However, in Macau, only a total of 84 EMs are in use (DSEC, 2016; Songet al., 2018). Currently, the government is still lack of correspondingpolicy and guidelines for EM promotion.It is a remarkable fact that most researches mainly focused on theEM's technology and emission reduction. However, there is still a lackof the researches on consumers’ attitudes and their willingness to buy(WTB) and willingness to pay (WTP). To promote the use of EM, re-levant policies’ establishment and implementation require a clear un-derstanding on the consumers’ attitudes and their WTB and WTP forEM. Currently, there are no studies related to the demand pattern of EMfrom the consumers’ perspective in Macau. It is important to analyzedemand characteristics and consumers’ preferences according to thedifferences among vehicle markets in various countries or regions(Roche et al., 2010; Lai et al., 2015).In recent years, contingent valuation method (CVM) has becomeone of the most widely used valuation techniques because of its flex-ibility and its ability to estimate total values. The CVM is a type ofstated-preference approach that employs a hypothetical market systemto extract WTP or willingness to accept for environmental goods(Carson, 2000; Song et al., 2016; Hadker et al., 1997). Song et al.(2012) used CVM to analyze the relationship between residents' beha-viors, attitudes, and their WTP for recycling e-waste. Jin et al. (2006)examined residents' preferences for alternative solid waste managementpolicy changes in Macau using the double-bounded dichotomous choiceCVM method. Some researchers have also begun to focus on the WTP ofelectric vehicles and their attributes. Wiser (2007) used a CVM surveyof 1574 residents to explore WTP for renewable energy under collectiveand voluntary payment vehicles in America. Potoglou and Kanaroglou(2007) examined the factors and incentives that are most likely to in-fluence households’ choice for cleaner vehicles in the metropolitan areaof Hamilton, Canada. Hidrue et al. (2011) estimated the WTP for fiveelectric vehicle attributes: driving range, charging time, fuel costsaving, pollution reduction, and performance. Zhang et al. (2011)analyzed consumers’ awareness towards electric vehicle (EV) and ex-amined the factors that are most likely to affect consumers’ choice forEV in China. Bjerkan et al. (2016) described the role of incentives forpromoting battery electric vehicle (BEV), and determined what in-centives are critical for deciding to buy a BEV in Norway. Lin and Tan(2017) investigated the environmental values of BEV based on a CVMsurvey conducted in the four biggest cities of Beijing, Shanghai,Guangzhou, and Shenzhen in China.Currently, EM has become an important branch of new energy ve-hicles. In order to promote the wide use of EM, this study aims to in-vestigate the influencing factors of consumers’ WTB and WTP for EM inMacau through the CVM survey. It is worth pointing out that the ex-ploration of the influencing factors or incentives that could encouragefamilies to purchase EM is very important. The results obtained in thestudy will help policy makers and managers to better understand thepurchase behaviors and preferences of consumers, and can providesome effective supports for policies designed to promote EM in Macau.2. Methods2.1. Questionnaire designThe respondents of the questionnaire survey are the motorcycledrivers in Macau, considering that the drivers are the most potentialconsumers. The questionnaire survey can be divided into the followingfive parts. The first part is designed to understand the respondents’perceptions on the overall environmental status in Macau. The secondpart is used to investigate the respondents’ knowledge of the char-acteristics of EM. The third part is mainly to identify the factors thatlead to the consumer's decision to buy EM. The fourth part is to analyzethe respondents’ WTB and WTP for EM. The fifth part is to collect therespondents’ personal information for understanding whether the per-sonal characteristics of respondents will affect their purchase decisionsabout EM.In this study, we select the Likert Scale to accomplish the first threeparts of the questionnaire survey. The five-point scale can exactly showall the contents of the survey questions, and ensure the respondents caneffectively distinguish and express their opinions.2.2. Data collectionPrior to the formal questionnaire survey, this study conducted pre-surveys and simulation assessments. In the pre-survey, after completingthe questionnaire, the respondents were also asked what other influ-ence factors or questions will be considered when making their con-sumption decisions, which helps us improve the questionnaire contentsand ensures that the form and content of the questionnaire are rea-sonable and acceptable. The pilot test was conducted in November2014 to ensure the validity and reliability of the questionnaire. A totalof 16 questionnaires were distributed and 15 valid questionnaires wereobtained. Based on the feedback from the pilot test, the content of thequestionnaire has been modified. The formal