South Korean Public Acceptance of the Fuel Transition from Coal to Natural Gas in Power Generation

Abstract

South Korea has set up a plan to convert 24 coal-fired power plants into natural gas-fired ones by 2034 in order to reduce carbon dioxide (CO₂) emissions. This fuel transition can succeed only if it receives the public support. This article seeks to investigate the public acceptance of the fuel transition. For this purpose, data on South Koreans’ acceptance of the fuel transition were gathered on a nine-point scale from a survey of 1000 people using face-to-face individual interviews with skilled interviewers visiting households. The factors affecting acceptance were identified and examined using an ordered probit model. Of all the interviewees, 73.6 percent agreed with and 12.2 percent opposed the fuel transition, respectively, agreement being about six times greater than opposition. The model secured statistical significance and various findings emerged. For example, people living in the Seoul Metropolitan area, people who use electricity for heating, people with a low education level, young people, and high-income people were more receptive of the fuel transition than others. Moreover, several implications arose from the survey in terms of enhancing acceptance.

1. Introduction

Since coal has a higher carbon content than other fossil fuels such as oil and natural gas (NG), it emits more carbon dioxide (CO₂), a greenhouse gas, during the combustion process. As a result, countries around the world are making various efforts to reduce coal power generation to prevent climate change [1,2]. In other words, energy transition is being pushed around the world to change from coal, which is high-carbon energy, to low-carbon or zero-carbon energy [3,4,5].

For example, the United States will shut down one-fifth of all coal-fired power plants by 2025 [6,7]. Germany plans to abolish all coal-fired power plants by 2038 by enacting the so-called ‘de-coal law’ and establishing a three-stage de-coal schedule [8,9]. As of December 2019, the plan is to reduce the capacity of 43.9 GW coal-fired power plants in Germany to 30 GW by 2022, 17.8 GW by 2030, and zero by 2038. In China, NG-fired power plants, replacing coal-fired power plants, are expected to grow at an average annual rate of 10% from 2025 to 2030, with a capacity of 235.7 GW by 2030 [10]. Japan also has a plan to abolish 100 of its 140 coal-fired power plants by 2030 [11].

South Korea is no exception to this trend [12]. The Ninth Basic Plan for Electricity Demand and Supply (2020–2034), which was finalized at the end of December 2020, proposed a goal to reduce the proportion of coal-fired power generation from 40.4 percent in 2019 to 29.9 percent in 2034 [13]. This is quite challenging as it will abolish 30 coal-fired power plants, half of the total of 60 coal-fired power plants in operation, by 2034. Of the 30 coal-fired power plants that will be abolished, six will be shut down completely, while the remaining 24 will be converted to NG-fired power plants. In producing the same amount of electricity, an NG-fired power plant emits less than half of the greenhouse gases emitted by a coal-fired power plant. In addition, if this trend continues, the remaining 30 coal-fired power plants that are supposed to operate without being abolished by 2034 are expected to be converted into NG-fired power plants or abolished as early as possible.

South Korea’s total CO₂ emissions in 2018 stood at 728 million tons, ranking eighth in the world; its per capita CO₂ emissions ranked higher, at sixth in the world. More specifically, the electricity and heat sector, industry sector, transportation sector, and commercial and other sector accounted for 287 million tons (39.4%), 243 million tons (33.4%), 98 million tons (13.5%), and 100 million tons (13.7%), respectively. Total CO₂ emissions in 2018 reached 149 percent of the amount in 1990, but CO₂ emissions in the electricity and heat sector reached 480 percent of the amount in 1990. That is to say, since the increase in CO₂ emissions in the electricity and heat sector is due to the increase in coal-fired power generation, reducing coal-fired power generation has emerged as the most basic challenge to reducing CO₂ emissions [14,15].

Under the Paris Agreement signed in 2015, South Korea submitted a nationally determined contribution (NDC) with a goal of reducing its 2018 greenhouse gas emissions by 26.3% by 2030 to the United Nations Framework Convention on Climate at the end of December last year. Currently, raising the reduction target on the NDC is discussed in preparation for the 26th Conference of the Parties to be held in Glasgow, United Kingdom in November 2021. For the successful implementation of the NDC, the power generation sector should abate its 2018 greenhouse gas emissions (269.6 million tons of CO₂e) to 192.7 million tons of CO₂e by 2030. To this end, the South Korean government intends to change the mix of the power generation sources in two directions [13].

First, South Korea will increase the share of renewable energy (RE) generation from 6.5% in 2019 to 20.8% by 2030. In particular, current global trends suggest that RE has become more reliable and cost-effective [16]. In South Korea, RE-related costs have been gradually decreasing. However, not a small amount of subsidies are still required for RE generation projects. This is because the price of RE facilities continues to fall, but the costs of compensation for local residents are increasing significantly as the acceptance of the residents in areas where RE facilities are being installed is getting worse. Therefore, the implementation of the NDC is almost impossible with the expansion of RE alone.

Second, replacing some of the coal-fired power plants with NG-fired ones will be promoted. Some argue that a significant portion of the coal-fired power plants being abolished should be substituted with RE. However, there are many limitations in terms of intermittency and variability to significantly expanding RE immediately. Moreover, unfortunately, South Korea’s power grid is not linked to that of neighboring countries, and its technological capabilities and market systems are not fully equipped to cope with the variability of RE. Thus, most of the coal-fired power plants that are abolished will be replaced by NG-fired ones for the time being. Of course, since RE will expand significantly in the long term, NG-fired power plants are expected to serve as bridges in the era of energy transition [17,18,19].

Three major problems have been raised in this regard, causing some controversy. First, South Korea is one of the highest-cost countries in the world to consume NG, as it must liquefy, transport, vaporize, and use NG produced overseas. In other words, converting power generation fuel from coal to NG could increase costs and eventually lead to higher electricity bills [20], which could increase the burden on the people and weaken industrial competitiveness. Second, almost all of the NG consumed in the country depends on imports, and due to its high dependence on the Middle East, such as Qatar and Oman, NG supply and price stability are weak depending on international political circumstances [21]. Therefore, it is pointed out that it may be not desirable to expand the use of NG in terms of improving energy security [22]. Third, comparing the number of people employed at coal-fired power plants with those employed at NG-fired power plants of equal capacity, the fuel transition is expected to reduce the number of jobs as the latter is about half of the former.

