Disguised Protectionism? Environmental Policy in Japanese Car Market∗ Taiju Kitano† Institute of Innovation Research, Hitotsubashi University March 2015

Abstract US government criticized a Japanese subsidy policy on the promotion of eco-friendly cars as a disguised protection by arguing that a fuel economy standard for the subsidy qualification was designed to be more beneficial to domestic firms. This paper examines the Japanese car market from 2005–2009 and shows that the fuel economy standard was too lax in achieving the environmental goal, and the lax standard was beneficial to Japanese firms. Although these indicate policy distortion, this paper further shows that the distortion was unlikely to hurt the US firms’ profits. Keywords: Car market; Cost-effectiveness analysis; Discrete choice model; Disguised protection; Scrap incentives JEL Classification: F18; L62; Q56

∗

This study was conducted as part of the Basic Research for a New Industrial Policy project undertaken at the Research Institute of Economy, Trade, and Industry (RIETI). The author would like to thank Minoru Kitahara, Hiroshi Mukunoki, Satoshi Myojo, Masaki Nakabayashi, Ryo Nakajima, Ryo Ogawa, Hiroshi Ohashi, Hideo Owan, Minsoo Park, seminar participants at the University of Tokyo, Osaka University and Hitotsubashi University, and attendees at the European Association for Research in Industrial Economics (EARIE) Annual Conference 2012 in Rome, and the Asian Meeting of the Econometric Society 2013 in Singapore for their helpful comments. Financial support from the Japanese Society for the Promotion of Science (JSPS) is gratefully acknowledged. All remaining errors are my own. † 2-1 Naka, Kunitachi, Tokyo 186-8607, JAPAN. E-mail: [email protected]

1

1

Introduction

With increasing concern for greenhouse gas emissions from car use, a number of countries have introduced environmental policies to diffuse low emission and fuel-efficient automobiles. In Japan, the government has employed several forms of environmental policy, including tax incentives and subsidies for eco-friendly car (hereafter, eco-car) purchases. The set of policies introduced in 2009 that included subsidies for eco-car purchases with scrap incentives had a significant impact on the market. In 2010, one year after the introduction of the policies, the average age of existing passenger cars decreased for the first time since 1993.1 Although this type of environmental policy seems to contribute to the resolution of negative externalities in car use,2 trade experts often express concern on the use of environmental policies as the secondary means of trade barriers. With respect to this concern, an earlier theoretical work by Markusen (1975) shows that the government does not always have an incentive to distort its domestic policies (including environmental policies) for the purpose of protection because the efficient means to achieve multiple policy goals, for example, trade and environment, are multiple policies that manage the relevant problems directly. However, World Trade Organization (WTO) member countries have an incentive to distort domestic policies for the purpose of trade goals, such as the terms of trade gains (Copeland (1990), Ederington (2001) and Ederington (2002)) or domestic firms’ competitive edge (Barrett (1994), Conrad (1993), and Kennedy (1994)) because countries cannot use their trade policies freely under the WTO system. Therefore, concern for disguised protection is a problem in the WTO system. In the case of the Japanese subsidy policy, US trade representatives criticized the policy suggesting that it was designed to provide more benefits to Japanese manufacturers. In fact, compared to Japanese car models, only a few US models qualified for the subsidy.3 To resolve the situation, the US government requested laxer fuel economy standards to expand the number of US car models that would meet the subsidy qualification. This paper assesses whether the subsidy for eco-cars in Japan was designed to be more beneficial for Japanese firms rather than to achieve the environmental goal. The assessment is based on a comparison between the actual subsidy policy and possible alternatives. This paper focuses on two parameters of the subsidy policy, the fuel economy standard for subsidy provision and the amount of the subsidy per unit. I construct alternative subsidy 1

The average car age was 2.93 in 1993 and continued to increase until 2009. In 2010, the average car age decreased to 7.48 from 7.49 in 2009. 2 See Parry, Walls, and Harrington (2007) for a recent survey on the evaluation of the externality of car use. 3 See “Kirk Comments on Release of List of US Autos Models That Qualify For Japan’s Cash for Clunkers Program.” http://www.ustr.gov/about-us/press-office/press-releases/2010/february/kirkcomments-release-list-us-autos-models-qualif

2

provision rules for different parameter values. Based on the alternatives, I implement a costeffectiveness analysis (CEA) whereby the average fuel economy of new cars sold is selected as the effect. Because the policy assessment requires the counterfactual outcomes for the alternatives, I employ a structural econometric model of imperfect competition as in Berry, Levinsohn, and Pakes (1995), which allows me to obtain the outcomes by simulation. The structure of the model is static; however, by exploiting the recent development of dynamic demand models (Adda and Cooper (2000) and Schiraldi (2011)), the model allows the consumer’s choice to depend on the state of car ownership—the age of the car that the consumer owns—to reveal the effects of the scrap incentives in the subsidy policy. The dataset used in the econometric analysis is based only on Japanese car model information. Unfortunately, the dataset does not contain the information on import cars because of the unavailability of import car sales data. Therefore, I cannot analyze the effects of the subsidy policy on foreign firms’ profits, although it is a core element in assessing whether the policy was a disguised form of protection. Despite this limitation, it is still possible to analyze the effects on the environment or the average fuel economy of new cars sold to some degree of accuracy by focusing only on Japanese cars because of the small share of import import cars. Therefore, the analysis can provide some valuable insights concerning the disguised protection problem by assessing the validity of the US request, which is clearly relevant to the US firms’ profits, from the environmental perspective; that is, laxer fuel economy standards. The findings of this paper are summarized as follows. First, the results of the CEA indicate that the actual fuel economy standard was too lax from an environmental perspective. To achieve a higher level of environmental quality, the government should narrow the subsidy targets by setting a higher fuel economy standard and increasing the amount of the subsidy per unit. Second, based on the comparison between the actual and optimal rules, I find it possible that the environmental policy was distorted for the purpose of domestic firms’ profits: the Japanese firms enjoyed greater profits from the lax fuel economy standard. Third, despite the possibility of the policy distortion, these results indicate that the US request concerning the fuel economy standard was not valid: the standard should be tightened rather than relaxed. This paper contributes to the literature that analyzes the validity of policy interventions with international trade acts. Previous studies examined the conditions justifying the use of trade remedies, such as the safeguard measures. One of the requirements in the implementation of trade remedies is called injury determination; that is, whether or not the domestic industries are injured from the increase in imports. These studies assessed the validity of the injury determinations made by a government by investigating whether or not a decline 3

in domestic firms’ performance was due to import competition. See Grossman (1986) for US steel, and Kelly (1988) for US wood shakes, shingles, and nonrubber footwear. Additionally, Kitano and Ohashi (2009) employed the structural econometric method, similar to this study, to investigate injury determination on US motorcycle safeguards in the 1980s. In contrast to the studies on the injury determination, this paper examines the subsidy policy in terms of the National Treatment principle of the WTO or, more specifically, the GATT Article XX general exemption, which specifies the prohibition of using environmental policies as “disguised restriction on international trade.” Although the Japanese subsidy was not investigated under the WTO rule, the empirical analysis in this paper provides an assessment of the validity of policy interventions in terms of GATT Article XX from an economic perspective. Substantial literature examines the effects of policy interventions in automobile markets. Among them, several previous studies employed structural econometric models similar to the model used in this study to assess policy interventions in automobile markets. For example, there are studies on Corporate Average Fuel Economy (CAFE) regulation in the US (Goldberg (1998); Bento, Goulder, Jacobsen, and von Haefen (2009); Klier and Linn (2012)), tax incentives for hybrid cars in the US (Beresteanu and Li (2011)), and on replacement subsidies in France (Adda and Cooper (2000)) and Italy (Schiraldi (2011)). Although this study examines the environmental policies similar to these previous studies, the focus of this paper is the role of the environmental policies as a secondary means to trade barriers, which has been widely discussed among economists and policy makers but rarely examined based on economic analyses. Hence, this is the first quantitative analysis on the international conflicts concerning environmental policies based on the structural econometric model. The remainder of this paper is organized as follows. The following section introduces disguised protectionism in the literature for international trade, and I assess this subject in this paper in detail. Section 3 describes Japanese car market characteristics and the evolution of environmental policies on car purchases and holdings from fiscal year 2005 to 2009. Section 3 introduces the structural model of demand and supply. Section 4 introduces moment conditions used in the generalized method of moments (GMM) estimation with a discussion on the identification of the structural parameters and shows the estimation results. Section 5 presents simulation results on the impacts of environmental policies in Japan and implements a CEA to assess the subsidy policy from the environmental perspective. Section 6 concludes the paper.

4

2

Japanese car market and environmental policies

The Japanese car market includes the sale of approximately three million passenger cars annually and is the second largest market in the world after the US. Toyota has been the largest manufacturer, owning approximately 43% of the market share in 2009, and Nissan and Honda are the second and third largest car manufacturers, owning approximately 17% and 15% of the market, respectively. Daihatsu, Mitsubishi, Mazda, Suzuki, and Subaru follow. In total, the share of Japanese cars exceeded 90% and, thus, import cars were less prevalent. With respect to import car sales, the share of US manufacturers has been small; the US sold approximately 4,000 cars. Because of the large market size, car use is a major source of greenhouse gas (GHG) emissions in Japan. According to the Ministry of Land, Infrastructure, Transport, and Tourism, the transportation sector accounts for approximately 20% of total CO2 emissions. To address the need for CO2 reduction, the Japanese government has employed environmental policies to promote the sales of low emission and fuel-efficient cars. This section explains the policies concerning passenger cars that were in effect from the year 2005 to the year 2009.4 Before a detailed explanation of the policies, I specify the target of this study and note the limitations of the available data. First, this study focuses on standard-sized cars with engine displacement of over 660cc that are considered major competitors of import cars because of the similarity of the characteristics. Hence, I exclude mini-vehicles and passenger cars with an engine displacement below 660cc,5 which represent a specific class in the Japanese market. Second, this study does not examine the effects of environmental policies on import cars because of limited import car sales data. Although I cannot investigate the effects of the policies on foreign firms’ profits, this paper tries to obtain some insight on the US firms’ profits by focusing on the US request concerning Japanese fuel economy standards used in the subsidy provision. To be more precise, I discuss whether the Japanese subsidy provision rule was designed to be unprofitable for US firms based on a comparison with an efficient subsidy provision rule from an environmental perspective. Note that the exclusion of import cars has limited impact on the assessment from an environmental perspective because of the small market share for import cars.