questionnaire survey wasconducted from March 10, 2015 to April 30, 2015. As shown in Table 1,a total of 450 questionnaires were distributed to motorcycle drivers inthe seven administrative districts of Macau. For each district, simplerandom sampling was conducted and the sampling size was propor-tional to the district's relative frequency in the population. Notably, 37questionnaires were invalid because of the lack of critical informationor the inconsistency in survey results. Finally, a total of 413 validquestionnaires were screened in this survey, with an effective rate of92%. The high effective rate of the questionnaire survey is due to theuse of face-to-face interview which makes it easier for the respondent toeither clarify answers or ask for clarification for some of the items onthe questionnaire.2.3. Contingent valuation method (CVM)This study applied the contingent valuation method (CVM) toquantify respondents’ WTB and WTP for EMs in Macau. This conceptrepresents the amount people would be willing to pay to avoid a spe-cified environmental damage, achieve a stated improvement in en-vironmental quality, or receive a specified supply of a public good.Existing methods for estimating WTP include Continuous CVM (re-presented by open-ended questions) and Discrete CVM (represented byDichotomous Choice questions) (Zhen et al., 2011). In ContinuousCVM, the respondents are free to answer the open-ended questions byfilling out the maximum amount they are willing to pay, which is easyfor data analysis. In Discrete CVM, on the other hand, the respondentsare required to show their willingness only by choosing either “Yes” or“No”, and they don’t need to indicate the specific amount they are ableto pay, thereby avoiding the problem of inconsistency between thestated WTP and actual amount the respondent is willing or able to pay(Hoehn and Randall, 1987).To obtain the necessary data for both WTP and the associated spe-cific amount to pay, the present study adopted the double-boundedDichotomous CVM method, because of its advantages of effectivelyTable 1Sampling in the seven administrative districts of Macau.Administrative districts Sample size Valid sampleNossa Senhora de Fátima 171 155Santo António 86 78São Lázaro 27 27Sé 28 22São Lourenço 33 31Taipa 83 80Coloane 22 20Total 450 413L. Zhu et al. Energy Policy 127 (2019) 64–7265
simulating the market pricing behaviors of consumers (Loomis andWalsh, 1997). This assessment method has the advantages of highstatistical efficiency, high real market simulation, incentives for re-spondents to speak the truth, reduction of protest responses, and in-crease of questionnaire efficiency (Hanemann, Loomis and Kanninen,1991; Hoehn and Randall, 1987). The double-boundary DichotomousCVM method does not directly obtain the respondent's specific WTPfrom the results of the survey inquiry. Firstly, it needs to obtain therespondent's “willingness” or “unwillingness” feedback on a certain bidvalue; then it can establish a functional relationship between theprobability of the respondent's response to the item and the amount ofthe bid value; finally, we can get the respondent's specific willingness topay (Hanemann, 1999). According to the double-bounded answer, theconsumer's WTB and WTP for EM can be estimated. Fig. 1 presents theflow chart of the double-bounded questionnaire. In the questionnaire,the respondents are asked two questions for their WTB and WTP for EM.The first question is “Are you willing to buy EM in Macau?”. The secondquestion is “Are you willing to pay more money for buying EM inMacau?”. If the respondents say that they want to pay more for buyingEM, then they will be asked whether they will pay more 1000 MOP forEM. If yes, the respondents will be asked a higher price until 2000 MOP;if not, a lower price (500 MOP) will be asked. In addition, for the re-spondents with no WTP, the possible reasons will be also investigatedand analyzed in the study.For the double-boundary question, there are four different answerresults (shown in Table 2), and their probability can be expressed as:= −+ + +∑expPi(YY) 111 α βBID γ x( )U k k k (2.1)=+−++ +∑ + +∑Piexp exp(YN)1111α βBID γ x α βBID γ x( ) ( )U k k k I k k k (2.2)=+−++ +∑ + +∑Piexp exp(NY)1111α βBID γ x α βBID γ x( ) ( )I k k k L k k k (2.3)=+ + +∑expPi(NN)11 α βBID γ x( )L k k k (2.4)where Pi(YY), Pi(YN), Pi(NY), and Pi(NN) show the probabilities ofanswer “Willing-Willing”, “Willing-Unwilling”, “Unwilling-Willing”,and “Unwilling-Unwilling”, respectively; BIDl is the initial value (1000MOP); BIDU is the higher bid value; BIDL is the lower bid value; α, β, γkare parameters to be determined; xk are the respondents’ socio-eco-nomic characteristics.The log-likelihood function can be established as:∑= + + +=L I P YY I P YN I P NY I P NNln [ ( ) ( ) ( ) ( )]DBiNYY i YN i NY i NN i1(2.5)where IYY, IYN, INY, and INN are binary-value indicator variables, whichare 1 if the argument is true, and 0 otherwise (see Table 2). Usingmaximum likelihood method, α, β and γk can be estimated and theconsumers’ WTP value for EM can be determined by=+ ∑−α γ xβWTP k k(2.6)2.4. Binary