Which Is Preferred between Electric or Hydrogen Cars for Carbon Neutrality in the Commercial Vehicle Transportation Sector of South Korea? Implications from a Public Opinion Survey

Abstract

South Korea has drawn up plans to reduce greenhouse gases by 29.7 million tons by supplying 4.5 million electric and hydrogen cars by 2030 to implement the “2050 carbon neutrality” goal. This article gathers data on public preferences for electric cars (ECs) over hydrogen cars (HCs) in the commercial vehicle transportation sector through a survey of 1000 people. Moreover, the strength of the preference was evaluated on a five-point scale. Of all respondents, 60.0 percent preferred ECs and 21.0 percent HCs, the former being 2.86 times greater than the latter. On the other hand, the strength of the preference for HCs was 1.42 times greater than that for ECs. Factors influencing the preference for ECs over HCs were also explored through adopting the ordered probit model, which is useful in examining ordinal preference rather than cardinal preference. The analyzed factors, which are related to respondents’ characteristics, experiences, and perceptions, can be usefully employed for developing strategies of promoting carbon neutrality in the commercial vehicle transportation sector and preparing policies to improve public acceptance thereof.

1. Introduction

In South Korea, a lot of greenhouse gases (GHGs) are emitted, mainly from power generation and industrial sectors. The next sector that emits a lot of GHGs is transportation. For instance, 13.5% of the country’s GHG emissions occurred in the transportation sector in 2018. Consequently, the South Korean government established an ambitious plan to reduce GHG emissions by 37.1 million tons by 2030, down from 98.1 million tons in 2018 [1]. In this case, GHG emissions from the transportation sector in 2030 will be 61.0 million tons, a reduction of 37.8% from 2018 values. In particular, roads account for 95.5% of GHG emissions from the transportation sector comprising roads, aviation, and sea. In response, the government has formulated a plan to reduce GHG emissions by 29.7 million tons by supplying 4.5 million electric cars (ECs) and hydrogen cars (HCs) by 2030.

Internal combustion engine vehicles in the commercial transportation sector, such as buses and trucks, will be converted to ECs and HCs [2,3,4]. ECs and HCs will be mandatory when old vehicles are replaced early. In order to promote the conversion to ECs and HCs, the government will strengthen systems, such as GHG emission standards, and expand the charging infrastructure, such as EC and HC charging stations. South Korea is a leading automobile producer and has launched the world’s first hydrogen truck. Currently, both ECs and HCs are produced in the country, although their production is not yet sufficient for these plans. Therefore, it is necessary to expand the production capacity of ECs and HCs. It is thus time for South Korean automakers to decide whether to focus on ECs or HCs. When this decision is made, the direction of construction of the related charging infrastructure will also be changed.

For example, if automakers largely choose to supply ECs, the EC charging infrastructure should increase intensively over HC charging stations. However, if HCs are chosen, the opposite is true. Each local government is considering whether to choose between ECs and HCs when replacing existing old internal combustion engine buses. The central government is also considering what level of subsidies should be provided for ECs and HCs. This is because, for the time being, the cost involved in manufacturing ECs and HCs will be higher than that for manufacturing internal combustion engine vehicles. As a result, automobile manufacturers and central and local governments want to ascertain the public preference for ECs versus HCs in the commercial transportation sector, such as buses and trucks.

One thing to emphasize here is that the production of hydrogen passenger cars has stopped in South Korea. In other words, hydrogen passenger cars are not sold in the country. The most important reason for this is that hydrogen passenger cars are significantly less competitive than electric passenger cars. On the other hand, in the case of commercial vehicles, such as buses and trucks, ECs and HCs are competing with each other. Of course, it is clear that electric commercial vehicles began to be distributed earlier and are being utilized more than hydrogen commercial vehicles; however, there is no guarantee that this trend will be maintained in the future. In particular, the number of hydrogen buses is increasing rapidly, and an automaker is directly producing hydrogen buses in the country. Therefore, this study analyzes people’s preference for ECs versus HCs, targeting commercial vehicles, not passenger cars. In other words, this study attempts to raise the following question: which is preferred between electric or hydrogen cars for carbon neutrality in the commercial vehicle transportation sector of South Korea?

Accordingly, this article seeks to gather data on the public preference for ECs over HCs in the commercial vehicle transportation sector through a survey of 1000 adults and to analyze the factors affecting this preference by applying the ordered probit model (ORM). The authors think that this article makes a novel and meaningful contribution to the literature in three respects.

First, as far as the authors are aware, this study is the first to delve into the public preference for ECs versus HCs. Previous related studies dealing with ECs or HCs compared their environmental performance or explored public preferences for one of the two. For example, Contestabile et al. [5] analyzed the effect of car fuel types on the total cost of ownership to solve the economic and environmental sustainability problem of road transport. Messagie et al. [6] looked into the environmental impact of European car fuel types and compared the environmental performance of new energy vehicles. Nocera et al. [7] assessed the CO₂ emissions of ECs and HCs in South Tyrol, France, helping policymakers to establish the right environmentally friendly transport policies. Kolbe et al. [8] found that the replacement of conventional vehicles in Beijing, China, by ECs can reduce urban heat island intensity. Liu et al. [9] compared the energy consumption and GHG emissions of gasoline cars and HCs. Xu et al. [10] revealed that the new energy vehicles had been widely implemented in many aspects. For example, hybrid EVs have the advantages of achieving high efficiency, eliminating restrictions on driving range, and reducing both fuel usage and pollutant emissions.

