Characteristics of potential buyers of low-pollution vehicles: the case of Santiago de Chile

Abstract

Due to the high emissions produced by the transport sector, one of the most implemented initiatives in developed countries in the last decade is the electrification of the public transport fleet and the private vehicle fleet. The purchase of electric and hybrid vehicles in developing countries is negligible compared to developed countries and little has been indicated about the characteristics of potential buyers of this type of technology. In this paper, we aim to cover this aspect. Chile is a good case to study, as it is an emerging country with the highest level of penetration of electric and hybrid vehicles in the market and with better import scenarios according to the free trade agreements signed with the USA, Europe, and Asia. The results show that the socio-demographic characteristics with the greatest impact on the likelihood of purchasing low-emission vehicles, in the absence of any incentives, are household size, educational level, and gender. Particularly the marginal effects indicate that of the individuals surveyed who belong to a two-person household, they are 9% more likely to choose to buy a low-pollution vehicle compared to single people, and 8.96% less likely to choose a high-pollution vehicle compared to a single person.

IMpact STATEMENT

This paper examines which socio-demographic aspects are most relevant to the purchase of electric and hybrid vehicles. The results suggest that income, schooling, and household size are three of the key aspects that should be considered in the adoption strategies for electric and hybrid vehicles.

Keywords:

• Stated preference
• ordered probit model
• buyer characteristics
• low-emission vehicles
• transport decarbonization

REVIEWING EDITOR:

• Robert Read
• Economics
• University of Lancaster
• Lancaster
• UK

SUBJECTS:

• Politics & Technology
• Public Administration & Management
• Regulatory Policy

1. Introduction

The United Nations Framework Convention on Climate Change has encouraged UN member states to establish actions and public policies to mitigate the effect of climate change. In particular, it calls for efforts to reduce greenhouse gas emissions from private cars. The Paris Agreement commits countries to keep global warming below 2 °C and encourages them to double their efforts to keep it below 1.5 °C. This agreement was ratified by the 197 member countries during the Framework Convention on Climate Change. Electric mobility is a key component to achieve this goal because the transport sector is responsible for 25% of global CO₂ carbon dioxide emissions (McCollum et al., 2018; Tran et al., 2012; Urrutia-Mosquera et al., 2023; Urrutia-Mosquera & Flórez-Calderón, 2023).

Developed countries such as Norway, the United States, the Netherlands, France, Japan, South Korea, Germany, and England have been pioneers in testing various policies and incentives to stimulate the adoption and use of electric and hybrid vehicles. The increase in the electric vehicle fleet in these countries aims to reduce the current levels of CO₂ produced by private car use. This has led these countries over the last decade to develop strong campaigns to disseminate and monitor the increase in the number of vehicles on the market and in use, and to regularly monitor the diffusion and use of electric and hybrid vehicles (Rietmann & Lieven, 2019; Hardman, 2019; Urrutia-Mosquera & Fábrega, 2021).

The literature on promotion policies is extensive (Hardman, 2019; Van Eck et al., 2019), however, the literature on the characteristics of potential buyers of low-pollution vehicles is much scarcer. Recently a couple of papers have shed valuable light on the socio-demographic, economic, and buyer characteristics that influence the adoption of electric vehicles in the Nordic countries and India (Chen et al., 2020; K V et al., 2022). However, the social and economic realities of these countries are far from the social and economic realities of Latin American countries. For example, of the OECD countries, Chile has a higher proportion of households that are economically vulnerable compared to OECD countries (Garda & Arnold, 2022).

The literature on the characteristics of potential buyers of low-pollution vehicles is much scarcer, and this is a relevant element for the design of marketing strategies that seek to increase the adoption of this type of technology, especially in developing countries. The literature addressing the impact of the characteristics of potential consumers of low-emission vehicles is focused on assessing aspects such as attitudes or psychological aspects (Alzahrani et al., 2019; Jing et al., 2019; Mohamed et al., 2016; Rezvani et al., 2015), environmental awareness (Alzahrani et al., 2019; Carley et al., 2019; Globisch et al., 2018; Han et al., 2017), however, empirical evidence on the importance of the demographic characteristics of potential buyers is scarce.

The literature addressing the impact of the characteristics of potential consumers of low-emission vehicles focuses on assessing aspects such as attitudes or psychological aspects (Alzahrani et al., 2019; Jing et al., 2019; Mohamed et al., 2016; Rezvani et al., 2015), environmental awareness (Alzahrani et al., 2019; Carley et al., 2019; Globisch et al., 2018; Han et al., 2017), however, empirical evidence on the importance of demographic characteristics of potential buyers is scarce in Latin American countries.

We believe that focused analysis of these elements helps to improve understanding of potential consumer behaviour. It contributes to the empirical evidence and sheds some light on the most important demographic and socio-economic elements in the Chilean context and allows to support for the development of a social or commercial marketing strategy contextualised to the economic realities of the country. This paper aims to address these issues by examining the importance of the socio-demographic characteristics of potential buyers of electric and hybrid vehicles in the Chilean context. To do so, the preferences of potential buyers are modelled under the random utility theory approach. A stated preference survey is used, and different ordered probit models are estimated.

