Impacts of Growth of a Service Economy on CO₂ Emissions: Japan’s Case

Abstract

The structural transition to a service economy has clearly contributed to decreasing direct (or territorial) greenhouse gas emissions. Nevertheless, the role of this structural transition on direct greenhouse gas emissions is not well understood quantitatively. This study applied the additive decomposition method and decomposed the change in CO₂ emissions from domestic industries into five components: changes in the overall scale of the economy, changes in the industrial composition of the various economic sectors, energy intensity changes, changes in import composition, and changes in the import scale. The decomposition results show that during the 15-year period from 1990 to 2005, structural change effects under the domestic technology assumption (which include industrial composition effects, import scale effects, and import composition effects) totaled −35 Mt CO₂, or 3 % of total CO₂ emissions in 1990. It is concluded that the CO₂ reduction due to the transition to a service economy was not negligible during 1990–2005 and that the structural transition to a service economy was much more important than the material dependence of service industries.

JEL Classification: O14, O44, Q56.

1 Introduction

Increased environmental loads can be understood as arising from a variety of economic factors. For example, the environmental Kuznets curve describes an inverted-U relationship between economic growth (including structural changes) and environmental pollution (Grossman and Krueger [1991, 1995, 1996]; Carson [2010] for a literature overview). In particular, this article sheds light on the relationship between structural changes and environmental load in a specific country. As in Levinson ([2009]), I will focus on influences on CO₂ emissions. In this study, I consider not only the economic scale, but also another factor that exhibits significant influence: changes in industrial composition. In Japan, the percentage of domestic Japanese production attributable to secondary industries (manufacturing), which exhibit high rates of CO₂ emissions per unit production (i.e., large direct emissions coefficients), fell drastically, from 49 % in 1990 to just 39 % in 2005. In contrast, the percentage of domestic Japanese production attributable to tertiary industries (service industries), which exhibit low coefficients of direct CO₂ emissions, rose significantly, from 48 % in 1990 to 60 % in 2005.¹ This also implies that Japan’s transition toward a service-oriented economy has contributed in reducing CO₂ emissions, but the extent to which this has slowed the pace of global warming remains unclear.

Important studies on the relationship between the transition to a service economy and CO₂ emissions include those of Suh ([2006]) and Nansai et al. ([2009]). Suh ([2006]) demonstrated that household consumption of services, excluding electric utilities and transportation services, accounts for 37.6 % of total industrial GHG emissions in the United States. Nansai et al. ([2009]) analyzed the factors governing life-cycle CO₂ emissions in Japanese service industries between the years 1990 and 2000 and concluded that increased inputs of energy and resources (including materials and components) led to significantly increased CO₂ emissions.

However, the studies of Suh ([2006]) and Nansai et al. ([2009]) did not quantify the transition to a service economy in terms of the increasing industrial composition attributable to service industries and also did not analyze the impact of the transition to a service economy on production-based CO₂ emissions.² In addition, their studies did not argue that the transition to a service economy spurs an increase in imports of CO₂-intensive commodities and that consequently this structural change contributes to global warming. Therefore, in the present study, I apply the Shapley–Sun additive decomposition method (Shapley [1953]; Sun [1998]) and decompose the change in production-based CO₂ emissions from domestic industries into five components: that due to changes in the overall scale of the economy, that due to changes in the industrial composition of the various economic sectors, that due to energy intensity (i.e., technical) changes, which measures CO₂ emissions per unit of domestic production, that due to changes in the import composition of the various commodities, and that due to changes in the import scale. Using this index decomposition method, I will analyze the impact of Japan’s transition to a service economy on Japanese CO₂ emissions between 1990 and 2005, and finally argue the environmental benefits of its structural transition.

The rest of this paper is organized as follows: Sect. 2 presents the decomposition method, Sect. 3 describes the data source, Sect. 4 presents a case study of Japan, and Sect. 5 concludes the paper.