She et al. [11] evaluated Tianjin consumer perceptions of EC adoption in China. Habich-Sobiegalla et al. [12] studied the factors that influence the introduction of ECs to address the sustainability problem in China. Liao et al. [13] examined Dutch consumer preferences for business models in the introduction of ECs. Li et al. [14] investigated the marginal willingness to pay for improvements in each attribute of HCs in China. Jreige et al. [15] explored consumer preferences for hybrid cars and ECs in Lebanon. Anwar et al. [16] identified the factors affecting consumers’ intention to switch from conventional to green vehicles in Pakistan.

Second, this article seeks to provide useful information for South Korean automobile manufacturers by ascertaining the public preference for ECs versus HCs for the first time. South Korea is a country that manufactures automobiles domestically, and it is particularly concerned about whether to focus on ECs or HCs in its commercial vehicle production. In this situation, this article at least offers hints on where to focus in terms of public preference. Of course, public preference is only one of many determining factors. Nevertheless, the government and automobile manufacturers may explore whether to increase production lines for ECs or HCs.

Third, this article strives to identify the factors determining the public preference for ECs versus HCs and to analyze the magnitude of that determination. This is also the first attempt in the literature regarding ECs versus HCs. As will be described later, several variables related to the respondent’s perception are considered as factors, along with personal characteristic variables, such as gender, education level, age, income, and residential area. By applying the ORM, the identification was successfully performed, and all factors were found to have a statistically significant impact on the preference. This is quite an interesting point of the study. The subsequent content of the article will comprise the methodology, results and discussion, and conclusions.

2. Methodology

2.1. Conducting the Survey

As mentioned earlier, this study collects data on preferences by adopting a survey of people. Four things should be decided in relation to conducting the survey. The first is the choice of the interviewees. For convenience, the interviewees in this study consisted of adults between the ages of 20 and 65. In other words, those who were younger than 20 or older than 65 were excluded from the interviewees. Of course, this limitation may make the implications of this study’s results restrictive, but the limitation was inevitable for four practical reasons. First, it was not easy to respond to the one-to-one individual interviews because people under 20 were attending school and people over 65 were retired. For this reason, public opinion polls conducted in South Korea’s public sector and media almost always use samples of people between the ages of 20 and 65. The sample selection method adopted in this study is consistent with that in the public opinion survey. Second, variables used as covariates include whether to agree to increase electricity rates to consume renewable energy, and it is reasonable for decisions related to this to be made by people with income. Most of the economically active population in the country is between the ages of 20 and 65. Originally, the economically active population included those aged 15 to 20, but most people of that age in the country are not engaged in economic activities because they are attending school. Third, the population between 20 and 65 years old actually represents the opinions of more than half of the South Korean population. As of 2022, out of the country’s total population of 51,692,272, those aged 20 to 65 account for 34,382,410, or 66.5%. Fourth, the sample composition of this study is consistent with that of most previous studies analyzing the South Korean public’s opinions on specific issues. For example, studies that examined opinions on renewable energy projects [17], hydrogen production using nuclear energy [18], construction of hydrogen charging stations near respondents’ residences [19], and construction of a combined heat and power plant [20] have always selected sample between the ages of 20 and 65. In short, the authors believe that the sample between 20 and 65 years of age does not itself pose much of a problem in representing the opinions of South Korean adults.

Second, the survey method should be decided upon. Since postal surveys are rarely used in South Korea, these will not be familiar to people. Random digital dialing telephone surveys could also be considered. However, it was difficult to ask questions while providing sufficient information over the phone. Moreover, the response rate to telephone surveys is quite low. Nowadays, internet surveys are a useful method; however, they are likely to suffer from sample selection bias. Eventually, the authors decided to implement face-to-face individual interview surveys. In crowded places, such as train stations, bus terminals, and department stores, one-to-one individual interviews can be conducted relatively easily. However, problems with sample selection can still arise. Therefore, in this study, data were collected through household visits.

Face-to-face individual interviews through household visits incur higher costs compared to other survey methods. Despite this, the authors made an effort to secure a sufficient budget for the survey. This is because the survey results and the analysis thereof can be utilized to provide recommendations to the government regarding ECs versus HCs. To derive meaningful insights, appropriately developed statistical models should be applied to the data collected through a reliable method. Therefore, the findings from this study can be considered the result of a rational survey approach.

The third aspect concerns who implements the survey. Although the authors can conduct the survey in person, three issues could arise as a result. First, in order to secure the representativeness of the sample, sampling must be performed to reflect the population’s characteristics accurately. In this regard, the most recent census data from Statistics Korea can be considered. However, the authors did not have scientific or professional skills in extracting a sample that properly represents the population. Second, it is difficult for interviewers hired by the authors to have sufficient experience and skill in conducting interviews and to collect reliably completed questionnaires. Third, a national survey should be carried out, but the authors could not afford to spend time traveling around the country. Therefore, a professional survey firm conducted the entire survey process. The firm has conducted dozens of national public opinion surveys per year commissioned by the public and private sectors. In other words, the firm has a lot of experience in surveys.