Because the above three issues have become disputable problems and no one is willing to take responsibility, public consensus on fuel transition is insufficient due to a lack of public debate and discussion [23]. In some cases, the fuel transition may not be carried out and may only have a declarative meaning, or it could run aground in the face of public opposition. Thus, it is necessary to draw clear implications after determining the public’s acceptance of the fuel transition from coal to NG at this point in time [24,25]. This is because it is the people who bear the social costs of fuel transition, and energy policymakers desperately need information about their acceptance.

With countries around the world struggling to reduce greenhouse gas emissions, the policy of converting power generation fuels from coal to NG is a global trend [26]. This fuel transition can be seen from two perspectives. First, both coal and NG are fossil fuels, but the fuel transition is inevitable because in generating the same amount of electricity combustion of NG emits much less both CO₂ and particulate matters than that of coal. For example, de Gouw et al. [27] reported that emissions of greenhouse gases and air pollutants decreased by 56% and 40–44%, respectively, when converting fuel from coal to NG. Eventually, in order to reduce emissions of air pollutants as well as greenhouse gases, the transition of power generation fuel from coal to NG will accelerate [14,28]. As mentioned earlier, the representative countries that are pursuing this fuel transition are the United States, Germany, China, and Japan [26].

Second, because NG also emits CO₂ during combustion, NG plays a role as a bridge energy that maintains an intermediate stage to a complete RE society [18,19,29,30,31]. Of course, since NG is also a fossil fuel, the expansion of its use may be limited at some point in the future. However, it is expected that its role as a bridge energy will expand [32,33]. Regardless of which of the two perspectives is more important, the fuel transition will continue for the time being.

However, so far as the authors know, there is no literature that analyzes the public acceptance of the fuel transition and the factors affecting it. Since the fuel transition leads to an increase in power generation costs, the public acceptance of the transition needs to be clearly examined. This study aims to meet this need. In other words, this study intends to add this analysis using the specific case of South Korea to the literature at a time when coal as a power-generation fuel is being replaced by NG, but investigation of the public acceptance of the replacement and the factors influencing it remains scarce.

The public acceptance covered in this paper is not about technology or the fuel itself. The fuel transition can have several effects, but the most important effect is an increase in electricity bills. Therefore, the main target of the public acceptance dealt with in this study can be said to be the increase in electricity rates caused by the fuel transition. This is because abolishing coal-fired power plants and converting them into NG-fired power plants will reduce greenhouse gas emissions while increasing electricity rates. In the South Korean situation, NG is more expensive than coal.

After simply examining the public acceptance of the transition of power generation fuel from coal to NG, this study attempts to analyze the factors that affect its acceptance. To this end, a survey of 1000 people nationwide was conducted and the results reported. To the best of the authors’ knowledge, since this attempt is the first in South Korea and there are few cases internationally, this research is believed to be able to contribute significantly to the related literature. The remainder of this paper consists of four sections. The next section reports a brief literature review, and the history and the present situation of coal-fired power generation in South Korea. Section 3 presents a description of the materials and methods adopted in this work. Section 4 explains and discusses the results. Section 5 deals with conclusions.

2. Brief Literature Review and History of Coal-Fired Power Generation in South Korea

2.1. Literature Review

Public acceptance, as well as expert opinion, is an important consideration in the establishment and implementation of energy policies [34]. Therefore, a number of studies have been conducted to analyze the public acceptance of various RE sources and energy policies. For example, the public acceptance of hydrogen technology [35], hydrogen charging stations [36,37], solar energy [38], wind energy [38,39,40,41], geothermal energy [42], hydroelectric energy [43], energy infrastructure [44,45], various energy sources [46], RE cooperatives [47], RE policies [48], and energy transition policy [4] have been analyzed in the literature.

Some of these studies identified factors influencing public acceptance and then examined them as independent variables. For instance, Huijts et al. [37] used variables on individual perceptions such as individual behavior, perception of the effectiveness of hydrogen charging stations, subjective norms, personal norms, trust in industries, trust in local governments, awareness of environmental issues, and awareness of fairness in hydrogen charging station deployment. Mistur [46] employed health level, political orientation, ideology, gender, education, age, race, religion, and residential area as variables. Tabi and Westenhagen [43] adopted gender, age, income, education, political views, residential areas, and membership of environmental organizations as variables.

Kim et al. [40] utilized gender, education, age, income, residence in the metropolitan area, home solar power retention, average monthly electricity consumption per household, perception of the proportion of NG-fired power generation in the nation’s total energy generation, and political tendencies as variables. Fischer et al. [47] included risk-taking perception, patience, political orientation, environmental perception, age, gender, education, income, rural area residence, and West Germany as variables. Kim et al. [4] deployed gender, age, education, income, residence in the metropolitan area, electric heating, cooktop use, environmental awareness, and prior knowledge of the renewable energy 100% campaign as variables. Furthermore, Venkatesh et al. [49] formulated the unified theory of acceptance and use of technology, and analyzed user acceptance of information technology. They found that the socio-demographic factors and individual experience affect individual acceptance of information technology.

Looking closely at these factors, they are largely divided into four categories. The first category is the socioeconomic variables of the respondent. Most of the studies that analyzed the public acceptance considered variables such as gender, age, education, and income of respondents. The second category is related to respondents’ perception and includes respondents’ perception of energy policy, RE-related technology, or prior knowledge of the object to be investigated. The third category is the variables concerning the respondent’s living environment, the location of the respondent’s residence, and the characteristics of the respondent’s house. The fourth category relates to the characteristics of respondents. For example, whether respondents engage in environmental group activities or use eco-friendly products. The variables used in previous studies are summarized in

Table 1. Factors affecting public acceptance used in previous studies.

