.98) 35%(0.48) 35%(0.47) 15.6%(0.364) 1%(0.1) 78%(0.41) 2.6(1.93) Ranking: important factors when purchasing a car Fuel type 591 Operating costs 591 Space 591 Price 591 CO2 591 Autonomy 591 3.79(1.63) 3.72(1.47) 3.02(1.53) 4.73(1.56) 2.81(1.62) 2.94(1.62) Ranking: incentives to purchase EV Car-sharing Subsidy Free parking Leasing Bus line Free highway Better charging infrastructure 3(2.08) 5.51(1.72) 4.99(1.59) 3.37(1.59) 2.94(1.76) 3.78(1.68) 4.4(2.01) 591 591 591 591 591 591 591 Baseline Salience Norm % EV choices Treatment-baseline comparison 51.2% [46.4%–56%] 59.4% [54.5%–64.4%] 53% [48.2%–58%] 0.0223* 0.5729 p-Values of Wilcoxon rank-sum test: ***p < 0.001. **p < 0.01. o p < 0.1. 95% Confidence Intervals in parentheses [64]. * p < 0.05. Table 3 Logit regression models: EV choices. (1) (2) (3) Variables Norm Salience Big size Willingness to risk Willingness to delay Altruism Reciprocity Trust Social ID Pro-environment ID Green beliefs Individual characteristics Constant EV choice 0.0784(0.141) 0.334(0.143)* No EV choice 0.0909(0.163) 0.382(0.163) * 1.596(0.125)*** −0.0156(0.0485) 0.185(0.0682)** 0.0220(0.0633) −0.132(0.0900) 0.00475(0.0487) 0.00636(0.0662) 0.428(0.0752) *** −0.0721(0.0573) No EV choice 0.166(0.173) 0.413(0.167) * 1.739(0.140) *** −0.0299(0.0539) 0.223(0.0722) ** 0.00640 (0.0708) −0.210(0.0954) * 0.00201(0.0504) 0.0454(0.0754) 0.329(0.0805) *** −0.0979(0.0601) Yes 0.0486 (0.0986) −0.769(0.563) −4.807(1.744) ID Observations Pseudo R2 591 1182 0.0035 591 1182 0.1378 584 1168 0.1953 ** Robust standard errors in parentheses. *** p < 0.001. ** p < 0.01. * p < 0.05. o p < 0.1. This insight is confirmed by the logit regression Model 1 (Table 3), in which we estimate the probability that the EV is chosen using as explanatory variables the two treatments (Norm is equal to 1 when respondents are assigned to the norm-based treatment and zero otherwise; Salience is equal to 1 when respondents are assigned to the salience treatment and zero otherwise). To measure the positive effect of Salience, we need to convert the coefficients βi of the logistic regression from log-odds units into odds-ratios Exp(βi). In particular, the value of odds ratios for Salience is 1.396 in Model (1), meaning that, compared to the Baseline, the odds ratio of choosing EVs increases by 1.396 times in Model (1). Therefore, results: 4. Results In this section, we first answer the three research questions. Then, we examine results with a more exploratory approach. 4.1. Confirmatory analysis 1. do not confirm that informing about what similar others do is an effective intervention for promoting preferences for EVs; 2. confirm that making future cost savings salient is an effective intervention for promoting preferences for EVs. To answer the first two questions, we assess whether EV choices differ across treatments. Table 2 shows that EV choices are affected only by the salience treatment. We find that, compared to the Baseline (51.2%) the percentage of EV choices is significantly higher for the salience intervention (59.4%) (Table 2). On the other hand, while the percentage of EV choices in the Norm treatment(53%) is higher than that in the Baseline, this is not statistically significant. To answer the third question, we eliminate potential sources of unobserved heterogeneity that might correlate with EV choices. In particular, we estimate the probability that the EV is chosen using as 51 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Table 4 Conditional marginal effects. Delta-method 1. Salience Below median WT delay No Yes Observations 0.0895(0.041) 0.037(0.0453) 591 * Below median proenvironment ID Big size 0.105(0.0394)** 0.037(0.0439) 591 0.095(0.044) * 0.04996(0.0392) 591 Robust standard errors in parentheses. ***p < 0.001. ** p < 0.01. * p < 0.05. o p < 0.1. explanatory variables not only the two treatment dummies, but also the measures of pro-environment identity, key economic preferences (willingness to risk, willingness to delay benefits, altruism, reciprocity and trust), and size. Moreover, we add measures of beliefs about other southern Tyroleans’ green behaviour and social identification as these might confound the effect of the norm-based treatment. First, the logit regression (Model (2) – Table 3) confirms that there are positive significant differences only between EV choices made in the baseline and in the salience