The final issue to address is the size of the sample. This study decided to collect 1000 observations for three reasons. First, in a general poll conducted in South Korea, a sample size of 1000 is usually selected [21,22,23,24,25,26,27,28]. Second, the Korea Development Institute [29] recommends 1000 as the appropriate sample size for public opinion surveys. Third, the budget secured by the authors allowed up to 1000 observations. The survey firm successfully collected 1000 completed questionnaires from all over the country, reflecting the characteristics of the population. The total number of households visited was 6000 due to resident absence, survey refusal, and failure to answer certain important questions, among others.

The population of South Korea is approximately 52 million people; some may therefore question the validity of surveying opinions based on a sample of only 1000 people. However, in most public opinion surveys carried out in the country, around 1000 observations are used, so 1000 is not considered a small number. Furthermore, in order to maintain consistency with the census data conducted by the Korean Statistical Office in 2020, the sample was constructed without arbitrary sampling. Therefore, the authors believe that their sample size of 1000 is sufficient to reflect the opinions of the South Korean population adequately.

2.2. Preparation of the Questionnaire

The authors aimed to design a questionnaire that was easy for respondents to answer while capturing all the necessary information in two stages. In the first stage, a list of required information was compiled, followed by the formulation of questions to extract that information. In the second stage, the questions were administered to undergraduate and graduate students, during which difficult or redundant parts were identified. Through this process, the survey items were refined, and the questionnaire draft was prepared to be handled over to the survey firm.

The supervisors at the survey company thoroughly reviewed and made significant revisions to the draft questionnaire created by the authors. In the course of the modification, the parts that would have been difficult for respondents to understand were sufficiently corrected. The final questionnaire used in the field survey was composed of three components. First, as the introduction to the survey, several questions were included about respondents’ perceptions, for example, whether they have ever driven an EC, HC, or hybrid car; whether they are aware that ECs are heavier than internal combustion engine vehicles because of the weight of their batteries; whether they are willing to pay more to consume electricity from renewable energy instead of electricity from fossil fuels; whether their political orientation is progressive; and their overall satisfaction with life.

The second component asked the interviewees about their preference in two steps. In the first step, an explanation was presented: “The transportation sector, such as roads, aviation, and sea, should reduce GHG emissions drastically. Among them, the platform that emits the most GHGs is the roads. That is, roads account for 96.5%. In response, the government wants to convert fuel used in commercial vehicles, such as trucks and buses, from diesel or natural gas to electricity or hydrogen that does not emit GHGs. ECs and HCs have their own advantages and disadvantages. Since it is hard to increase both dramatically at the same time, we have to focus more on one of the two”.

Table 1. Explanation of electric cars and hydrogen cars.

	Electric Cars	Hydrogen Cars
Definition	A car that generates a driving force by supplying electrical energy from a high voltage battery to an electric motor	A car that uses a fuel cell producing electric power by directly reacting hydrogen with oxygen in the air
Advantages	(i) Charging stations everywhere (ii) Low charging cost (iii) Relatively low price for a car	(i) Long mileage (approximately 600 km after one charge) (ii) Short charging time (approximately 5 min)
Disadvantages	(i) Short mileage (approximately 400 km after one charge) (ii) Long charging time (approximately 30 min for fast charging and 5 h for slow charging)	(i) Lack of charging stations (ii) High charging cost (iii) Relatively high price for a car

, describing the advantages and disadvantages of ECs and HCs, was presented to the interviewees during the course of the survey.

In the second step, the question was asked: “Please check with V which one you prefer to convert existing commercial vehicles using diesel or natural gas (trucks and buses) that emit GHG into electric cars or hydrogen cars?” More specifically, Figure 1 was presented to the interviewees. Each interviewee selected one of the 11 scores and checked it with V. The interviewees first determined whether they preferred ECs or HCs and then selected one of the five scores in relation to the strength of this preference. Five scores were used because the five-point scale is most commonly adopted in ordinary polls implemented in South Korea.

The third component contained questions about interviewees’ characteristics, such as age, gender, household head status, education level, household income, personal income, and residence. Some interviewees were particularly reluctant to provide numerical information in the case of income. In this case, 11 examples related to the interval to which income belongs were presented, and one was selected. As a result, the interviewees chose one without difficulty. An equation can be estimated using the response to the question contained in the second component as a dependent variable and the responses to the questions contained in the first and the third components as independent variables. Then, the effect of independent variables on the preference, which is the dependent variable, can be analyzed.

2.3. Analysis of the Data

It is necessary to examine the nature of the dependent variable more comprehensively. The preferences obtained from Figure 1 are ordinal rather than cardinal. In other words, numbers represent orders, not sizes. Therefore, a classical regression analysis that considers the dependent variable to be a continuous value should not be used. A representative model that can handle ordinal dependent variables is the ORM.