We focus our empirical analysis on Chile, particularly in the city of Santiago, the capital city of the country, which has the largest private automotive fleet in Chile. Chile represents an interesting case for this study because Chile is one of thirteen countries currently participating in the Electric Vehicle Initiative, a multi-governmental policy forum coordinated by the International Energy Agency dedicated to accelerating the introduction and adoption of electric vehicles (International Energy Agency, 2020). Although a member of the Electric Vehicles Initiative, Chile still has a very low market penetration of electric vehicles. Sales of battery electric vehicles and plug-in hybrid electric vehicles in 2019 represented only 0.1% of total new car sales (International Energy Agency, 2020, Table A.7). The total number of battery electric vehicles and plug-in hybrid electric vehicles in Chile in 2019 was 6500 (International Energy Agency, 2020, Tables A.2 and A.3), against a total car fleet of 5.718.409vehicles (INE, 2019).

According to the latest report of February 2023 of the Chilean National Automotive Agency (ANAC), there are currently a total of 1229 conventional hybrid vehicles (HEV); 582 electric vehicles (EV); 181 plug-in hybrid vehicles (PHEV); 570 micro-hybrid vehicles (MHEV); 10 extended range vehicles (E-REV) and 3 hydrogen cell electric vehicles (FCEV). Figure 1 shows the evolution of the total number of vehicles accumulated per year.

Figure 1. Evolution of vehicles of the total number of low emission vehicles per year. Source: Own elaboration based on the ANAC 2023 report.

The Chilean government expects that 40% of their total car fleet will be electric by 2050 (National Electric Mobility Strategy, 2018). Their strategy revolves around five pillars: standards and regulations, public transport as the backbone of sustainable mobility, research and development, knowledge transfer, and initial support to electromobility development (National Electric Mobility Strategy, 2018).

However, this initial support does not include any plans to introduce financial incentives of the type that many other countries implemented in the initial stages. Given the lack of economic incentives for private vehicle owners, the purchase of this type of vehicle can be explained in terms of the characteristics of the potential buyers. Against this backdrop, it is important to understand whether the sale of hybrid and battery electric cars in Chile is explained by the characteristics of the buyers and whether it could take off if the government were to provide financial incentives.

Understanding the impact of socio-demographic characteristics on the potential purchase of electric and hybrid vehicles in developing countries is an interesting research question given the differences in levels of economic wealth in these countries compared to Europe and the United States. Consequently, understanding the differential impact of the socio-demographic characteristics of the population segments in these countries offers potential benefits, such as providing the basic inputs for the development of a differentiated and contextualised social marketing strategy, in which different types of electric and hybrid vehicles can be offered according to the diversity of potential buyer profiles. In this context, the main objective of this paper is to assess how demographics impact or influence consumer preferences for low-emission vehicles.

The contribution of this research lies in two aspects. First, it assesses the impact of socio-demographic characteristics on the preference for low-emission vehicles. This adds empirical evidence to the scarce existing literature. Second, it specifically assesses the differential impact of different income segments in Chile on the purchase of electric and hybrid vehicles. Besides, we provide some implications for policy and marketing strategies. To our knowledge, this is the second paper to use data from Chile, which provides an important input in the context of the region. The first focuses on assessing the effectiveness of different economic incentives and their implementation mechanism. Here we consider the socio-demographic aspects as a central point.

The rest of the article is organised as follows. Section 2 presents a review of previous work related to the objective of this research. Section 3 describes the methodology used, explains details of the stated preference survey, the estimated ordered probit model and the results obtained. Section 4 discusses the results obtained from the estimated ordered probit model. Section 5 presents the conclusions and practical implications derived from the research.

2. Literature review

Consumers can be considered widely diverse or heterogeneous in their perceptions, tastes, preferences, motivations as well as in their socio-demographic characteristics, therefore, it is essential to identify, characterise and measure how different socio-demographic characteristics in population groups or segments impact the purchase of low- vehicles. Understanding these aspects is key for national and regional policy makers, as well as consumer marketing strategists, who remain interested in identifying the key elements to incentivise and support the growth of the electric and hybrid vehicle market (Axsen et al., 2016), and ensure that the technological transition has an impact on social welfare (Madl & Radebner, 2021). According to the work of (Axsen et al., 2018; Axsen & Kurani, 2013a, 2013b), early adopters of electric and hybrid vehicles, in the absence of government incentives, tend to have a different demographic distribution than conventional vehicle buyers.

A consensus among several studies indicates that early adopters of electric vehicles tend to be predominantly male, older, highly educated, and with higher incomes compared to conventional vehicle buyers (Axsen & Kurani, 2013a, Plötz et al., 2014, Tal & Nicholas, 2013; Peters & Dütschke, 2014), while these studies highlight important demographics of potential buyers and precursors, they are conducted in a developed country such as the USA and Germany where the motorisation rate is more than one vehicle per person (World Bank, 2021), their socio-economic conditions are significantly different and their purchasing power higher than in developing countries.