Factors	Sources
Gender	Kim et al. [40], Tabi and Wuestenhagen [43], Mistur [46], Fischer et al. [47], Kim et al. [4], Venkatesh et al. [49], Seo et al. [19]
Age	Kim et al. [40], Tabi and Wuestenhagen [43], Mistur [46], Fischer et al. [47], Kim et al. [4], Venkatesh et al. [49], Seo et al. [19]
Education	Kim et al. [40], Tabi and Wuestenhagen [43], Mistur [46], Fischer et al. [47], Kim et al. [4], Seo et al. [19], Kim et al. [28]
Income	Kim et al. [40], Tabi and Wuestenhagen [43], Fischer et al. [47], Kim et al. [4], Seo et al. [19], Kim et al. [28]
Residential area	Kim et al. [40], Tabi and Wuestenhagen [43], Mistur [46], Fischer et al. [47], Kim et al. [4], Seo et al. [19]
Personal life characteristics	Huijts et al. [37], Kim et al. [40], Tabi and Wuestenhagen [43], Kim et al. [4], Seo et al. [19]
Health and faith	Mistur [46]
Personal Perception	Huijts et al. [37], Kim et al. [40], Tabi and Wuestenhagen [43], Mistur [46], Fischer et al. [47], Kim et al. [4], Venkatesh et al. [49], Seo et al. [19]

.

2.2. History of Coal-Fired Power Generation in South Korea

Coal-fired power plants, along with nuclear power plants, have contributed significantly to enhancing the industrial competitiveness of the export-driven South Korean economy by serving as a source of low electricity bills over the past 30 years. With the country’s successful localization of coal-fired power plants with a capacity of 500 MW, coal-fired power plants have not only played a role in offering a cheap and stable power supply, but also helped create jobs. South Korea suffered a nationwide rolling power outage on 15 September 2011 due to a lack of power supply facilities. A stable power supply emerged as an important task, and the Sixth Basic Plan for Electricity Demand and Supply (2013–2027), announced in February 2013, reflected the new construction of 10.5 GW capacity of coal-fired power plants [50].

In the course of establishing the Seventh Basic Plan for Electricity Demand and Supply (2015–2029), announced in July 2015, reducing the proportion of coal-fired power plants was discussed for the first time to reduce emissions of CO₂ and particulate matter [51]. However, the trend of expanding coal-fired power plants remained, as a stable electricity supply was considered more important than the environment. Until April 2017, a policy of continuously expanding coal-fired power generation had been implemented since the cost of coal-fired power generation was lower than that of NG-fired power generation or that of RE generation. In particular, the Ministry of Trade, Industry and Energy—a South Korean government department in charge of electric power policy—judged that maintaining cheap electricity bills through coal-fired power plants was more important than supplying eco-friendly power.

However, with the launch of a new government advocating energy transition in May 2017, policies began to be implemented to control the pace of the increase in coal-fired power plants. The coal-fired power plants under construction would be completed to stabilize the supply and demand of electricity, but all of the planned new coal-fired power plants would be scrapped, and older coal-fired power plants that reach 30 years of operation would be abolished [52]. The coal-fired power plants which were abolished were old and had a small capacity of 500 MW, but the coal-fired power plants under construction were the latest model, with a capacity of more than 1000 MW. Therefore, the speed of increase in the number of coal-fired power plants was reduced, but it was not planned to reduce the overall capacity of the coal-fired plants significantly.

Since then, severe particulate matter problems occurred in March and April 2019. People’s anxiety about particulate matter overwhelmed other social problems. In April 2019, the National Climate and Environment Council was launched as a state agency to take responsibility for and deal with particulate matter under the President’s direct control. Ban Ki-Moon, who served as the Secretary-General of the United Nations, was appointed chairman of the Council, to take charge of international cooperation to reduce particulate matter. The Korea Ministry of Environment announced that about 15 percent of particulate matter generated in the country was emitted from coal-fired power plants, and the issue of how to deal with coal-fired power plants was seriously discussed.

As a first step, the Council proposed the introduction of a particulate matter seasonal management system that would stop or minimize the operation of coal-fired power plants for four months from December 2019 to March 2020. The Korea Ministry of Trade, Industry and Energy opposed the introduction of the system, citing a surge in electricity demand for heating during winter, but the system was introduced and implemented in the name of reducing particulate matter. Under the system, coal-fired power plants should not operate as much as possible. If coal-fired power plants are inevitably operated to meet the increased demand for electricity, their power generation should be lowered to 80% of normal operation. This is because it is impossible for South Korea’s coal-fired power plants to reduce their power generation to less than 80% due to their design. Therefore, unlike other countries that aimed to reduce CO₂ emissions, South Korea has begun to push for reducing coal-fired power generation to abate serious particulate matter emissions.

In November 2020, the year after the Council was launched, it proposed to remove all coal-fired power plants by 2045. Considering Germany’s push to eliminate coal-fired power generation in 2038, a vote was taken among selected people in the nation with three proposed dates to cease coal-fired power generation: 2040, 2045 and 2050; the majority chose 2045. These people, called national representatives, were 500 people selected from all over the country by the Council in terms of age, income, gender, and region. In the meantime, China declared its intention to reach carbon neutrality by 2060 in September 2020, and Japan declared its intention to reach carbon neutrality by 2050 in October 2020. South Korea, geopolitically located between the two countries, also declared carbon neutrality by 2050 in October 2020. The specific carbon neutrality route and implementation means will be determined through further discussions, which are currently active.

The Ninth Basic Plan for Electricity Demand and Supply (2020–2034), which began in January 2019, was finalized and announced in December 2020, about two years later [13]. The reason it took such a long time was to reflect the greatly strengthened goal to reduce CO₂ emissions compared to the Eighth Basic Plan for Electricity Demand and Supply (2017–2031). The Ninth Basic Plan for Electricity Demand and Supply (2020–2034) calls for an additional reduction in CO₂ emissions in the power generation sector of 34.1 million tons by 2030 compared to the Eighth Basic Plan for Electricity Demand and Supply (2017–2031).