The conventional probit model is one of the regression methods applied when the dependent variable is binary data and not continuous. For example, it is used when the survey response is divided into “preferred” and “not preferred”. On the other hand, the ORM can be applied when the data of the dependent variable are of three or more discrete types. In other words, it can be employed when the survey response is divided into “preferred”, “slightly preferred”, “slightly not preferred”, “not preferred”, etc. Indeed, ORM has been applied in a number of studies in which qualitative data observed as categories were used as the dependent variable [30,31,32,33]. In summary, ORM is suitable for analyzing the survey data conducted in this study. In addition, its application is in line with several previous studies.

Therefore, the ORM is applied in this research. When converting the value selected in Figure 1 into a dependent variable, the leftmost value is set to be 11, and the rightmost value is set to be 1. The larger the number, the greater the preference for ECs over HCs. However, since this number is ordinal, it is difficult to say that the preference strength of the number 6 is twice as large as that of the number 3.

In order to formulate the ORM, two variables representing preference, a latent variable and an observed variable, must be defined. Let the former and the latter be $S_i^{*}$ and $S_i$ for the interviewee $i$, respectively. Let the vector of the independent variables that affects the preference and the corresponding parameter vector be $Y_i$ and $\theta$, respectively. The preference equation may thus be expressed as follows:

$$S_i^{*}=Y_i\theta +{\lambda }_i$$

(1)

where ${\lambda }_i$ is assumed to follow a standard normal distribution with a mean of 0 and a standard deviation of 1. The observed variable, $S_i$, can have one of a total of 11 values from 1 to 11, and the probability associated with $S_i$ is defined as follows:

$$S_i=\left\{\begin{array}{c}1 \mathrm{if} S_i^{*}\le {\omega }_0\hfill \\ 2 \mathrm{if} {{\omega }_0<S}_i^{*}\le {\omega }_1\\ 3 \mathrm{if} {{\omega }_1<S}_i^{*}\le {\omega }_2\\ 4 \mathrm{if} {{\omega }_2<S}_i^{*}\le {\omega }_3\\ 5 \mathrm{if} {{\omega }_3<S}_i^{*}\le {\omega }_4\\ 6 \mathrm{if} {{\omega }_4<S}_i^{*}\le {\omega }_5\\ 7 \mathrm{if} {{\omega }_5<S}_i^{*}\le {\omega }_6\\ 8 \mathrm{if} {{\omega }_6<S}_i^{*}\le {\omega }_7\\ 9 \mathrm{if} {{\omega }_7<S}_i^{*}\le {\omega }_8\\ 10 \mathrm{if} {{\omega }_8<S}_i^{*}\le {\omega }_9\\ 11 \mathrm{if} {{\omega }_9<S}_i^{*}\hfill \end{array}\right.$$

(2)

where $\omega$s are thresholds to be estimated, and ${\omega }_0=0$ is usually assumed. Thus, the thresholds to be estimated are nine, ${\omega }_1,\dots ,{\omega }_9$.

Each case in Equation (2) can be modeled as probability. For $n$ from 1 to 9, the probability that $S_i^{*}$ is less than or equal to $w_n$ can be expressed as Equation (3):

$$\begin{array}{cc}\mathrm{Pr}\left(S_i\le w_n\right)\hfill & =\mathrm{Pr}\left(Y_i\theta +{\lambda }_i<w_n\right)\hfill \\ & =\mathrm{Pr}\left({\lambda }_i<w_n-Y_i\theta \right)\hfill \\ & =\mathrm{Pr}\left(Z_i<w_n-Y_i\theta \right)\hfill \\ & =G(w_n-Y_i\theta )\hfill \end{array}$$

(3)

where $Z_i$ is a standard normal random variable and $G(·)$ is a standard normal cumulative distribution function. Using Equation (3), each probability in Equation (2) can be expressed as follows so that it can be entered into the likelihood function.

$$\left\{\begin{array}{c}\mathrm{Pr}\left(S_i=1\right)=G\left({\omega }_0-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=2\right)=G\left({\omega }_1-Y_i\theta \right)-G\left({\omega }_0-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=3\right)=G\left({\omega }_2-Y_i\theta \right)-G\left({\omega }_1-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=4\right)=G\left({\omega }_3-Y_i\theta \right)-G\left({\omega }_2-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=5\right)=G\left({\omega }_4-Y_i\theta \right)-G\left({\omega }_3-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=6\right)=G\left({\omega }_5-Y_i\theta \right)-G\left({\omega }_4-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=7\right)=G\left({\omega }_6-Y_i\theta \right)-G\left({\omega }_5-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=8\right)=G\left({\omega }_7-Y_i\theta \right)-G\left({\omega }_6-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=9\right)=G\left({\omega }_8-Y_i\theta \right)-G\left({\omega }_7-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=10\right)=G\left({\omega }_9-Y_i\theta \right)-G\left({\omega }_8-Y_i\theta \right)\hfill \\ \mathrm{Pr}\left(S_i=11\right)=1-G\left({\omega }_9-Y_i\theta \right)\hfill \end{array}\right.$$

(4)