Hardman (2019) provides the most recent review of low-emission vehicle adoption strategies for first world countries; however, although socio-demographic elements are considered, the study focuses on government policy strategies and the efficiency of incentives and subsidy programmes. Rezvani et al. (2015) also present a review of consumer characteristics that affect the adoption of electric vehicles, however, their focus of analysis concentrated on examining the effect of pro-environmental attitudes, symbolic meanings, identity, innovation and emotions. Not aspects related to socio-demographic characteristics.

Coffman et al. (2017) also analysed the factors affecting electric vehicle adoption, however, their analysis focused on examining the effectiveness of government incentives and highlighted that there is still a gap that needs to be addressed to address further research and address the relationship between attitude and action. Meanwhile, Adnan et al. (2018) studied consumer behaviour of plug-in hybrid vehicles and electric vehicle adoption in Malaysia. However, although their study is based on a non-developed country, their study did not consider socio-economic factors and did not investigate how these impact electric vehicle adoption.

Li et al. (2017) classify the factors that influence consumers' intentions to adopt battery electric vehicles, while achieving a classification of three groups of factors, defined as: Demographic, situational and psychological. While the demographic component is considered, and the paper is an excellent framework, it merely presents a general description of these factors without providing measurements of them. In the work of Javid and Nejat (2017), socio-demographic aspects are also considered, finding that income and education significantly influence the adoption of electric vehicles. However, the study uses data from the United States, which, as indicated in the introduction, has quite different socio-demographic and economic conditions than developing countries such as Chile.

K V et al. (2022) analyses the factors influencing the adoption of electric vehicles in an Indian city. For this they analysed aspects related to Financial Barriers, Vehicle Performance Barriers, Lack of charging infrastructure, Environmental Conservation, Social Influence. However, the socio-demographic variables they consider relevant are limited to the educational aspect and vehicle knowledge.

An important work with data from Latin America that studies the adoption of electric vehicles in Brazil is the work of Luna et al. (2020), which focuses on determining how charging schemes and governance impact the adoption of electric vehicles. However, the focus of their study is not on the characteristics of power purchasers.

In Chen et al. (2020), the socio-demographic, technical, economic and behavioural factors of electric vehicle adoption in the Nordic countries are assessed. Their main results suggest that young men, higher income, more children and experience with electric vehicles are related to the potential adoption of electric vehicles. This work adds to the international empirical evidence, but its target audience is people in wealthy countries with a higher rate of electrification of urban transport, thanks to the development of public policies more than 10 years old.

Singh et al. (2020), as well as Li et al. (2017), provide a classification of the factors influencing electric vehicle consumption based on a literature review study, including the demographic component. Their findings highlight the importance of demographic aspects such as age, gender and education. However, the paper does not provide statistics on the impact of these components. In Soto et al. (2018), the importance of personal characteristics as a key element for the adoption of low-emission vehicles is also highlighted. However, their work focuses especially on the impacts of attitudinal characteristics of potential buyers as does the work of Herberz et al. (2020) which also considers demographic aspects as one of the key factors explaining low-emission vehicle consumption.

An important work on the role that advertising should play in attracting new buyers is the work of Bennett et al. (2016), which indicates the importance of the demographic component in social marketing campaigns initiated by government and private agencies. They indicate that this component is an instrument that allows reflecting the preferences of various segments of motorists.

Finally, two scarce works found in the literature, whose main focus of the study is the social, demographic, and economic characteristics of consumers of electric vehicles are the work of Oliveira and Dias (2019), and the work of Higueras-Castillo et al. (2020). The former provides evidence on how consumer demographics influence their preferences with respect to electric vehicle consumption and the latter provides the necessary inputs to establish the profiles of early adopters of electric vehicles in the Spanish context. While both papers specifically focus on examining the importance of demographics, the paper by Oliveira and Dias (2019), merely points to findings from other papers in developed countries, and the paper by Higueras-Castillo et al. (2020), uses data from Spain, which is a first-world country.

In summary, the main findings of the literature indicate that the individual characteristics and/or demographics of consumers influence their preferences as one of the relevant explanatory factors in the consumption of low-emission vehicles. However, as evidenced in the literature, very few studies have addressed the influence of demographics on preferences for low-emission vehicles, especially those whose modelling technique is based on discrete choice models and the elicitation of individual consumer preferences through a stated preference survey. Such as for example the work of (Assum et al., 2014; Beresteanu & Li, 2011; Bjerkan et al., 2016, Langbroek et al., 2016; Caulfield et al., 2010; Chandra et al., 2010; Diamond, 2009; Ewing & Sarigöllü, 1998, 2000; Figenbaum, 2017; Figenbaum & Kolbenstvedt, 2013; Fridstrøm et al., 2014; Gallagher & Muehlegger, 2011; Gass et al., 2014; Horne et al., 2005; Jenn et al., 2013; Jin et al., 2014; Potoglou & Kanaroglou, 2007; Tal & Nicholas, 2016; Wee et al., 2018).

