Ⅰ. Introduction

Currently, the global climate is deteriorating, posing a significant challenge to sustainable development worldwide. Reducing greenhouse gas emissions is fundamental way to solve this problem. In recent years, the Chinese government has actively formulated relevant energy conservation and emission reduction plans to reduce CO₂ emissions. For example, National Energy Administration announced that the government will strengthen the development and utilization of clean coal and speed up the construction of natural gas storage systems. The National Energy Administration has announced plans to strengthen the development and utilization of clean coal, promote oil and gas exploration and development, and speed up the construction of natural gas storage systems. China is also accelerating the development and utilization of wind, solar, biomass, and other renewable energy sources. The Central Committee of the Fifth Plenary Session of the 19th Communist Party of China (CPC, 2020) had put forward a Five-Year plan for national economic and social development to 2035, which aims to achieve a green way of production and life, stabilize carbon emissions, improve the ecological environment and achieve the goal of a beautiful China.

Additionally, the Chinese government has been promoting green vehicles to reduce air pollution. Most previous studies have focused on

HEVs (hybrid-electric vehicles), PHEVs (plug-in hybrid-electric vehicles) and BEVs(battery－electric vehicles) while only a few studies investigated preferences for the HFCVs. Thus, this paper contributes to the literature by investigating the consumers’ preference for HFCVs in Shanghai using a choice experiment (CE).

We consider purchase price, distance to the hydrogen refuelling stations, driving range, government regulation for CO₂ emissions in the transportation sector, and green vs. grey hydrogen as main attributes affecting consumers’ purchase of HFCVs. A conditional logit model (CLM) was used to analyze consumers’ preferences for different types of HFCVs. This study also explores whether the observable heterogeneity of individuals such as education, income, age, and gender affects people’s preference for HFCVs.

The next section discusses previous studies on consumer preferences for green vehicles. The third section introduces the theoretical aspects of the choice experiment and the estimation models, such as the CLM. The fourth section consists of the questionnaire design, and the descriptive statistics of the collected data. The fifth section includes parameter and WTP estimation results. The last section represents the conclusion with policy implications.

Ⅱ. Literature review

The growing concerns over GHG emissions have led to increasing attention from policy makers and scholars toward alternative fuel vehicles. Table 1 provides a summary of the literature that examined consumers’ preferences for green vehicles using the CE approach. While there is a number of studies related to WTP for green vehicles, most studies focused on developed countries, and there is an insufficient number of studies on China. The CE approach has identified major attributes related to vehicle characteristics that affect consumers’ WTP for green vehicles. The most common attributes include charging/refueling time and driving range (Li et al., 2020; Hackbarth and Madlener, 2016; Noel et al., 2019), some studies used emission reduction of alternative vehicles (Li et al., 2020; Potoglou and Kanaroglou, 2007; Ito et al., 2013; Bae and Jung, 2013), acceleration (Noel et al., 2019; Potoglou and Kanaroglou, 2007), vehicle size (Hackbarth and Madlener, 2016; Ziegler, 2012) and other vehicles’ features.

In addition, socio-demographic characteristics play an essential role including age and education (Hackbarth and Madlener, 2016; Potoglou and Kanaroglou, 2007; Li et al., 2020; Ziegler, 2012), gender (Hackbarth and Madlener, 2016; Potoglou and Kanaroglou, 2007; Ziegler, 2012), income (Hackbarth and Madlener, 2016; Potoglou and Kanaroglou, 2007; Li et al., 2020).

Another important factors that can affect consumers’ WTP for HFCVs are their awareness and knowledge of the technology. For example, Sierzchula et al. (2014) found that consumers who were more familiar with HFCVs had a higher WTP compared to those who were less knowledgeable about the technology. Similarly, a study by He et al. (2021) found that consumers who had a positive attitude towards HFCVs and perceived them as environmentally friendly had higher WTP premiums. The availability and accessibility of hydrogen refueling stations are other important factor that influences consumers’ WTP for HFCVs. A study by Ha et al. (2016) and Daziano et al. (2018) found that consumers were less willing to pay a premium for HFCVs if they had to travel a long distance to access hydrogen refueling stations.

Government incentives and policies can also influence consumers’ WTP for HFCVs. Several studies have found that consumers have higher WTP if government provides subsidies or tax credits for HFCVs purchases (Kim et al., 2019; Sierzchula et al., 2014).

Finally, this study is unique as it includes two additional attributes that have not been previously examined in the literature: government regulation on CO2 emission in China and sources of making hydrogen (green vs. grey).

