• Tidak ada hasil yang ditemukan

Chapter 5. Simulated Analysis

A. Consumer Module

For the purposes of this study, “consumers” refers to those who own and use a vehicle. Consumers typically make vehicle purchase decisions by considering numerous factors, including: fuel type, driving distance, mileage, vehicle price, vehicle size, charging/fueling time, and carbon dioxide emissions. In our study, vehicles were limited to five types of models:

gasoline vehicles, diesel vehicles, LPG vehicles, HEVs, PHEVs, and BEVs. Among the five types, consumers were allowed to choose the vehicle size: compact, small-size, mid-size, mid-to-large size, large-size, and sports utility vehicle. Our model assumes that each consumer only owns and drives one vehicle, knows the date of manufacture of their vehicle, and remains alive for the entire duration of the simulated period. It was also assumed that consumers only choose and purchase vehicles as first-time car buyers or when their old car needs to be replaced.

The consumers in our model undergo the following decision-making stages when making a vehicle purchase.

Stage 1. Decision to replace an old car or buy a car for the first time.

Consumers with older cars must decide to either replace their vehicle or continue using it until it can no longer run. This probability distribution—the probability of a consumer having to remove his/her car from the market—can be seen as follows.

Figure 5-2. Total Number of Owned Vehicles and Number of New Vehicles Sold (in Millions)

Source: KEEI, Long-Term Energy Demand Projection Model.

총 자동차 대수 Total number of vehicles owned

판매 차량 Number of vehicles sold

The probability distribution used in our model is borrowed from the Korea Energy Economics Institute (KEEI)’s long- term energy demand projection model for vehicles. The same model is used to estimate increases in the total number of vehicles owned. Figure 5-2 illustrates the trends in the total number of vehicles owned and the number of new vehicles sold, predicted using the KEEI’s long-term projection model.

When there are more new vehicles being sold than old vehicles being scrapped, we can assume that new consumers are entering the vehicle market. If the reverse occurs, we can assume that existing consumers have exited the vehicle market.

The fact that the total number of vehicles owned continually increased throughout the simulated period means that more vehicles were sold than gotten rid of, with more consumers entering the market than exiting it.

Stage 2. Selection of car size

Each consumer who decided to purchase a car must then decide the size of the vehicle they wish to purchase. The probability distribution for vehicle size can be seen as follows.

Table 5-1. Distribution of Sizes of Domestic Vehicles Sold in Korea

Size 2015 2016 September 2017

Compact 13.1 12.9 10.7

Small 15.8 16.0 15.2

Midsize 16.0 15.1 13.5

Medium-to-large 9.6 9.4 14.3

Large 4.2 6.4 5.1

SUVs 41.3 40.3 41.1

Source: KAMA.

Note: Compact: below 1,000 cc Small: 1,000 to 1,599 cc Midsize: 1,600 to 1,999 cc

Medium-to-large: 2,000 to 2,999 cc Large: 3,000 cc or greater

Table 5-1 shows the trends in the sale of sedans and SUVs in Korea. The demand for compact cars (below 1,000 cc) has been dwindling, while the demand for small vehicles (below 1,600 cc) and midsize vehicles (below 2,000 cc) has been stagnant or has dropped slightly. The demand for medium-to-large vehicles (2,000 to 3,000 cc), however, has risen steadily.

SUVs made up 22.6 percent of all new vehicles sold in 2010, and 41.3 percent in 2015, but the demand for SUVs has remained stagnant over the past three years.59

Our model assumes that the percentages of different sized vehicles sold will remain the same throughout the entire period from 2017 to 2030. A vast majority of EVs fall in the categories of small and compact vehicles. Our model thus supposes that only consumers choosing small and compact vehicles are likely to choose EVs.60 As for consumers choosing larger vehicles, the same percentages of vehicles sold in 2017 continue to apply.61

59 The percentages of vehicles sold by size are from the KAMA.

60 Although Tesla has made its way into the Korean market, relatively few Tesla vehicles have been sold thus far. For the simplicity of our model, we thus limited our analysis to compact and small cars that are popularly sold in Korea at the present time.

61 As the purpose of this study is to determine how differences in purchase subsidies affect the demand for EVs in Korea, there is little meaning in discussing the sales trends for EV vehicles that are not generally available in Korea. We thus excluded EV vehicles larger than midsize from our discussion.