To construct the likelihood function with Equations (1)–(4), eleven dummy variables are defined as follows:

$$\left\{\begin{array}{c}{D}_{1i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{E}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{a}\mathrm{b}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{2i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{E}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{e}\mathrm{x}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{3i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{E}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{v}\mathrm{e}\mathrm{r}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{4i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{E}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{5i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{E}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{s}\mathrm{l}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{6i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{I}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{t}.\mathrm{”}\right)\hfill \\ D_{7i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{H}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{s}\mathrm{l}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{8i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{H}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{9i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{H}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{v}\mathrm{e}\mathrm{r}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{10i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{H}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{e}\mathrm{x}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \\ D_{11i}=1\left(\mathrm{T}\mathrm{h}\mathrm{e} \mathrm{r}\mathrm{e}\mathrm{s}\mathrm{p}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s} \mathrm{“}\mathrm{H}\mathrm{C} \mathrm{i}\mathrm{s} \mathrm{a}\mathrm{b}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{l}\mathrm{y} \mathrm{p}\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{d}.\mathrm{”}\right)\hfill \end{array}\right.$$

(5)

where 1(∙) is an indicator function. This function returns 1 when the content in parenthesis is correct and 0 otherwise.

Finally, the derived log-likelihood function is expressed as follows:

$$\begin{array}{cc}\mathrm{l}\mathrm{n}L\hfill & ={\displaystyle \sum _{i=1}^N}{\displaystyle \sum _{j=1}^{11}}{D}_{ji}\mathrm{l}\mathrm{n}\left[\mathrm{Pr}\left(S_i=j\right)\right]\hfill \\ & ={\displaystyle \sum _{i=1}^N}\left\{\begin{array}{c}{D}_{1i}\ln \left[G\left({\omega }_0-Y_i\theta \right)\right]\hfill \\ +D_{2i}\ln \left[G\left({\omega }_1-Y_i\theta \right)-G\left({\omega }_0-Y_i\theta \right)\right]\\ {+D}_{3i}\ln \left[G\left({\omega }_2-Y_i\theta \right)-G\left({\omega }_1-Y_i\theta \right)\right]\\ {+D}_{4i}\ln \left[G\left({\omega }_3-Y_i\theta \right)-G\left({\omega }_2-Y_i\theta \right)\right]\\ {+D}_{5i}\ln \left[G\left({\omega }_4-Y_i\theta \right)-G\left({\omega }_3-Y_i\theta \right)\right]\\ {+D}_{6i}\ln \left[G\left({\omega }_5-Y_i\theta \right)-G\left({\omega }_4-Y_i\theta \right)\right]\\ {+D}_{7i}\ln \left[G\left({\omega }_6-Y_i\theta \right)-G\left({\omega }_5-Y_i\theta \right)\right]\\ {+D}_{8i}\ln \left[G\left({\omega }_7-Y_i\theta \right)-G\left({\omega }_6-Y_i\theta \right)\right]\\ {+D}_{9i}\ln \left[G\left({\omega }_8-Y_i\theta \right)-G\left({\omega }_7-Y_i\theta \right)\right]\\ {+D}_{10i}\ln \left[G\left({\omega }_9-Y_i\theta \right)-G\left({\omega }_8-Y_i\theta \right)\right]\\ {+D}_{11i}\ln \left[1-G\left({\omega }_9-Y_i\theta \right)\right]\hfill \end{array}\right\}\hfill \end{array}$$

(6)

The estimates for $\theta$ and $\omega$s are obtained through maximizing the log-likelihood function given in Equation (6). The estimated size of $w_n$ is not of great significance. However, if these secure statistical significance, it can be determined that the ORM has been applied successfully. In addition, it is possible to check how the characteristics of respondents influence their preferences by using the sign of the estimated coefficients.

3. Results and Discussion

3.1. Results from Conducting the Survey

The survey began in early June 2022 and was conducted nationwide for about a month before ending in early July. The sample was extracted scientifically and elaborately to represent the population. Figure 2 compares the characteristics of the population and the sample using three pieces of information. The first is the proportion of households residing in a specific area. For example, the proportion of households residing in Seoul was 19% of the population and 22% of the sample. The second piece of information is the gender composition ratio for each group. The third piece of information is the average household income. As shown in Figure 2, there are not significant differences between the values of the population and the sample. Therefore, the representativeness of the sample obtained for this survey is sufficiently secured.

The public opinions for ECs versus HCs are summarized in

Table 2. Summary of public views on electric cars and hydrogen cars.

	Electric Cars	Hydrogen Cars
Slightly preferred	70	14
Preferred	274	20
Very preferred	177	46
Extremely preferred	55	101
Absolutely preferred	24	29
Total	600	210
Indifferent	190

. A total of 600 interviewees preferred ECs to HCs, and 210 interviewees preferred HCs to ECs. These account for 60.0% and 21.0% of the total survey population, respectively. The former is about 2.9 times the latter. A total of 190 interviewees expressed indifference, accounting for 19% of the total. Figure 3 represents the distribution of public opinions on ECs versus HCs according to the characteristics of the interviewees. People responded by comprehensively considering the characteristics of ECs and HCs presented in Table 1, and ECs were evaluated as being more important than HCs for carbon neutralization of commercial vehicles. This can be said to be an interesting discovery of this study.

Table 3. Explanation of the eleven variables chosen as factors affecting respondents’ preferences.