Ⅲ. Estimation strategy

Since the HFCV market in China is not yet mature, non-market valuation methods such as the stated Preference Method (SP) should be used. The SP can be divided into two approaches contingent valuation (CV) and choice experiment (CE). Both methods involve a questionnaire survey to derive consumers’ preferences and WTPs for the non-market value of goods and services. However, we chose the CE method over the CV method because it includes alternatives in the choice sets, which provides more realistic assumptions for respondents to choose the most preferred option.

The CE method is based on Lancaster’s (1966) consumer choice model and random utility theory. In this method, consumer utility is determined by the attributes of goods, not the value of goods themselves. Respondents are asked to choose one of several options based on their characteristics and the level of commodity attributes. According to the random utility theory, respondents will choose “the best combination of choice sets” to maximize their utility, transforming the selection problem into a utility comparison problem.

The random utility model (RUM) can be expressed as:

where χnis and μnis represent the observable and unobservable factors in a choice set. The utility of each individual n choosing option i in the choice set s is derived from the vector of attributes (χnis), and the parameter vector (𝛽）represents the relative preference weight of the utility, as shown in the equation. According to the studies such as Batt and Katz (1997), the same as Hensher (1994) and Hanley et al. (1998), and Slothuus et al. (2002).

In choice situation s, a consumer n chooses an alternative i if his utility (Uni) from choosing i exceeds the utility (Unj) from choosing an alternative j. The choice probability can then be converted into a conditional logit model (McFadden, 1974). The observable utilityVni is affected by both individual-specific variables (Xni) and alternative-specific attributes (Yi).

Once the attribute parameters have been estimated using the CLM, the formula for a consumer’s willingness to pay (WTP) for each attribute can be expressed as follows:

where MUa , and MUc are marginal utility from the non-monetary and monetary attributes respectively. βa and βc are parameter estimates for a non-monetary and monetary attributes respectively.

Ⅳ. Design of the Choice Experiment

A survey was conducted by the Chinese research company (https://www.wjx.cn/) with a sample size of 300 Shanghai residents, between September 5th and 25th, 2022. Respondents were selected based on gender, age, education, and income level to match the average profile of Shanghai residents. The choice sets considered small private cars only. The SPSS software was used to generate 8 eight choice sets, which were divided into two groups, each group contains four choice sets. Within each choice set, respondents were asked to choose their preferred alternative among the presented options.

Each choice set’s alternative consisted of five attributes: the purchase price of the HFCV in RMB, the driving range of the HFCV, the distance to the hydrogen station, the government plans for CO₂ emissions, and the source of the hydrogen. Shanghai was chosen for the survey location due to its relatively fast developing economy and more stringent promotion of HFCVs compared to other provinces.

The price levels for HFCVs were based on statistics from the official website of the domestic green vehicle market (https://chejiahao.m.autohome.com), with levels set at $50,000/60,000/70,000. The usual driving range of HFCVs in China is between 500 and 700 km, so the levels for the driving range were set at 500, 600, and 700 km. The hydrogen sources, were catigorized as either gray hydrogen, made from coal or natural gas, or green hydrogen, made from renewable energy sources, based on the Catalogue of Industries to Encourage Foreign Investment (National Development and Reform Commission, Ministry of Commerce, 2022). The Chinese government’s regulation for the CO₂ emission level was set at 150g and 176g per km, in accordance with China’s Automobile Low-carbon Action Plan (China Automotive Technology Research Center, 2021).

Each choice set consists of three options (A, B, and C). Options A and B differ based on the levels of the attributes, while Option C serves as an opt-out option, indicating that the respondent rejected the idea of purchasing an HFCV.

Table 4 shows that 97% of the survey participants are under the age of 40, which is similar to the population structure of Shanghai. Of the respondents, a higher proportion (55.7%) are male, which is consistent with the gender structure of Shanghai in 2022 (where males make up 51.8% and females make up 48.2%). The education level of the respondents is concentrated in the first two groups, which are high school or below (30.3%) and junior college or Bachelor’s degree (58.7%). In terms of income, the respondents are mostly concentrated in the range of $900 or below (58%) and $900 - $2221 (35%). It is worth noting that the average income of Shanghai’s population in 2022 was around ￥9580 ($1400). <Table 5> presents the descriptive statistics of the main socio-demographic variables.