2 Methodology

2.1 Estimating CO₂ Emissions Originating from Industrial Activities

Let $e_{k,i}^t$ denote the energy consumption (Gigajoules: GJ) of fuel type k ($k=1,2,\dots ,M$) associated with 1 unit (¥1 million) of production in industry sector i ($i=1,2,\dots ,N$) during year t. Here, N is the number of industry sectors and M is the number of types of fuel. Also, let $c_k$ denote the CO₂ emissions (t CO₂) generated directly from the consumption of 1 GJ worth of fuel type k in the specific industry sector. Then the quantity of CO₂ emitted in conjunction with unit production in industry sector i in year t can be expressed in the form $c_k\times e_{k,i}^t$ (t CO₂/million yen).

If ${\theta }_i^t$ denotes the industrial composition showing the fraction of output of industry sector i of total production across all industries, and $X_d^t$ denotes total industrial output summed over all industry sectors, the total amount of domestic production contributed by industry sector i in year t is then represented as ${\theta }_i^t\times X_d^t$ (million yen).

Multiplying the CO₂ emission coefficient of industry sector i, $c_k\times e_{k,i}^t$, by the domestic output of industry sector i, ${\theta }_i^t\times X_d^t$, yields $c_k{e}_{k,i}^t{\theta }_i^t{X}_d^t$ as an estimate of CO₂ emissions arising from the use of fuel type k in industry sector i. Summing these estimates over all industry sectors and all fuel types, we obtain the following estimate of total domestic production-based emissions $Q_d^t$ (t CO₂):

$$Q_d^t=\sum _{i=1}^N\sum _{k=1}^M{c}_k{e}_{k,i}^t{\theta }_i^t{X}_d^t$$

(1)

2.2 Changes in CO₂ Emissions: Factor Decomposition

We now use the Shapley–Sun decomposition method to analyze changes in the quantity of CO₂ emissions originating from industrial activities (i.e., the quantity $Q_d^t$) into three sources: technical effects, industrial composition effects, and economic scale effects (Levinson [2009]). (For details on the decomposition method, see Ang [2004]; Ang et al. [2003]; Wood and Lenzen [2006] and see e.g., Ma and Stern [2008]; Kagawa et al. [2012] for the energy decomposition analysis.)

Let $\mathrm{\Delta }{Q}_d$ denote the change from year t to year $t+1$ in CO₂ emissions originating from industrial activities, expressed as follows:

$$\begin{array}{rl}\mathrm{\Delta }{Q}_d& =Q_d^{t+1}-Q_d^t\\ =\sum _{i=1}^N\sum _{k=1}^M{c}_k{e}_{k,i}^{t+1}{\theta }_i^{t+1}{X}_d^{t+1}-\sum _{i=1}^N\sum _{k=1}^M{c}_k{e}_{k,i}^t{\theta }_i^t{X}_d^t\\ =\mathbf{c}\cdot {\mathbf{E}}^{t+1}\cdot {\mathit{\theta }}^{t+1}\cdot X_d^{t+1}-\mathbf{c}\cdot {\mathbf{E}}^t\cdot {\mathit{\theta }}^t\cdot X_d^t\end{array}$$

(2)

Here, c is a $(1\times M)$ row vector whose k th element, $c_k$, is the emission coefficient of fuel type k; E is an $(M\times N)$ matrix whose $(k,i)$ element, $e_{k,i}$, is the energy consumption (i.e., energy intensity) for fuel type k used to produce one unit of output in industry sector i; and θ is an $(N\times 1)$ column vector whose i th element, ${\theta }_i$, is the industrial composition of industry sector i. The superscripts t and $t+1$ indicate the year.

The changes in $\mathbf{E}=(e_{k,i})$, $\mathit{\theta }=({\theta }_i)$, and X can be expressed as follows:

$$\mathrm{\Delta }\mathbf{E}={\mathbf{E}}^{t+1}-{\mathbf{E}}^t$$

(3)

$$\mathrm{\Delta }\mathit{\theta }={\mathit{\theta }}^{t+1}-{\mathit{\theta }}^t$$

(4)

$$\mathrm{\Delta }{X}_d=X_d^{t+1}-X_d^t$$

(5)