Due to time constraints, the Ninth Plan did not reflect carbon neutrality by 2050, but the Tenth Plan, which is scheduled to be finalized at the end of 2022, decided to reflect it. The Ninth Plan decided to abolish coal-fired power plants when they reached the age of 30 and to introduce a price-bidding mechanism on coal-fired power plants from April 2022 by setting an upper limit on coal-fired power generation [13]. The amount of coal-fired power generation decided annually will decrease year by year, and South Korea’s coal-fired power plants will be shut down in the near future.

3. Materials and Methods

3.1. How to Investigate Public Acceptance

Public acceptance of a particular policy pursued by the government means the quantity of people who agree with the implementation of the policy. The implementation could gain momentum if many people vote for it. However, it is difficult to secure momentum for the policy if many people disagree with its implementation. Thus, as mentioned in Section 2, there are quite a number of studies analyzing public acceptance of a newly introduced policy or technology in the field of energy. Of course, people’s approval is not the only consideration in pursuing a particular policy, but it must be one of the most important considerations [53].

The first thing to do here is to determine the methodology for analyzing public acceptance. This study aims to collect data by asking about public acceptance through a survey of a large number of people selected at random. Surveys are an effective means of collecting people’s opinions directly [4,34,40]. The study will investigate whether, in order to reduce CO₂ emissions, they are in favor of or against the policy of abolishing coal-fired power generation early and replacing the corresponding capacity with NG-fired generation. Not only are the pros and cons examined, but they are also tallied. In fact, comprehensively aggregating pros and cons is a very simple task. A more complex and meaningful task is to derive the implications by analyzing the determinants of those pros and cons, which is also performed in this work.

First, various issues related to survey data collection, such as the survey target, sampling method, sample size, and survey method, are reviewed below, given that a survey is used for this study. Next, how to construct specific questions to identify public acceptance is explained. Finally, an econometric model is presented that can derive implications by analyzing the determinants while considering the nature of the data collected on public acceptance.

3.2. How to Gather the Data through a Survey

Regarding the collection of data, the survey method, survey target, sampling method, sample size, and survey area are determined. First, the survey method used in this study is a costly person-to-person individual interview with households. Of course, other low-cost survey methods were also available, such as postal interviews, telephone interviews, and internet interviews; however, person-to-person individual interviews were essential to fully explain the background to the transition of power generation fuel from coal to NG. In the case of postal interviews, there is no guarantee that people will properly look at the enclosed data, and the collection rate of questionnaires is extremely low in South Korea. In the case of telephone interviews, it is difficult to fully explain background information to respondents. While internet interviews have the advantage of being the lowest cost of the survey methods, they can cause problems that make random sampling difficult, leading to an increased probability of sample selection bias.

Second, the survey target in this study was selected as adults aged between 20 and 65 years. Of course, the opinions of people under 20 or over 65 may be important, but to obtain a responsible answer, the survey target was limited to those who could engage in economic activities. In South Korea, one graduates from high school at the age of 19 and becomes a true adult from the age of 20, becoming a university student or getting a job, and engaging in economic activities. In addition, people aged 65 usually retire from economic activity. Meanwhile, the proportions of men and women in the survey were the same.

Third, the sample size in this study was 1000. If a relatively large sample size is used, it is desirable in that it reflects many people’s opinions, but it also creates disadvantages that increase the cost of the survey. In the end, it was necessary to determine the sample size that could collect people’s opinions with some representation and avoid the problem of a sharp increase in survey costs. In this regard, Arrow et al. [54] pointed out that a sample size of 1000 may be appropriate to collect people’s opinions. The Korea Development Institute [55] also proposed that the sample size be 1000 when conducting a survey for public sector decision-making. The sample size of 1000 used in this study is consistent with the suggestions made in these studies.

Fourth, the interviews were conducted by experienced interviewers belonging to a professional survey company. The authors first fully discussed the content of the questionnaire with the company’s supervisors. Next, the supervisors trained the interviewers. In the course of the training, interviewers practiced questioning each other with the questionnaire. Interviewers who completed the training visited households and had them complete the questionnaire. The interviewers checked that there was no response to the main questions in the questionnaire and asked respondents to correct and supplement the questionnaire if necessary. Respondents who completed the questionnaire received simple household items such as a shopping bag, toothpaste, or a portable sewing kit as gifts.

Fifth, out of a total of 17 provinces in South Korea, 16 provinces were targeted, excluding Jeju Island. This was because Jeju-do, an island far from the mainland, had the lowest population among the 17 provinces, while the unit price of the survey was the highest. For these reasons, Jeju-do is usually excluded when conducting person-to-person individual interview surveys in South Korea. The Korea Development Institute [55] also suggests excluding Jeju-do when conducting a national opinion survey.

3.3. How to Prepare the Questionnaire

The final version of the questionnaire used in this study consisted of three main parts. After explaining the purpose of the survey, the first part asked about basic perceptions of several things. The second part asked about acceptance of the policy of converting 24 currently operating coal-fired plants into equal-capacity NG-fired power plants by 2034. The third part contained questions about the general characteristics of respondents. Answers to these questions are considered as candidates for factors affecting acceptance. The questions were about residential area, heating system, household income, personal income, education level, age, gender, etc.

For the second part, which is a key part of the questionnaire, how to ask questions about public acceptance had to be decided. For example, Ono and Tsunemi [56] used four views: “absolutely disagree,” “slightly disagree,” “slightly agree,” and “absolutely agree.” However, as a result of requesting a preliminary survey of 30 people from a professional survey company, two points were raised. First, those who participated in the preliminary survey asked to add a “neutral” view. In practice, the most widely used Likert scale adopts five levels. For example, it would be appropriate to use the five options “strongly disagree,” “disagree,” “neutral,” “agree,” and “strongly agree.“

Second, participants required more granularity in the five views. For example, instead of two views, “strongly disagree“ and “disagree,“ it was proposed to use more granular views. Thus, these two were divided into four: “absolutely disagree,“ strongly disagree,“ “disagree,“ and “slightly disagree.“ The same was true of “strongly agree“ and “agree.“ In fact, the analytic hierarchy process, developed by Saaty [57] and widely used in decision analysis, suggests the use of a nine-point scale rather than a five-point scale in value judgment. Consequently, this research finally used a total of nine views, from opposition to affirmation, as measures of acceptance. In other words, 1 to 9 correspond to “absolutely disagree,” “strongly disagree,” “disagree,” “slightly disagree,” “neutral,” “slightly agree,” “agree,” “strongly agree,” and “absolutely agree.”

