Economic Performance, Greenhouse Gas Emissions, Environmental Management, and Supply Chains in India: A Comparison with Japan

Using input–output tables and data on wastes from the Japanese industrial sectors, we have provided empirical evidence that, in Japan environmental performance of their upstream suppliers contributes positively to the performance of their final product assembly firms or economic sectors. In this paper, we propose to investigate the same hypothesis for firms and other establishments in manufacturing and other sectors in India. Indian supplier firms that sell goods and services to their client assembler firms are not generally structured in the form of efficient supply chains as in advanced economies. So, the environmental performance of these suppliers may not have positive impacts on the performance of their assembler firms or economic sectors, but this is yet to be verified empirically. In our study on Japan, we measured supply chains’ environmental performance using various amounts of waste materials and also CO 2 -equivalent greenhouse gas emissions generated in their production processes. Unfortunately, the only environmental performance data we have for the Indian economic sectors is their CO 2 emissions. So, we investigate the impact of CO 2 emissions by supplier firms on the economic performance of their assembler firms in India.


Introduction
Limiting the amounts of industrial wastes generated in firms' manufacturing processes has been of policy interest in recent years. A type of waste of our interest in this paper is greenhouse gases (represented by the carbon dioxide equivalent below). Even though it is not harmful to human health, CO 2 is being regulated like toxic industrial wastes in many developed countries including Japan. More recently, the importance of limiting CO 2 emissions globally has been recognized by both developed and developing nations, and an international treaty to strengthen the former Kyoto protocol was signed in Paris. 1 One of the topics of research interest, which has not received much empirical attention, is the extent to which CO 2 emissions, as an industrial waste, are generated along firms' supply chains. Although we see large corporations (e.g., 3M, Sony) promoting green procurement policies and claiming to use environment-friendly suppliers, we have little empirical evidence yet to suggest how such environmental management methods based on supply chains might benefit large downstream firms economically. We do not have much empirical evidence either about the impacts on final products of environmental management policies conducted by firms in their supply chains emerging in developing countries like India.
In this paper we present empirical estimates for the amounts of greenhouse gas (GHG) emissions generated by Indian manufacturing and other economic sectors, and their supply chains. (GHG emissions are measured in carbon dioxide (CO 2 ) equivalent in this paper.) We then estimate their contributions to firm performance measured in terms of value added. Figures 1 and 2 show CO 2 emissions per person and per income, respectively, in India and  1 The 2015 United Nations Climate Change Conference, held in Paris, France, from November to December 2015 was the 21st yearly session of the Conference of the Parties (COP) to the 1992 United Nations Framework Convention on Climate Change (UNFCCC) and the 11th session of the Meeting of the Parties to the 1997 Kyoto Protocol. The Paris Agreement, a global agreement on the reduction of climate change, the text of which represented a consensus of the representatives of the 196 parties attending it, was signed. It needs to be ratified to become a world treaty [1]. Japan over time. We see from these figures that while Japan emits more CO 2 than India per capita, Japan generates less CO 2 emissions per dollar than India does. The rest of the paper is organized as follows. After a brief review of earlier studies in Section 2, we discuss our method and approach toward the analysis of the generation of industrial waste (CO 2 emissions here) by supply chains in Section 3. Our data are briefly introduced in Section 4. We show and analyze certain patterns that are found in the generation of CO 2 emissions in Indian and Japanese industries in Sections 5 and 6. Section 6 presents our empirical results that relate the value added to the generation of wastes by downstream and upstream firms. Section 7 concludes.

Literature
There are relatively few research studies that use nations' input-output (I-O) tables as the data source for analyzing the relationships between supply chains and firms' environmental performance. Hayami et al. [6] present a framework in which I-O tables can be used for analyzing the effects, at the sector level, of the environmental management performance of firms in supply chains on their downstream assembly firms' performance. They present references on the literature that discusses many aspects of environmental management at upstream supply chains as related to their downstream customer firms [7,8]. Discussions on supply chains in India are also found [9][10][11]. Details of I-O analysis and applications to the Indian economy and environmental management are found in papers contained in [12].