Using Eqs. (3), (4), and (5), Eq. (2) can be transformed as follows:

$$\begin{array}{rl}\mathrm{\Delta }{Q}_d=& \mathbf{c}\cdot {\mathbf{E}}^{t+1}\cdot {\mathit{\theta }}^{t+1}\cdot X_d^{t+1}-\mathbf{c}\cdot {\mathbf{E}}^t\cdot {\mathit{\theta }}^t\cdot X_d^t\\ =& \mathbf{c}\cdot ({\mathbf{E}}^t+\mathrm{\Delta }\mathbf{E})\cdot ({\mathit{\theta }}^t+\mathrm{\Delta }\mathit{\theta })\cdot (X_d^t+\mathrm{\Delta }{X}_d)-\mathbf{c}\cdot {\mathbf{E}}^t\cdot {\mathit{\theta }}^t\cdot X_d^{t+1}\\ =& \mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathit{\theta }}^t{X}_d^{t+1}+\mathbf{c}{\mathbf{E}}^t\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}{\mathbf{E}}^t{\mathit{\theta }}^t\mathrm{\Delta }{X}_d+\mathbf{c}\mathrm{\Delta }\mathbf{E}\mathrm{\Delta }\mathit{\theta }{X}_d^t\\ +\mathbf{c}{\mathbf{E}}^t\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d+\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathit{\theta }}^t\mathrm{\Delta }{X}_d+\mathbf{c}\mathrm{\Delta }\mathbf{E}\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(6)

The first term on the right-hand side of Eq. (6) represents the influence on emissions of changes in the energy intensity in the industrial sector. The second and third terms represent the influence on emissions of changes in the industrial composition of the industrial sector and the total industrial output, respectively. The simplified additive decomposition method (e.g., Park [1992]) ignores second-order interaction terms (such as the fourth, fifth, and sixth terms on the right-hand side of Eq. (6)) and third-order interaction terms (such as the seventh term). As a result, the sum of the contributions of the first three terms on the right-hand side will not be equal to total change in emissions $\mathrm{\Delta }{Q}_d$. The important question is how to treat the influence of the interaction terms (Sun [1998]).

In the present study, following Sun ([1998]), I consider the second-order interaction terms and the third-order interaction term, and employ the following additive decomposition formulation:

(7)

We refer to the first, second, and third terms on the right-hand side of Eq. (7) respectively as the technical effect, the industrial composition effect, and the economic scale effect, which we denote by $\mathrm{\Delta }{Q}_d^{\mathrm{Tech}}$, $\mathrm{\Delta }{Q}_d^{\mathrm{Comp}}$, and $\mathrm{\Delta }{Q}_d^{\mathrm{Scale}}$. The effect expressed by Eq. (7) is the total effect, representing the sum of the effects across all industries; thus, for example, it is not possible to isolate from Eq. (7) the industrial composition effect in the service industry or the technical effect in the manufacturing industry. For this reason, we will further decompose Eq. (7) into the effect in each industry.

We will classify our N industry sectors into four industry groups:

• (1) primary industries,

• (2) secondary industries,

• (3) electricity, gas, and water supply industries, and

• (4) tertiary industries (service industries).

For industry sector i belonging to the group of primary industries (i.e., $i\in \mathrm{primary}\mathrm{industry}$), we define ${\mathbf{S}}_a$ to be the $(N\times N)$ diagonal matrix with i th diagonal element equal to 1 and all other elements equal to 0. Here, the subscript a indicates primary industries (i.e., agriculture, forestry, and fishery industries). The technical effect (i.e., that from changes in the energy intensity) in industry sectors belonging to the group of primary industries and the effect from changes in industrial composition in industry sectors belonging to the primary industries can be quantified using Eqs. (8) and (9) below:

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,a}^{\mathrm{Tech}}=& \mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a{\mathit{\theta }}^t{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a{\mathit{\theta }}^t\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(8)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,a}^{\mathrm{Comp}}=& \mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_a\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(9)

Similarly, the technical effects and industrial composition effects in secondary industries, electricity, gas, and water supply industries, and tertiary industries can be estimated as in Eqs. (10) through (15) below:

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,m}^{\mathrm{Tech}}=& \mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m{\mathit{\theta }}^t{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m{\mathit{\theta }}^t\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(10)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,m}^{\mathrm{Comp}}=& \mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_m\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(11)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,g}^{\mathrm{Tech}}=& \mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g{\mathit{\theta }}^t{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g{\mathit{\theta }}^t\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(12)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,g}^{\mathrm{Comp}}=& \mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_g\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(13)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,s}^{\mathrm{Tech}}=& \mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s{\mathit{\theta }}^t{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s{\mathit{\theta }}^t\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(14)

$$\begin{array}{rl}\mathrm{\Delta }{Q}_{d,s}^{\mathrm{Comp}}=& \mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }{X}_d^t+\frac{1}{2}(\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }{X}_d^t+\mathbf{c}{\mathbf{E}}^t{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d)\\ +\frac{1}{3}\mathbf{c}\mathrm{\Delta }\mathbf{E}{\mathbf{S}}_s\mathrm{\Delta }\mathit{\theta }\mathrm{\Delta }{X}_d\end{array}$$

(15)

Here, ${\mathbf{S}}_m$, ${\mathbf{S}}_g$, and ${\mathbf{S}}_s$, where the subscripts m, g, and s, respectively, denote secondary industries, electricity, gas, and water supply industries, and tertiary industries, are $(N\times N)$ diagonal matrices whose i th diagonal element is 1 for all i in the corresponding industry group and all other elements are zero.

3 Data

I used CO₂ emissions data obtained from industrial tables contained in the Embodied Energy and Emission Intensity Data for Japan Using Input–Output Tables: 3EID data book released by the Center for Global Environmental Research at the National Institute for Environmental Studies of Japan (2012). In addition, I used the 1990–1995–2000–2005 linked environmental input–output tables (396 industry sectors) (Nansai et al. [2007, 2009]).

Using the 3EID data book allows energy intensity data for joules of 32 types of raw fuel directly consumed by producing one unit of output in each of 396 industry sectors in the years 1990, 1995, 2000, and 2005 (see Table 1 for the 32 raw fuel types). From this database, we can obtain values of $e_{k,i}^t$. In addition, from the same database, we can obtain data on the quantity $c_k$ (Table 1).

Table 1 The classification of fuel types

From the 1990–1995–2000–2005 linked input–output tables (which are evaluated in terms of 2005 producer prices), we can obtain not only data on the total production in each industry sector in each year, but also data on the quantity $X_d^t$. This, in turn, allows us to easily compute ${\theta }_i$, which measures the industrial composition of industry sector i. For details on the categorization of industry sectors, see Table 2.

Table 2 The categorization of industrial sectors

4 Results

4.1 Macro-level Decomposition Results

According to the 1990–1995–2000–2005 linked input–output tables, Japan’s total industrial output was ¥841 trillion in 1990, ¥886 trillion in 1995, ¥922 trillion in 2000, and ¥962 trillion in 2005. Meanwhile, CO₂ emissions originating from industrial activity were 1.04 billion t CO₂ in 1990, 1.10 billion t CO₂ in 1995, 1.13 billion t CO₂ in 2000, and 1.17 billion t CO₂ in 2005. The increase in CO₂ emissions can be attributed to the growth in total industrial output. However, the CO₂ intensity, which can be defined by dividing CO₂ emissions originating from each year’s industrial activity by total industrial output, was 1.24 t CO₂/million yen in 1990, 1.25 t CO₂/million yen in 1995, 1.22 t CO₂/million yen in 2000, and 1.22 t CO₂/million yen in 2005. Thus, Japan’s CO₂ intensity has been gradually improving, indicating that factors such as technological progress and the transition to cleaner fuels have contributed to reducing CO₂ emissions.