4

The market structure and policies concerning Japanese car markets are documented in The Motor Industry of Japan, an annual publication of the Japanese Automotive Manufacturers Association (JAMA). 5 A car is classified as a mini-vehicle if (1) the size of engine displacement ≤ 660cc, (2) length ≤ 3.4 m, width ≤ 1.48 m, and height ≤ 2 m. If one of the conditions is violated, the car is classified as a standard-sized car.

5

2.1

Tax reduction

Car users in Japan pay various taxes at the car purchase and ownership stage.6 At the time of purchase, car users must pay 4.5% automobile acquisition tax7 in addition to 5% consumption tax. During ownership, consumers must pay tonnage tax and automobile tax. The tonnage tax depends on the car’s weight, 6,300 JPY (approximately 70 USD8 ) per 500 kg, whereas the automobile tax depends on the size of engine displacement; for example, the tax on cars with less than 1,000cc is 29,500 JPY, and the tax on cars with 1,000 to 1,500cc is 34,500 JPY. While acquisition tax must be paid once at the time of car purchase, the automobile and tonnage taxes must be paid on an annual basis during ownership. The tonnage tax is assessed at the time of new car purchase and car inspections, and car buyers must deposit the sum of the taxes until the next car inspection. For instance, since the first car inspection must be conducted three years after a new car is purchased, car users must deposit three years’ tax on the new car registration. The Japanese government has provided tax incentives for eco-cars. From fiscal year 2005 to 2009, these taxes were reduced or exempted for individuals who purchased cars that met a fuel economy and emission control standard. The tax incentives were slightly revised over time, as is summarized in Table 1. For example, the automobile tax was reduced by up to 50%, while the acquisition tax was reduced by up to 150,000 JPY for gas cars and by up to 44% for hybrid-vehicles in 2005.9 Major changes in policies occurred in 2009, when the tonnage tax became the target of tax reductions, and the amount of tax reduction was increased. After this amendment, the acquisition tax for hybrid vehicles was exempted, and the tax on other vehicles was reduced by up to 75% depending on the amount of emissions and the fuel economy of the car. The reductions in the automobile tax and the tonnage tax were not perpetual, but the automobile tax reduction was effective for one year, whereas the tonnage tax reduction was effective until the first car inspection; that is, for three years.

6

In addition to taxes due at these stages, car users are subject to gasoline taxes during car use. This study does not focus on the gasoline tax because this is beyond the scope of environmental policies. 7 The acquisition tax is 5% but this is imposed on the tax base of a car that is around 90% of the car price. Hence, real tax rate becomes 4.5%. 8 1 USD was approximately equal to 90 JPY at the end of 2009. 9 Table 1 shows that the automobile tax incentives took the form of a deduction. The amount of deduction was 300,000 JPY and, thus, the amount of the tax reduction was 15,000 JPY at its maximum because the acquisition tax rate was 5%. During the study period, the prices of all car models exceeded 300,000 JPY. Therefore, the amount of tax reduction should be 15,000 JPY for the car models that satisfy the criteria for eco-cars in Table 1.

6

2.2

Replacement subsidy

In addition to the tax reduction, the Japanese government passed the “Green” Vehicle Purchasing Promotion Measures that were in effect from April 10, 2009 for the purpose of eco-car diffusion. The program had two components: one for consumers replacing an older passenger car with a new eco-car (“replacement program”), and one for consumers purchasing a new eco-car without replacing an older car (“non-replacement program”). To qualify for the replacement program, consumers had to scrap a passenger car that was first registered 13 years ago or earlier. Under the replacement program, these consumers were eligible for 250,000 JPY for the purchase of a new car that complies with the 2010 fuel economy standards in Table 2. However, the non-replacement program imposed higher standards for the subsidy provision: the cars eligible for the subsidy must comply with ♠ in Table 1. The subsidy payment per unit was 100,000 JPY and, hence, much smaller than the payment under the replacement program. The policy was initially scheduled to end on March 31, 2010, or earlier if the total subsidy payments reached the budget amount for this policy. However, in December 2009, the government extended the policy until September 30, 2010 and expanded the budget from 360 billion JPY to 590 billion JPY. The policy ended slightly earlier than the scheduled period because of budget exhaustion.10 For an overall view of the changes in sales along with the policy changes, Table 3 shows the sales by firms from the year 2005 to the year 2009. Sales decreased substantially in 2008, which might have been caused by a depression from the financial crisis at that time. The set of policies introduced in 2009 appeared to contribute to sales recovery, but the effects on sales were not uniform across firms. Honda, Mitsubishi, and Toyota experienced a substantial increase in their sales, whereas the other firms experienced a smaller increase or decrease in sales. Hereafter, I investigate the effects of policy in greater detail based on a structural econometric model.

3

Model

This section introduces the structural econometric model of demand and supply to assess the role of environmental policies on a market outcome. I employ a discrete choice method to model consumer behavior and a multi-product oligopolistic competition model to back out marginal costs for each car model. 10

Note that this study investigates the policy until the end of fiscal year 2009; that is, before the termination. In addition, the budget is not only for the passenger cars for private use but also those for commercial cars and others.

7

3.1

Demand

The demand model is similar to Goldberg and Verboven (2001), who allows a consumer choice to depend on their income in a nested logit framework. In this study, I allow a consumer choice to depend on the age of the car that the consumer owns as well as their income to consider the scrap incentives. I consider a household as a unit that makes a car choice as is typical in the literature. Hence, a market size Mt represents the number of households in Japan at time t. Each unit chooses one alternative that gives the highest utility from #Jt + 1 alternatives: Jt is the set of car models offered at time t, and the additional alternative is an outside option representing the decision not to purchase in Jt . I focus on standard-sized cars that have an engine displacement of over 660cc. Therefore, Jt includes only standard-sized car models while the outside option includes the purchase of mini-vehicles and import cars. Consumer i’s utility obtained from alternative j at time t is specified as follows. uijt = vijt + ϵijt ,

(1)

Here, vijt represents the deterministic part of the utility obtained from product j, and ϵijt is a random part of the utility. The deterministic part of the outside option, vi0t , is normalized to be zero. I further decompose vijt into two parts; that is, vijt = δjt + µijt ,

(2)

where δjt is common to all consumers, while µijt varies across individuals. The common part of the utility is specified as follows. δjt = xjt β + ξjt ,

(3)

where xjt is a 1 × K vector of the characteristics of car j and β is a K × 1 vector of the parameters to be estimated. ξj represents the characteristics and demand shock specific to car j that are unobservable to researchers but observable to consumers and producers. µijt depends on individual characteristics; that is, household i’s income yit , the car’s age ait , and the age of the car that household i owns at the beginning of time t: µijt = −αit [(1.05 + T1jt )pjt − Sjt (ait ) + T2jt ],

(4)

where pjt is the price of car j at time t, and αit = yαit is the price sensitivity of consumer i. Under this setting, high-income consumers are less sensitive to car prices for positive α; that 8

is, the parameter to be estimated. T1jt is an acquisition tax that has to be paid at the time of purchase in addition to the 5% consumption tax. T2jt is the sum of tonnage and automobile taxes. Although these taxes are paid on an annual basis, I focus on the payment at the time of purchase. As is mentioned in Section 2, the automobile tax is a one-year payment, whereas the tonnage tax is the sum of a three-year payment. Sjt (ait ) is the subsidy for the purchase and replacement of low emission and fuel-efficient cars that came into effect in April 2009, which is expressed as a function of the age of a car that consumer i holds:    250, 000 if ait ≥ 13, t = 2009, and the car j meets fuel economy standards,   Sjt (ait ) = 100, 000 if ait < 13, t = 2009, and the car j meets ♠ in Table 1,    0 otherwise. (5) ϵijt represents taste heterogeneity for car models. I assume ϵijt to follow a generalized extreme value that allows the substitution pattern of the products to depend on the group to which the car belongs. I classify all car models into five groups: compact, sedan and wagon, minivan, sports utility vehicle (SUV), and specialty cars. Additionally, the outside option is defined as one group in the choice set. Under this framework, the probability of consumer i choosing a car j at time t can be decomposed into consumer i’s choice probability of the car j conditional on choosing a group g(j), sij|g(j) , and the probability of choosing group g(j), the group to which the car j belongs, sig(j) : sijt = sijt|g(j) sig(j)t .

(6)

The first term in the above equation is given by: sijt|g(j) = ∑

evijt /λ evijt /λ = , vilt /λ eIig(j)t l∈g(j) e 

where

Iig(j)t = ln 

∑

(7)

 evilt /λ  ,

(8)

l∈g(j)

which is a logit inclusive value; that is, the expected utility obtained from choosing group g(j). The second term in eq. (6) is given by sig(j)t =

evi0t

eλIig(j)t ∑ . + g∈G eλIig

9

(9)

where G = {compact, sedan & wagon, minivan, SU V, specialty}. Here, a deterministic part of utility obtained from the outside option vi0t is normalized to be zero. Then, the first term in the denominator becomes 1. λ is the distributional parameter of the nested logit and captures the pattern of dependency across products in the same group. To be consistent with random utility maximization, λ must lie between zero and one.(McFadden (1978)) Particularly, if λ = 1, the nested logit structure reduces to a logit model and, thus, the substitution pattern among products becomes independent of the groups to which the products belong. On the other hand, if λ is close to zero, the dependency across products in the same group becomes stronger and, at the extreme, the model reduces to the elimination by aspects model by Tversky (1972). In the estimation, I will test whether the estimate of λ lies in the range consistent with the utility maximization problem. To derive market share function sjt , I integrate the individual choice probabilities in eq. (6) over income yit and car age ait distribution: ∫ ∫ sjt =

sijt dPy (y)dPa (a), a

(10)

y

where Py (·) and Pa (·) represent the distributions of income and car age. I use the empirical distributions of income and car age to approximate the demographics of Japanese households. Constructing the empirical distribution, I assume that each household owns no more than one car. Then, the data on the number of cars by car age correspond to the car age distribution for the households that own their cars in Japan. Note that this assumption is clearly problematic because some households own multiple cars in Japan. However, Wakamori(2011) indicated that it is often the case that households tend to own a standard-sized car and a mini-vehicle rather than multiple standard-sized cars. Because this paper focuses only on standard-sized cars, the assumption is a reasonable approximation of the real situation.