Variables	Definitions	Means	Standard Deviations
Gender	Whether the interviewee is female (0 = no; 1 = yes)	0.50	0.50
Education	Whether the interviewee’s number of years of education is 12 or more (0 = no; 1 = yes)	0.65	0.48
Age	Whether the interviewee is in her/his 40s or older (0 = no; 1 = yes)	0.74	0.44
Head	Whether the interviewee is the head of the household (0 = no; 1 = yes)	0.54	0.50
Income	The interviewee household’s monthly income (unit: million Korean won)	5.28	2.18
Metro	Whether the interviewee dwells in the Seoul Metropolitan area (0 = no; 1 = yes)	0.53	0.50
Drive	Whether the interviewee has experience driving eco-friendly cars, such as electric cars, hydrogen cars, and hybrid cars (0 = no; 1 = yes)	0.22	0.42
Electric	Whether the interviewee is aware that electric cars are heavier than internal combustion engine cars due to the weight of the battery (0 = no; 1 = yes)	0.48	0.50
Renewable	Whether the interviewee is willing to consume renewable energy instead of fossil fuels, even if it costs more (0 = no; 1 = yes)	0.38	0.49
Political	The interviewee’s political inclination (0 = conservative or moderate; 1 = progressive)	0.27	0.44
Satisfaction	Whether the interviewee is satisfied with her/his life (0 = no; 1 = yes)	0.67	0.47

explains the eleven variables selected as factors influencing people’s preference for ECs over HCs. The first column of the table shows the definitions of the variables. The table also presents the averages and standard deviations. The variables are largely composed of three categories. The first category relates to the individual characteristics of interviewees. Four variables were included: gender, education level, age, and whether the respondent was a homeowner. The second category concerns the characteristics of interviewee households and included two variables: the income of interviewee households and whether they live in the Seoul Metropolitan area. The third category reflects the interviewee’s experiences or perceptions. In this regard, a total of five variables in the lower part of Table 3 were used.

3.2. Results from Analyzing the Data

As addressed in the previous section, this article estimated the ORM to delve into some factors influencing public preference for ECs over HCs.

Table 4. Results from estimating the model.

Variables a	Estimates	t-Values
Constant	1.0251	4.36 *
Gender	0.5271	3.23 *
Education	−0.1312	−1.77 *
Age	−0.1409	−1.71 *
Head	0.4536	2.77 *
Income	0.0517	3.16 *
Metro	0.2047	3.03 *
Drive	0.1956	2.41 *
Electric	0.1737	2.48 *
Renewable	0.1444	2.10 *
Political	−0.1413	−1.90 *
Satisfaction	0.1212	1.75 *
${\omega }_1$	0.7984	10.52 *
${\omega }_2$	0.9977	12.65 *
${\omega }_3$	1.0731	13.46 *
${\omega }_4$	1.1230	14.00 *
${\omega }_5$	1.6861	19.99 *
${\omega }_6$	1.8714	21.91 *
${\omega }_7$	2.6361	29.23 *
${\omega }_8$	3.4181	34.16 *
${\omega }_9$	3.9960	33.46 *
Scaled-$R^2$	0.06
Log-likelihood	−1994.53
Likelihood ratio statistic (p-value) b	62.58 (0.000)
Sample size	1000

^a They are described in Table 3. ^b The statistic is derived under the null hypothesis that the model is meaningless. * indicates that the estimate is statistically significant at the 10% level.

shows the observations from the estimation. First, the statistical significance of the model is explored. The likelihood ratio test statistic under the null hypothesis that the model is meaningless is computed as 62.58. This statistic is distributed as chi-squared under the null hypothesis. Its degree of freedom is 11. The corresponding p-value was 0.000. Consequently, the hypothesis is dismissed at the 5% level. Statistical significance is secured by the estimated model. Next, the scaled-$R^2$ presented in Estrella [34] was calculated to be 0.06. Since this value is relatively small, the goodness-of-fit of the model seems low. However, low $R^2$ is often observed when using cross-sectional data. Moreover, the low goodness-of-fit has little to do with the significance of the model.

Statistical significance is guaranteed for all 12 estimated coefficients at the significance level of 10%. All nine threshold values were also statistically significant. Since the dependent variable is the preference for ECs compared to HCs, the signs of independent variables have meaning, for example, individuals who were younger, had lower education levels, had higher household incomes, or had a higher preference for ECs compared to HCs. People who were household heads, were residents of the Seoul Metropolitan area, or had experience driving eco-friendly cars were more supportive than others of ECs versus HCs. Furthermore, the respondents’ perceptions and inclinations have significant effects on their preferences.

3.3. Discussion of the Results

Since the survey identified that the three kinds of factor have an effect on the preference for ECs over HCs, it is necessary to discuss this further. The first factor is interviewees’ characteristics. Women were more supportive than men of ECs over HCs. Education level also had a negative correlation with the preference for ECs over HCs. Similarly, both younger people and homeowners supported ECs more than HCs. Therefore, men, more educated people, older people, and non-homeowners preferred HCs to ECs in relation to implementing carbon neutrality in the commercial vehicle transportation sector. Given that HCs are now more expensive than ECs, the identification of these groups is interesting.