treatment also when we control for sources of heterogeneity (the value of odds ratios for Salience is 1.465 in Model (2) meaning that, compared to the Baseline, the odds ratio of choosing EVs increases by 1.465 times when controlling for sources of heterogeneity). Second, it suggests treatment effect heterogeneity. In particular, those who display higher capacity to value future events, i.e. higher willingness to delay benefits, and those who identify themselves as pro-environment individuals, are more likely to choose EVs. Moreover, when the choice is between a big conventional fuelled vehicle and a big EV, the likelihood to choose EVs increases, compared with when the choice is between a small conventionally fuelled vehicle and a small EV. To confirm treatment effect heterogeneity, we look at heterogeneous responses to Salience with respect to the measures of preferences to delay benefits, pro-environment self-identity and size. In particular, to see whether the salience treatment also has a positive impact on individuals with low time-preferences and pro-environment self-identity values, we categorize individuals depending on whether their level of willingness to delay benefits and pro-environment selfidentity is below the median level of willingness to delay and pro-environment self-identity in the sample. Moreover, to see whether the salience treatment also has a positive impact on individuals who are not motivated by a potential conspicuous consumption effect, we also see the marginal effect of Salience for different vehicle sizes. The marginal effects of Salience (Table 4 and Fig. 4) show that the difference between responses of those assigned to the baseline and those assigned to the salience treatment is significantly different only for high values of pro-environment identity and willingness to delay, and when the vehicle size is big. Hence, compared to the baseline, the positive difference in the probability to choose EVs between those who are assigned to the salience treatment and to the baseline increases significantly conditionally on size, individual pro-environment selfidentity and time preferences. Therefore, results confirm that: Fig. 4. Conditional marginal effects. 4.2. Exploratory analysis The choice to adopt EVs can be influenced by further factors that might differ across individuals, even after the random assignment to different treatment groups. Therefore, in Model 3 (Table 3), we estimate the probability that the EV is chosen using as explanatory variables other individual characteristics, in addition to the variables used in Model 2 (see Table A2 in the Appendix for more detailed results). We find that the positive effect of Salience is still significant (the value of odds ratios for Salience is 1.511 in Model (3) meaning that, compared to the Baseline, the odds ratio of choosing EVs increases by 1.511 times when controlling also for individual characteristics). Interestingly, we find that when we account for individual characteristics, the negative coefficient associated with preferences for Reciprocity becomes significant: without adjusting for individual characteristics, unexplained variability is large and the effect from higher preferences for reciprocity is too small to emerge. As suggested in Section 2.3, displaying preferences for reciprocity can provide the basis 1. Controlling for economic preferences, pro-environment self-identity, and size improves identification of (salience) treatment effects on preferences for EVs. 52 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva for individuals to decide to contribute to a public good. This implies that reciprocal individuals are more willing to contribute to a public good when others are also willing to contribute [65], but when they expect others to free ride from contributing, they “punish” others by free riding too [66]. Looking at individual characteristics, we find that, compared to German speakers, those who speak languages other than Ladin and Italian are more likely to choose EVs. Similarly, we find that compared to those who are employed, those who are not currently working and – only slightly – those who are self-employed are more likely to choose EVs. Not surprisingly, those who are already using EVs are more likely to choose EVs. Finally, we find that for those whose CO2 emissions and fuel type are very important