3.2

Multi-product oligopolistic competition

The supply side of the model is specified under the assumption that all car manufacturers compete with respect to price. The variable profit function of firm f is πf t =

∑

[pjt qjt − cjt (qjt )] ,

(11)

j∈Jf t

where Jf t is the set of models produced by firm f , and cj (qjt ) is a cost function of product j.

10

Solving this profit maximization problem, I obtain the following first order condition for each car model j: mct = pt − ∆−1 (12) t st , where pt = (p1t , . . . , pJt ,t )′ , st = (s1t . . . . , sJt ,t )′ , and mct = (c′1t , . . . , c′Jt ,t )′ . c′jt are the first derivatives of cjt and the marginal cost of product j. I assume that the marginal costs are constant over quantity. ∆t is a #Jt × #Jt matrix, whose (j, r)-th element is ∆∗jrt × Hjrt : ∆∗jrt is an (j, r) element of #Jt × #Jt and the substitution matrix of the demand system; that is, ∫ ∫ [ ( ) ]   αi (1.05 + T1jt )sijt λ1 − 1−λ sijt/g(j) − sijt dPy (y)dPa (a) if j = r  λ a y [( ) ] ∂srt  ∫ ∫ = − a y αi (1.05 + T1jt )sirt 1−λ s + s dPy (y)dPa (a) if j ̸= r, r ∈ g(j) ijt ijt/g(j) λ ∂pjt   ∫ ∫  − α (1.05 + T1jt )sirt sijt dPy (y)dPa (a) if j = ̸ r, r ̸∈ g(j), a y i (13) and, under the price competition assumption, Hjrt takes a value of one if both j and r are produced by the same firm, and zero otherwise. Note that ∆t can be computed from the demand estimates. Hence, the (unobserved) marginal cost vector can be backed out from eq. (12). In the simulation analyses, I use the demand estimates and the uncovered marginal costs to obtain counterfactual outcome.

4

Estimation

Estimation of the model is based on a moment assumption on ξjt that represents the unobserved demand shock and characteristics. A problem here is that ξjt should be correlated with pjt because the positive unobservable characteristics or demand shocks induce higher prices. To resolve the problem, I use the set of instruments obtained from the following moment condition proposed in Berry, Levinsohn, and Pakes (1995) (hereafter, BLP): E[ξjt |x1t , . . . , x#Jt t ] = 0 for all j. This assumption is justified if firms specify the observed characteristics of their car models before realizing ξjt , although there are some concerns on the assumption as is discussed in the literature of demand estimation, e.g. Ackerberg, Benkard, Berry, and Pakes (2007). Given the identification assumption, the characteristics of all other products are valid instruments for prices because the pricing of each car model depends on the location of the model in the characteristics space: if the characteristics of the model are located in a crowded area in the characteristics space, the markup should be smaller, and vice versa.11 Based on this assumption, I use a set of instruments similar to that 11

See Bresnahan, Stern, and Trajtenberg (1997) for a discussion on the validity of the instruments.

11

of Goldberg and Verboven (2001): for j ∈ Jf t , (1) the sum of the characteristic k of other ∑ products belonging to the same group, r∈{g(j)\j} xrkt , (2) the sum of the characteristic k of ∑ products belonging to other groups, r∈{Jt \g(j)} xrkt , and (3) the sum of the characteristic k ∑ of products belonging to the same group and made by the same firm, r∈{Jf t ∩g(j)} xrkt . Based on the set of instruments, I implement the two-step efficient generalized GMM estimation of Hansen (1982). The estimation incorporates the derivation of the mean utility for each car model δjt by contraction mapping, given the parameters in µijt and the GEV distribution function, that is, (α, λ). The contraction mapping is slightly different from the BLP contraction mapping because ϵijt does not follow type I extreme value, but follows GEV that generates the two-stage nested logit structure. The following series meets the condition of contraction mapping: h+1 h h ; α, λ) − ln sajt ], δjt = δjt + λ[ln sjt (δjt

(14)

h where h indicates the number of iterations, sjt (δjt ; α, λ) represents the market share coma puted from the introduced model, and sjt represents the actual market share in the data.12

4.1

Data

The dataset used in this paper includes data for fiscal years 2005 to 2009. As is mentioned, this paper focuses only on Japanese standard-sized car models (>660cc). I constructed the dataset based on several independent sources. Price and characteristics data are obtained from a car magazine, Saishin Kokusan & Yunyuu-sha Konyuu Guide (Current Domestic and Import Cars Purchase Guide) published by JAF Publishing Company Limited (JAF). Quantity data are obtained from Jidousha Touroku Tokei Jouhou: Shinsha-hen (New Car Registration Statistics) published monthly by the Japan Automobile Dealers Association. The price and quantity data are available at the variant level, and the quantity data is available at the model (nameplate) level. The standard practice in constructing the database is to use the base variant price and characteristics for each model, as in Berry, Levinsohn, and Pakes (1999). However, the base variant matching might be problematic in this study because the subsidy target and the amount of tax exemption vary across variants within a model. In this paper, I use “recommended variant” for each model used by JAF in matching the price (characteristics) and the quantity data. This is because magazine experts are likely to choose the recommendation given the information on environmental policies. I collect the information on the individual characteristics distribution: the number of cars 12

See Iizuka (2007) for the proof. See also Davis and Schiraldi (2014), who show the contraction mapping methods for a broader class of GEV models.

12

by car age from Sho-do Touroku Nen-betsu Jidousha Hoyuu Sharyou-suu Toukei (Number of Vehicle Holdings by First Registration Years) published annually by the Automobile Inspection and Registration Information Association, and income distribution from Kokumin Seikatsu Kiso Chosa (Comprehensive Survey of Living Conditions of the People on Health and Welfare) released annually by the Ministry of Health, Labor, and Welfare. Table 4 shows the summary statistics for the estimation variables. I include fuel cost, which is the gasoline price divided by the fuel economy, rather than fuel economy, in the estimation to account for the variation in gasoline price over time. The summary statistics for fuel economy are included to explain how the firms adjust fuel economy for their models. As shown, the average fuel economy increased by approximately 6% in 2009, and firms continued to strive to improve fuel economies. I show the amount of automobile tax reductions to see the changes in the environmental policies. Section 2 explains that tax incentives for automobile tax had been in effect since before 2005; these incentives are an effective measure to capture policy evolution. Although the maximum amount of automobile tax reduction increased in 2009, the average reduction was not significantly increased because of changes in the fuel economy standard for the automobile tax reduction from 2008, as shown in Table 1.

4.2

Estimation results

In addition to the model introduced in Section 3, I also implement standard demand estimation based on a standard nested logit framework as in Berry (1994). The estimation equation is as follows: ln(sjt ) − ln(s0t ) = −αpjt − (1 − λ) ln(sjt|g(j) ) + δjt = −αpjt − (1 − λ) ln(sjt|g(j) ) + xjt β + ξjt

.

(15)

For the estimation of the standard nested logit, I do not directly include the volumes of tax reduction and subsidies but include a dummy variable for the eco-car xjt , which takes a value of one if the model is the target of the tax reduction and/or the subsidy, to see the effects of the environmental policies. I first implement the estimation of the standard nested logit using OLS and GMM to see whether the bias in parameter estimates are corrected by instrumenting. There are two endogenous variables, price pjt and the conditional share ln(sjt|g(j) ). The price and unobserved characteristics are positively correlated; therefore, the price coefficient should be upwardly biased. In addition, positive (negative) ξjt induces a higher share within the group to which product j belongs: given a positive correlation between ξjt and sjt|g(j) , the coefficient of ln(sjt|g(j) ) should also be upwardly biased, which indicates a downward bias 13

in the estimate of λ. As shown in Table 5 (i) and (ii), the price coefficients −α and λ get lower after instrumenting. The coefficients on the car characteristics are also reasonably estimated; for example, the coefficient on car size and fuel cost takes a positive and negative value, respectively. In particular, the coefficient on the eco-car dummy variable is positive, which suggests that the environmental policies have a positive impact on car demand. The next section presents the results of the estimation with individual characteristics, income and car age. The estimation result is summarized in Table 5 (iii). As expected, the price coefficient −α is negative and significant. λ is positive, less than one, and statistically significant, which implies that the model is consistent with the random utility maximization problem. In addition, λ less than one indicates that the model is not the logit: the substitution pattern among products depends on the group to which the products belong. The coefficients on car characteristics also have a reasonable sign; for example, the coefficient on fuel cost is negative and significant.

5

Simulation

Based on the estimates, this section shows the results of counterfactual simulations for the assessment of the effects of the environmental policies in Japan. I first implement the simulation to reveal what would happen in the absence of the environmental policies; that is, the tax incentives and the subsidy. Next, focusing on the subsidy that was the subject of debate between Japan and the US, I investigate the efficiency of the actual subsidy provision rule from an environmental perspective. To do this, I implement some counterfactual simulations to obtain outcomes under alternative subsidy qualification rules and then search for the efficient subsidy provision rule based on the CEA. Finally, I assess the validity of the US critique on the subsidy qualification rule in Japan.

5.1

Effects of the environmental policies on the Japanese car market

This section shows the effects of the environmental policies on the market outcome. To do this, I derive a counterfactual situation in the absence of environmental policies by simulation based on the estimation results from the previous section. I focus on the sales and variable profits for each firm and the average fuel economy of new cars sold as a measure of the environmental goal. Average fuel economy might be an inadequate target in assessing the environmental policies because an increase in the average fuel economy does not always lead

14

to environmental improvements, such as lower CO2 emissions.13 Despite the problem with the measurement, it is sometimes employed as a goal of environmental policy, such as the Corporate Average Fuel Economy (CAFE) standard in the US. Because this paper focuses on the dispute between Japan and the US, average fuel economy is a reasonable choice in analyzing the disguised protection problem. Table 6 reports the effects on firms’ profits and sales. From the year 2005 to the year 2008, the period when only tax incentives for acquisition tax and automobile tax were applied, the effects of the policies were negligible; the policies increased sales and profits by approximately 2% in total. However, the introduction of the subsidy and the expansion of the tax incentives had significant effects on the market, and sales and profits increased by 20.7% and 16.9%, respectively. The effects are not uniform across firms. Honda earned substantial profits, whereas Suzuki and Subaru experienced limited benefit or loss as a result of the policies. The losses experienced by Subaru are because of the tax reduction and subsidy provisions on car models that represent close substitutes for Subaru models. Table 7 shows the impacts of the contribution of tax incentives and the subsidy focusing on the policies in 2009. The table shows that each of the policies does not explain the overall impacts on sales and profits; both policies were equally important in stimulating firms’ performance. Table 8 reports the effects on average fuel economy. Similar to the impacts on sales and profits, the policies from the year 2005 to the year 2008 contributed minimally to improvements in average fuel economy. However, the policies introduced in 2009 had some impact on average fuel economy.