The second kind of factor is the interviewee household’s characteristics. The higher the income of the interviewees’ households, the more they supported ECs over HCs. This discovery is not encouraging in terms of expanding the supply of HCs. In order to expand the supply, more subsidies must be provided to HCs than to ECs, and the financial resources will be taxes borne by the general public. Given that people with higher incomes pay more taxes, the discovery that the richer people are, the more negative they are about HC is cause for the government worry. Respondents living in the Seoul Metropolitan area preferred ECs over HCs as a carbon neutrality implementation plan in the commercial vehicle transportation sector more than respondents who did not. The region occupies only 12% of South Korea’s land area but has half its population. Consequently, the views of the Seoul Metropolitan area’s residents are treated as important.

The third factor type was the interviewee’s experiences and perceptions. Those who had experiences of driving eco-friendly vehicles such as ECs, HCs, and hybrid cars preferred ECs to HCs. In the case of HCs, as the charging infrastructure is still insufficient, there are many difficulties in charging HCs. The third factor seems to be a discovery that reflects this point. Those who knew that ECs were heavier than internal combustion engine cars were more favorable toward ECs than HCs. Because people who know this fact have a high understanding of ECs, it seems that they have a friendly attitude toward ECs.

People who intended to pay more for electricity from renewable energy instead of electricity from ordinary fossil fuels thought more favorably than others toward ECs over HCs. Since they value renewable energy, they seem to have given higher scores to ECs that can directly use renewable energy electricity than to HCs that need to rely on gray hydrogen from fossil fuels for the time being. Those who judged themselves to have progressive political inclinations were more favorable toward HCs than ECs. This discovery is noteworthy. As progressive forces in South Korea tend to value the expansion of renewable energy, it was expected that they would prefer ECs to HCs.

3.4. Implications for Policy-Making

The South Korean government plans to supply 3.62 million ECs and 0.88 million HCs by 2030 to promote carbon neutrality in the transportation sector. For the time being, not only are HCs more expensive than ECs, but it is also difficult to secure sufficient hydrogen charging stations. Thus, the EC supply target is 4.11 times the HC supply target. The same goes for the commercial vehicle transportation sector. As a means of achieving carbon neutrality in the commercial vehicle transportation sector, 2.86 times more people preferred ECs than preferred HCs. The government’s policy to supply more ECs than HCs is consistent with the public preference. The government needs to steadily and continuously push for a carbon neutrality policy in the commercial vehicle transportation sector centered on ECs. Of course, it is not necessarily desirable to pursue a policy based solely on public preferences, but the fact that the current government policy meets public preferences must be an important basis for the policy to secure momentum.

It is necessary to examine the distribution status and prices of buses and trucks currently sold in South Korea. First, buses are largely divided into city buses and intercity buses. The fuel of most city buses is gradually being converted from natural gas to electricity. Hydrogen buses are very poorly distributed. The cost of city buses is in the order of hydrogen buses, electric buses, and natural gas buses. In particular, the price difference between hydrogen buses and electric buses is greater than that between electric buses and natural gas buses. The expansion of hydrogen buses is facing difficulties. Thus, both local governments and the central government are providing subsidies to expand the distribution of hydrogen buses and to expand the hydrogen charging infrastructure. Nevertheless, hydrogen buses and electric buses are competing with each other in the city bus sector, with electric buses currently taking the lead.

On the other hand, the fuel for intercity buses is mostly diesel. Since intercity buses operate long distances, it is not feasible to use electricity as the power source. Thus, the transition to electric buses is minimal. It is projected that in the future intercity buses will be replaced by hydrogen buses, which are more suitable than electric buses for long-distance operations. In line with this, the government is focusing on supporting the construction of hydrogen refueling stations for large commercial vehicles rather than small vehicles and encouraging domestic automakers to produce large hydrogen buses. Additionally, some major companies that emphasize environment, social, and governance (ESG) management are actively considering the conversion of fuel for commuting buses from diesel to hydrogen.

Second, freight trucks are broadly classified into small and large trucks. For small trucks, the adoption of electric freight trucks is increasing, while small hydrogen trucks are not yet widely distributed in South Korea. On the other hand, a South Korean automaker has produced the world’s first hydrogen freight truck in the large truck category. However, whether it is an electric or hydrogen large truck, the adoption is still limited. This is because the price of electric or hydrogen large trucks is high compared to conventional ones. Since large trucks generally have long transportation distances, hydrogen trucks are expected to be more advantageous in the long run compared to electric trucks. Moreover, companies, such as department stores and courier services, are actively exploring the transition from diesel trucks to hydrogen trucks as part of their ESG management.

In summary, although the adoption of electric vehicles in the commercial transportation sector is expanding, the adoption level of hydrogen vehicles is relatively low. For the time being, electric vehicles are expected to grow at a faster pace than hydrogen vehicles. Nevertheless, thanks to various government support measures, hydrogen vehicles are expected to take the lead in the city bus sector. In the long-distance intercity bus sector, involving commuting buses of large corporations, the freight truck sector, department stores, and courier companies, there is a possibility that small hydrogen trucks will gain traction over electric trucks. While many people have preferred ECs over HCs for carbon neutrality in the commercial vehicle transportation sector, there is a possibility that hydrogen vehicles may see more growth in certain commercial vehicle sectors.