when purchasing a vehicle, the likelihood to choose EVs increase substantially, and slightly for those whose operating costs, purchasing price and autonomy are very important when purchasing a vehicle. On the other hand, compared to those who come from a big city, respondents coming from small towns, are less likely to choose EVs. buy EVs rather than actual purchasing choices. However, while there is some evidence suggesting that purchasing intentions do not predict well actual purchasing choices [69], we believe that this study provides useful methodological insights to policy makers on how to design complementary interventions aimed at promoting EV adoption. Future research could attempt to validate the findings from this study in the field. Secondly, to isolate the effect of our treatments, we controlled for sources of heterogeneity drawing from the behavioural economic literature. Although the focus of the study was to identify treatment effects, accounting for other factors that might correlate the willingness to choose EVs, such as choice overload, perceived behavioral control, pioneering spirit and personality traits such as openness [23], would improve the validity of the findings. Of course, this would make the model less parsimonious. 6. Conclusions To be effective on the environment, EVs have to be adopted by a large mass of individuals. However, the market, even in regions (such as South Tyrol) that are ready to tackle some material barriers related to EV adoption, is still dominated by conventional vehicles. Several policy mechanisms are being implemented to incentivize a market for EVs. However, these have been proven to be costly and not fully effective. One reason lies in the fact that these policies tackle most of the material barriers related to EV diffusion, but often disregard not only the role of cognitive biases associated with the decision problem of adopting EVs, but also potential heterogeneous responses to interventions. In this study, we examined to what extent two interventions enriched with behavioral economic insights are effective at simplifying the decision problem related to EV adoption. In particular, we tested the effect of a norm-based intervention, i.e., providing information about EV choices made by the reference group, and that of a salience intervention, i.e., making future cost savings salient, on EV choices. Moreover, to better isolate treatment effects, we controlled for major sources of individual heterogeneity that might affect EV choices, namely economic preferences, pro-environment self-identity and size. We find that while providing information about the choices of the reference group has a positive coefficient on the likelihood to choose EVs, it does not reach a significant level. On the other hand, we find that making future cost savings salient is an effective strategy to increase the likelihood to choose EVs, and that heterogeneity in behavior affects responses to the treatment. While this result confirms previous studies already testing the role of salience on EV choices, it provides new insights by proving that it can be effective on individuals who are already motivated to choose EVs. Overall, results suggest the need to always test the type of information embedded in norm-based interventions, since individuals’ willingness to change depends on the information provided. At the same time, they provide evidence of the need to measure heterogeneity in behavior before implementing information strategies aimed at shifting preferences for EVs. Therefore, they highlight the importance of always scanning the target context before implementing the intervention in the field [70]. Practically, in the context of large-scale surveys, governments might promote the inclusion of items, such as those developed by [63], to capture predictors of heterogeneity in behavior. This would allow a better grasp of behavioral levers (barriers) that might enhance (inhibit) the efficacy of a particular information strategy. In particular, governments can start targeting the subgroups displaying a higher probability to positively react to the salience intervention. Thereafter, they might implement a large-scale norm-based intervention, as soon as individuals will start reacting to the salience intervention (the norm-based 5. Discussion In this study, we experimentally tested how a salience and