**3. Data**

In this study, the vehicle models of each company sold in 2015 were as follows: Toyota, 42; Nissan, 21; Honda, 17; Mitsubishi, 10; Mazda, 9; Subaru, 9; and Suzuki, 8. The number of vehicles of each model sold by each company, which is necessary for calculating the CAFE and CAFE target, can be obtained from data on the number of vehicles sold by brand [25]. For the fuel economy of each vehicle model the fuel economy figures for each vehicle model in JC08 mode cycle was used, as published in the Automobile Fuel Economy List [26]. The vehicle weight categories for the CAFE standards due to be introduced in MY2020 are shown in Table A1.

The CO2 emission intensities per passenger vehicle in manufacturing, during travel, and in disposal were estimated using the Embodied Energy and Emission Intensity Data for Japan Using Input-Output Tables [24]. The passenger vehicle lifetime travel distance *d* was assumed to be 100,000 km and therefore estimated the emission intensity during travel *rg* to be 0.00231 t-CO2 and *rc* to be 0.00063 t-CO2. In addition, in accordance with a previous study [6], the emission intensity in disposal *fh,i* was set to be 0.0574 t-CO2.

In order to estimate the life cycle CO2 emission intensity of vehicles, the life cycle CO2 emission intensity derived from both manufacturing and driving must be estimated for each vehicle model. While by no means a simple task, in this study, the lifecycle CO2 emission intensity for each vehicle model was estimated by specifying a relationship for model sales prices and new vehicle weight. First, the sales price information was obtained for 82 gasoline-engine vehicle models sold by the automakers (Toyota, Nissan, Honda, Mitsubishi, Mazda, Subaru, and Suzuki) in 2015 from Autoc One [27], an informational site that releases comprehensive vehicle sales information. Vehicle weight information was also obtained for the same 82 models from the MLIT automotive information site [26]. From the sales price and vehicle weight data for the 82 models, a regression analysis was run and the following results were obtained.

$$\begin{array}{rcl}p\_i^\% = & 0.35w\_i^\% \quad - & 222\\ & & (7.46) \\ & & & \text{Adjusted}R^2 \text{ : } 0.38 \end{array} \tag{12}$$

where *wgi* (kg) is the vehicle weight for vehicle model *i* and *pgi* (10,000 s of Japanese yen) is the sales price for vehicle model *i*. The numbers in parentheses below the parameters are the *t*-values, and each of the estimated parameters is statistically significant at the 1% level in a two-sided test. The relationship given in Equation (12) shows us that an increase of 100 kg in vehicle weight corresponds to an increase of 350,000 yen in sales price.

From the 2005 Input-Output Tables, the average vehicle sales price in 2005 was 2.2 million yen. Given this, the relationship specified in Equation (12) can be used to estimate the average vehicle weight as *wg* = (220 + 222)/0.35 = 1264 kg. Meanwhile, from the Embodied Energy and Emission Intensity Databook (3EID) ([24]) as based on the 2005 Input-Output Table as released by the National Institute for Environmental Studies, the average lifecycle emission intensity for vehicle production is 1.93 t-CO2 per 1 million yen, and the lifecycle emission intensity for transportation and sales services incidental to sales price for one vehicle unit is 1.2 t-CO2 per 1 million yen. Accordingly, one can estimate a lifecycle CO2 emission intensity of 1.93 × 2.2 = 4.2 t-CO2 as derived from manufacturing one average vehicle in 2005 with a sales price of 2.2 million yen and vehicle weight of 1264 kg. Next, the lifecycle CO2 emission intensity was estimated, as derived from manufacturing a relevant vehicle model by taking the ratio of the vehicle weight of that model to the average vehicle weight (1264 kg) and multiplying by the unit intensity derived from manufacturing. To estimate the lifecycle CO2 emission intensity incidental to transportation and sales services for one unit of a relevant vehicle model, the lifecycle emission intensity for transportation and sales services was taken as 1.2 t-CO2 per 1 million yen and multiplied this quantity by the sales price of the relevant vehicle model. The lifecycle CO2 emission intensity *f gm*,*i* for a single gasoline vehicle model *i* was then solved for by adding up the lifecycle CO2 emission intensities derived from manufacturing and from transportation and sales for the relevant model. It is important to note that although we can estimate the lifecycle CO2 emissions by multiplying the average lifecycle emission intensity for vehicle production (1.93 t-CO2 per 1 million yen) by each vehicle price, and that the estimated emissions are not consistent with the vehicle weight important for the CAFEs.

Similarly, a separate regression analysis for 42 hybrid vehicle models was run and the following relationship for sales price and vehicle weight was obtained:

$$\begin{array}{ccccc}p\_i^h = & 0.41w\_i^h & - & 282\\ & & \text{(8.12)} & & \text{(-3.54)}\\ & & & \text{Adjusted} \mathbb{R}^2 \text{ : 0.62} \end{array} \tag{13}$$

where *whi* (kg) is the vehicle weight for hybrid vehicle model *i* and *phi* (10,000s of yen) is the sales price for hybrid vehicle model *i*. Again, the numbers in parentheses below the parameters are the *t*-values, and each of the estimated parameters is statistically significant at the 1% level in a two-sided test. The relationship given in Equation (13) shows us that an increase of 100 kg in vehicle weight for hybrid vehicles corresponds to an increase of 410,000 yen in sales price. The lifecycle CO2 emission intensity *f hm*,*i* derived from manufacturing and from transportation and sales for a single hybrid vehicle model *i* was solved for with the same methods described above for calculating unit intensity for a gasoline vehicle model. The detailed lifecycle CO2 emission intensity data by vehicle model as estimated in this study are described in Table S3 of the Supporting Information.