logistic regression analysisIn this study, the binary logistic regression model was employed toinvestigate how respondents' personal characteristics and their knowl-edge of EM affect the WTB and WTP decisions for EM. The binary lo-gistic regression model can be set as follows (Zhang et al., 2011):= + + +⋯+ + + + +⋯+β β x β x β x β β x β x β xP exp( )/(1 exp( ))n n n n0 1 1 2 2 0 1 1 2 2(2.7)where xi are independent variables referring to respondents’ socio-economic characteristics (e.g. gender, age, income, and education level,etc.), knowledges, and the considerations of buying EM (e.g. chargingfee, environmental benefits and safety); β0 is a constant term; βi areregression coefficients to be determined; P is dichotomous dependentvariable which represents respondents' WTP or WTB (if willing, thevalue will be 1; if not, the value will be 0). By taking WTB or WTP as thedependent variable, the study developed two binary logistic regressionmodels to predict the probabilities of consumers’ WTB and WTP for EM.General form of the logistic regression equation can be written as:⎟−⎞⎠= + + +⋯+PPβ β x β x β xlog (1 nn0 1 1 2 2(2.8)3. Results and discussions3.1. Respondents' socio-economic characteristics and perceptions on localenvironmentTable 3 presents the descriptive statistical results regarding re-spondents’ age, gender, living time in Macau, monthly income,If yes, we will ask a higherprices (2000MOP) WTP?If not, we will ask a lower prices (500MOP) WTP?Why you are not WTPmore money to buy EM?Are you WTP 1000 MOP more for EM?Q1: Are you WTB for EM in Macau?Yes, I am No, I am notQ2: Are you WTP more money to buy EM in Macau?No, I am notYes, I amSix possiblereasonsFig. 1. Flow chart of the double-bounded questionnaire.Table 2Double-boundary numerical assignment.Options Pi IYY IYN INY INNWilling-Willing P(YY) 1 0 0 0Willing-Unwilling P(YN) 0 1 0 0Unwilling-Willing P(NY) 0 0 1 0Unwilling-Unwilling P(NN) 0 0 0 1L. Zhu et al. Energy Policy 127 (2019) 64–7266
education level, number of family members, and number of motorcyclesowned. As the table indicates, it can be known that the younger drivers(18–40 years old) prefer to accept the questionnaire survey, and themale respondents have also more intentions to answer our questions.About 74.8% of respondents had been settled for more than ten years inMacau. The average education level of respondents was between highschool and college, closer to college level. Specifically, 59.1% of re-spondents finished college education. Most of the respondents’ familiesin Macau had more than one motorcycle according to our survey.The respondents' perceptions on local environmental issues inMacau are shown in Table 4. It is observed that more than half of re-spondents thought that air pollution and noise pollution in Macau wereserious or very serious, and 45.5% of respondents believed that thelarge use of motorcycles will result in serious environmental pollutionissues. The respondents’ behaviors for driving motorcycle are presentedin Fig. 2. It is indicated that 12.1% of respondents considered themotorcycle as the most important travel way and 20% of respondentsdrove motorcycle often in their daily life. In addition, more than 55% ofrespondents were willing to give up or reduce the use of motorcycles forenvironmental protection.3.2. Respondents' knowledge of EMsTable 5 reports the respondents' knowledge of EM with respect toenvironmental benefits, low charging cost, high price, poor batterylifetime, immature technology, high safety, and low maintenance andrepair costs. Noted that most of the respondents expressed theiragreement for the environmental benefits of EM. However, on thewhole, the results indicate that the respondents had relatively less un-derstanding and knowledge of EM. Only 46.8% and 49.2% ofTable 3Respondents’ socioeconomic information.Basic information Group Population Proportion Description Sample averageAge 18–30 127 30.75% 24 36.4431–40 143 34.62% 3541–50 97 23.5% 4551–60 35 8.5% 55> 60 11 2.6% 65Gender Male 236 57.1% 1 0.57Female 177 42.9% 0Living time in Macau (years) < 1 12 2.9% 0.5 13.541–5 57 13.8% 36–10 35 8.5% 811–15 29 7.0% 13> 15 280 67.8% 17Monthly income (MOP) < 8000 5 1.2% 4000 33,4708001–16000 60 14.5% 12,00016,001–30,000 169 40.9% 23,00030,001–50,000 130 31.5% 40,00050,001–100,000 45 10.9% 75,000> 100,000 4 1.0% 150,000Education level None or primary school 2 0.5% 1 3.54Middle school 29 7.0% 2High school 138 33.4% 3College 188 45.5% 4Master or above 56 13.6% 5Number of family members 1 2 0.5% 1 4.322 74 17.9% 23 71 17.2% 34 130 31.5% 4≥ 5 136 32.9% 5Motorcycle number 0 4 1% 0 2.131 94 22.8% 12 176 42.6% 23 123 29.8% 3≥ 4 16 3.8% 4Table 4Respondents' perceptions on local environmental issues in Macau.Item Level Number Percentage （%） Mean value (Standard deviation)Air pollution Not serious at all 0 0 3.70 (0.807)Not serious 30 7.3Neutral 124 30.0Serious 198 47.9Very serious 61 14.8Noise issue Not serious at all 0 0 3.54 (0.874)Not serious 48 11.6Neutral 150 36.3Serious 157 38.0Very serious 58 14.1Potential pollution from motorcycle Not serious at all 5 1.2 3.42 (0.677)Not serious 16 3.9Neutral 204 49.4Serious 175 42.4Very serious 13 3.1L. Zhu et al. Energy Policy 127 (2019) 64–7267