a normbased intervention affect individual willingness to choose EVs in South Tyrol. Using an online survey study, we randomly assigned residents in South Tyrol to hypothetical scenarios to buy either a conventionally fuelled vehicle or an EV, which were informing either about EV choices made by South Tyroleans in the last year, or which were making salient future cost savings. In addition, we isolated treatment effects by controlling for major sources of heterogeneity that might impact EV choices. While we found a positive effect of the salience intervention on EV choices, there was no effect for the norm-based intervention. While there might be several reasons why the norm-based intervention was ineffective, we speculate that it is related to asymmetry in what individuals inferred from the normative information we provided [17]. In particular, different individuals have not only different empirical expectations about the number of people following the prescribed descriptive norm, but also different thresholds for how many people are necessary to conform to the norm [67,68]. Therefore, we believe that the descriptive norm we provided was ineffective, because it did not reach the threshold of people required to change empirical expectations. In particular, individuals might have perceived EV registrations made by South Tyroleans in the last years still as too low. Future research testing norm-based interventions should investigate whether individuals perceive that the normative information provided represents a sufficient threshold to change their empirical expectations. For what concerns the salience intervention, we confirmed the findings of previous studies supporting that it is effective at shifting preferences for EVs [44,47]. In addition, by controlling for major sources of heterogeneity, we showed that the effect is limited to those who display high values in pro-environment self-identity, and preferences for big size and for delaying benefits. Finally, when looking at the effect of individual characteristics, we found reciprocal individuals are less likely to choose EVs. This suggests that while they might share the belief that EVs are good for the environment, they might believe that too few individuals are adopting them: the beliefs that there are non-contributors to the public good induce reciprocal types to make selfish choices, i.e. not to contribute to protect the environment by choosing EVs. Overall, our results contribute to the research on EV adoption, by experimentally testing the impact of two cost-effective interventions on EV choices, and by isolating the effect through the account of major sources of heterogeneity in behavior. Nevertheless, our study presents some limitations. Firstly, this study measured hypothetical choices to 53 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Acknowledgements intervention might be more effective when it provides information about a high percentage of individuals already engaging in the target behavior). Hence, while we conclude that norm-based interventions should be implemented at a later stage, future research should examine how other behaviorally informed interventions affect EV adoption. Overall, our study shows that information strategies acknowledging individuals’ imperfect information-processing capacities can be costeffective interventions for promoting EV adoption. Therefore, it is of practical relevance to local governments that face an increasing need to develop complementary interventions to promote alternative-fuelled vehicles’ diffusion. The research leading to these results is part of the project Culture Building and territorial Development. Come preparare un territorio alla rivoluzione disruptive dell’e-mobility, funded by Jaguar Land Rover Italia, in collaboration with Giulia Isetti, Philipp Corradini, Mirjam Gruber and Gerhard Vanzi (Eurac Research). We are especially grateful to the editors and the anonymous referees for their encouragement and constructive advice. We would also like to acknowledge the colleagues of the Urban and Regional Energy System at the Institute for Renewable Energy - Eurac Research, who supported this research by providing their feedback. The authors thank the Department of Innovation, Research and University of the Autonomous Province of Bozen/Bolzano for covering the Open Access publication costs. Conflict of interest The author declare that there are no conflict of interests. Appendix A Figs. A1–A3 Fig. A1. Baseline. treatment. 