3.4. How to Identify the Factors Affecting Public Acceptance

Public acceptance will be affected by a variety of characteristics of respondents. This research considers a total of 11 variables in relation to these characteristics. The names and definitions of these are shown in

Table 2. Information on variables in the model.

Variables	Definitions	Mean	Standard Deviation
Metro	Dummy for interviewees living in the Seoul Metropolitan area (0 = no; 1 = yes)	0.534	0.499
Heating	Dummy for interviewee households using electricity for heating (0 = no; 1 = yes)	0.013	0.113
Income	Dummy for interviewee households’ monthly income being larger than KRW 4.88 million (USD 5.75 thousand) (0 = no; 1 = yes)	0.478	0.500
Education	Interviewees have more than twelve years’ education (0 = no; 1 = yes)	0.633	0.482
Age	Interviewees’ age	48.009	9.417
Know1	Dummy for interviewees knowing about energy transition policy well before the survey (0 = no; 1 = yes)	0.408	0.492
Know2	Dummy for interviewees knowing about hydrogen vehicles well before the survey (0 = no; 1 = yes)	0.323	0.468
Environment	Interviewees’ subjective judgment about which is more important: jobs or the environment (0 = jobs; 1 = environment)	0.456	0.498
Forest	Dummy for interviewees being in favor of the utilization of unused forest biomass (0 = no; 1 = yes)	0.482	0.500
Fsolar	Dummy for interviewees being in favor of the expansion of floating solar power facilities (0 = no; 1 = yes)	0.510	0.500
H2-car	Dummy for interviewees being in favor of the expansion of hydrogen vehicles (0 = no; 1 = yes)	0.359	0.480

. In addition, basic statistics such as average and standard deviation are included in the table. In determining these 11 variables, the previous studies that investigated some factors affecting the public acceptance presented in Table 1 were referred to as important. As shown in Table 1, the main variables used in previous studies were income, residential area, gender, age, education, personal life characteristics, health and faith, and personal perception. These variables are generally composed of three categories: the characteristics of respondent households, the individual characteristics of respondents, and the perception and judgment of respondents. Therefore, the 11 variables presented in Table 2 are similarly divided into three categories.

The first category is the characteristics of respondent households. This category contains three variables: Metro, Heating, and Income. The Metro, Heating, and Income variables are a dummy for the interviewee’s living in the Seoul Metropolitan area (0 = no; 1 = yes), a dummy for the interviewee households’ using electricity for heating (0 = no; 1 = yes), and a dummy for the interviewee household’s monthly income being larger than KRW 4.88 million (USD 5.75 thousand) (0 = no; 1 = yes), respectively. The Income variable is defined as a dummy that identifies whether the interviewee household income is greater or less than the average value of the sample. That is, considering that the average monthly household income in the sample is KRW 4.88 million, the Income variable has a value of one if the interviewee household’s income is greater than KRW 4.88 million and zero otherwise.

In the second category, two variables, Education and Age, were used as individual characteristics. The Education and Age variables are dummies for interviewees having more than twelve years’ education (0 = no; 1 = yes) and interviewees’ age in years, respectively. Other personal characteristic variables, such as the gender of the respondents and whether they are household owners, were also considered candidates for reflection, but were eventually excluded from Table 2 as the analysis indicated that they had little effect on acceptance.

Six variables related to respondents’ recognition and judgment were utilized as the third category. Know1, Know2, Environment, Forest, Fsolar, and H2-car are dummies for interviewees knowing about the energy transition policy well before the survey (0 = no; 1 = yes), a dummy for interviewees knowing about hydrogen vehicles well before the survey (0 = no; 1 = yes), interviewees’ subjective judgment on which is more important between jobs and the environment (0 = jobs; 1 = environment), a dummy for interviewees being in favor of the utilization of unused forest biomass (0 = no; 1 = yes), a dummy for interviewees being in favor of expanding floating solar power facilities (0 = no; 1 = yes), and a dummy for interviewees being in favor of expanding hydrogen vehicles (0 = no; 1 = yes), respectively.

3.5. How to Model the Data

Two important points should be taken into account in establishing a model in which acceptance is a dependent variable and the factors affecting this acceptance are independent variables. First, the observed dependent variable has a range of only 1 to 9—that is, the minimum is 1 and the maximum is 9. Therefore, the range is not the whole real number, but a natural number between 1 and 9. Second, the acceptance values are ordinal, not cardinal. Looking at the score, the measure of acceptance, it is clear that the larger the number, the greater the acceptance. However, this value is not cardinal. Performing classical regression without reflecting these two points can produce misleading analysis results.

To simplify the analysis, Ono and Tsunemi’s [56] work transformed the collected data on a four-point scale into a two-point scale for pros and cons, and then applied a discrete logit model to the transformed data. However, adopting this approach results in the loss of important information from data collected on a nine-point scale. Therefore, it is necessary to apply a model that fully utilizes the collected data. The ordered probit model is useful in dealing with data on acceptance evaluated on the Likert scale as in this study [58,59]. The model defines a latent variable distributed over the whole real number instead of an observed variable on a nine-point scale and sets it as a dependent variable. In addition, the likelihood function reflects the relationship between the observed and latent variables.

The ordered probit model used in this research can be formulated as follows. For respondent $i$ ($i=1,\cdots ,I)$, the latent variable variable, $A_i^{*}$, and the observed variable, $A_i$, are:

$$\{\begin{array}{c}{A}_i^{*}={y_i}^{\prime }\beta +{\omega }_i \\ A_i=J \mathrm{if} {\sigma }_{J-2}<A_i^{*}\le {\sigma }_{J-1} \mathrm{for} J=1,\cdots ,9 \end{array}$$

(1)

where $y_i$ is a vector of a constant term and the variables given in Table 2, ${y_i}^{\prime }$ means the transpose matrix of $y_i$, $\beta$ is a vector of the parameters matching $y_i$, ${\omega }_i$ is the disturbance term, and $\sigma$ is a threshold value that is not known and should be estimated.