respondents agreed the low charging cost and high sale price of EM,respectively. For poor battery lifetime, immature technology, highsafety, and low maintenance cost, less than 25% of the respondentsshowed their agreements. The possible reasons for the observed lack ofunderstanding of EM are that currently the motorcycle drivers in Macauare still lacking EM experience and the government still has no effectivemeasures to promote EM.3.3. Influencing factors on buying EM in MacauIn the survey, eleven possible influencing factors on buying EMwere considered as shown in Table 6. The sale price of EM was the firstconsideration when the respondents wanted to buy EM in Macau. About82.5% of respondents paid more attention to sale price of EM. Mean-while, charging fee, repair fee, battery lifetime and cost, battery en-durance and tax incentives were also top considerations. In contrast,driving speed and load capacity were less considered. These resultsindicate that, when purchasing the EM, the consumer will pay moreattention to the EM's actual cost such as sale price, charging fee, repairfee, battery lifetime and cost, battery endurance and tax incentives,while the product features of EM such as driving speed and load ca-pacity attract very little attention. If the government wants to promotethe use of EM, how to lower the cost of EM needs to be addressed as thefirst step in future.3.4. Binary logistic analysis about WTB and WTP3.4.1. Respondents’ WTB and WTPTable 7 lists the respondents’ options for WTB and WTP. It is in-teresting to find that when asked if they are willing to buy EM, therespondents who chose “yes” reached to 66.8%. However, when re-ferring to whether they are willing to pay more to buy EM, the re-spondents who answered “yes” declined to 45%.As shown in Table 7, some respondents’ WTB did not transfer intothe actual WTP. In the survey, possible reasons about unwillingness topay more on EM were also asked and the results are shown in Table 8.About 29.1% of respondents stated that “the existing income is in-sufficient to cover this cost” is the most primary reason for not willingto pay. 20.7% of respondents thought that the companies or enterprisesshould undertake the treatment fees and the relevant responsibility.14.9% of respondents stated that the governments should be re-sponsible for the air pollution issues. 10.1% of respondents were lack ofconfidences on the air pollution control under the current environ-mental management policy. Meanwhile, 9.3% of respondents even be-lieved that the exhaust emissions and noise pollution of ordinary fuelmotorcycles have little impacts on their life.3.4.2. Determinants of WTB and WTP for EMThis study established two binary logistic regression models to es-timate the probabilities of consumers’ WTB and WTP for EM. Model 1used WTB as the dependent variable and Model 2 used WTP as thedependent variable. The independent variables included the re-spondents’ knowledge of EM (Table 5), respondents’ considerations ofEM (e.g. charging fee, environmental benefits and safety), and personalsocial-economic information. The regression analyses were conducted. Veryseldom18.4%Seldom 18.9%Normal30.3%Often20.3%Always12.1%(A) De nitelywould not consider1.0%Unwilling11.1%Neutral31.2%Willing41.6%Verywilling15.0%(B)Fig. 2. Public behaviors for driving motorcycles: (A) Driving frequency; (B) Giving up or reducing the use of motorcycles for environmental protection.Table 5Respondents' knowledge of EM in Macau.Item Level Number Percentage (%) Mean (Standarddeviation)Environmentalbenefits1 0 0 4.02 (0.689)2 2 0.53 88 21.34 223 54.05 100 24.2Low charging cost 1 1 0.2 3.51 (0.752)2 16 5.33 183 47.74 140 375 34 9.8High price 1 0 0 3.44 (0.737)2 41 9.93 169 40.94 183 44.45 20 4.8Poor battery lifetime 1 0 0 2.99 (0.521)2 57 13.83 304 73.64 51 12.45 1 0.2Immature technology 1 0 0 2.99 (0.684)2 95 233 229 55.54 86 20.85 3 0.7High safety 1 1 0.2 2.97 (0.689)2 100 24.23 226 54.74 84 20.45 2 0.5Low maintenance cost 1 3 0.7 2.80 (0.704)2 143 34.63 201 48.74 66 165 0 0Note: “1=Strongly disagree, 2=Disagree, 3=Neutral, 4=Agree, 5=Stronglyagree”.L. Zhu et al. Energy Policy 127 (2019) 64–7268
by using SPSS 19.0 software and the results are shown in Table 9. Here,in order to highlight the key points, we just list the important influencefactors with statistical significances.For Model 1, there is a significant positive relationship betweenrespondents’ knowledge of EM and WTB at the 1% significance level.The respondents’ considerations of charging fee and environmentalbenefits were statistically significant at 1% level. The coefficient ofcharging fee was negative, which makes sense because when con-sidering more on the charging cost, the respondents will have the lowerWTB. For the respondents’ socioeconomic information, the results inTable 9 show that only education level and income level were statis-tically significant at the 1% significance level. Moreover, the coeffi-cients of education level and income