54 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Fig. A2. Salience treatment. 55 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Fig. A3. Norm treatment. Table A1 Table A1 Description of costs in the Salience treatment. Infrastructural costs Fuel/electricity costs Variable costs Fixed costs Total costs Price Car type: small EV Conventionally fuelled Difference 686 0 686 2776 8409 −5633 1812 3062 −1250 6339 7408 −1069 11,613 18,879 −7266 39,150 28,150 11,000 Car type: big EV Conventionally fuelled Difference 686 0 686 3409 9074 −5665 2593 4059 −1466 7685 8961 −1276 14,373 22,094 −7721 88,000 68,000 20,000 56 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva Table A2 Logit regression model (3) – individual characteristics. (3) Variables Norm Salience Big size Willingness to risk Willingness to delay Altruism Reciprocity Trust Social ID Pro-environment ID Green beliefs Female Age [18–25] Age [26–40] Age [41–65] Language [Italian] Language [Ladin] Language [Other] City size [ < 2.000] City size [2.000–10.000] City size [10.001–100.000] Job [Unemployed] Job [Self-employed] Student Education [High school] Education [Graduate] Education [Post graduate] Income [12.001–35.000] Income [35.001–60.000] N house inhabitants Has children N cars Already uses EV Has small car Has big car N flights last year Ranking [Fuel type] Ranking [CO2] Ranking [Operating costs] Ranking [Price] Ranking [Autonomy] Ranking [Car-sharing] Ranking [Subsidy] Ranking [Free parking] Ranking [Leasing] Ranking [Bus line] Ranking [Better charging infrastructure] Constant ID Observations Pseudo R2 EV choice 0.166(0.173) 0.413(0.167) * 1.739(0.140)*** −0.0299(0.0539) 0.223(0.0722) ** 0.00640(0.0708) −0.210(0.0954) * 0.00201(0.0504) 0.0454(0.0754) 0.329(0.0805) *** −0.0979(0.0601) 0.0600(0.152) −0.107(0.704) 0.00900(0.654) −0.137(0.647) 0.247(0.181) 0.612(0.520) 0.930(0.435) * −0.412 (0.244) * −0.118(0.209) −0.252(0.193) 1.044(0.422) * 0.398(0.220)o 0.210(0.319) −0.237(0.326) −0.143(0.357) −0.0742(0.382) 0.239(0.207) 0.134(0.253) 0.0705(0.0670) 0.151(0.212) −0.00352(0.0846) 2.098(1.043) * −0.139(0.171) −0.321(0.208) 0.00821(0.0427) 0.296(0.0617) *** 0.345(0.0613) *** 0.130(0.0645) * 0.153(0.0651) * 0.125(0.0572) * −0.0394(0.0510) 0.0488(0.0620) 0.0954(0.0659) 0.0352(0.0626) −0.0453(0.0737) −0.00921(0.0582) −4.807(1.744) ** 584 1,168 0.1953 Robust standard errors in parentheses *** p < 0.001. ** p < 0.01. * p < 0.05. o p < 0.1. 57 N. DellaValle and A. Zubaryeva WTD How willing are you to give up something that is beneficial for you today in order to benefit more from that in future? WTR How willing are you to take risks? Altruism How willing are you to donate to causes without expecting anything in return? Reciprocity When someone does me a favour I am willing to return it Trust I assume that people have only the best intentions 58 Environmental ID I think of myself as someone who is very concerned with environmental issues Respecting the environment is an important part of my identity I turn off lights when I do not use them I buy energy- efficient appliances I drive economically (e.g. braking or accelerating gently) I walk, cycle or take public transport for short journeys (i.e. trips of < 3 miles) I share a car journey with someone else I cut down on the amount you fly Normative beliefs Many people in South Tyrol engage in environmentally friendly behaviour Disagree Disagree Somewhat Neutral Agree somewhat Agree Strongly agree 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 5 5 5 6 6 6 6 6 7 7 7 7 7 1 1 2 2 3 3 4 4 5 5 6 6 7 7 1 2 3 4 5 6 7 1 1 2 2 3 3 4 4 5 5 6 6 7 7 1 1 2 2 3 3 4 4 5 5 6 6 7 7 Energy Research & Social Science 55 (2019) 46–61 CSR Overall, my social groups are considered well by others Overall, my group memberships have very little to do with how I feel about myself In general, I’m glad to be a member of the social groups I belong to In general, belonging to social groups is an important part of my self image Strongly disagree N. DellaValle and A. Zubaryeva Transport questions How many cars do you have in your households? Do you have a driving license? Have you recently bought a new/ nearly new car or a used car in the last 24 months? Which car type do you use? Which fuel do you use? 59 1 Yes