Figure 1 shows the results of decompositions, using Eq. (7), of the changes in Japanese CO₂ emissions originating from industrial activity over the 15-year period from 1990 to 2005, as decomposed into three factors: technical effects, industrial composition effects, and economic scale effects. Between 1990 and 1995, the change in CO₂ emissions was +64 Mt CO₂; from the figure, we see that this number breaks down into −2 Mt CO₂ arising from technical effects, +8 Mt CO₂ arising from industrial composition effects, and +58 Mt CO₂ arising from economic scale effects. Next, between 1995 and 2000, the change in CO₂ emissions was +25 Mt CO₂; this number breaks down into −99 million t CO₂ arising from technical effects, +78 Mt CO₂ arising from industrial composition effects, and +46 Mt CO₂ arising from economic scale effects. Finally, between 2000 and 2005, the change in CO₂ emissions was +46 Mt CO₂; this number breaks down into +98 Mt CO₂ arising from technical effects, −102 Mt CO₂ arising from industrial composition effects, and +50 Mt CO₂ arising from economic scale effects.

Fig. 1

CO₂ decomposition result using the Shapley–Sun decomposition method (units: Mt CO₂)

Thus, we see that, during the 10-year period from 1990 to 2000, economic scale effects and industrial composition effects both contributed to increasing CO₂ emissions, while technical effects contributed to reducing CO₂ emissions. However, this trend reversed itself in the years between 2000 and 2005, during which technical effects contributed significantly to increasing CO₂ emissions, whereas industrial composition effects contributed significantly to reducing CO₂ emissions.

Because the results presented in Fig. 1 are aggregate totals over all industry sectors, they do not allow us to identify the particular industry sectors in which technical effects and industrial composition effects influenced CO₂ emissions. To investigate these questions, we use Eqs. (8) through (15) to analyze technical effects and industrial composition effects in each of our four industry groups: primary industries, secondary industries, electricity, gas, and water supply industries, and tertiary industries.

4.2 Technical Effects for the Four Industry Groups

Within each industry, the technical effect measures the impact on CO₂ emissions of changes in the industrial energy intensity. A negative technical effect for an industry signifies that the industry has successfully reduced energy consumption or shifted its use of energy in a way that reduces CO₂ emissions. Figure 2 shows technical effects for the four industry groups considered in this study. As shown, electricity, gas, and water supply industries exhibited a negative technical effect throughout the 10-year period from 1990 to 2000 but crossed over to a large positive technical effect (+102 Mt CO₂) during the interval between 2000 and 2005.

Fig. 2

Technical effects for the four industry groups (units: Mt CO₂)

Thus, we see that, in the past 15 years, the technical effects in electricity, gas, and water supply industries have varied widely. In particular, one factor contributing to the increase in emissions during the 5-year period from 2000 to 2005 was the high technical effect of +62 Mt CO₂ observed for the commercial electric power sector. The primary cause of this phenomenon in the commercial electric power sector is the fact that, although the energy intensity for crude oil decreased during this period, the energy intensity for coal, lignite, and anthracite increased, and an energy shift to these fuels, which exhibit relatively higher concentrations of CO₂ emissions, has occurred.

Figure 2 also reveals that technical effects in tertiary industries led to a significant decrease in CO₂ emissions between the years 2000 and 2005. Considering the technical effects in specific sectors, we see that the technical effect in the ocean cargo transport industry was −8 Mt CO₂ and that in the road cargo transport industry was −7 Mt CO₂. Improved fuel efficiency in both these sectors significantly reduced the quantity of heavy oil needed to power ships and the quantity of light oil needed to power trucks, accounting for 88 % of the technical effects observed in tertiary industries.

4.3 Industrial Composition Effects for the Four Industry Groups

Within each industry, the industrial composition effect measures the impact of changes in the fraction of the overall industry accounted for by the various sectors. A negative value for this effect indicates that an industry sector contributed to reducing CO₂ emissions by decreasing the industrial composition. Figure 3 displays industrial composition effects for the four industry groups. As indicated in the figure, both primary and secondary industries exhibited negative industrial composition effects throughout the 15-year period from 1990 to 2005, whereas tertiary industries exhibited an overall positive effect throughout this period.

Fig. 3

Industrial composition effects for the four industry groups (units: Mt CO₂)

The total industrial composition effect for primary, secondary, and tertiary industries was −18.8 Mt CO₂ between 1990 and 1995, −15.8 Mt CO₂ between 1995 and 2000, and −30.4 Mt CO₂ between 2000 and 2005. These observations indicate that, throughout this 15-year period, the market for primary and secondary industries contracted, whereas the market for tertiary industries expanded (indicating the transition to a service economy); these changes consequently reduced CO₂ emissions by 65 Mt CO₂.