level were positive, which supportthe hypothesis that the probability of the respondents’ saying “yes” tothe WTB question increases with education level and income level.In Model 2, the relationship between respondents’ knowledge of EMand WTP was statistically significant at the 1% level. If the respondentsown more EM knowledge, their WTP will be higher. The environmentalbenefits and safety considerations also showed a positive influence onthe WTP at the 1% and 10% level, respectively. However, the re-lationship between charging fee and WTP was not significant. Besides,education level, income level, number of family member and motor-cycle number were also statistically significant at the 1% level. Inparticular, if the respondents’ households owned more motorcycles,they were more willing to pay more for EM. However, it is noticed thatthe coefficient of number of family member was negative. One possiblereason is that if the respondents’ family had more members, they couldconsider to buy a passenger car instead of motorcycle. Therefore, thenumber of family member showed negative effects on the WTP for EM.Many scholars’ results support the conclusion that the respondents’knowledge, income and education level have positive relationship withthe willingness of residents for environmental issues (Yoo and Kwak,2009; Tonglet et al., 2004), which is in accordant with the present re-search results. In addition, some similar results on the respondents’considerations were also obtained in the relevant literature. For ex-ample, Adamson (2005) determined that the reliability, safety, en-vironment benefits, and the purchase price of the vehicle were theTable 6Influence factors on buying EM.Project Level Percentage (%) Mean (Standard deviation) Project Level Percentage (%) Mean (Standard deviation)Sale price 1 0 4.08 (0.641) Environmental benefits 1 0 3.65 (0.831)2 0.2 2 4.43 7.3 3 45.34 27.1 4 31.75 55.4 5 18.6Charging fee 1 0 3.93 (0.808) Charging convenience 1 0 3.47 (0.677)2 2.7 2 2.73 28.1 3 554 42.3 4 34.65 26.9 5 7.7Repair fee 1 0 3.84 (0.664) Fuel price 1 1.7 3.30 (0.901)2 0.5 2 15.53 29.8 3 43.64 55 4 29.85 14.7 5 9.4Battery lifetime and cost 1 0 3.75 (0.696) Driving speed 1 8.7 2.84 (1.014)2 0.5 2 29.13 38.2 3 374 47 4 19.95 14.3 5 5.3Battery endurance 1 0 3.74 (0.681) Load capacity 1 9.2 2.75 (0.968)2 3.6 2 32.23 34.9 3 36.34 47.7 4 19.45 13.8 5 2.9Tax incentives 1 0 3.72 (0.744)2 2.43 404 47.25 10.4Note: “1=Definitely would not consider, 2=Would not consider, 3=Neutral, 4=Would consider, 5=Definitely would consider”.Table 7Frequency Distribution of Respondents’ WTB and WTP.Item Options Number of people Percentage (%)WTB YES 276 66.8NO 137 33.2WTP YES 186 45NO 227 55Table 8Reasons for unwillingness to pay more.Reason Number of people Percentage (%)Existing income levels are insufficient to cover this cost 66 29.1%The exhaust emissions and noise pollution of ordinary fuel motorcycles have little impact on my life 21 9.3%The companies should undertake the treatment fees and the relevant responsibility 47 20.7%Air pollution treatment belongs to public services and should be provided by the government 34 14.9%There is no confidence in air pollution control under current environmental management 23 10.1%others 36 15.9%Total 227 100%L. Zhu et al. Energy Policy 127 (2019) 64–7269
important factors affecting consumers' choice of new energy vehicles.Mabit and Fosgerau (2011) believe that the cost of use and maintenancecosts are the key factors determining consumers' willingness to pur-chase.For the Model 1 and Model 2, the chi-square values were 210.861and 289.193 (significance< 0.001), respectively, indicating that thetwo logistic regression models had statistical significance. The like-lihood values were 313.971 and 279.270, respectively and the overallpredictive accuracies of the two models were 83.3% and 85.7%, re-spectively. In addition, the models’ R2 values were relatively good. Allthese results indicate that the present models and results are reasonableand acceptable.3.5. Contingent valuation analysis of WTP valueTable 10 summarizes the detailed WTP values from the respondentsthrough the closed double-bound dichotomous method. It is found that186 respondents were willing to pay more for purchasing EM. With theWTP value increasing from the initial 1000 MOP to 8000 MOP, thenumber of respondents saying “yes” showed a significant downwardtrend and decreased from 36 to 17. In addition, the respondents’ sup-port for the initial value of 1000 MOP was 91.7%, while the supportrate of 8000 MOP decreased to 17.6%. This is consistent with consumerbehavior. The respondents’ answers in Table 10 are reasonable for thebid amounts adopted, which shows that the present research resultscould reflect the actual consumption choices of the potential consumersin Macau.Using STATA statistical software, this study estimated the de-terminants of WTP value under the double-bound dichotomous CVM,shown in Table 11. It was observed that the respondents' considerationof EM's safety had positive effects on the WTP value at the 1% level.Safety recognition is very