Yes 2 Not Not 3 4 5 6 7 Small Petrol Medium Diesel Large gas Sport Hybrid SUV Electric (not plugin) VAN electric Other other Do you own or lease the car that you use? Own Lease For how many km is this car driven? Less than 2000 km 2.000–10.000 km Someone in the household owns it 10.001–20,000 Someone in the household leases it 20.0001–30,000 30,001–40,000 40.001–50.000 For which activity do you most use this car? Commute to work/ school 1 Car Car Car Car Daily activities Shopping Business trips Leisure 2 Bicycle Bicycle Bicycle Bicycle 3 Scooter Scooter Scooter Scooter 4 Train Train Train Train 5 Bus/tram/metro Bus/tram/metro Bus/tram/metro Bus/tram/metro Going on holiday 6 Airplane Airplane Airplane Airplane More than 50.0001 other 26–40 Male 2001–10.000 2 Not Not Not Employed 41–65 More than 65 10,001–100.000 3 More than 100.000 4 More than 4 Student Voluntary worker Retired Family career High school 24.001–35.000 Bachelor 35.001–60.000 Master More than 60.000 € PhD How many flights have you made last year? Which transport vehicle do you use for going on holidays? Which transport vehicle do you use for commuting to work/school? Which transport vehicle do you use for shopping? Which transport vehicle do you use for leisure (sport, social visit, day trip)? Sociodemographics Which is your age? Which is your gender? How big (in terms of inhabitants) is the place where you live? How many people live in your house? Are you married or do you live with your partner? Do you have disabilities? Do you have children? What describes you best? Which is your highest educational qualification? Which is approximately your gross household annual income? 18–25 Female Up to 2000 1 Yes Yes Yes Unemployed/Job seeker Elementary Less than 12,000 € Middle 12.001–24.000 7 Other Other Other Other Energy Research & Social Science 55 (2019) 46–61 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva References experiment on curbside recycling, Basic Appl. Soc. Psychol. 21 (1999) 25–36. [32] W.P. Schultz, A.M. Khazian, A.C. Zaleski, Using normative social influence to promote conservation among hotel guests, Soc. Infl. 3 (2008) 4–23. [33] P.J. Ferraro, J.J. Miranda, M.K. Price, The persistence of treatment effects with norm-based policy instruments: evidence from a randomized environmental policy experiment, Am. Econ. Rev. 101 (2011) 318–322. [34] J. Jansson, A. Nordlund, K. Westin, Examining drivers of sustainable consumption: the influence of norms and opinion leadership on electric vehicle adoption in Sweden, J. Clean. Prod. 154 (2017) 176–187. [35] P.W. Schultz, J.M. Nolan, R.B. Cialdini, N.J. Goldstein, V. Griskevicius, The constructive, destructive, and reconstructive power of social norms, Psychol. Sci. 18 (2007) 429–434. [36] H. Allcott, C. Knittel, Are Consumers Poorly Informed about Fuel Economy? Evidence from Two Experiments, Technical Report, Natl. Bureau Econ. Res. (2017). [37] S. Frederick, G. Loewenstein, T. O’donoghue, Time discounting and time preference: a critical review, J. Econ. Liter. 40 (2002) 351–401. [38] T.S. Turrentine, K.S. Kurani, Car buyers and fuel economy? Energy Policy 35 (2007) 1213–1223. [39] H. Allcott, N. Wozny, Gasoline prices, fuel economy, and the energy paradox, Rev. Econ. Stat. 96 (2014) 779–795. [40] A.B. Jaffe, R.N. Stavins, The energy-efficiency gap what does it mean? Energy Policy 22 (1994) 804–810. [41] S.L. Heinzle, Disclosure of energy operating cost information: a silver bullet for overcoming the energy-efficiency gap? J. Consum. Policy 35 (2012) 43–64. [42] R.G. Newell, J. Siikamäki, Nudging energy efficiency behavior: the role of information labels, J. Assoc. Environ. Resour. Econ. 1 (2014) 555–598. [43] M. Andor, A. Gerster, S. Sommer, Consumer inattention, heuristic thinking and the role of energy labels, Ruhr Economic Papers (2017). [44] R. Wüstenhagen, K. Sammer, Wirksamkeit umweltpolitischer anreize zum kauf energieeffizienter fahrzeuge: eine empirische analyse schweizer automobilkunden, Zeitschrift für angewandte Umweltforschung (J. Environ. Res.) 18 (2007) 61–78. [45] J. Kaenzig, R. Wüstenhagen, The effect of life cycle cost information on consumer investment decisions regarding eco-innovation, J. Ind. Ecol. 14 (2010) 121–136. [46] K. Sammer, R. Wüstenhagen, The influence of eco-labelling on consumer behaviourresults of a discrete choice analysis for washing machines, Bus. Strat. Environ. 