Some contextual differences of the studies reported in the literature are related to (1) Most of the reported studies are developed in developed countries that have pioneered transport electrification policies, and also maintain a robust middle class. This is a significant difference with respect to the Chilean case. (2) The studies on developing countries are carried out in India, which although it is a developed country, has great social, economic and political differences that are far from the reality of Latin American countries.

Our work adds to the scarce literature that provides empirical evidence of the influence of demographic characteristics on consumers' preferences for low-emission vehicles. This is the case for a developing country such as Chile. To our knowledge, this is the second work of its kind to be done with data from Chile.

3. Methodology

3.1. Questionnaire and survey

We conducted a stated preference survey in Santiago de Chile. Santiago is the capital and largest city of Chile, with a population of 6.3 million, or 7 million when the whole of Santiago Metropolitan Region is considered (Instituto Nacional de Estadísticas [INE], 2017). It also has the biggest automobile fleet, with around 39.3% of the country total (INE, 2019). For a further explanation of why Santiago de Chile is a good case study, see Urrutia-Mosquera and Fábrega (2021). The questionnaire consisted of six parts. Part 1 focused on questions related to driving frequency (i.e. the number of times the car is used in a week, measured on an ordinal scale), driving distance (i.e. the average daily distance travelled, measured in kilometres) and the main purpose for which the car is used (work, study, etc., measured on an ordinal scale). All participants signed an informed consent form in order to participate in the research.

Part 2 contained the indicated choice experiment. The stated preference experiment consisted of a choice between three alternatives: a conventional car, an electric vehicle, and a hybrid vehicle. The alternatives were described in terms of the purchase price, tax incentive offered by the government (subsidy) for electric and hybrid cars, and discounted purchase for conventional vehicles offered by dealers (these are not subsidies). We also include the driving range. The attributes associated with the choice alternative (purchase price, driving range, and incentive), were defined with three levels each. In the table they are shown as: [1], [2], [3].

The attributes for the choice experiment were selected based on the work of (Peters & Dütschke, 2014; Plötz et alt., 2014; Potoglou & Kanaroglou, 2007). Part 3 asked additional questions about policies and additional barriers to electric vehicle adoption (dummy variable). Part 4 focused on perceptions and values related to the environment (dummy variable), and Part 5 focused on key socio-economic information such as marital status, age, educational attainment, household size, and income of the respondent. A summary of the variables and their description is shown in Tables 1 and 2. As our objective is to examine the importance of socio-economic and demographic characteristics on the potential purchase of an electric and hybrid vehicle, our analyses focus on parts 1, 2, 4 and 5 of the choice questionnaire. The stated choice experiment was based on a fractional factorial experimental design, allowing for interactions and quadratic effects. In the design, three attributes with three levels each were considered. The attributes and levels used are presented in Table 1. There was a total of nine choice cards per experiment organised into three blocks (Appendix A). After being piloted, the survey was run between October and December 2017. Study participants were selected through quota sampling, based on three segmentation variables. Gender, age, and income. A total of 525 participants potential buyers of private cars were interviewed in person. 23 of the 525 observations were discarded because they were not complete. Respondents were men and women between the ages of 30 and 65, with a monthly income greater than or equal to US$1200. Respondents were contacted in their homes, workplaces, and malls. Participation was on a voluntary basis; no incentives were given to participate. The only condition to be eligible was that respondents needed to be 30 years old or more, own a car, or express the intention to buy a car in the next coming months and have a net monthly income of at least 800 thousand Chilean pesos (Equivalent to USD 1200).

Table 1. Attributes and levels in the discrete choice experiment.

Attributes		[Level] Description
Purchase price^a	Conventional car	[1] 15,453 USD
		[2] 20,478 USD
		[3] 24,946 USD
	Electric car	[1] 30,304 USD
		[2] 37,056 USD
		[3] 39,691 USD
	Hybrid car	[1] 24,106 USD
		[2] 28,285 USD
		[3] 31,263 USD
Driving range (km per fill/charge)	Conventional car	[1] 450 km
		[2] 550 km
		[3] 650 km
	Electric car	[1] 199 km
		[2] 250 km
		[3] 280 km
	Hybrid car	[1] 350 km
		[2] 450 km
		[3] 550 km
Incentive	Conventional car	[1] 10%
	(Discount on purchase	[2] 20%
	price)	[3] 30%
	Electric car	[1] 30%
	(% of purchase price as	[2] 35%
	income tax rebate)	[3] 40%
	Hybrid car	[1] 20%
	(% of purchase price as	[2] 30%
	income tax rebate)	[3] 35%

^aPurchase prices are shown in US dollars (USD), but they were presented in Chilean pesos (CLP) in the survey. The average exchange rate in 2017 was 1 USD = 649 CLP (International Monetary Fund, 2021).

Table 2. Socio-economic and demographic characteristics of the sample.