Stage 3. Consideration of probability distribution for choosing an EV

Once a consumer decides to purchase a small or compact vehicle, they are even more likely to purchase an EV. Using our indirect utility function estimates from Chapter 4 (Table 4-4), let us now estimate the probability distribution for consumers choosing to purchase an EV after having decided to purchase a small or compact vehicle.

Table 5-2. Default Settings for Estimating the Probabilities of Choosing Compact Vehicles

Fuel type Attribute Default

Gasoline

Price KRW 13.15 million

Mileage 15.4km/l

CO2 emissions 192g/km

Driving distance 693km

Charging time 5 minutes

Electric

Price KRW 59.5 million

Mileage 52.7km/l

CO2 emissions 94g/km

Driving distance 130km

Charging time 30 minutes

Sources: Kia Morning Luxury (automatic transmission) for gasoline,

(http://auto.daum.net/newcar/make/model/main.daum?modelid=4359), accessed November 20, 2017; BMW i3 94 LUX for electric (because the Chevrolet Spark EV, the only model of a comparable class available in Korea, is no longer in production), (https://www.bmw.co.kr/ko/all-models/bmw-i/i3/2013/at-a-glance.html), accessed November 20, 2017; Table 4-2 for CO2 emissions; mileage for the electric vehicle represented was calculated by converting the mileage listed on www.fueleconomy.gov into kilometer-per-liter.62

Note: The driving distance for the represented gasoline vehicle is based upon a tank size of 45 liters. The gas-charging time for ICE vehicles is assumed to be five minutes. The charging time for EVs is assumed to be 30 minutes.

In order to estimate the distribution of vehicle choice probabilities, we first established initial (default) settings for mileage, price, charging time, driving distances, and carbon dioxide emissions for gasoline and diesel vehicles, HEVs, PHEVs, and BEVs.63

In the compact segment, our model assumes that consumers choose only either gasoline vehicles or EVs. Since LPG vehicles made up only 2.7 percent of all compact vehicles sold in Korea in 2017,64 we assumed that this minuscule share of LPG compact vehicles would remain the same throughout the period of analysis. Consumers are thus assumed to choose either gasoline compact vehicles or compact EVs with a combined probability of 97.3 percent.

The probability of consumers choosing gasoline compact cars can be projected using the estimates shown in Tables 5-2 and 404, calculated using Equation (3). The probability of a consumer choosing a compact EV is 11.8 percent, and the probability of a consumer choosing a compact gasoline vehicle is 85.5 percent. This estimated distribution of probabilities, however, does not accurately reflect choices in the real world. As of September 2017, gasoline vehicles represented 97.1

62 The average mileage for EVs, which is calculated by converting the battery capacities, is 23 to 24 kilometers per liter. American estimates have been used in this study because the mileage for PHEVs listed in Table 5-3 is difficult to convert into the required unit due to a lack of available data. Accordingly, we used American data on the mileage of PHEVs. The mileage estimates may vary, but do not significantly alter the conclusion of this study, as this study is not primarily focused on projecting EV demand. For more information, see:

https://www.fueleconomy.gov/feg/PowerSearch.do?action=noform&path=1&year1=1984&year2=2018&vtype=Electric

63 The prices consumers pay for their vehicles consist of the prices set by automakers/dealers and taxes. Our model applies a seven- percent sales tax to every vehicle purchased, with the exception of EVs.

64 See Table 5-1, based on data provided by the KAMA, for vehicle sales numbers by vehicle size and fuel type.

percent of all compact vehicles sold in Korea, while EVs represented only 0.3 percent. This indicates the need to readjust our estimated probability distribution.

In our conjoint method analysis, fuel type was not among the factors considered by consumers when purchasing a vehicle.

Table 4-4 therefore does not reflect the utility that consumers would derive from choosing different fuel types.65 In estimating the probabilities of different consumer choices, we therefore need to factor in consumer fuel type preferences.

There are two main sources from which we can obtain information on consumers’ preferred fuel types—the percentages of vehicles of different fuel types sold in the market and fuel type preferences as stated by consumers themselves in opinion polls. We thus readjusted the probabilities of fuel type choice using information from both of these sources. This allowed us to estimate the effects of fuel types on both current and future choices made by consumers. To estimate choice probabilities, we applied Equation (3), using the percentage of compact vehicles sold in 2014 and consumers’ stated preferences, to determine the fuel type of their next vehicle purchase, as shown in Figure 4-14.