4.4 Role of the Service Economy and International Trade on CO₂ Emissions

Figure 4 compares the total technical effect for primary, secondary, and tertiary industries to the total industrial composition effect for these three industry groups.³ Considering the overall effect (that is, the sum of the technical effect and the industrial composition effect), we see that, in the years between 1990 and 1995, technical effects and industrial composition effects together accounted for an increase in CO₂ emissions of 880 kt CO₂ (the sum of the technical effect and the industrial composition effect for 1990–1995 shown in Fig. 4). On the other hand, between 1995 and 2000, technical effects and industrial composition effects led to a decrease in CO₂ emissions of 50.7 Mt CO₂, and between 2000 and 2005 these effects led to a further decrease of 34.2 Mt CO₂. Thus, the overall decrease was particularly significant between 1995 and 2000; from the figure, we can see that this is largely attributable to the relatively large technical effects exhibited by tertiary industries during this interval.

Fig. 4

Overall effects for three industry groups (units: Mt CO₂)

The 1990–1995 overall effect of +880 kt CO₂ corresponds to 0.1 % of total emissions in 1990, which is the base year of the Kyoto Protocol. Whereas the industrial composition effect during this period was a large negative effect due to the transition to a service economy, the technical effect contributed significantly to increased CO₂ emissions. Between 1995 and 2000, the overall effect was −50.7 Mt CO₂, corresponding to 4.6 % of total emissions in 1995; between 2000 and 2005, the overall effect was −34.2 Mt CO₂, or a 3 % decrease compared to total emissions in 2000. Nansai et al. ([2009]) analyzed the domestic CO₂ emissions associated with the energy and material goods absorbed by services through the supply chain during the decade 1990–2000. They found that the CO₂ emissions contributed by way of the material goods absorbed by service industries rose from 68 Mt CO₂ in 1990 to 87 Mt CO₂ in 2000. As a result, the material dependence of service industries increased by 19 Mt CO₂ during 1990–2000. On the other hand, this study found that the CO₂ reduction due to the transition of a service economy was 35 Mt CO₂.⁴ This reveals that the structural transition to a service economy was much more important than the material dependence of service industries.

Over the past 15 years, the declining share of domestic output by Japan’s manufacturing industries has contributed to the mitigation of global warming, but the corresponding increase in the share of manufactured goods imported from overseas has increased CO₂ emissions in foreign countries. This leads to the question of whether it is possible that the net impact has been to exacerbate the phenomenon of global warming. To address this question, we considered the impact on CO₂ emissions of the changing share of imports; we decomposed import-based CO₂ emissions into three sources, as formulated in the Appendix.⁵ Figures 5 and 6 present the results of this decomposition analysis. As shown in Fig. 5, over the past 15 years, the absolute quantity of imports from foreign countries to Japan rose and at the same time domestic CO₂ emissions rose by the equivalent of 38 Mt CO₂ (the total import scale effect). In contrast, as shown in Fig. 6, changes in the import composition decreased domestic CO₂ emissions by 8 Mt CO₂. These results demonstrate that Japan’s increasing dependence on imports during the past 15 years has accelerated global warming.

Fig. 5

Import scale effects for three industry groups (units: Mt CO₂)

Fig. 6

Import composition effects for three industry groups (units: Mt CO₂)

In this study, we have employed the domestic technology assumption to estimate import-based CO₂ emissions by multiplying Japanese import volumes by Japanese CO₂ emission coefficients for each of 396 industries. For this reason, we might have underestimated CO₂ emissions due to imports from developing countries with relatively high emission coefficients. As the Japanese economy transitions from agricultural and manufacturing industries to service-based industries, it depends increasingly on imports of agricultural products and manufactured goods; on the basis of the domestic technology assumption, these imports changes (especially, the increase in the import scale of manufacturing products) and the previous industrial composition changes (i.e., the transition to a service economy) have consequently brought about a reduction in production-based CO₂ emissions of 35 Mt CO₂, or approximately 3 % of total emissions in 1990.