Variable	Description	Sample (%)
Socio-economic information
	Gender
	Female	47
	Male	53
	Age
	30–39	47
	40–49	24
	50–59	21
	>60	8
	Level of education
	Elementary school or middle school	11
	Incomplete university education	10
	Technical education complete	15
	Completed university education	50
	Postgraduate studies	14
	Average income
	678.3 $ USD–1204.7 $ USD	29
	1206.1 $ USD–$20052.6 USD	34
	2054 $ USD–3435.6 $ USD	24
	3437.1 $ USD–5789.4 $ USD	11
	Over 5789.4 $ USD	2
Use of the vehicle	Driving frequency
	Once or twice a week	33
	Three to four times a week	62
	Four to five times a week	4
	More than five times a week	1
Variable	Description	Sample (%)
Purpose of the trip	To go to work
	Very frequently	49
	Frequently	15
	Occasionally	17
	Never	18
	Don't know/Not sure	2
	To go shopping
	Very frequently	35
	Frequently	39
	Occasionally	21
	Never	4
	Don't know/Not sure	1
	For leisure activities
	Very frequently	34
	Frequently	40
	Occasionally	19
	Never	6
	Don't know/Not sure	1
	For long trips
	Very frequently
	Frequently	29
	Occasionally	31
	Never	11
	Don't know/Not sure	1
How often you replace your car?	Years in use
	After 1 or 2 years	7
	After 3 or 4 years	36
	After 5 or 6 years	37
	After 7 or more years	20
Possible barriers to Electric Car Uptake	If it is considered a barrier
	Absence of charging stations	67
	Domestic refuelling time	79
	Commercial cost of vehicles	49
	Range per charge	67
	Cost of future battery replacement.	46
Attitudes Towards The Environment	Identify with the statement
	I don't think I should do anything to protect the environment	5
	My behaviour can help protect the environment	48
	I would like to play an active role in protecting the environment	13
	I am willing to make personal concessions to protect the environment	27
	I am willing to adopt clean technologies	8

This income is representative of the income in high income communes in Santiago de Chile. These high-income communes are Providencia, La Reina, Vitacura, Las Condes, Maipú, Lo Barnechea, Nuñoa and Santiago. Consumers that buy hybrid or battery electric cars tend to be older than 30 (Bjerkan et al., 2016), and this was the reason why there was a screening question on age at the beginning of the questionnaire. Consumers that buy hybrid or battery electric vehicles also tend to belong to higher income groups (Christidis & Focas, 2019). A target number of observations was assigned to each commune, depending on its average income level. The average income level for each of the communes, the target number of observations, and the actual number of observations are presented in Table 3.

Table 3. Average income level and target and actual number of observations per commune.

	Average monthly income in 2017 (USD)^a	Target number of observations	Actual number of observations
La Reina	3564	55	40
Las Condes	4127	60	70
Lo Barnechea	2975	50	45
Maipú	1553	75	90
Nuñoa	3481	58	55
Providencia	3555	52	52
Santiago	2158	90	100
Vitacur	4897	62	50

^aThe average monthly income in CLP was taken from Gobierno de Chile (2017) and converted to USD using the average exchange rate in 2017 (1 USD = 649 CLP) from the International Monetary Fund (2021).

3.2. Descriptive analyses

This section provides a descriptive analysis of the information collected in the survey. Table 2 shows a summary of the socio-demographic characteristics of the sample and Figures 2–6 report the percentage of respondents who chose low-emission vehicles by socio-demographic characteristics. As reported in Table 2, 79% of the respondents have a professional education, and only 37% of them report earning a salary of more than 200 US dollars. This fact shows that although by sample design, people with higher purchasing power were selected, it is unlikely that in the absence of tax incentives these people would be willing to buy an electric or hybrid vehicle due to its high cost given their income levels. It is also observed that only 5% of the sample participants use the vehicle more than 4 times a week. 64% of them report frequent use of the vehicle for commuting to work. 74% of the respondents reported frequent use for shopping and leisure activities, and only 60% reported using the vehicle for long distance trips. One characteristic of electric and hybrid vehicle buyers reported in the literature is their level of environmental awareness. However, only 13% said they strongly agreed to play an active role in protecting the environment, and only 28% said they were willing to make personal concessions to protect the environment, in contrast to 8% who said they were willing to adopt clean technologies. Figure 2 shows the percentage of people who chose each of the three vehicle options according to their affinity with the five expressions reflecting environmental attitudes. As can be seen, the percentage of people who chose an electric vehicle ranges from 12% to 27%, while the percentage who chose a hybrid vehicle ranges from 36% to 49%. This is striking since, in the absence of vehicle characteristics, the percentage of choosing an electric vehicle would be expected to be higher. As can be seen in Figures 3 to 6, in none of the cases that discriminate by any socio-demographic characteristic does the choice of electric vehicles exceed 20%. In contrast, the choice of hybrid vehicles exceeds 34% in each case.

Figure 2. Choice of vehicles according to environmental awareness.

Figure 3. Choice of vehicles by income bracket.