However, following the usual practice in the literature, ${\sigma }_{-1}=-\infty$, ${\sigma }_0=0$, and ${\sigma }_8=\infty$ are assumed. Furthermore, the disturbance term is assumed to be distributed as normal with a standard deviation of one. Therefore, the probability that the observed variable has one value of 1 to 9 can be induced as:

$$\mathrm{Prob}\left(y_i=J\right)=F\left(\frac{{\sigma }_{J-1}-{y_i}^{\prime }\beta }{1}\right)-F\left(\frac{{\sigma }_{J-2}-{y_i}^{\prime }\beta }{1}\right)$$

(2)

where $F\left(·\right)$ indicates the standard normal cumulative distribution function. The finally derived likelihood function is:

$$L={\displaystyle \prod }_{i=1}^I{\displaystyle \prod }_{J=1}^9{D}_t^s\mathrm{Prob}\left(y_i=J\right)$$

(3)

where $D_t^s={\displaystyle 1}\left(y_i=J\right)$ for $J=1, 2, \cdots , 9$ where ${\displaystyle 1}\left(·\right)$ is a function that returns one if the argument is true and zero otherwise. The maximum likelihood estimates can be obtained by finding the parameter values that maximize Equation (3).

4. Results and Discussion

4.1. Data

The authors sought to focus on three points in conducting field surveys. First, scientific sampling should reflect the characteristics of the population. Second, experienced professional interviewers should obtain reliable responses from respondents. Third, the person-to-person interviews should be carried out maintaining the distancing rules in the pandemic situation caused by COVID-19. To meet these three points, a professional survey company took charge of the entire process of the survey. The survey was conducted for one month from mid-March to mid-April 2021. Judging from the comments of the interviewers belonging to the company, respondents responded to the survey without any difficulties. In particular, people were actively involved in the survey because of the controversy over coal-fired power generation due to issues such as particulate matter and CO₂ emissions.

The results from interviewees selecting one of the nine views are presented in

Table 3. Summary of responses regarding acceptance of replacing coal-fired power plants with natural gas-fired ones.

Responses	Frequency	Percentage (%)
Absolutely disagree	2	0.2
Strongly disagree	16	1.6
Disagree	63	6.3
Slightly disagree	41	4.1
Neutral	142	14.2
Slightly agree	140	14.0
Agree	383	38.3
Strongly agree	153	15.3
Absolutely agree	60	6.0
Totals	1000	100.0

. Four views—“absolutely disagree,” “strongly disagree,” “disagree,” and “slightly disagree”—can be aggregated as “disagree with implementing the fuel transition;” while four views—“slightly agree,” “agree,” “strongly agree,” and “absolutely agree”—can be aggregated as “agree with implementing the fuel transition”. Of the 1000 respondents, 122 opposed the fuel transition and 736 supported the fuel transition, the latter (73.6%) being about six times the former (12.2%). Overall, therefore, the approval rate was higher than the disapproval rate. A total of 142 respondents said “neutral.” It was interesting that 14.2 percent of people were neutral or indifferent to fuel transition.

4.2. Estimation Results of the Model

The estimation results of the ordered probit model are given in

Table 4. Estimation results of the ordered probit model.

Variables a	Coefficient Estimates	t-Values
Constant	2.6779	8.30 *
Metro	0.4915	6.92 *
Heating	0.5862	1.99 *
Income	0.1262	1.86 #
Education	−0.1381	−1.71 #
Age	−0.0085	−2.08 *
Know1	0.1689	2.47 *
Know2	0.1306	1.81 #
Environment	0.2248	3.33 *
Forest	0.2658	3.53 *
Fsolar	0.2096	2.63 *
H2-car	0.1712	2.23 *
${\sigma }_1$	0.7698	3.74 *
${\sigma }_2$	1.4839	6.83 *
${\sigma }_3$	1.7282	7.92 *
${\sigma }_4$	2.3110	10.51 *
${\sigma }_5$	2.7474	12.45 *
${\sigma }_6$	3.9078	17.50 *
${\sigma }_7$	4.7369	20.68 *
Sample size	1000
Log-likelihood	−1671.95
Peusdo-R2	0.165
Log-likelihood ratio test statistic (p-value)	175.62 (0.000)

^a The variables are described in Table 2. * and ^# indicate statistical significance at the 5% and 10% levels, respectively. $\sigma$’s are parameters to be estimated in the model.

. $R^2$, which indicates goodness of fit, was 0.165. In the analysis using cross-sectional data, the value of $R^2$ is usually low [60,61]. In particular, Gans [62] pointed out that it is a kind of norm for $R^2$ to be between 0.1 and 0.2 when analyzing data obtained from the survey. Thus, it can be seen that the value of 0.165 for $R^2$ obtained in this study is not particularly low. A likelihood ratio test can be applied for the specification test of the model. In this case, the null hypothesis is that all estimated coefficients except the constant term are zero—that is, the model is mis-specified. The computed likelihood ratio test statistic was 175.62, which corresponds to a p-value of 0.000. Thus the statistical significance of this model, determined at a significance level of 1 percent, is ascertained.

All of the $\sigma$ values used to define the observed variable, $A_i$, were estimated to be positive, having statistical significance at the 1 percent level. Moreover, the estimation results of the coefficients corresponding to all the variables defined in Table 4 had statistical significance at the 10 percent level. Thus, the estimation results of the ordered probit model are significant. Interestingly, the model provides reasonably good performance. The application of the ordered probit model in this study was an appropriate strategy.

The estimated coefficients for the Metro, Heating, Income, Know1, Know2, Environment, Forest, Fsolar, and H2-car variables had a positive sign. The positive sign of each estimated coefficient implies that the greater the value of each variable, the greater the acceptance of the fuel transition. For example, those who lived in the Seoul Metropolitan area, those who used electricity for heating, those whose household income was high, those who knew about the energy transition policy before the survey, those who knew about hydrogen vehicles before the survey, those who considered the environment more important than jobs, those who were in favor of utilizing unused forest biomass, those who were in favor of expanding floating solar power facilities, and those who were in favor of expanding hydrogen vehicles were more prone to the fuel transition than others.