5.2

Alternative subsidy qualification rules

This paper studies whether the environmental policies introduced in Japan were effective in achieving environmental goals. To investigate, I implement some counterfactual simulations to assess what would happen if alternative subsidy provision rules were introduced. To do this, I focus on the two parameters in the policy: the fuel economy standard for subsidy provision and the amount of subsidy per unit. The qualification for the subsidy was based on the fuel economy standard shown in Table 2. In the counterfactual analyses, I consider situations where the level of fuel economies for each weight category is 0.5 to 2.5 times as large as the original fuel economy. Analogous to the fuel economy standard, I consider situations 13

Environmental policies, such as subsidies in Japan, do not always decrease CO2 emissions although they induce higher average fuel economy in new cars sold for several reasons. First, the policies may induce a higher car ownership rate than the replacement of existing cars. Second, the effects depend on the pattern of replacement: the policies might be ineffective if the replacements are conducted by owners of fuel-efficient cars. Third, the miles driven depend on the fuel efficiency of the cars; therefore, an increase in the sales of fuel-efficient cars might increase miles driven, which increase CO2 emissions.

15

where the amount of subsidy, 100,000 or 250,000 JPY is specified in eq. (5) and is changed by a factor of 0.1 to five times. Hereafter, I denote the scale factors for the level of fuel economy and the amount of the subsidy as γf e and γsub , respectively. In the counterfactual simulations, I discretize γf e and γsub with step size 0.1. Given this discretization, I derive outcomes for all possible combinations of γf e and γsub . Potentially, there are many candidates for parameters for the counterfactual experiments. For example, because fuel economy standards depend on car weights, as shown in Table 2, the subsidy rules based on alternative weight categories might be interesting to assess. However, because car weights can be adjusted relatively easily, firms are likely to adjust the car weights of their models in response to changes in the weight categories. Therefore, I have to introduce a model that allows firms to choose car weights endogenously to make a policy assessment if the alternatives with different weight categories are considered. To avoid this complexity, I fix the weight categories.14 All the simulation results are reported in Appendix A. In the following section, I implement CEA based on the results of the counterfactual simulation.

5.3

Cost-Effectiveness analyses

Based on the simulation results, I assess whether the subsidy for eco-car purchase introduced in Japan was designed to achieve the environmental goal effectively. The assessment here is based on a CEA, where the cost is the sum of the subsidy payment and the effect is the average fuel economy of new cars sold. In particular, I focus on the alternative that achieves the highest fuel economy given a budget constraint.15 The budget constraint is the sum of the subsidy under the actual policy, which equals 253.3 billion JPY.16 Note that the sum of payment is smaller than the amount of subsidy budget shown in Section 2. This is because the periods of this study does not cover the whole subsidy period and the subsidy was not only for the passenger cars for private use that I investigate in this paper. To implement the analysis, I use all the combinations of γf e and γsub where the amount of subsidy payments is below the budget. Within these combinations, Table 9 shows the value of γsub that achieves the maximum average fuel economy for each γf e and the level of the average fuel economy at the maximum. The table shows that the rules for subsidy provision 14

Klier and Linn (2012) construct a structural econometric model of endogenous weight choice to assess the effects of the US CAFE regulation in the medium run. 15 See Boardman, Greenberg, Vining, and Weimer (2010) for details of the CEA. 16 Note that welfare is clearly another candidate of the effects. However, in this application, the welfare is not appropriate, or, the cost-benefit analysis is not an appropriate approach in assessing the policy issues concerning disguised protection. This is because the policy choice based on welfare might support the subsidy rule that gives domestic firms a competitive edge.

16

under the actual policy do not achieve the highest average fuel economy, but the combination (γf e , γsub ) = (1.6, 2.8) is the most effective. The result indicates that the government should provide greater subsidies with a narrower model range. The difference between the actual and the optimal is not negligible: by setting the optimal policy, average fuel economy would increase by 1.45 km/l, or 7.52%. The amount of subsidy is different across the values of γf e . Because the effects on fuel economy depend on the amount of the subsidy, I also focus on the cost-effectiveness ratio to assess the alternative subsidy provision rules. The fifth column in Table 9 reports the results. The most effective rule is the combination (γf e , γsub ) = (1.9, 3.2). Although this is different from the combination in the previous analysis, these results indicate that the actual rules for the fuel economy standard were too lax to achieve the environmental goal effectively. Hereafter, I treat the combination (γf e , γsub ) = (1.6, 2.8) as the optimal rule from the environmental perspective and compare the outcomes between the optimal and actual rules to assess the disguised protection problem. Next, I analyze how the lax standard affected the Japanese firms’ profits. The sixth column in Table 9 shows the total variable profits for every combination of γf e and γsub . Total variable profit is greater under the actual policy than under the optimal, which implies that the policy might be designed to provide more benefit for Japanese firms rather than to achieve the environmental goal. Additionally, the profit under the actual policy is greatest among the values of γf e in the table. Similar to the discussion for the CEA, the greater profits under the actual policy might be caused by the larger subsidy payment rather than the distortion of the policy. Particularly, the subsidy payment under the actual rule is always greater than or equal to the subsidy payment under the alternatives because the budget constraint is the subsidy payment under the actual. To account for the differences in the total amount of subsidy, I focus on the ratio between the subsidy payment and profit. Given the ratio, I assess whether or not the actual policy was the most effective way to improve domestic firms’ profits. The last column in Table 9 shows the results. The ratio under the actual (0.598) is still larger than the ratio under the optimal (0.534). This indicates that the Japanese government might distort the environmental policy for the purpose of domestic firms’ profits. The profit-subsidy ratio under the actual is not the largest, but the combination (γf e , γsub ) = (0.8, 0.7) achieves the maximum profits. The actual rule is close to the rule that maximizes the firms’ profits. To explain the effects on profits, I compute the effects on the variable profits under the actual and the optimal according to firms. Table 10 shows that the benefits from the subsidy under the optimal rule are biased toward some firms. Honda and Toyota could earn substantially, whereas the other firms could earn only a little or lose profit because of 17

the subsidy provision. Additionally, compared to the optimal policy, the actual policy was not more beneficial to all firms; only Toyota lost profits because of the introduction of the actual policy rule instead of the optimal. This indicates that the government might avoid the targets of the subsidy biased toward Toyota’s models, and tried to equalize the effects of the policy among the Japanese firms. In summary, the results imply the possible distortion of the policy. Finally, I assess the primary question: Did the subsidy policy constitute a disguised form of protection? The distortion of the subsidy provision rule does not always indicate that the policy represents a disguised protection policy. This is because the distortion of the policy could be beneficial to foreign firms as well as Japanese firms. I cannot directly analyze the effect of the distortion on foreign firms’ profits because the data on import cars are not available. However, it is possible to infer the effects on foreign firms’ profits, particularly US firms’ profits, based on the request made by the US government upon its critique of the subsidy policy. The US requested laxer fuel economy standards to expand the target car models that qualified for the subsidy provision. This request implies that US firms could earn more from a relaxed standard; however, as is shown, this is not valid from the environmental perspective because the subsidy provision should be based on a higher fuel economy standard. Instead, the lax standard in the actual policy might also be favorable to the US firms because no US model qualified for the subsidy under the tighter standard, that is, the optimal.17

6

Conclusion

Tariff barriers have been dropped substantially through negotiations under GATT and the subsequent WTO system. There is also emerging concern that domestic policies could be used as secondary trade barriers. Although the WTO system prohibits the use of domestic policies in this way, whether a particular policy is designed to protect a domestic industry is not easy to prove. This paper demonstrated the assessment of disguised protection for the case of the subsidy policy for eco-cars in Japan. Based on the estimates of the structural econometric model for demand and supply, I implemented counterfactual simulations to derive the outcomes under alternative subsidy qualification rules and conducted a CEA to assess whether the policy was designed to achieve the environmental goal effectively. This study found that the actual fuel 17

Note that the tighter standard may increase US firms’ profits if the alternative standard effectively excludes competing Japanese car models from the subsidy target and increases the demand for US car models through substitution effects. However, because the substitution effects are usually smaller than the direct effects of the subsidy, it is unlikely that US firms could benefit financially from the optimal standard.

18

economy standard was too lax from an environmental perspective, and the subsidy targets should be narrowed by setting a higher fuel economy standard with an increase in the amount of subsidy per unit. In addition, Japanese firms enjoyed higher profits thanks to the lax fuel economy standard, which implies a possible case of disguised protection. However, policy amendments for achieving the environmental goal are unlikely to meet US firms’ interests because no US car models qualified for the subsidy under the optimal fuel economy standard. Instead, the distortion was likely to be beneficial to the US firms.