Ⅴ. Results

The estimation results of two models, CLM1 and CLM2, are presented in <Table 6>. CLM1 model includes only the attributes of the HFCV choice sets and the alternative specific constant (ASC), while the CLM2 model adds interaction terms between ASC and selective socio-demographic variables such as sex, age, education, and income. the ASC is defined as equal to one if respondents chose option A or B, and zero if they chose the opt-out option. Therefore, the coefficient for the ASC represents the relative preference for choosing the HFCV.

In both models (CLM1 and CLM2), the coefficient of the ASC is significant and positive, indicating that purchasing the HFCV increases consumers’ utility. The estimated parameters of the price attribute in both models are also statistically significant and have expected negative signs, as an increase in the sales price of the HFCV decreases consumers’ utility. The coefficients for the driving range are positive and statistically significant at a 1% level, indicating that an improvement in the driving range of the HFCV increases consumers’ utility when choosing the HFCV. In contrast, the estimation results show that an increase in distance to a hydrogen station reduces people’s utility of choosing the HFCV, as the coefficient of the distance to the hydrogen station attribute is negative and statistically significant. The coefficient of the government regulation on CO₂ emission is significant and negative, which indicating that more stringent regulation on CO₂ emission encourages people to choose HFCV. Furthemore, the coefficient for the sources of producing hydrogen is significant and positive, implying that Shanghai citizens prefer the green hydrogen to grey hydrogen.

According to the CLM2 model with the interaction terms of income and ASC as well as age and ASC, younger and high-income respondents are more likely to choose the HFCV.

The results presented in <Table 7> show the AIC (Akaike information criterion) and BIC (Bayesian information criterion) values for the two models (CLM1 and CLM2), which are commonly used to select the best model. According to the AIC value, CLM2 model is preferred over CLM1 as it has a lower AIC value. However, based on the BIC value, the CLM1 model is preferred. Since the results are inconclusive, we chose to use the CLM2 model as it contains more parameters, including interaction terms between attributes and socio-demographic variables, which could improve the model’s accuracy.

<Table 8> shows the mean, lower limit, and upper limit for the coefficient of the ASC, which represents people’s WTP for HFCV options in the choice sets.

The current market prices for most HFCVs in China range from $60,000 to $149,617, which is higher than the estimated mean WTP for the HFCV based on the CE ($56,555). Therefore, a government subsidy policy should be maintained to bridge the gap between the WTP and the selling price of HFCVs in China and encourage people to purchase them.

Ⅵ. Policy Implication and conclusion

According to the CE analysis on Shanghai citizens’ preferences for HFCVs, it is evident that consumers are more likely to choose an HFCV with a longer driving range, shorter distance to the hydrogen station and cheaper price. Moreover, citizens prefer to choose a HFCV when its hydrogen is produced from renewable energy sources, and they are more likely to choose the HFCV under more stringent CO2 emission regulations on transportation. The observable heterogeneity analysis reveals that younger and high-income individuals are more likely to choose the HFCV.

To encourage HFCV adoption, the central as well as local government subsidy policy should be maintained steadily. Moreover, to facilitate HFCV expansion, the government should support the installation of more hydrogen stations and promote green hydrogen production. It is also important to note that the current subsidy levels are sufficient to incentivize consumers to purchase HFCVs. It is crucial to note that the Shanghai MAXUS EUNIQ7 costs between $50,000 and $59,800 after receiving a total subsidy of 400,000 yuan ($57,200) from the Shanghai and national governments. In contrast, foreign brands like Toyota Mirai receive only a subsidy of 100,000 yuan ($14,300). Therefore, the Chinese government should consider subsidizing foreign HFCV brands more to make them more accessible to consumers.

The CE survey summary indicates that the average monthly income is around $1,300, and the mean WTP for HFCVs is $56,555. Therefore, the WTP is approximately three times higher than the average annual income of the respondents. However, with an appropriate borrowing system or payment in instalments within a five-year period, consumers can afford to purchase an HFCV with the current subsidy. The Chinese government should promote a green loan program to help people buy HFCVs.

Our study contributes to the current literature on WTP for green vehicles by examining HFCVs, while most of the other studies focused on HEVs, PHEVs, and BEVs. We are also cognizant of the limitations of our study. A major drawback is related with attributes that were included in the choice sets. For instance, respondents’ awareness of possible benefits and risks of using HFVC may also have an impact on WTP for HFVCs including environmental damage of fossil-fuel cars, the risk of hydrogen explosions etc. Thus, as a future research we will extend this study in terms of additional variables and various attributes.