Figure 4. Choice of vehicles by age range.

Figure 5. Choice of vehicles by educational level.

Figure 6. Choice of vehicles by household size.

3.3. Theoretical basis and modelling approach

This paper considers the foundations of discrete choice models (Train, 2009; Ortúzar & Willumsen, 2011), which are supported by Random Utility theory. Random Utility theory (Manski, 1977; McFadden, 1981), is more in line with consumer theory.

In this approach, choice probabilities are derived by assuming a joint probability distribution for the set of random utilities. Manski (1973) identified four distinct sources of randomness: 1. unobserved attributes, 2. unobserved taste variations, 3. unobserved taste variations, and 4. unobserved taste variations. 2. unobserved taste variations, 3. measurement errors, and imperfect information, 4. instrumental (or proxy) variables.

The Discrete Choice family of models operationalise Random Utility theory by proposing functional forms of econometric models to estimate preferences and choice (McFadden, 1981; Ortúzar & Willumsen, 2011).

In a discrete choice context. It assumes that an individual q, when choosing among a finite set of j alternatives, will evaluate all the characteristics of each alternative and choose the option that provides the highest utility. The evaluation of the alternative may differ between individuals depending on their socio-economic (SE) characteristics and other factors that are unknown to the modeller and/or to the respondents themselves.

Regarding the use of discrete choice models to model the preferences and choices of individuals, Ortúzar and Willumsen (2011), present a new model of discrete choice and random utility based on the assumption that a population of individuals makes the same choice over a set of alternatives, and determines the fraction of the population that chooses a given alternative, dividing the population into H observable socio-economic groups (income, age, profession, etc.).

To measure the impact of socio-economic characteristics on the willingness to purchase low-pollution vehicles, we use an ordered probit model, where P_q is the probability that individual q chooses a vehicle alternative with a low pollution degree (Electric Vehicle), with medium pollution degree (Hybrid Vehicle), or with higher pollution degree (conventional vehicle), according to socio-economic characteristics. The model assumes the form: (1) $$\begin{array}{c}{P}_q\left(A=1\right)=\Phi \left(A_q\left(S{E}_q,{\eta }_A\right)\right)\\ P_q\left(A=2\right)=\Phi \left(A_q\left(S{E}_q,{\eta }_A\right)\right)-\Phi \left(A_q\left(S{E}_q,{\eta }_{A-1}\right)\right)\\ P_q\left(A=3\right)=1-\Phi \left(A_q\left(S{E}_q,{\eta }_{A-1}\right)\right)\end{array}$$(1) where η_A are thresholds defined respectively as: low pollution degree; this corresponds to $P_q(A=1)$; medium degree of contamination, this corresponds to $P_q(A=2)$; and high pollution degree, this corresponds to $P_q(A=3)$. A comparison of the pollution levels of each type of vehicle can be found in Dey et al. (2020).

3.4. Models results

Table 4 reports the results of the estimated ordered probit model. The response variable is the choice of one of the vehicle alternatives presented in the stated choice experiment and the explicative variables are the socio-economic characteristics of the individuals (age, education level, income, etc.). Column two of Table 4 reports the coefficients and test statistics for the variables that were significant in the model. The first value represents the coefficient with its respective sign, and the value of the z-statistic is indicated in brackets.

Table 4. Predicted probabilities and marginal effects from the estimated ordered probit model.

Predicted probabilities		Pq (A = 1) = 0.15	Pq (A = 2) = 0.40	Pq (A = 3) = 0.44
Variable	Value	Marginal eff	Marginal eff	Marginal eff
Age
40–49 years old	0.153 (2.95)	−4.91%	0.09%	4.82%
50–59 years old	0.124 (2.1)	4.46%	0.28%	4.18%
More of 60 years old	–	–	–	–
Academic background
Incomplete Education T/U	0.53 (3.3)	2.73%	2.43%	1.29%
Complete Technical Education	0.46 (2.96)	1.17%	2.34%	1.83%
University Education	0.55 (3.59)	2.65%	2.45%	1.2%
Postgraduate	0.61 (4.44)	2.02%	2.34%	2.68%
Household size
Two people	0.23 (3.29)	9.25%	0.36%	−8.96%
Three people	–	–	–	–
four people	–	–	–	–
five people	5.50 (3.74)	1.95%	−1.74%	−1.21%
Six or more people	−0.66 (−2.11)	−2.49%	−3.11%	−2.60%
Gender
Female	0.08 (2.11)	3.23%	0.07%	3.16%
Income
1206.1 $ US–20,052.6 $US	0.359 (3.510)	2.8%	2.1%	3.6%
2054 $ US–3435.6 $ US	0.466 (4.730)	3.5%	5%	1.8%
3437.1 $ US – 5789.4 $ US	0.22 (2.1)	4.2%	6.2%	5%
More than 5789.4 $ US	–	–	–	–
Number of obs	4364
Final log likelihood:	−4324.6128