important in the marketing of new energyvehicles (Adamson, 2005). At the same time, when considering therespondents' family characteristics, education level, age, living time inMacau, number of family member, and income level were statisticallysignificant. For details, the coefficients of education level, living time inMacau and income level were positive, which support the hypothesisthat the respondents’ WTP value increases with education level, livingtime in Macau, and income level. Similar findings were obtained byother researchers in Turkey (Erdem et al., 2010), Taiwan (Liu, 2009)and the Philippines (Francisco, 2010). However, the coefficient of agewas negative at the 5% level, which is consistent with the existingfindings that young people were more willing to pay more for newenergy vehicles (Ewing and Sarigöllü, 2000). In the present model, thechi-square test value was 44.02 (significance<0.001), and the max-imum likelihood value was −252.454. All the results indicate that themodels and results are reasonable and acceptable.The respondents’ mean WTP value was estimated in this study ac-cording to Eq. (2.6). The result shows that the motorcycle drivers inMacau were willing to pay an additional amount of 2921.07 MOP onaverage for buying the EM. The result was significant (P < 0.001), andits standard error was 127.94. With the 95% confidence interval, thelower and upper bounds of mean WTP were 2670.31 MOP and 3171.83MOP, respectively. The above result is the mean WTP value for 186respondents who were willing to pay more for EM. Considering all 413respondents in this survey, the mean WTP will be 1135.54 MOP. ThisWTP value could be used as a reference value to design a conservationpayment scheme and determine the total funding required for pro-moting the EM use in Macau.4. Conclusions and suggestionsThis study investigated the residents’ knowledge, perceptions, WTBand WTP for EM based on data from 413 face-to-face surveys. Twobinary logistic regression models were established to predict theprobabilities of consumers’ WTB and WTP for EM in Macau. The de-terminants of WTP value were studied by the double-bound dichot-omous CVM and the consumers’ WTP value for EM was estimated usingSPSS 19.0 software and STATA software.Overall, the respondents had relatively less understanding of EM.Most of respondents knew that the EM is environmentally friendly, butless than 25% of respondents showed their agreements with other EM'sfeatures (poor battery lifetime, immature technology, high safety, andlow maintenance and repair costs). The sale price of EM was the firstconsideration when the respondents wanted to buy EM in Macau. WhenTable 9Determinants of WTB and WTP for EM.Variables Model I: Factors Affecting the WTB for EM Model II: Factors Affecting the WTP for EMB p-value B p-valueRespondents' knowledge and their considerations ofEMRespondents' knowledge of EM 1.690 0.002*** 2.899 0.000***Charging fees −0.705 0.007*** – –Environmental benefits 0.746 0.009*** 1.221 0.000***Safety 0.512 0.060*Socioeconomic information Education level 0.939 0.002*** 1.186 0.000***Family members – – −0.621 0.002***Motorcycle number – – 0.838 0.009**Income level 0.864 0.001*** 1.338 0.000***Constant −12.906 0.000 −22.932 0.000Chi-square 210.861 0.000 289.193 0.000Log likelihood 313.971 279.270Nagelkerke R2 0.556 0.674Forecast accuracy 83.3% 85.7%Notes: ***1% significance level; *10% significance level.Table 10Detailed WTP values from the respondents.Bid amount Number of samples Question answerInitial Higher Lower YY YN NY NN1000 2000 500 36 21 12 2 12000 3000 1000 34 19 12 2 13000 4000 2000 26 8 14 1 34000 5000 3000 23 5 6 6 65000 6000 4000 19 6 5 1 76000 7000 5000 16 0 2 4 107000 8000 6000 15 0 1 1 138000 9000 7000 17 0 3 0 14Total 186 59 55 17 55Note: YY=Willing-Willing; YN=Willing-NOT Willing; NY=NOT Willing-Willing; NN= NOT Willing- NOT Willing.L. Zhu et al. Energy Policy 127 (2019) 64–7270
purchasing the EM, the respondents paid more attention to the EM'sactual cost such as sale price, charging fee, repair fee, and tax in-centives, while the product features of EM such as driving speed andload capacity attracted very little attention.When asked if they are willing to buy EM, 66.8% of the respondentschose “yes”. However, when referring to WTP for EM, the respondentswho answered “yes” declined to 45%. The “existing income is in-sufficient to cover this cost” was the most primary reason for not willingto pay. The logistic regression models indicated that the respondents’knowledge of EM and their education level and income level were thedeterminants of WTB and WTP for EM.The respondents’ mean WTP value was estimated to be 1135.54MOP through the double-bounded dichotomous CVM assessment. Thesafety consideration, education level, age, living time in Macau,number of family member and income level were statistically sig-nificant in influencing the WTP value of EM. In particular, the WTPvalue increased with the safety consideration, education level, livingtime in Macau and income level, while it decreased with the re-spondents’ age.Based on the investigation of the