On the other hand, the sign of the estimated coefficients for the Education and Age variables was negative. A negative sign indicates that the greater the value of each variable, the lower the acceptance of the fuel transition. Respondents with more than twelve years’ education and those who were older than 48 years were less receptive to the fuel transition than others.

4.3. Discussion of the Results

After data on acceptance of the fuel transition were collected on a nine-point scale, the determinants of acceptance were identified and analyzed. The results derived from this work can contribute to the literature, having various implications in terms of both research and policy. First of all, the usefulness of applying an ordered probit model to ordinal data collected on a nine-point scale was ascertained. The specification test confirmed the statistical significance of the model, and both the threshold values appearing in the model and the estimated coefficients for the eleven variables of interest were statistically significant. Thus, it is possible to derive various implications from the results.

The fact that the approval rate for the fuel transition (73.6%) exceeded the disapproval rate (12.2%) by about six times was a positive finding in promoting the fuel transition in South Korea. In order to achieve the goal of reducing CO₂ emissions declared to the international community, the country should carry out the fuel transition continuously and robustly while maintaining a stable power supply [13,14,28]. This finding can be recognized as an encouraging sign for the country. Without public support, the fuel transition cannot succeed. It is also worth noting that 14.2 percent of the respondents considered themselves neutral or indifferent.

Interestingly, three interviewee household characteristic variables had a significant impact on acceptance. First, those living in the Seoul Metropolitan area were more receptive to the fuel transition than those who were not. The Seoul Metropolitan area is home to about half of the population, an important area that determines public opinion in South Korea. Therefore, it is quite difficult to implement policies that residents in the area oppose. The finding that the acceptance of residents in the area of the fuel transition is secure is quite encouraging in promoting the fuel transition.

Second, those who use electricity for heating were more receptive than those who do not. The main fuel for residential heating in South Korea is city NG since electricity for heating is more expensive than city NG. Nevertheless, some people use electricity instead of city NG for heating because they prefer electricity to other fuels. This is because electricity is partially made from RE, and even if it is produced using fossil fuels, fossil fuel-fired power plants greatly reduce particulate matter through air pollutant reduction facilities, while city NG boilers are not equipped with these facilities. In other words, since people who use electricity for heating have a high interest in the environment, they seem to make more positive judgments on the fuel transition.

Third, the household income of the respondent was positively correlated with acceptance. The transition of power generation fuel from coal to NG will inevitably lead to higher electricity bills. From an individual household’s point of view, income is limited, and an increase in electricity costs means spending on other goods should be reduced. In fact, the high-income group is more likely to accept an increase in electricity bills than others. In particular, low-income people may be opposed to an increase in electricity bills caused by the fuel transition rather than to the fuel transition itself. Therefore, even if the fuel transition is made, measures will need to be in place to alleviate the burden by continuously applying the electricity rate discount system for low-income people.

Moreover, two individual characteristics of interviewees had a significant impact on acceptance. The education level of the respondent had a negative impact on acceptance of the fuel transition, while older interviewees were less receptive to the fuel transition than younger interviewees. In fact, in South Korea, the higher the level of education and the older the person is, the more they tend to settle for the present situation. This is because fuel transition can not only increase electricity bills, but also reduce jobs and cause problems in the stability of electricity generation fuel supply. To help increase the acceptance of fuel transition among people with a high education level or older age, the fuel transition should be promoted in more persuasive and appealing ways.

Three issues concerning the fuel transition were addressed in the introduction: rising electricity bills due to increased power generation costs, reduced fuel supply stability due to increased NG use, and job losses [20,21,22,23]. Interviewees responded after hearing full explanations of the issues during the survey, and it was found that the approval rate for the fuel transition was six times higher than the opposition rate. However, if these three issues become a reality, the public acceptance of the fuel transition may drop significantly, which could place South Korea in a difficult situation. It may happen that coal-fired power generation facilities are reduced but NG-fired power facilities are not increased in time, which could give rise to a crisis in the supply and demand of electricity. Therefore, the three issues need to be discussed here in conjunction with qualitative implications obtained from respondents in the process of the survey.

First, since the unit cost of NG-fired power generation is higher than that of coal-fired power generation, the fuel transition will bring about an increase in electricity bills, which could negatively affect acceptance of the fuel transition. As of March 2021, South Korea’s unit cost of NG-fired power generation was 99.72 KRW per kWh, about 10 percent higher than that of coal-fired power generation, which is 90.19 KRW per kWh. However, because NG prices are linked to oil prices, rising international oil prices could have a significant impact on NG prices, which could widen the gap. Usually, NG prices soar during periods of high oil prices and plunge or stabilize during periods of low oil prices. Recently, as oil prices have been rising, voices opposing the fuel transition have already begun to appear.

In addition, since NG will serve as a bridge in an era of energy transition, NG prices are likely to rise further in the future as demand for NG increases around the world. As coal-fired power generation decreases and NG-fired power generation continues to increase, the cost burden for electricity distribution companies is expected to increase. This could eventually lead to higher electricity bills, which could place a burden on consumers. Therefore, it is necessary to persuade the public by fully informing them that fuel transition is inevitable and that they must endure an increase in electricity bills to implement the fuel transition. Fuel transition will gain momentum only when there is a social consensus that electricity produced from NG-fired power plants is inevitably more expensive than that from coal-fired ones, just as organic products are more expensive than regular products.

Second, given the geopolitical situation of South Korea, NG is more vulnerable to supply instability than coal. Almost all of the NG consumed in the country is imported from abroad. In particular, the country relies heavily on the Middle East, including Qatar and Oman, as NG suppliers. Therefore, if political instability occurs in the Middle East, South Korea’s NG-fired power plants may have to be shut down. In addition, liquefied NG produced in the Middle East is transported to South Korea, and the country’s NG procurement costs are the highest in the world due to excessive liquefaction, transportation, and vaporization costs.