Meanwhile, the public preference for ECs versus HCs, may be further investigated. Responses to preferences are clearly aggregated in Table 2, but they can be synthesized in consideration of the strength of preferences. This consists of two steps. In step 1, for convenience, a score from 1 to 5 can be assigned to preference responses in ascending order of preference strength. In other words, 1 is assigned to “slightly preferred” and 5 “absolutely preferred”, while 2, 3, and 4 are assigned to the responses in ascending order of preference strength. In step 2, the scores given to those who prefer ECs and to those who prefer HCs are averaged. As a result, the average vales of preference strength was 2.48 for ECs and 3.53 for HCs. That is, although more people preferred ECs than HCs, the strength of preference was greater for HCs than for ECs. Although there are relatively few people who prefer HCs, the preference for HCs was stronger. This is also an interesting finding from this research.

South Korea is a representative automobile manufacturer. Most of the internal combustion engine buses currently in operation are domestic, but more than half of the ECs are foreign. This is because foreign electric buses are cheaper than domestic ones, and the domestic EC production infrastructure itself is insufficient. At present, few electric trucks are in operation. Instead, hydrogen buses and hydrogen trucks are being produced by local carmakers, and their production will gradually increase. Therefore, in terms of domestic employment and value-added creation, some argue that it is more desirable to expand the supply of commercial HCs than commercial ECs. The government agrees with this. However, this study found that the public preferred ECs to HCs in the commercial vehicle transportation sector.

There is therefore a gap between the argument and the discovery from this study. We need to examine the causes of this gap in terms of policy. Judging from the comments from the interviewers, there were two reasons why people showed a higher preference for ECs than HCs: the insufficient hydrogen charging infrastructure and concerns about the safety of hydrogen fueling stations and HCs. If the government intends to increase the supply of commercial HCs, its policy must concentrate on addressing these two concerns. In particular, these two concerns are closely linked rather than being wholly distinct. In other words, safety issues mean residents’ acceptance of the construction of a hydrogen charging infrastructure remains low; thus, the infrastructure is insufficient. Therefore, priority should be given to addressing safety issues.

It is necessary to discuss the appropriateness and limitations of using survey data, without employing engineering or macroeconomic analytical models, to provide policy recommendations for ECs to HCs. As emphasized earlier, the authors do not claim that the preferences of the public, as revealed through surveys, are the definitive answer to solving the decision on the policy direction. Public preferences are merely one important piece of information in addressing the problem at hand. Nonetheless, public preferences cannot be ignored because government policies require public support to ensure their implementation. In particular, the public is the entity directly bearing the costs associated with government policy implementation. Considering the opinions of taxpayers is an essential element for the successful implementation of policies.

In other words, while public preferences are not the ultimate answer, they undoubtedly should be a key factor reflected in policy decisions. The purpose of this study was to identify and accurately determine those factors, to analyze various determinants of those factors, and derive and present implications. If the results obtained from engineering or economic analytical models could be directly compared and aligned with the findings of this study, it would further enhance the significance of this research. However, if any discrepancies were found, investigating the causes of such discrepancies would also be meaningful. To use an analogy, the objective of this study was not to cook a delicious meal but to provide the ingredients that can be used to make that meal.

4. Conclusions

This article makes two major contributions to the literature. First, the authors have delved into the public preference for ECs over HCs in South Korea. This was the first trial in the literature as the authors failed to find relevant previous research. It was discovered that there are 2.86 times more people who prefer ECs to HCs. Although this discovery is specific to the country, the main findings from this research can provide a useful reference for countries struggling to decide between ECs and HCs. Moreover, if the results applied to countries other than South Korea are compared, differences between countries may be identified. If the causes of these differences are identified, new implications will be drawn.

Second, the factors affecting the public preference for ECs over HCs have been effectively investigated in this article. In particular, the ORM, which is widely accepted in the literature, was successfully applied while explicitly reflecting that the preferences derived from the survey were not cardinal but ordinal. This model passed the specification test while securing statistical significance. In addition, statistical significance was secured for all estimated coefficients. The implications of the signs of the estimated coefficients were also appropriately addressed. Therefore, this study is significant in confirming the applicability and usefulness of the ORM in analyzing these preferences.

There will be two follow-up works to this study. First, when implementing carbon neutrality in the commercial vehicle transportation sector, the appropriate proportion of ECs and HCs should be derived. About 20 million vehicles are currently registered in South Korea. Future increases in the EC to HC ratio can be determined by reflecting public preferences. As explained earlier, South Korea has established a plan to supply 3.62 million ECs and 0.88 HCs by 2030, and it should be considered whether this supply ratio is appropriate. Of course, in this regard, various factors, such as employment-inducing effects, supply stability, the economy, and GHG reduction effects, should be comprehensively considered from the perspective of lifecycle assessment.

Second, since preferences may vary from region to region, additional research should be conducted to identify regional differences in the preference for ECs over HCs. To achieve this, the number of observations should be increased. The regional preference identified may enable us to determine the supply ratio of ECs over HCs differently by region. For example, if ECs are preferred over HCs in a particular region, implementation of ways to expand the supply of ECs rather than HCs may be considered in that region. In areas where HCs are preferred to ECs, investment in expanding the HC charging infrastructure should be increased while allocating more subsidies to HCs compared to ECs in order to expand the supply of HCs.