19

References Ackerberg, Daniel, C. Lanier Benkard, Steven Berry, and Ariel Pakes. 2007. “Econometric tools for analyzing market outcomes.” In Handbook of Econometrics, vol. IIIA, edited by James J. Heckman and Edward E. Leamer. North Holland. Adda, Jerome and Russell Cooper. 2000. “Balladurette and Juppette: a discrete analysis of scrapping subsidies.” Journal of Political Economy 108 (4):778–806. Barrett, Scott. 1994. “Strategic environmental policy and international trade.” Journal of Public Economics 54:325–338. Bento, Antonio M., Lawrence H. Goulder, Mark R. Jacobsen, and Roger H. von Haefen. 2009. “Distributional and efficiency impacts of increased US gasoline taxes.” American Economic Review 99 (3):667–699. Beresteanu, Arie and Shanjun Li. 2011. “Gasoline prices, government support, and the demand for hybrid vehicles in the United States.” International Economic Review 52 (1):161– 182. Berry, Steven. 1994. “Estimating discrete-choice models of product differentiation.” RAND Journal of Economics 25 (2):242–262. Berry, Steven, James Levinsohn, and Ariel Pakes. 1995. “Automobile prices in market equilibrium.” Econometrica 63 (4):841–890. ———. 1999. “Voluntary export restraints on automobiles: evaluating the trade policy.” American Economic Review 89 (3):400–430. Boardman, Anthony, David Greenberg, Aidan Vining, and David Weimer. 2010. Cost-benefit analysis: concepts and practice. Prentice Hall, 4th edition ed. Bresnahan, Timothy F., Scott Stern, and Manuel Trajtenberg. 1997. “Market segmentation and the sources of rents from innovation: personal computers in the late 1980s.” RAND Journal of Economics 28:17–44. Conrad, Klaus. 1993. “Taxes and subsidies for pollution intensive industries as trade policy.” Journal of Environmental Economics and Management 25:121–135. Copeland, Brian R. 1990. “Strategic interaction among nations: negotiable and nonnegotiable trade barriers.” Canadian Journal of Economics 23 (1):84–108.

20

Davis, Peter and Pasquale Schiraldi. 2014. “The flexible coefficient multinomial logit (FCMNL) model of demand for differentiated products.” RAND Journal of Economics 45 (1):32–63. Ederington, Josh. 2001. “International coordination of trade and domestic policies.” American Economic Review 91 (5):1580–1593. ———. 2002. “Trade and domestic policy linkage in international agreement.” International Economic Review 43 (4):1347–1367. Goldberg, Pinelopi Koujianou. 1998. “The effects of the corporate average fuel economy standards in the automobile Industry.” Journal of Industrial Economics 46:1–33. Goldberg, Pinelopi Koujianou and Frank Verboven. 2001. “The evolution of price dispersion in the European car market.” Review of Economic Studies 68:811–848. Grossman, Gene M. 1986. “Imports as a cause of injury: the case of the US steel industry.” Journal of International Economics 20:68–105. Hansen, Lars Peter. 1982. “Large sample properties of generalized method of moments estimators.” Econometrica 50 (4):1029–1054. Iizuka, Toshiaki. 2007. “Experts’ agency problems: evidence from the prescription drug in Japan.” RAND Journal of Economics 38 (3):844–862. Kelly, Kenneth. 1988. “The analysis of causality in escape clause cases.” Journal of Industrial Economics 37 (4):187–207. Kennedy, Peter W. 1994. “Equilibrium pollution taxes in open economies with imperfect competition.” Journal of Environmental Economics and Management 27:49–63. Kitano, Taiju and Hiroshi Ohashi. 2009. “Did US safeguards resuscitate Harley-Davidson in the 1980s?” Journal of International Economics 79:186–197. Klier, Thomas and Joshua Linn. 2012. “New-vehicle characteristics and the cost of the Corporate Average Fuel Economy standard.” RAND Journal of Economics 43 (1):186– 213. Markusen, James R. 1975. “International externalities and optimal tax structures.” Journal of International Economics 5:15–29.

21

McFadden, Daniel. 1978. “Modelling the choice of residential location.” In Spatial Interaction Theory and Planning Models, edited by Anders Karlqvist, Lars Lundqvist, Folke Snickars, and Jorgen W. Weibull. Amsterdam: North-Holland, 75–96. Parry, Ian W. H., Margaret Walls, and Winston Harrington. 2007. “Automobile externalities and policies.” Journal of Economic Literature 45:373–399. Schiraldi, Pasquale. 2011. “Automobile replacement: a dynamic structural approach.” RAND Journal of Economics 42 (2):266–291. Tversky, Amos. 1972. “Elimination by aspects: a theory of choice.” Psychological Review 79:281–299.

22

23

Emissions down by 50% from 2005 standards

Emissions down by 75% from 2005 standards

Emissions down by 75% from 2005 standards

Emissions down by 75% from 2005 standards

Emissions down by 75% from 2005 standards

2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 75% reduction Tonnage Tax: 75% reduction

2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 300,000 JPY deductible Tonnage Tax: None

2010 target fuel economy standards +20% Automobile Tax: 50% reduction Acquisition Tax: 300,000 JPY deductible Tonnage Tax: None

2010 target fuel economy standards +20% Automobile Tax: 50% reduction Acquisition Tax: 300,000 JPY deductible Tonnage Tax: None

2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 36% reduction Tonnage Tax: None ♠2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 100% reduction Tonnage Tax: 100% reduction

♠2010 target fuel economy standards +15% Automobile Tax: 25% reduction Acquisition Tax: 50% reduction Tonnage Tax: 50% reduction

2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 40% reduction Tonnage Tax: None

2010 target fuel economy standards +25% Automobile Tax: 50% reduction Acquisition Tax: 44% reduction Tonnage Tax: None

Hybrid-Vehicle 2010 target fuel economy standards +25% or better Automobile Tax: 50% reduction Acquisition Tax: 44% reduction Tonnage Tax: None -

2010 target fuel economy standards +15% Automobile Tax: 25% reduction Acquisition Tax: 150,000 JPY deductible Tonnage Tax: None

2010 target fuel economy standards +10% Automobile Tax: 25% reduction Acquisition Tax: 150,000 JPY deductible Tonnage Tax: None

2010 target fuel economy standards +10% Automobile Tax: 25% reduction Acquisition Tax: 150,000 JPY deductible Tonnage Tax: None

Note: All the consumers can obtain subsidies when purchasing the cars that satisfy ♠.

Environmental Performance Certification

FY2009

Environmental Performance Certification

FY2008

Environmental Performance Certification

FY2007

Environmental Performance Certification

FY2006

Environmental Performance Certification

Emissions down by 75% from 2005 standards

FY2005

Fuel Economy Certification Gas-Vehicle 2010 target fuel economy 2010 target fuel economy standards +5% or better standards or better Automobile Tax: 50% reduction Automobile Tax: 25% reduction Acquisition Tax: 300,000 JPY deductible Acquisition Tax: 200,000 JPY deductible Tonnage Tax: None Tonnage Tax: None Automobile Tax: 25% reduction Acquisition Tax: 200,000 JPY deductible Tonnage Tax: None

Table 1: Tax incentives, fiscal year 2005 to 2009

24

– 703 kg 703 – 828 kg 828 – 1,016 kg 1,016–1,266 kg 1,266–1,516 kg 1,516–1,766 kg 1,766–2,016 kg 2,016–2,266 kg 2,266 kg–

Weight

Fuel Economy Standards+10% 23.3 20.7 19.7 17.6 14.3 11.6 9.8 8.6 7.0

Fuel Economy Standards+15% 24.4 21.6 20.6 18.4 15.0 12.1 10.2 9.0 7.4

Table 2: 2010 target fuel economy standards (km/l)

Fuel Economy Fuel Economy Standards Standards+5% 21.2 22.3 18.8 19.7 17.9 18.8 16.0 16.8 13.0 13.7 10.5 11.0 8.9 9.3 7.8 8.2 6.4 6.7

Fuel Economy Standards+20% 25.4 22.6 21.5 19.2 15.6 12.6 10.7 9.4 7.7

Fuel Economy Standards+25% 26.5 23.5 22.4 20.0 16.3 13.1 11.1 9.8 8.0

25

Total

Daihatsu Honda Mazda Mitsubishi Nissan Subaru Suzuki Toyota

3,055,793

100

2005 Sales (units) Share (%) 12,960 0.4 458,592 15.0 192,637 6.3 78,127 2.6 615,506 20.1 105,650 3.5 79,648 2.6 1,512,673 49.5 2,761,394

100

2006 Sales (units) Share (%) 19,771 0.7 399,902 14.5 171,779 6.2 68,657 2.5 504,746 18.3 87,937 3.2 84,560 3.1 1,424,042 51.6 2,689,401

100

2007 Sales (units) Share (%) 9,971 0.4 409,512 15.2 170,390 6.3 77,659 2.9 471,281 17.5 87,196 3.2 85,851 3.2 1,377,541 51.2

2,325,228

100

2008 Sales (units) Share (%) 7,098 0.3 387,163 16.7 141,311 6.1 49,862 2.1 402,960 17.3 77,353 3.3 82,143 3.5 1,177,338 50.6

Table 3: Sales and shares (within Japanese firms’ sales), 2005 to 2009

2,716,434

100

2009 Sales (units) Share (%) 7,066 0.3 499,601 18.4 153,059 5.6 62,232 2.3 432,620 15.9 80,542 3.0 61,392 2.3 1,419,922 52.3

26

Mean

Num. Obs.

Sales 23344 Price (million JPY) 2.498 3 Interior Space = Length*Width*Height (m )) 3.840 HP (ps)/Weight (kg) 0.082 Engine Displacement (1,000 cc) 2.113 Wheelbase (m) 2.666 Fuel Cost = Gasoline Price (JPY)/Fuel Economy(km/l) 10.62 Fuel Economy (km/l) 14.05 Automobile Tax Reduction (%) 15.36

Variables

464

28138 1.389 1.161 0.025 0.816 0.185 2.98 4.18 20.05

Max

531 152185 0.983 9.671 1.139 9.351 0.044 0.203 0.995 4.968 2.000 3.750 3.61 21.90 6.60 35.50 0.00 50.00

Before FY 2009 Std. Dev. Min

Table 4: Summary statistics

24039 2.579 3.987 0.079 2.104 2.681 9.16 14.88 16.15

Mean

113

38537 1.371 1.199 0.022 0.894 0.208 2.62 4.54 24.98

Max 512 277485 1.004 9.739 1.139 9.351 0.044 0.166 0.996 6.400 2.000 3.750 3.54 19.02 6.60 35.50 0.00 100.0

FY 2009 Std. Dev. Min

Table 5: Estimation results

Linear Model Variables Car Space HP/Weight Engine Displacement (cc) Wheelbase Fuel Cost Eco-car dummy Constant −α λ

(i) OLS Coef. S.E. 0.377 -0.855 0.204 0.598 -0.220 0.254 -7.580

0.047 2.600 0.094 0.234 0.018 0.072 0.553

Non-Linear Model (iii) GMM Coef. S.E.