However, this reduction effect may be considerably overestimated due to differences in CO₂ emission intensities between Japan and other countries. Based on the World Input–Output Database (40 countries and 35 industrial sectors),⁶ the Japanese industrial CO₂ intensities are approximately half those of China (one of the more CO₂-intensive countries) on average. Although the Chinese CO₂ emission intensities from the World Input–Output Database cannot be easily used for our study due to the highly aggregated sectoral classifications, it is clear that if we simply assume all the Japanese CO₂ intensities for a particular year (1990, 1995, 2000, and 2005) to be double their actual values, both the import scale effect and the import composition effect would be also double, accounting for 76 Mt CO₂ and −16 Mt CO₂, respectively. As a result, this assumption leads to the findings that the imports change effect, including their scale and composition effects, is 60 Mt CO₂ and the reduction effect due to the industrial composition changes over the entire 15-year period was offset by the imports change effect (see Sect. 4.3 for the industrial composition effects). Thus, the CO₂ emission leakage of Japan might not be negligible.

Under the terms of the Kyoto Protocol, Japan’s target was to reduce domestic emissions by 6 % of total emissions in 1990; thus, if we consider only the domestic industrial composition effect (−65 Mt CO₂) discussed in Sect. 4.3, then we must conclude that this structural transition has contributed significantly to Japan’s attainment of its emissions-reduction goals under the Kyoto Protocol. Moreover, the CO₂ emissions tax under consideration by Japan’s Ministry of the Environment is 289 yen/t CO₂, and, based on this tax rate, the environmental benefit of the transition to a service economy will amount to ¥18.7 billion ($=289\mathrm{yen}/\mathrm{t}{\mathrm{CO}}_2\times 65\mathrm{Mt}{\mathrm{CO}}_2$). Thus, we cannot ignore these structural change effects when considering the mitigation of domestic greenhouse gas emissions. Industrial policies that accelerate Japan’s transition to a service economy are an effective means of reducing Japanese domestic CO₂ emissions. However, such policies may result in increased emissions overall, by steering the production of manufactured industrial goods to foreign producers exhibiting high concentrations of CO₂ emissions. The important point is to strive for the dematerialization of society as a whole, thereby reducing CO₂ emissions from manufacturing sectors both in Japan and abroad.

5 Conclusions

In this study, I considered the Japanese economy during three time periods, from 1990 to 1995, from 1995 to 2000, and from 2000 to 2005, and I decomposed changes in CO₂ emissions originating from detailed industrial activities into five contributing factors, technical effects, industrial composition effects, economic scale effects, import scale effects, and import composition effects.

The major findings of this study are as follows.

• (1) During the 15-year period from 1990 to 2005, technical effects in the ocean and road cargo transport sectors (including, among other factors, increased fuel efficiency for ships and trucks) helped to ensure an overall technical effect of −29 Mt CO₂ for tertiary industries as a whole, thus contributing significantly to a reduction in CO₂ emissions.

• (2) The industrial composition changes during the period from 2000 to 2005 contributed to a decrease in CO₂ emissions, while those changes during the 10-year period from 1990 to 2000 led to an increase in CO₂ emissions. The main reason is that the Japanese economy experienced a significant decarbonization due to structural changes toward a service economy during 2000 to 2005.

• (3) During the 15-year period from 1990 to 2005, structural change effects under the domestic technology assumption (which include industrial composition effects, import scale effects, and import composition effects) totaled −35 Mt CO₂, or 3 % of total CO₂ emissions in 1990. These effects were instrumental in allowing Japan to attain its emissions-reduction target under the Kyoto Protocol, which was a 6 % reduction from 1990 emissions levels.

• (4) I demonstrated that the domestic environmental benefit arising from the transition to a service economy would amount to ¥18.7 billion.

Appendix

Using the same decomposition as in Eq. (7), the decomposition formula regarding the CO₂ emissions induced by imports can be obtained as

where π is an $(N\times 1)$ column vector whose i th element, ${\pi }_i$, is the import composition of imported commodity i, and $X_m$ is the total amount of imports to Japan.

Author’s Contributions

SO proposed the SDA method, conducted data analysis, and provided policy implications.