As this is an ordered probit model, these values are not interpretable, which is why the marginal effects of the significant variables are presented in columns 3, 4 and 5. As shown in Table 4, the probability of choosing an electric vehicle, based on socio-demographic characteristics and in the absence of an incentive policy, is only 15%, compared to hybrid and conventional vehicles, which have a probability of choice of 40% and 44% respectively. In this context, the variables marital status, and the variables associated with vehicle use and renewal time, are not significant. The marginal age effects indicate that a potential buyer aged 40–49 years is 4.91% less likely to choose a low-pollution vehicle (electric vehicle) compared to potential buyers aged 30–39 years. However, this group is 0.09% and 4.82% more likely to choose a low pollution (hybrid vehicle) and high pollution (conventional vehicle), respectively, compared to the base age category. And being in the 50–59 age range, increases the probability of choosing a low pollution vehicle by 4.46%, and by 0.28% of choosing a low pollution vehicle, compared to the base age range.

The marginal contributions of education level in increasing the probability of choosing a low-pollution vehicle range from 1.17% to 2.73%. In particular, having a university education increases the probability of choosing a low-pollution vehicle by 2.65%, and increases the probability of choosing a medium-pollution vehicle by 2.45%. An interesting socio-demographic characteristic is household size. The marginal effects indicate that of the individuals surveyed who belong to a two-person household (childless couples), they are 9% more likely to choose to buy a low-pollution vehicle compared to single people, and 8.96% less likely to choose a high-pollution vehicle compared to a single person. On the other hand, being a woman is 3.23% more likely to choose an electric vehicle and 0.07% more likely to choose a hybrid vehicle than being a man. These results are in line with those reported in the literature. Women tend to identify more with actions aimed at preserving the environment than men.

On the other hand, belonging to higher income brackets, compared to the base category, progressively increases the probability of choosing a low and medium pollution vehicle. In particular, having an income between 3437 US dollars and 5789 US dollars increases the probability of choosing an electric and hybrid vehicle by 4.2% and 6.2% respectively.

4. Discussion of the results

The results of measuring the impact of socio-demographic characteristics on the consumption of low-emission vehicles reported in the descriptive analyses and the estimated ordered probit model are consistent with each other and are in line with what is reported in the literature. The results show that the socio-demographic characteristics with the greatest impact on the likelihood of purchasing low-emission vehicles, in the absence of any incentives, are household size, educational level, and gender. These findings are consistent with the work of (Axsen et al., 2018; Axsen & Kurani, 2013a, 2013b) which indicates that older, highly educated men with higher incomes have a high willingness to purchase a low-emission vehicle. In the Chilean context, the marginal effects of the estimated ordered probit model indicate that, if a person belongs to an age range between 50 and 59 years old, the probability of buying an electric vehicle increases by 4.46% compared to people in the age range 30 to 39 years old. In the case of Axsen and Kurani (2013a), the proportion of EV buyers in this age range is reported to be 43% of their sample. As mentioned above, our results are along the same lines. In Plötz et al. (2014) however, the highest proportion of consumers interested in purchasing electric vehicles is the population group belonging to the age range 41–50 years old, accounting for 53.1% of the buyers in their sample.

Another demographic characteristic with a high incidence is educational attainment. The probability of a potential buyer choosing or buying a vehicle increases by 2.65% if the potential buyer reports having completed university education, compared to people who report only a high school education. In the case of the work of Axsen and Kurani (2013a), the percentage of buyers of electrified vehicles with university education corresponds to 43.3% being the highest percentage of buyers in the sample. In the Chilean context, people with these characteristics have the second highest marginal effect value. The highest value is presented by the group of people who reported incomplete university or technical education, with a marginal effect of 2.7%. That is, the probability that a potential buyer chooses or buys a vehicle increases by 2.7% if the potential buyer reports having a university or incomplete technical education, compared to people who only report secondary education.

Regarding household size, the marginal effects indicate that of the individuals surveyed who belong to a two-person household (childless couples), they are 9% more likely to choose to buy a low-pollution vehicle compared to single people, and 8.96% less likely to choose a high-pollution vehicle compared to a single person. There are several explanations for this in the literature. For example, the economic factor associated with the cost. For a single person it can be a significant financial burden on their income level. Another aspect may be the social aspect, related to the level of environmental awareness as indicated in Axsen et al. (2016).

Our results are partially different from the results reported in Javid and Nejat (2017). In both papers this variable is significant. However, the way in which this variable was specified in both papers was different. They modelled household size as a continuous variable, we modelled it as a dummy variable.

In the case of the paper by Javid and Nejat (2017) the impact of the household size variable is negative and the probit likelihood ratio is 0.96. In the case of our paper, the impact of the household size variable is negative for the category of households with a household size larger than 6 persons. For households of sizes 2 and 5 persons, the impact is positive. Household size categories 3 and 4 were not significant for the model.