On the other hand, as coal is imported from all over the world without the need for liquefaction, it has higher supply stability than NG and its procurement cost is also low. Thus, it is quite important to secure a stable supply of NG in order for fuel transition to succeed. In Europe, NG is supplied stably mainly in the form of pipeline NG, but South Korea is introducing NG through liquefied NG carriers only. Currently, efforts are being made to import NG produced from other regions besides the Middle East. For example, South Korea is trying to increase NG imports from Australia, the United States, Indonesia, and Mozambique. These efforts must yield significant results. If South Korea fails to secure stability in NG supply, it will be difficult for the fuel transition to succeed and receive public support.

Third, if jobs are lost due to the fuel transition, this can lead to serious social conflict. Currently, a 500 MW coal-fired power plant in South Korea has a total of 168 to 200 employees in the operating and maintenance sectors, while the number of workers in an NG-fired power plant of the same capacity is between 90 and 110. In other words, the latter is about half the former. Consequently, converting all 24 (12.6 GW) coal-fired power plants to NG-fired power plants could result in the loss of between 2000 and 2400 jobs. Since all 24 coal-fired power plant operators are public companies, not private ones, responding to the fuel transition by reducing working hours and increasing the retraining of staff may not actually reduce jobs; however, this approach will be neither long-term nor stable.

The decline in the number of jobs will not only increase unemployment, making workers’ lives difficult, but also cause considerable damage to the local economies where power plants are located. The 24 coal-fired power plants are located on the shores of rural areas, not urban areas, because they are a form of “not in my backyard” facility and need to use seawater as cooling water. Since most of them are located in underdeveloped areas, they contribute greatly to the local economy through local taxes, donations, public utility expenditures, and procurement of resources within the region. However, because it is advantageous for NG-fired power plants to be located in urban areas or industrial complexes adjacent to the demand for electricity, they are likely to change their location during the transition. Consequently, the areas where coal-fired power plants used to be located could face economic difficulties due to the fuel transition.

Due to these problems, during a recent visit to Chungnam Province, where the largest number of coal-fired power plants are located, President Moon Jae-In stressed that “The energy transition, which replaces existing coal-fired power plants with NG-fired ones, should be done in a just way so that no-one loses a job and the local economy is not damaged.” Therefore, the just and fair transition of fuel is an important issue that South Korea will face in the future. There should be in-depth consideration of this and immediate preparation of concrete measures to implement this in practice. The measures currently proposed include the following.

• Send workers on master’s and doctorate courses in graduate schools for long-term re-training;
• Deploy minimum management personnel for coal-fired power plants that are not normally operated but are used as reserve power resources in case of an emergency in which demand increases;
• Relocate existing employees working at coal-fired power plants to new business sectors, such as RE plants and fuel cell generation plants, after sufficient re-education;
• Subsidize living expenses and education expenses for many years so that a worker can transfer to a completely different field.

5. Conclusions

In Introduction section, the authors looked at previous studies dealing with the transition of power generation fuel from coal to NG. An important implication was that this transition is inevitable to reduce CO₂ and air pollutant emissions, and is a strategy adopted by a number of countries. South Korea is no exception. The previous policy to expand coal-fired power generation has been changed to a reduction policy since May 2017. It was clearly pointed out that the use of NG will be expanded as an intermediate stage because it is difficult to immediately engage in a significant expansion of RE. It was also mentioned that the speed of fuel transition will be faster for the implementation of carbon neutrality by 2050. In the end, concerning coal-fired power generation, the current status and prospects of South Korea are consistent with those of other countries presented in several previous studies.

In Section 2, previous studies that analyzed energy-related public acceptance were examined. It was found that, to the best of the authors’ knowledge, this study was the first to analyze the public acceptance of converting power generation fuel from coal to NG. In particular, it is an important discovery, differentiated from previous research, that the public acceptance of the fuel transition policy is secured to some extent and that many people support the transition even though there is an increase in cost. In addition, this study not only identified various factors affecting the public acceptance of the fuel transition, but also derived various implications by proposing and applying a framework for analyzing the impacts of the factors on the acceptance. The results of this study could be important information for South Korean policymakers. Although the main findings of this study are unique to South Korea, the structure of this study can be extended to other countries as much as possible. The main qualitative findings of this study are not so different from those of the research that has analyzed the public acceptance of RE expansion policy [63], large-scale offshore wind power generation construction [40], and energy transition policy [4] in South Korea.

South Korea’s CO₂ emissions in 2018 totaled about 728 million tons. However, the 2030 CO₂ emissions target submitted to the United Nations Framework Convention on Climate Change is 536 million tons. Ultimately, South Korea needs to reduce its CO₂ emissions drastically, and the power generation sector should play an important role in this as it is very difficult to reduce CO₂ emissions in other sectors such as transportation, industry, building, and agriculture. To that end, South Korea has set out a plan to convert 24 coal-fired power plants into NG-fired power plants by 2034. This study sought to ascertain and analyze public acceptance of this through a survey of 1000 people nationwide. The results were statistically significant and reveal a number of useful implications.

The finding that the approval rate for the fuel transition was six times the opposition rate suggests that the government-led fuel transition should be carried out continuously. However, since respondents were concerned about three challenges that could arise in the process of fuel transition, the authors have attempted to discuss them above. These were higher electricity bills due to rising power generation costs, reduced fuel supply stability due to the increased use of NG, and job losses and a consequent negative impact on the local economy. Ultimately, dealing with these three challenges effectively will determine whether or not the fuel transition succeeds.

Of course, further challenges remain to be addressed. For example, is it socially desirable to tear down a coal-fired power plant that has only been in operation for 30 years? Given that NG-fired power plants are also fossil-fuel-utilizing facilities and there is strong opposition from residents to new construction of these, how will we facilitate fuel transition? It is clear that NG is the bridge energy to a complete RE society, but for how long will it play that role? In other words, if NG-fired power plants become stranded assets in a not-too-distant future in which carbon neutrality is realized, is it reasonable to invest in constructing NG-fired power plants now? Subsequent studies should be able to answer these questions.