(ii) GMM Coef. S.E. *** ** ** *** *** ***

-0.001 0.001 * 0.597 0.025 ***

0.299 0.058 8.483 3.407 0.266 0.157 0.819 0.340 -0.191 0.023 0.762 0.104 -11.051 0.852 -0.005 0.892

*** ** * ** *** *** ***

0.348 0.059 *** 14.934 3.843 *** 1.476 0.420 *** 0.830 0.172 *** -0.187 0.029 *** -10.571 0.767 ***

0.002 *** 0.040 ***

-16.289 3.884 *** 0.808 0.014 ***

Note: ***, **, and * indicate statistical significance at 1, 5 and 10% level, respectively.

27

28

60,465

Daihatsu Honda Mazda Mitsubishi Nissan Subaru Suzuki Toyota

Total

Total

39.26

2005 ∆π (billion JPY) Daihatsu 0.20 Honda 8.51 Mazda 2.47 Mitsubishi 0.48 Nissan 5.92 Subaru -0.63 Suzuki -0.40 Toyota 22.72

2005 ∆Sales (units) 443 14,662 4,956 979 11,044 -770 -811 29,961 27,438

17.49

0.98

(%) 0.29 1.73 0.46 1.30 0.18 -0.46 -0.22 1.23

1.00

(%) 0.42 1.99 0.37 1.41 0.32 -0.41 -0.30 1.21

24.20

2007 ∆π (billion JPY) 0.09 6.71 0.36 0.54 6.28 -0.41 -0.14 10.76

42,170

2007 ∆Sales (units) 225 13,128 598 1,105 11,967 -686 -338 16,172 33,775

2008 ∆Sales (units) 94 13,064 1,529 524 854 -308 -479 18,497

1.40

20.66

2008 (%) ∆π (billion JPY) 1.84 0.04 2.95 6.52 0.39 0.69 1.15 0.26 2.25 0.10 -0.81 -0.21 -0.33 -0.22 1.08 13.48

1.59

(%) 2.31 3.31 0.35 1.44 2.61 -0.78 -0.39 1.19

Table 6: Effects of environmental policies, 2005 to 2009

2.00

2006 (%) ∆π (billion JPY) 3.39 0.03 3.23 3.88 2.32 0.43 1.06 0.53 1.61 0.54 -0.81 -0.27 -1.01 -0.09 2.15 12.44

2.02

(%) 3.54 3.30 2.64 1.27 1.83 -0.72 -1.01 2.02

2006 ∆Sales (units) 83 7,812 625 956 1,624 -365 -256 16,959

465,474

1.38

257.82

2009 (%) ∆π (billion JPY) 1.18 0.49 3.16 66.03 0.92 7.33 0.88 5.72 0.04 28.28 -0.47 -1.02 -0.56 -0.10 1.57 151.10

1.47

(%) 1.34 3.49 1.09 1.06 0.21 -0.40 -0.58 1.60

2009 ∆Sales (units) 1,573 137,874 20,257 11,782 61,727 -34 1,251 231,044

29 142,089

Daihatsu Honda Mazda Mitsubishi Nissan Subaru Suzuki Toyota

Total

6.31

151.47

Tax incentives only (%) ∆π (billion JPY) 3.66 0.41 13.67 37.57 1.64 6.26 6.76 3.56 4.25 19.17 -2.63 0.45 -2.01 0.49 6.26 83.57 9.31

465,474

20.68

257.82

(ii) Tax incentives and subsidy (%) ∆Sales (units) (%) ∆π (billion JPY) 15.68 1,573 28.64 0.49 15.90 137,874 38.12 66.03 8.64 20,257 15.25 7.33 10.29 11,782 23.35 5.72 8.14 61,727 16.64 28.28 0.95 -34 -0.04 -1.02 1.73 1,251 2.08 -0.10 8.61 231,044 19.43 151.10

Note: (%) indicates the rate of change in profits in the case of no subsidy.

(i) ∆Sales (units) 201 49,462 2,178 3,411 15,760 -2,119 -1,210 74,406

Table 7: Decomposition of the effects of tax incentives and subsidy in 2009

16.94

(%) 19.64 31.77 10.26 17.62 12.49 -2.09 -0.36 16.73

Table 8: Effects on average fuel economy

2005 2006 2007 2008 2009

Actual(km/l) Counterfactual(km/l) Difference 15.686 15.646 0.041 15.938 15.868 0.070 16.414 16.335 0.079 17.054 16.964 0.090 19.234 18.566 0.668

30

Rate of Change (%) 0.259 0.440 0.483 0.531 3.600

31

γsub 0.7 0.7 0.7 0.7 0.8 1.0 1.3 1.7 2.0 2.4 2.4 2.8 3.0 3.0 3.2 4.3 4.3 4.4

Fuel Economy (km/l) 18.985 18.985 18.985 18.988 19.042 19.234 19.462 19.900 20.143 20.433 20.432 20.680 20.647 20.647 20.648 20.254 20.254 20.338

Subsidy (billion JPY) 221.7 221.7 221.6 221.0 246.4 253.3 237.9 246.3 243.5 245.3 244.8 252.2 247.3 247.3 244.9 241.4 241.4 242.6

∆F E/Subsidy 0.397 0.397 0.399 0.411 0.588 1.332 2.376 4.074 5.119 6.261 6.273 7.072 7.078 7.078 7.149 5.624 5.624 5.942

∆π (billion JPY) 137.53 137.53 137.48 137.16 150.89 151.47 140.92 135.63 135.74 129.21 129.00 134.57 133.52 133.52 138.21 101.38 101.38 97.12

∆π/Subsidy 0.620 0.620 0.620 0.621 0.612 0.598 0.592 0.551 0.558 0.527 0.527 0.534 0.540 0.540 0.564 0.420 0.420 0.400

Notes: ∆F E is the difference in the fuel economy with no subsidy, 18.897 km/l. γsub is the value that achieves the highest average fuel economy given γf e . The highest fuel economy is achieved under (γf e , γsub ) = (1.6, 2.8). The results for γf e ≥ 2.3 are not shown because the subsidy payments are below the budget constraint for all γsub ∈ [0, 5.0]. See the tables in the Appendix.

γf e 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2

Table 9: Simulation results for the cost-effectiveness analysis

Table 10: Variable profits under the actual and optimal rule for subsidy provision

(i) Actual rule (ii) Optimal rule (iii) (i) - (ii) ∆π (billion JPY) (%) ∆π (billion JPY) (%) ∆π (billion JPY) Daihatsu 0.49 19.64 0.04 1.58 0.45 Honda 66.03 31.77 61.57 29.63 4.46 Mazda 7.33 10.26 -1.75 -2.46 9.08 Mitsubishi 5.72 17.62 1.09 3.36 4.63 Nissan 28.28 12.49 0.14 0.06 28.14 Subaru -1.02 -2.09 -4.47 -9.13 3.44 Suzuki -0.10 -0.36 -1.12 -3.87 1.01 Toyota 151.10 16.73 185.41 20.53 -34.31 Total

257.82

16.94

240.91

15.83

Note: (%) indicates the rate of change in profits in the case of no subsidy.

32

16.90

A

All simulation results

The CEA shown in Section 5 is based on the counterfactual outcomes under the different subsidy qualification rules that are characterized by the factors on the fuel economy standard, γf e , and the subsidy per unit, γsub . This section reports all the simulation results for γf e ∈ {0.5, 0.6, . . . , 2.5} and γsub ∈ {0.0, 0.1, . . . , 5.0}. The outcomes shown in Table 11 and Table 12 are the average fuel economy and the total amount of subsidy that are central to CEA implementation. I choose the subsidy qualification rule that maximizes average fuel economy under the budget constraint. The budget constraint is the subsidy payment under the actual; that is, (γf e , γsub ) = (1.0, 1.0), which is 253.3 billion JPY as shown in Table 12. Additionally, the average fuel economy without subsidy can be found in Table 11, which is 18.897 km/l.

33

34

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

0.0

19.559 19.559 19.560 19.568 19.682 20.122 20.535 21.149 21.350 21.401 21.401 21.252 21.012 21.012 20.813 19.701 19.701 19.698 18.898 18.898 18.897

3.4

19.130 19.130 19.130 19.136 19.213 19.474 19.650 19.900 19.925 19.890 19.889 19.836 19.717 19.717 19.655 19.149 19.149 19.145 18.897 18.897 18.897

1.7

18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897 18.897

19.606 19.606 19.607 19.614 19.728 20.171 20.594 21.225 21.450 21.514 21.514 21.357 21.111 21.111 20.900 19.750 19.750 19.747 18.898 18.898 18.897

3.5

19.146 19.146 19.147 19.152 19.234 19.509 19.698 19.967 19.996 19.962 19.961 19.904 19.777 19.777 19.710 19.171 19.171 19.167 18.897 18.897 18.897

1.8

18.909 18.909 18.909 18.909 18.915 18.930 18.938 18.949 18.948 18.945 18.945 18.943 18.936 18.936 18.933 18.907 18.907 18.907 18.897 18.897 18.897

0.1

19.656 19.656 19.657 19.664 19.778 20.221 20.654 21.300 21.553 21.630 21.630 21.465 21.213 21.213 20.989 19.802 19.802 19.800 18.898 18.898 18.897

3.6

19.163 19.163 19.164 19.170 19.254 19.543 19.746 20.035 20.069 20.035 20.035 19.973 19.839 19.839 19.766 19.193 19.193 19.189 18.897 18.897 18.897

1.9

18.921 18.921 18.921 18.922 18.933 18.964 18.979 19.001 19.001 18.994 18.994 18.989 18.976 18.976 18.971 18.918 18.918 18.917 18.897 18.897 18.897

0.2

19.709 19.709 19.710 19.717 19.830 20.273 20.713 21.373 21.659 21.750 21.750 21.576 21.318 21.318 21.081 19.856 19.856 19.855 18.898 18.898 18.897

3.7

19.181 19.181 19.182 19.187 19.276 19.578 19.795 20.105 20.143 20.111 20.110 20.044 19.903 19.903 19.824 19.217 19.217 19.213 18.897 18.897 18.897

2.0

18.934 18.934 18.934 18.935 18.951 18.997 19.021 19.055 19.055 19.045 19.044 19.037 19.017 19.017 19.009 18.929 18.929 18.928 18.897 18.897 18.897

0.3

19.763 19.763 19.764 19.771 19.883 20.325 20.771 21.444 21.768 21.873 21.874 21.690 21.427 21.427 21.175 19.914 19.914 19.913 18.899 18.899 18.897