15 (2006) 185–199. [47] C. Codagnone, G.A. Veltri, F. Bogliacino, F. Lupiáñez-Villanueva, G. Gaskell, A. Ivchenko, P. Ortoleva, F. Mureddu, Labels as nudges? An experimental study of car eco-labels, Econ. Polit. 33 (2016) 403–432. [48] J. Dumortier, S. Siddiki, S. Carley, J. Cisney, R.M. Krause, B.W. Lane, J.A. Rupp, J.D. Graham, Effects of providing total cost of ownership information on consumers’ intent to purchase a hybrid or plug-in electric vehicle, Transport. Res. A: Policy Pract. 72 (2015) 71–86. [49] J. Jansson, A. Marell, A. Nordlund, Green consumer behavior: determinants of curtailment and eco-innovation adoption, J. Consum. Market. 27 (2010) 358–370. [50] L. Whitmarsh, S. O’Neill, Green identity, green living? The role of pro-environmental self-identity in determining consistency across diverse pro-environmental behaviours, J. Environ. Psychol. 30 (2010) 305–314. [51] R. Bénabou, J. Tirole, Incentives and prosocial behavior, Am. Econ. Rev. 96 (2006) 1652–1678. [52] P. Kollock, Social dilemmas: the anatomy of cooperation, Annu. Rev. Sociol. 24 (1998) 183–214. [53] M. Farsi, Risk aversion and willingness to pay for energy efficient systems in rental apartments, Energy Policy 38 (2010) 3078–3088. [54] Y. Qiu, G. Colson, C. Grebitus, Risk preferences and purchase of energy-efficient technologies in the residential sector, Ecol. Econ. 107 (2014) 216–229. [55] R.G. Newell, J. Siikamäki, Individual time preferences and energy efficiency, Am. Econ. Rev. 105 (2015) 196–200. [56] C.D. Higgins, M. Mohamed, M.R. Ferguson, Size matters: how vehicle body type affects consumer preferences for electric vehicles, Transport. Res. A: Policy Pract. 100 (2017) 182–201. [57] S. Hardman, E. Shiu, R. Steinberger-Wilckens, Comparing high-end and low-end early adopters of battery electric vehicles, Transport. Res. A: Policy Pract. 88 (2016) 40–57. [58] S.E. Sexton, A.L. Sexton, Conspicuous conservation: The Prius halo and willingness to pay for environmental bona fides, J. Environ. Econ. Manage. 67 (2014) 303–317. [59] B.K. Sovacool, J. Axsen, S. Sorrell, Promoting novelty, rigor, and style in energy social science: towards codes of practice for appropriate methods and research design, Energy Res. Soc. Sci. (2018). [60] R.G. Cummings, L.O. Taylor, Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method, Am. Econ. Rev. 89 (1999) 649–665. [61] R. Luhtanen, J. Crocker, A collective self-esteem scale: self-evaluation of one's social identity, Pers. Soc. Psychol. Bull. 18 (1992) 302–318. [62] K. Ando, S. Ohnuma, E.C. Chang, Comparing normative influences as determinants of environmentally conscious behaviours between the USA and Japan, Asian J. Soc. Psychol. 10 (2007) 171–178. [63] A. Falk, A. Becker, T. Dohmen, D. Huffman, U. Sunde, The preference survey [1] International Energy Agency, Global EV Outlook 2016, (2016) Online (accessed 14.06.18) https://www.iea.org/publications/freepublications/publication/Global_ EV_Outlook_2016.pdf. [2] Z. Rezvani, J. Jansson, J. Bodin, Advances in consumer electric vehicle adoption research: a review and research agenda, Transport. Res. D: Transp. Environ. 34 (2015) 122–136. [3] N. Bergman, Stories of the future: personal mobility innovation in the United Kingdom, Energy Res. Soc. Sci. 31 (2017) 184–193. [4] J. Sousa Lourenco, E. Ciriolo, S. Rafael Rodrigues Viera De Almeida, X. Troussard, Behavioural Insights Applied to Policy – European Report 2016, Publications Office of the European Union, 2016. [5] European Commission, A Roadmap for moving to a competitive low carbon economy in 2050, (2011) Online (accessed 31.05.18) https://eur-lex.europa.eu/ legal-content/EN/TXT/PDF/?uri=CELEX:52011DC0112&from=EN. [6] A. Zubaryeva, C. Thiel, E. Barbone, A. Mercier, Assessing factors for the identification of potential lead markets for electrified vehicles in Europe: expert opinion elicitation, Technol. Forecast. Soc. Change 79 (2012) 1622–1637. [7] C. Thiel, J. Krause, P. Dilara, Electric vehicles in the EU from 2010 to 2014 – is full scale commercialisation near, JRC Sci. Policy Rep. (2015). [8] A. Zubaryeva, C. Thiel, Analyzing potential lead markets for hydrogen fuel cell vehicles in Europe: expert views and spatial perspective, Int. J. Hydrog. Energy 38 (2013) 15878–15886. [9] A. Zubaryeva, C. Thiel, N. Zaccarelli, E. Barbone, A. Mercier, Spatial multi-criteria assessment of potential lead markets for electrified vehicles in Europe, Transport. Res. A: Policy Pract. 