WTB and WTP for EM in Macau, itis suggested that the government and stakeholders should carry outmore publicity and education to enhance consumers' knowledgeable ofEM. Since the EM's environmental benefits and safety play a positiverole when the respondents purchase the EM, the government and sta-keholders can also advertise EM's advantages, such as good safety andlow emissions during use. As consumers pay more attention to the EM'sactual cost such as sale price, charging fee, repair fee, battery lifetimeand cost, battery endurance and tax incentives, the government andstakeholders need to focus on how to lower the cost of EM as the firststep to promote EM. Currently the price of EM is about 8000 MOPhigher than that of ordinary fuel motorcycles. Therefore, there is still abig gap between WTP value (1135.54 MOP) and the actual price var-iance. The government may provide some subsidies and preferentialpolicies, such as “EM Promotion Fund” to reduce the price gap. At thesame time, the government should also take the lead in purchasing EMas a demonstration, and further improve infrastructure construction ofEM. According to research results, sellers should organize some activ-ities (such as product launches, and public test drives) for attractingthose potential consumers. In addition, the seller should also perfecttheir product repairs, maintenance and consulting service for the gooduser experiences. The consumers who have higher education level orincome levels will be more likely to buy EM. Therefore, companies andpolicy-makers may foster positive attitudes among consumers throughcommunication and campaigns emphasizing price savings. And it isnecessary to enrich elementary education in environmental protectionto foster the green consumption habits from the child.AcknowledgementsThis work was supported by Research Grants for Macau Universityof Science and Technology (FRG-18-004-MERI) and the Foundation forDevelopment of Science and Technology of Macau (FDCT) (no. 0011/2018/A).ReferencesADB (Asian Development Bank), 2009. Electric Bikes in the People's Republic of China:Impact on the Environment and Prospects for Growth (Retrieved from). AsianDevelopment Bank.Adamson, K.-A., 2005. Calculating the price trajectory of adoption of fuel cell vehicles.Int. J. Hydrog. Energy 30 (4), 341–350.Bjerkan, K.Y., Nørbech, T.E., Nordtømme, M.E., 2016. Incentives for promoting batteryelectric vehicle (BEV) adoption in Norway. Transp. Res. D.—Transp. Environ. 43,169–180.Ching, T.W., Cheong, T.H., 2013. Cost analysis of electric vehicles in Macau. In:Proceeding of TENCON Spring Conference, 2013 IEEE, April 2013, pp. 342–346.ChinaDaily.2011.Motor vehicles in china hit 217m. Retrieved from: 〈http://www.chinadaily.com.cn/business/2011-07/20/content_12941888.htm〉.Carson, R.T., 2000. Contingent valuation: a user's guide. Environ. Sci. Technol. 34 (8),1413–1418.Cai, H., Wang, M.Q., 2014. Consideration of black carbon and primary organic carbonemissions in life-cycle analysis of greenhouse gas emissions of vehicle systems andfuels. Environ. Sci. Technol. 48 (20), 12445–12453.Domingues, A.R., Marques, P., Garcia, R., Freire, F., Dias, L.C., 2015. Applying multi-criteria decision analysis to the life-cycle assessment of vehicles. J. Clean. Prod. 107,749–759.DSEC, 2016. Yearbook of Statistics 2016. Statistics and Census Service of Macau.EEA (European Environmental Agency), 2011. Air Quality in Europe – 2011 Report(Retrieved from). European Environmental Agency.Erdem, C., Şentürk, İ., Şimşek, T., 2010. Identifying the factors affecting the willingness topay for fuel-efficient vehicles in Turkey: a case of hybrids. Energy Policy 38 (6),3038–3043.Ewing, G.O., Sarigöllü, E., 2000. Assessing consumer preferences for clean-fuel vehicles: adiscrete choice experiment. J. Public Mark. 19 (1), 106–118.Francisco, J.P.S., 2010. Are Metro Manila Households Willing to Pay for Cleaner PublicTransport? EEPSEA, IDRC Regional Office for Southeast and East Asia, Singapore, SG.Hanemann, M., Loomis, J., Kanninen, B., 1991. Statistical efficiency of double-boundeddichotomous choice contingent valuation. Am. J. Agric. Econ. 73 (4), 1255–1263.Hanemann, W.M., 1999. The economic theory of WTP and WTA. Valuing environmentalpreferences: Theory and practice of the contingent valuation method in the US, EU,and developing countries,42-96.Hadker, N., Sharma, S., David, A., Muraleedharan, T.R., 1997. Willingness-to-pay forBorivli National Park: evidence from a contingent valuation. Ecol. Econ. 21 (2),105–122.Hoehn, J.P., Randall, A., 1987. A satisfactory benefit cost indicator from contingent va-luation. J. Environ. Econ. Manag. 14 (3), 226–247.Hidrue, M.K., Parsons, G.R., Kempton, W., Gardner, M.P., 2011. Willingness to pay forelectric vehicles and their attributes. Resour. Energy Econ. 33 (3), 686–705.IEA, 2009. World Energy Outlook 2009. Paris, France.Jin, J., Wang, Z., Ran, S., 2006. Comparison of contingent valuation and choice experi-ment in solid waste management programs in Macao. Ecol. Econ. 57 (3), 430–441.JAMA, 2012. Worldwide Vehicle Ownership 2012. Japan Automobile ManufacturersAssociation, Inc (Retrieved from). 