3.8

19.199 19.199 19.200 19.206 19.297 19.613 19.844 20.174 20.219 20.188 20.188 20.117 19.968 19.968 19.883 19.242 19.242 19.238 18.897 18.897 18.897

2.1

18.946 18.946 18.946 18.948 18.969 19.031 19.063 19.109 19.110 19.096 19.096 19.086 19.060 19.060 19.049 18.940 18.940 18.939 18.897 18.897 18.897

0.4

19.818 19.818 19.819 19.826 19.937 20.376 20.827 21.513 21.879 22.000 22.000 21.808 21.539 21.539 21.273 19.975 19.975 19.974 18.899 18.899 18.897

3.9

19.218 19.218 19.219 19.225 19.319 19.649 19.893 20.245 20.296 20.268 20.267 20.191 20.035 20.035 19.944 19.268 19.268 19.264 18.898 18.898 18.897

2.2

18.959 18.959 18.959 18.961 18.987 19.065 19.106 19.165 19.165 19.149 19.149 19.136 19.103 19.103 19.089 18.952 18.952 18.951 18.897 18.897 18.897

0.5

19.872 19.872 19.873 19.880 19.989 20.426 20.882 21.580 21.993 22.129 22.129 21.929 21.655 21.655 21.373 20.039 20.039 20.039 18.899 18.899 18.897

4.0

19.237 19.237 19.238 19.245 19.342 19.684 19.943 20.317 20.374 20.349 20.349 20.268 20.104 20.104 20.006 19.295 19.295 19.291 18.898 18.898 18.897

2.3

18.972 18.972 18.972 18.974 19.005 19.098 19.149 19.221 19.222 19.203 19.203 19.187 19.147 19.147 19.130 18.965 18.965 18.964 18.897 18.897 18.897

0.6

19.925 19.925 19.926 19.933 20.040 20.473 20.935 21.646 22.108 22.257 22.257 22.054 21.776 21.776 21.477 20.107 20.107 20.107 18.899 18.899 18.897

4.1

19.258 19.258 19.259 19.266 19.365 19.720 19.994 20.389 20.454 20.433 20.432 20.346 20.175 20.175 20.070 19.324 19.324 19.320 18.898 18.898 18.897

2.4

18.985 18.985 18.985 18.988 19.023 19.132 19.192 19.278 19.280 19.258 19.258 19.240 19.193 19.193 19.172 18.978 18.978 18.977 18.897 18.897 18.897

0.7

19.974 19.974 19.975 19.982 20.088 20.518 20.985 21.710 22.224 22.383 22.383 22.183 21.901 21.901 21.584 20.178 20.178 20.180 18.899 18.899 18.897

4.2

19.280 19.280 19.281 19.287 19.390 19.756 20.045 20.463 20.536 20.518 20.518 20.427 20.248 20.248 20.136 19.354 19.354 19.350 18.898 18.898 18.897

2.5

18.998 18.998 18.998 19.001 19.042 19.166 19.236 19.336 19.339 19.315 19.315 19.293 19.239 19.239 19.216 18.992 18.992 18.990 18.897 18.897 18.897

0.8

20.020 20.020 20.021 20.027 20.133 20.559 21.032 21.772 22.336 22.504 22.504 22.316 22.030 22.030 21.695 20.254 20.254 20.257 18.899 18.899 18.897

4.3

19.303 19.303 19.304 19.310 19.415 19.793 20.096 20.537 20.619 20.606 20.606 20.509 20.324 20.324 20.203 19.385 19.385 19.381 18.898 18.898 18.897

2.6

19.012 19.012 19.012 19.015 19.060 19.200 19.281 19.395 19.400 19.373 19.373 19.348 19.287 19.287 19.260 19.007 19.007 19.004 18.897 18.897 18.897

0.9

20.060 20.060 20.061 20.068 20.172 20.596 21.076 21.832 22.443 22.618 22.619 22.454 22.164 22.164 21.810 20.335 20.335 20.338 18.899 18.899 18.897

4.4

19.327 19.327 19.328 19.335 19.442 19.830 20.148 20.612 20.704 20.696 20.696 20.594 20.401 20.401 20.272 19.418 19.418 19.414 18.898 18.898 18.897

2.7

19.025 19.025 19.026 19.029 19.079 19.234 19.325 19.455 19.461 19.432 19.432 19.404 19.336 19.336 19.305 19.022 19.022 19.019 18.897 18.897 18.897

1.0

20.096 20.096 20.097 20.104 20.207 20.629 21.116 21.888 22.543 22.724 22.725 22.596 22.303 22.303 21.928 20.420 20.420 20.425 18.899 18.899 18.897

4.5

19.352 19.352 19.354 19.361 19.469 19.868 20.200 20.687 20.791 20.789 20.789 20.680 20.481 20.481 20.343 19.453 19.453 19.449 18.898 18.898 18.897

2.8

19.039 19.039 19.040 19.043 19.097 19.268 19.371 19.516 19.523 19.493 19.493 19.462 19.386 19.386 19.351 19.038 19.038 19.035 18.897 18.897 18.897

1.1

20.127 20.127 20.128 20.135 20.237 20.658 21.152 21.942 22.634 22.822 22.823 22.742 22.447 22.447 22.051 20.511 20.511 20.517 18.899 18.899 18.897

4.6

19.380 19.380 19.381 19.388 19.499 19.907 20.254 20.764 20.879 20.884 20.884 20.769 20.563 20.563 20.416 19.489 19.489 19.485 18.898 18.898 18.897

2.9

19.053 19.053 19.054 19.058 19.116 19.302 19.416 19.578 19.587 19.555 19.555 19.521 19.438 19.438 19.399 19.054 19.054 19.051 18.897 18.897 18.897

1.2

20.153 20.153 20.154 20.161 20.263 20.683 21.185 21.991 22.716 22.913 22.914 22.894 22.597 22.597 22.178 20.607 20.607 20.614 18.899 18.899 18.897

4.7

19.410 19.410 19.411 19.418 19.530 19.947 20.308 20.840 20.969 20.982 20.982 20.861 20.647 20.647 20.491 19.528 19.528 19.524 18.898 18.898 18.897

3.0

19.068 19.068 19.069 19.073 19.135 19.336 19.462 19.640 19.652 19.619 19.619 19.581 19.491 19.491 19.448 19.071 19.071 19.068 18.897 18.897 18.897

1.3

Table 11: Average fuel economy (in km/l) for γf e = [0.5, 2.5] and γsub = [0, 5.0] 1.4

20.176 20.176 20.177 20.183 20.285 20.705 21.215 22.037 22.790 22.996 22.997 23.051 22.753 22.753 22.310 20.709 20.709 20.718 18.899 18.899 18.897

4.8

19.442 19.442 19.443 19.451 19.564 19.988 20.363 20.917 21.061 21.082 21.082 20.955 20.734 20.734 20.569 19.568 19.568 19.564 18.898 18.898 18.897

3.1

19.083 19.083 19.083 19.088 19.154 19.371 19.509 19.704 19.718 19.684 19.684 19.642 19.545 19.545 19.497 19.089 19.089 19.086 18.897 18.897 18.897

1.5

20.194 20.194 20.195 20.202 20.303 20.724 21.241 22.080 22.857 23.073 23.074 23.213 22.915 22.915 22.446 20.818 20.818 20.828 18.900 18.900 18.897

4.9

19.478 19.478 19.479 19.486 19.600 20.031 20.419 20.995 21.155 21.186 21.185 21.051 20.824 20.824 20.648 19.610 19.610 19.606 18.898 18.898 18.897

3.2

19.098 19.098 19.099 19.103 19.174 19.405 19.555 19.768 19.786 19.751 19.751 19.705 19.601 19.601 19.549 19.108 19.108 19.105 18.897 18.897 18.897

1.6

20.210 20.210 20.211 20.217 20.318 20.741 21.266 22.119 22.918 23.145 23.146 23.381 23.083 23.083 22.588 20.934 20.934 20.945 18.900 18.900 18.897

5.0

19.516 19.516 19.517 19.525 19.639 20.075 20.477 21.072 21.252 21.292 21.292 21.150 20.917 20.917 20.729 19.654 19.654 19.651 18.898 18.898 18.897

3.3

19.114 19.114 19.114 19.119 19.193 19.439 19.602 19.834 19.855 19.820 19.819 19.770 19.658 19.658 19.601 19.128 19.128 19.125 18.897 18.897 18.897

35

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5

XXX γsub XXX γf e

2,397.6 2,397.6 2,397.1 2,393.5 2,328.5 1,990.7 1,499.2 1,146.2 727.6 559.3 558.2 371.5 321.3 321.3 276.8 124.8 124.8 113.6 2.2 2.2 0.0

3.4

678.3 678.3 678.1 676.5 649.5 520.0 352.1 246.3 187.5 135.0 134.7 111.3 94.4 94.4 85.1 31.3 31.3 27.2 0.7 0.7 0.0

1.7

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0

2,600.3 2,600.3 2,599.7 2,596.0 2,528.4 2,175.3 1,659.2 1,280.9 789.7 613.3 612.2 395.7 342.7 342.7 294.1 134.3 134.3 122.5 2.3 2.3 0.0

3.5

736.7 736.7 736.5 734.8 705.9 566.8 385.1 270.4 205.0 147.9 147.5 121.2 102.8 102.8 92.6 34.4 34.4 30.0 0.8 0.8 0.0

1.8

28.0 28.0 28.0 27.9 26.5 20.4 13.1 8.7 6.9 4.9 4.9 4.4 3.7 3.7 3.4 1.0 1.0 0.9 0.0 0.0 0.0

0.1

2,827.5 2,827.5 2,826.9 2,823.1 2,752.8 2,384.4 1,841.4 1,436.0 859.3 675.5 674.3 421.3 365.5 365.5 312.3 144.5 144.5 132.2 2.5 2.5 0.0

3.6

798.1 798.1 797.8 796.0 765.2 616.2 420.2 296.1 223.7 161.6 161.2 131.5 111.7 111.7 100.4 37.7 37.7 33.0 0.9 0.9 0.0

1.9

57.1 57.1 57.1 56.9 54.1 41.7 27.0 17.9 14.2 10.1 10.1 8.9 7.5 7.5 6.9 2.2 2.2 1.8 0.1 0.1 0.0