46 (2012) 1477–1489. [10] E.H. Noppers, K. Keizer, J.W. Bolderdijk, L. Steg, The adoption of sustainable innovations: driven by symbolic and environmental motives, Glob. Environ. Change 25 (2014) 52–62. [11] F. Ecker, U.J. Hahnel, H. Spada, Promoting decentralized sustainable energy systems in different supply scenarios: the role of autarky aspiration, Front. Energy Res. 5 (2017) 14. [12] Klimaland, PIANO CLIMA Energia-Alto Adige-2050, (2011) Online (accessed 14.06.18) http://www.provincia.bz.it/agenzia-ambiente/download/PianoClima_ Energia_AA2050_Ansicht.pdf. [13] M.G. Prina, M. Cozzini, G. Garegnani, G. Manzolini, D. Moser, U.F. Oberegger, R. Pernetti, R. Vaccaro, W. Sparber, Multi-objective optimization algorithm coupled to EnergyPLAN software: the EPLANopt model, Energy 149 (2018) 213–221. [14] G. Loewenstein, D. Prelec, Anomalies in intertemporal choice: evidence and an interpretation, Q. J. Econ. 107 (1992) 573–597. [15] R.H. Thaler, C.R. Sunstein, Nudge: improving decisions about health, Wealth Happiness 6 (2008). [16] C.R. Sunstein, Misconceptions about nudges, J. Behav. Econ. Policy 2 (1) (2018) 61–67. [17] C. Bicchieri, E. Dimant, Nudging with care: the risks and benefits of social information, Public Choice (2019) Forthcoming. [18] M. Lehner, O. Mont, E. Heiskanen, Nudging – a promising tool for sustainable consumption behaviour? J. Clean. Prod. 134 (2016) 166–177. [19] A. Falk, A. Becker, T.J. Dohmen, B. Enke, D. Huffman, The nature and predictive power of preferences: global evidence, CEPR Discussion Papers (2015). [20] H. Allcott, D. Taubinsky, Evaluating behaviorally motivated policy: experimental evidence from the lightbulb market, Am. Econ. Rev. 105 (2015) 2501–2538. [21] W. Li, R. Long, H. Chen, Consumers’ evaluation of national new energy vehicle policy in china: an analysis based on a four paradigm model, Energy Policy 99 (2016) 33–41. [22] Y. Zhang, Y. Yu, B. Zou, Analyzing public awareness and acceptance of alternative fuel vehicles in china: the case of EV, Energy Policy 39 (2011) 7015–7024. [23] W. Li, R. Long, H. Chen, J. Geng, A review of factors influencing consumer intentions to adopt battery electric vehicles, Renew. Sustain. Energy Rev. 78 (2017) 318–328. [24] E.H. Green, S.J. Skerlos, J.J. Winebrake, Increasing electric vehicle policy efficiency and effectiveness by reducing mainstream market bias, Energy Policy 65 (2014) 562–566. [25] C. Bicchieri, Diagnosing Norms, Oxford University Press, 2016. [26] C. Bicchieri, The Grammar of Society, Technical Report, Cambridge University Press, 2006. [27] H. Tajfel, M.G. Billig, R.P. Bundy, C. Flament, Social categorization and intergroup behaviour, Eur. J. Soc. Psychol. 1 (1971) 149–178. [28] J.M. Nolan, P.W. Schultz, R.B. Cialdini, N.J. Goldstein, V. Griskevicius, Normative social influence is underdetected, Pers. Soc. Psychol. Bull. 34 (2008) 913–923. [29] H. Allcott, Social norms and energy conservation, J. Publ. Econ. 95 (2011) 1082–1095. [30] F. Fornara, G. Carrus, P. Passafaro, M. Bonnes, Distinguishing the sources of normative influence on proenvironmental behaviors: the role of local norms in household waste recycling, Group Process. Interg. Relat. 14 (2011) 623–635. [31] P.W. Schultz, Changing behavior with normative feedback interventions: a field 60 Energy Research & Social Science 55 (2019) 46–61 N. DellaValle and A. Zubaryeva [64] [65] [66] [67] schemes, Am. Econ. Rev. 97 (2007) 999–1012. [68] M. Granovetter, Threshold models of collective behavior, Am. J. Sociol. 83 (1978) 1420–1443. [69] V.G. Morwitz, J.H. Steckel, A. Gupta, When do purchase intentions predict sales? Int. J. Forecast. 23 (2007) 347–364. [70] N. DellaValle, A. Bisello, J. Balest, In search of behavioural and social levers for effective social housing retrofit programs, Energy Build. 172 (2018) 517–524. module: a validated instrument for measuring risk, time, and social preferences, Netspar discussion paper (2016). G. Cumming, The new statistics: why and how, Psychol. Sci. 25 (2014) 7–29. R. Sugden, Reciprocity: the supply of public goods through voluntary contributions, Econ. J. 94 (1984) 772–787. E. Fehr, S. Gächter, Fairness and retaliation: the economics of reciprocity, J. Econ. Perspect. 14 (2000) 159–181. D. Sliwka, Trust as a signal of a social norm and the hidden costs of incentive 61