〈http://www.jamabj.cn/world/world/world_2t1.asp〉.Lai, I.K., Liu, Y., Sun, X., Zhang, H., Xu, W., 2015. Factors influencing the behaviouralintention towards full electric vehicles: an empirical study in Macau. Sustainability 7(9), 12564–12585.Loomis, J.B., Walsh, R.G., 1997. Recreation Economic Decisions; Comparing Benefits andCosts, 2 edition. Venture Publishing Inc.Lin, B., Tan, R., 2017. Estimation of the environmental values of electric vehicles inChinese cities. Energy Policy 104, 221–229.Liu, Y.F., 2009. Study on the Willingness and Willingness to Pay for the Use of HydrogenFuel Buses in Taiwan Metropolitan Areas. National Kaohsiung Normal University,NKNU.Li, R., He, Q., Li, S., 2013a. Analyses of Automobile Ownership Development Trend inChina. Urban Transport of China.Li, X., Clark, C.D., Jensen, K.L., Yen, S.T., English, B.C., 2013b. Consumer purchaseTable 11Determinants of WTP values for EM.Variables B value Standard Deviation p-valueRespondents' knowledge and their consideration of EM Safety 580.060 188.064 0.002***Respondents' socioeconomic information Education level 456.502 240.246 0.057*Age −175.112 181.543 0.035**Living time in Macau 252.800 129.166 0.050**Number of family members −166.245 148.514 0.063*Income level 588.354 182.979 0.001***Constant −3476.918 1396.858 0.013Log likelihood Chi-square Test −252.45444.02 0.000Notes: *** 1% significance level; **5% significance level; *10% significance level.L. Zhu et al. Energy Policy 127 (2019) 64–7271
intentions for flexible-fuel and hybrid-electric vehicles. Transp. Res. Part D: Transp.Environ. 18, 9–15.Mabit, S.L., Fosgerau, M., 2011. Demand for alternative-fuel vehicles when registrationtaxes are high. Transp. Res. Part D: Transp. Environ. 16 (3), 225–231.Orsi, F., Muratori, M., Rocco, M., Colombo, E., Rizzoni, G., 2016. A multi-dimensionalwell-to-wheels analysis of passenger vehicles in different regions: primary energyconsumption, CO2 emissions, and economic cost. Appl. Energy 169, 197–209.Potoglou, D., Kanaroglou, P.S., 2007. Household demand and willingness to pay for cleanvehicles. Transp. Res. Part D., Transp. Environ. 12 (4), 264–274.Roche, M.Y., Mourato, S., Fischedick, M., Pietzner, K., Viebahn, P., 2010. Public attitudestowards and demand for hydrogen and fuel cell vehicles: a review of the evidence andmethodological implications. Energy Policy 38 (10), 5301–5310.Sung, Y.-C., 2010. Consumer learning behavior in choosing electric motorcycles. Transp.Plan Technol. 33 (2), 139–155.Song, Q., Wang, Z., Li, J., 2012. Residents' behaviors, attitudes, and willingness to pay forrecycling e-waste in Macau. J. Environ. Econ. Manag 106, 8–16.Song, Q., Wang, Z., Li, J., 2016. Exploring residents' attitudes and willingness to pay forsolid waste management in Macau. Environ. Sci. Pollut. Res. Int. 23, 16456–16462.Song, Q., Li, J., Duan, H., Yu, D., Wang, Z., 2017a. Towards to sustainable energy-efficientcity: a case study of Macau. Renew. Sust. Energy Rev. 75, 504–514.Song, Q., Wu, Y., Li, J., Wang, Z., Yu, D., Duan, H., 2017b. Well-to-wheel GHG emissionsand mitigation potential from light-duty vehicles in Macau. Int J. Life Cycle Assess.1–12.Song, Q., Wang, Z., Wu, Y., Li, J., Yu, D., Duan, H., Yuan, W., 2018. Could urban electricpublic bus really reduce the GHG emissions: a case study in Macau. J. Clean. Prod.172, 2133–2142.Sheng, N., Tang, U.W., 2011. Spatial analysis of urban form and pedestrian exposure totraffic noise. Int. J. Environ. Res. Publ. Health 8 (6), 1977–1990.Sheng, N., Zhou, X., Zhou, Y., 2016. Environmental impact of electric motorcycles: evi-dence from traffic noise assessment by a building-based data mining technique. Sci.Total Environ. 554, 73–82.Senzhu, F.R.Y., 2010. China's new standards for motorcycle fuel consumption and thestudy of energy saving technology. Motorcycle Technol. 1, 028.Tonglet, M., Phillips, P.S., Read, A.D., 2004. Using the Theory of Planned Behaviour toinvestigate the determinants of recycling behaviour: a case study from Brixworth, UK.Resour. Conserv. Recycl. 41 (3), 191–214.Weinert, J.X., 2007. The Rise of Electric Two-wheelers in China: Factors for Their Successand Implications for the Future. University of California, Davis.Wiser, R.H., 2007. Using contingent valuation to explore willingness to pay for renewableenergy: a comparison of collective and voluntary payment vehicles. Ecol. Econ. 62(3–4), 419–432.Weinert, J., Ma, C., Cherry, C., 2007. The transition to electric bikes in china: history andkey reasons for rapid growth. Transportation 34 (3), 301–318.Wang, X., Westerdahl, D., Wu, Y., Pan, X., Zhang, K.M., 2011. On-road emission factordistributions of individual diesel vehicles in and around Beijing, China. Atmos.Environ. 45 (2), 503–513.Yoo, S.H., Kwak, S.Y., 2009. Willingness to pay for green electricity in Korea: a contingentvaluation study. Energy Policy 37 (12), 5408–5416.Zhang, Y., Yu, Y., Zou, B., 2011. Analyzing public awareness and acceptance of alter-native fuel vehicles in China: the case of EV. Energy Policy 39, 7015–7024.L. Zhu et al. Energy Policy 127 (2019) 64–7272