0.2

3,081.2 3,081.2 3,080.5 3,076.6 3,003.6 2,619.9 2,046.9 1,612.1 938.0 747.8 746.6 448.4 389.9 389.9 331.5 155.5 155.5 142.6 2.6 2.6 0.0

3.7

862.7 862.7 862.4 860.5 827.8 668.5 457.6 323.7 243.5 176.2 175.8 142.3 121.0 121.0 108.5 41.2 41.2 36.2 0.9 0.9 0.0

2.0

87.4 87.4 87.3 87.1 82.9 64.0 41.5 27.7 21.9 15.6 15.6 13.7 11.5 11.5 10.6 3.4 3.4 2.8 0.1 0.1 0.0

0.3

3,361.8 3,361.8 3,361.2 3,357.1 3,281.3 2,882.4 2,275.8 1,809.3 1,028.0 833.0 831.7 477.3 415.8 415.8 351.8 167.3 167.3 153.8 2.8 2.8 0.0

3.8

930.8 930.8 930.5 928.5 893.8 724.0 497.6 353.4 264.5 191.7 191.3 153.8 130.8 130.8 117.1 45.0 45.0 39.6 1.0 1.0 0.0

2.1

118.9 118.9 118.9 118.5 112.9 87.4 56.8 38.0 30.0 21.4 21.4 18.7 15.7 15.7 14.5 4.6 4.6 3.9 0.1 0.1 0.0

0.4

3,669.1 3,669.1 3,668.5 3,664.2 3,585.4 3,171.5 2,527.8 2,026.9 1,132.1 933.9 932.5 508.1 443.5 443.5 373.3 180.0 180.0 165.9 2.9 2.9 0.0

3.9

1,002.8 1,002.8 1,002.5 1,000.4 963.5 783.0 540.3 385.2 286.9 208.4 207.9 165.7 141.2 141.2 126.1 49.0 49.0 43.3 1.1 1.1 0.0

2.2

151.8 151.8 151.7 151.3 144.2 111.9 73.0 49.0 38.5 27.5 27.5 24.0 20.2 20.2 18.5 6.0 6.0 5.0 0.2 0.2 0.0

0.5

4,001.7 4,001.7 4,001.1 3,996.7 3,914.9 3,486.0 2,802.2 2,263.9 1,253.7 1,053.5 1052.0 540.9 473.2 473.2 396.0 193.6 193.6 178.9 3.1 3.1 0.0

4.0

1,078.9 1,078.9 1,078.6 1,076.3 1,037.4 845.7 586.1 419.6 310.7 226.2 225.7 178.4 152.1 152.1 135.5 53.3 53.3 47.2 1.1 1.1 0.0

2.3

186.0 186.0 185.9 185.4 176.9 137.6 90.0 60.6 47.6 34.0 33.9 29.5 24.8 24.8 22.8 7.4 7.4 6.3 0.2 0.2 0.0

0.6

4,357.8 4,357.8 4,357.1 4,352.6 4,267.6 3,823.7 3,097.7 2,519.3 1,396.2 1,193.6 1,192.0 575.9 505.0 505.0 420.0 208.4 208.4 193.0 3.2 3.2 0.0

4.1

1,159.6 1,159.6 1,159.2 1,156.9 1,115.8 912.6 635.3 456.8 336.1 245.3 244.8 191.6 163.6 163.6 145.4 57.9 57.9 51.4 1.2 1.2 0.0

2.4

221.7 221.7 221.6 221.0 210.9 164.5 107.9 72.9 57.1 40.8 40.7 35.3 29.7 29.7 27.2 9.0 9.0 7.6 0.2 0.2 0.0

0.7

4,734.7 4,734.7 4,734.0 4,729.3 4,641.2 4,182.1 3,412.6 2,792.0 1,562.1 1,354.2 1,352.6 613.3 539.2 539.2 445.5 224.3 224.3 208.2 3.4 3.4 0.0

4.2

1,245.3 1,245.3 1,245.0 1,242.5 1,199.2 984.2 688.5 497.2 363.3 265.9 265.3 205.6 175.7 175.7 155.8 62.7 62.7 55.8 1.3 1.3 0.0

2.5

258.9 258.9 258.8 258.1 246.4 192.7 126.7 85.9 67.1 48.0 47.9 41.4 34.8 34.8 31.9 10.6 10.6 9.0 0.3 0.3 0.0

0.8

5,129.3 5,129.3 5,128.6 5,123.8 5,032.5 4,557.6 3,744.8 3,080.8 1,751.9 1,533.7 1,532.0 653.3 576.0 576.0 472.5 241.4 241.4 224.7 3.6 3.6 0.0

4.3

1,336.7 1,336.7 1,336.3 1,333.7 1,288.2 1,061.0 746.1 541.3 392.3 288.1 287.4 220.3 188.5 188.5 166.7 67.9 67.9 60.6 1.4 1.4 0.0

2.6

297.7 297.7 297.6 296.8 283.6 222.2 146.6 99.8 77.7 55.6 55.5 47.7 40.2 40.2 36.8 12.4 12.4 10.6 0.3 0.3 0.0

0.9

5,538.2 5,538.2 5,537.5 5,532.6 5,438.2 4,946.6 4,091.2 3,384.1 1,963.5 1,729.5 1,727.7 696.2 615.6 615.6 501.2 260.0 260.0 242.6 3.8 3.8 0.0

4.4

1,434.4 1,434.4 1,434.0 1,431.3 1,383.5 1,143.7 808.8 589.7 423.5 312.0 311.4 235.8 202.0 202.0 178.2 73.5 73.5 65.7 1.5 1.5 0.0

2.7

338.2 338.2 338.0 337.2 322.3 253.3 167.6 114.4 88.9 63.7 63.5 54.4 45.9 45.9 41.9 14.2 14.2 12.2 0.4 0.4 0.0

1.0

5,958.0 5,958.0 5,957.2 5,952.2 5,854.6 5,345.4 4,448.7 3,699.7 2,193.3 1,939.0 1,937.1 742.4 658.4 658.4 531.8 280.1 280.1 262.0 4.0 4.0 0.0

4.5

1,539.2 1,539.2 1,538.8 1,536.0 1,485.8 1,233.2 877.4 643.0 457.0 338.1 337.3 252.2 216.2 216.2 190.2 79.4 79.4 71.2 1.6 1.6 0.0

2.8

380.4 380.4 380.3 379.3 362.9 285.9 189.8 130.0 100.8 72.2 72.0 61.4 51.8 51.8 47.2 16.2 16.2 13.9 0.4 0.4 0.0

1.1

6,385.3 6,385.3 6,384.5 6,379.4 6,278.7 5,750.8 4,814.1 4,025.5 2,437.3 2,160.0 2,158.0 792.0 704.8 704.8 564.3 301.9 301.9 283.1 4.2 4.2 0.0

4.6

1,652.2 1,652.2 1,651.7 1,648.8 1,596.2 1,330.3 953.0 702.3 493.2 366.4 365.7 269.4 231.3 231.3 202.9 85.8 85.8 77.1 1.7 1.7 0.0

2.9

424.6 424.6 424.5 423.4 405.3 320.1 213.2 146.5 113.3 81.2 81.0 68.8 58.1 58.1 52.8 18.4 18.4 15.8 0.5 0.5 0.0

1.2

6,817.6 6,817.6 6,816.8 6,811.6 6,707.6 6,160.2 5,184.4 4,359.1 2,692.7 2,390.9 2,388.8 845.6 755.2 755.2 598.9 325.6 325.6 306.1 4.4 4.4 0.0

4.7

1,774.6 1,774.6 1,774.1 1,771.1 1,716.1 1,436.5 1,037.1 768.8 532.2 397.5 396.7 287.7 247.3 247.3 216.2 92.6 92.6 83.4 1.8 1.8 0.0

3.0

470.8 470.8 470.6 469.4 449.6 356.1 237.9 164.1 126.5 90.7 90.5 76.5 64.6 64.6 58.7 20.6 20.6 17.8 0.5 0.5 0.0

1.3

Table 12: Total subsidy payment (in billion JPY) for γf e = [0.5, 2.5] and γsub = [0, 5.0] 1.4

7,252.7 7,252.7 7,251.9 7,246.5 7,139.4 6,571.4 5,557.5 4,698.8 2,957.2 2,630.5 2,628.2 903.6 810.0 810.0 635.8 351.4 351.4 331.2 4.6 4.6 0.0

4.8

1,908.1 1,908.1 1,907.6 1,904.4 1,846.9 1,553.3 1,131.3 844.1 574.7 431.7 430.8 306.9 264.2 264.2 230.2 99.8 99.8 90.2 1.9 1.9 0.0

3.1

519.1 519.1 518.9 517.6 496.1 393.9 264.0 182.7 140.4 100.8 100.6 84.6 71.5 71.5 64.8 23.0 23.0 19.9 0.6 0.6 0.0

1.5

7,689.0 7,689.0 7,688.1 7,682.7 7,572.3 6,983.0 5,931.5 5,043.0 3,229.7 2,878.0 2,875.7 966.5 869.9 869.9 675.3 379.5 379.5 358.5 4.8 4.8 0.0

4.9

2,054.6 2,054.6 2,054.1 2,050.8 1,990.9 1,682.8 1,237.9 930.4 620.9 469.5 468.6 327.2 282.1 282.1 244.9 107.6 107.6 97.4 2.0 2.0 0.0

3.2

569.7 569.7 569.5 568.1 544.8 433.8 291.7 202.6 155.2 111.6 111.3 93.1 78.8 78.8 71.3 25.6 25.6 22.2 0.6 0.6 0.0

1.6

8,125.3 8,125.3 8,124.5 8,118.8 8,005.1 7,394.0 6,305.2 5,390.3 3,509.4 3,133.2 3,130.8 1,034.9 935.5 935.5 717.5 410.2 410.2 388.4 5.1 5.1 0.0

5.0

2,216.8 2,216.8 2,216.3 2,212.8 2,150.4 1,827.6 1,359.6 1,030.2 671.7 511.7 510.8 348.7 301.1 301.1 260.5 115.9 115.9 105.2 2.1 2.1 0.0

3.3

622.7 622.7 622.5 621.0 595.9 475.8 321.0 223.7 170.9 123.0 122.7 102.0 86.4 86.4 78.1 28.3 28.3 24.6 0.7 0.7 0.0