5. Conclusion and practical implications

This paper aimed to shed light on the impact of socio-demographic characteristics on the choice of low-pollution vehicles (electric and hybrid) in the absence of fiscal incentives in the context of Chile. The results reveal that while some socio-demographic characteristics are significant, on their own, their contribution to the choice of this type of vehicle does not exceed 15% probability. However, education level, household size and gender turn out to be key variables in the choice, rather than income level, in the absence of tax incentives.

The level of environmental awareness is a variable that was not significant in the ordered probit model, however, in the descriptive statistics, it is observed that no more than 15% of the people most identified with actions aimed at preserving the environment chose an electric vehicle. This could be explained by a lack of knowledge of the technical characteristics of this type of vehicle. Hybrid vehicles obtained a higher degree of choice under this characteristic of individuals.

International empirical evidence as reported in the works of (Assum et al., 2014; Figenbaum & Kolbenstvedt 2013; Figenbaum, 2017; Fridstrøm & Alfsen, 2014; Tal & Nicholas, 2016; Wee et al., 2018) shows that the increase of the electric vehicle fleet should be accompanied by subsidy and incentive policies focused on groups of people with some special characteristics.

In the Chilean case, based on our results, we consider that this discussion can be started. However, as this study is not based on a representative sample of the population, it should only be considered as a starting point for further studies, which deepen, corroborate or refute these findings.

In the Chilean context, in the work of Urrutia-Mosquera and Fábrega (2021), which is the only work present to date, it is possible to observe how the impacts of these variables change in the presence of a possible incentive policy. For the Chilean context, our results show that in the absence of such incentives, the probabilities of purchasing an electric and hybrid vehicle are only 15% and 40% respectively. In the presence of such incentives, the probabilities range between 54% and 57% for electric vehicles and between 70% and 73% for hybrid vehicles as reported by Urrutia-Mosquera and Fábrega (2021). These results present some practical implications for the design of social marketing strategy. For example, the focus of social marketing campaigns should be on showing the importance and impact for society, and especially for the reduction of CO₂ levels, of conventional vehicle users switching to low-emission vehicles. And indicate that, as the change of technology has an economic cost, high-income people are the ones called to be the pioneers of this transition as a kind of social responsibility, since in the context of Chile, people with incomes between 3.43 US $ and 5.789 US $ are the most likely to buy this type of technology. The experience of successful European countries in a transition to mobility, such as Norway, Sweden and England, indicates that the transition is achieved if electric and hybrid vehicles become competitive in price and the necessary infrastructure is available for their operation.

As this is not the case in Latin American countries, there is an opportunity for social change to be led by business organizations, by generating financing mechanisms for employees with specific demographic characteristics, as the work reveals, to gain access to this type of technology.

A successful example is the case of the Chilean energy company ENEL, which subsidized the purchase of electric vehicles for a significant number of its employees. (Enel Chile Sustainability Report, 2019). This type of actions can be channeled into social marketing policies that can be used to motivate changes and modify attitudes and behaviors of vehicle consumption.

It is worth mentioning that the design of social marketing strategies, as well as fiscal policies that aim to incentivise the consumption of low-emission vehicles, should initially consider the population segment that, given its demographic characteristics, is more likely to purchase in the absence of fiscal incentives. We consider that this research provides the first evidence on how socio-economic or socio-demographic aspects impact the consumption of low-emission vehicles in the context of Chile.

Finally, we recommend using this work as a baseline input to allow for the possibility of a broader study that considers a larger and more diverse sample in order to overcome the limitation of this work.

One of the limitations of this work is the scope of the sampling. Although the number of participants is considered adequate for the existing universe of vehicles, a larger sample would allow for representativeness at the national level.

Another limitation is associated with the survey and the estimated model. According to advances in modelling and the latest reviewed studies, estimating a hybrid choice model would help to improve the understanding of the different latent dimensions that affect the probability of purchase. This would imply changes in the data collection instrument. For example, considering more attributes and testing other levels for the attributes considered.

Acknowledgements

The corresponding author gratefully acknowledges project: D2325-1002 of the Universidad del Desarrollo-Chile, for providing access to the data source used in this work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Jorge Urrutia-Mosquera

Jorge Urrutia-Mosquera is an academic in the Department of Economics at the Universidad Católica del Norte. Chile. Her research interests cover issues of sustainable transport and urban sustainability.

Luz Flórez-Calderón

Luz Flórez-Calderón is academic in the Department of Industrial Engineering, Universidad Católica del Norte and Researcher at the Department of Transport and Logistics Engineering of the Pontificia Universidad Católica de Chile. Her research interests cover issues of transport, emissions, logistics, and vehicle routing.

Appendix A

CHOICE CARDS

Assume you need to replace your car in the next few months, and you need to choose a car amongst three types of cars: Conventional car, hybrid car, and electric car. Please choose a car on the basis of the following:

• Purchase price

• Driving range per tank fill/battery charge

• Financial incentives (for hybrid and electric cars the financial incentives are provided by the government; for conventional cars, the financial incentives are provided by the car dealers, who also bear the cost)

Please choose which car out of the three cars you would buy in each of the following 9 different situations. Please consider each choice situation INDEPENDENTLY from the other choice situations.