In our prior estimation of choice probabilities, we disregarded the accessibility of charging stations for EVs. As our survey indicates, the shortage of charging stations is the biggest obstacle to raising EV demand. The accessibility of EV charging stations would therefore be likely exert a significant effect on whether or not consumers are likely to choose an EV. Our ABM model (see the following equation) is thus designed so that consumers’ probability of choosing an EV can be analyzed according to the level of charging station distribution.

4)

Here m and a are parameters, and N is the number of EVs sold. We modified the m and a according to the number of EVs sold in Korea in 2017.66 Using this formula, charging station accessibility was determined to be only 20 percent of the accessibility of gas stations for ICE vehicles. The lower the charging station accessibility, therefore, the less likely consumers are to choose an EV.

65 Fuel type as a purchasing consideration was omitted from our conjoint method model because, as explained earlier, fuel type is such a significant factor that it could easily overwhelm all other considerations. We made this decision based on the advice of other experts before running our conjoint method analysis.

66 The charging station accessibilities used in this study are arbitrary. This is because there are neither any studies nor any time-series data on charging station accessibilities to help us estimate the distribution of probable charging station accessibilities with any degree of reliability. The distribution of charging station accessibilities, however, does not significantly alter the conclusion of this study.

The primary purpose of this study is not to project EV demand, but rather to determine consumer responses to changes in EV purchase subsidies so as to help policymakers determine the most effective subsidy levels to achieve policy goals.

Figure 5-3. Distributions of Charging Station Accessibilities

All of the forgoing discussion in this section can be summarized using the following probability equation:

5)

Here, i represents consumers; x, a vector consisting of consumers’ idiosyncrasies, mileage, price, charging time, driving distance, and carbon dioxide emissions; j, the vehicle type (i.e., gasoline, diesel, HEV, PHEV, or BEV); , a vector consisting of the estimated coefficients of all the attribute variables; p(jc), the percentages of vehicles of different types sold in 2017; and p(jf), consumers’ stated fuel type preferences. If j in A(j) is an EV, A(j) is equivalent to Equation (4). If j is not an EV, A(j) = 1. The readjusted probabilities are estimated according to complex equations, which are simulated with precision by our model program.

The same process of estimating choice probabilities was applied to consumers purchasing small cars. This time, however, a greater variety of vehicle types (i.e., gasoline and diesel vehicles, HEVs, PHEVs, and BEVs) were included in the analysis. As with the compact vehicle market, the percentage of LPG vehicles sold in the small vehicle market is also extremely small and is assumed to remain constant throughout the period of analysis at 1.5 percent.

The default settings needed to estimate the probabilities of consumers choosing different types of vehicles in the small vehicle market are shown in Table 5-3.

Table 5-3. Default Settings for Probabilities of Choosing Small Vehicles

Fuel type Attribute Default

Gasoline Price KRW 20.14 million

Mileage 13.7km/l

CO2 emissions 192g/km

Driving distance 685km

Charging time 5

Diesel Price KRW 20.20 million

Mileage 18.4km/l

CO2 emissions 189g/km

Driving distance 920km

Charging time 5

HEV Price KRW 25.33 million

Mileage 22.4km/l

CO2 emissions 141g/km Driving distance 1,008km

Charging time 5

PHEV Price KRW 33.73 million

Mileage 48.1km/l

CO2 emissions 141g/km

Driving distance 927km

Charging time 15

EV Price KRW 42.60 million

Mileage 57.8km/l

CO2 emissions 94g/km

Driving distance 191

Charging time 30

Sources: Hyundai Avante 1.6 GDi Modern for gasoline and 1.6 VGT Smart for diesel (http://auto.daum.net/newcar/make/model/main.daum?modelid=4407), accessed November 25, 2017; Hyundai Ioniq N models for EV, HEV, and PHEV (https://www.hyundai.com/kr/ko, all models/eco-friendly), accessed November 25, 2017;

Table 4-2 for CO2 emissions; mileage for the electric vehicle was calculated by converting mileages listed on www.fueleconomy.gov into kilometer-per-liter.

Note: The driving distances for the gasoline and diesel vehicles are based upon a tank size of 50 liters. The gas-charging time for ICE vehicles is assumed to be five minutes. The charging time for EVs is assumed to be 30 minutes. A penalty was applied to the CO2 emissions of the diesel vehicle due to its additional particulate matter emissions.