Our approach to estimating output and waste along the stages of a supply chain
As noted earlier, certain downstream producers in developed countries are beginning to practice "green procurement," by which upstream suppliers with greener production proc-Economic Performance, Greenhouse Gas Emissions, Environmental Management, and Supply Chains in India: A Comparison with Japan http://dx.doi.org/10.5772/62535 esses become the preferred suppliers of their downstream customer firms. For example, Cisco, NEC, Sony, and Toshiba discuss their corporate green procurement guidelines in [13][14][15][16]. We apply this notion to India and investigate empirically the extent to which the same notion holds in India.
In order for the government to evaluate the potential benefits (i.e., the greening) of upstream firm production processes resulting from promoting downstream instruments, it is essential that we estimate relationships that describe the generation of waste materials at both upstream and downstream firms in a national economy. However, to our knowledge, only Hayami et al. present an empirical framework to achieve this objective using available data [6]. They also present an empirical model that allows us to estimate downstream firms' benefits of reduction of their suppliers' environmental wastes.
We apply the above model to India and derive some preliminary empirical estimates that evaluate the relative importance of the waste materials generated along the supply chain. Our findings in this chapter provide complementary evidence to the importance of environmental management in supply chains reported, for example, for individual firms, obtained using survey data and methodologies different from ours [8,17]. 2 , 3

Estimation of output along a supply chain
Our methodology is based on the input-output (I-O) analysis originally developed by Leontief [19,20]. (Applications of the I-O analysis to waste management and other environmental issues are found, for example, in [12,[21][22][23]. Additional uses of input-output analysis in environmental management are found in [24]. We divide an economy into industrial and other economic sectors where production of goods and services takes place. We define I-O technical coefficients a ij (i,j = 1,2,…,n) to be the amount of input from sector i per unit amount of output from sector j. To ensure positive output values, it is customary to assume the Hawkins-Simon condition [25] that a ij lie between 0 and 1 and their column sums are less than 1.
Suppose x j denotes the output from sector j. Then a ij are estimated as follows: where X ij denotes the amount of input from sector i that is required for the production of x j . Using supply chain terms, we say a ij connect downstream output from sector j to its immediate predecessor upstream input from sector i.
We denote by A an n × n matrix with elements a ij , and by x an n × 1 vector in which each component x j represents domestic production (output) of sector j (j = 1,2,..,n). We also denote by f i the final downstream demand for sector i. We denote by f the corresponding n × 1 final downstream demand vector. For example, f i = 1 means a unit final downstream demand for output from sector i. (For simplicity, we ignore the impacts of international trade.) In order to produce the final downstream demand f, the total amount of input required from sector i in the immediate predecessor stage (denoted by k = 1) is given by (1) is also interpreted to be the indirect demand for the previous stage (k = 1) production process, which is induced by final demand f, because without the production of x (1) , the final demand cannot be met. In order to produce x (1) , the total amount of input required from sector i in the immediate predecessor stage (denoted by k = 2) in the supply chain is given by the ith element of the following vector: Generally, we can trace production activities along the supply chain backward, starting from the final demand, and we get We call x (k) the kth stage indirect effect of final demand f (k = 1,2,…) in the supply chain.
In order to be able to produce final demand f, the following total indirect output must be produced: where (I -A) -1 is the Leontief inverse matrix which exists provided that the a ij satisfy the Hawkins-Simon condition given above.
We have shown that our input-output analysis identifies the successive upstream production processes that are followed by the average supply chain for the final demand vector f. This is summarized as follows. The input-output analysis describes all economic activities of the average supply chain in a national economy by following input-output transactions for all goods and services. The analysis typically starts from the final stage of downstream demand as shown above and moves backward by backtracking all predecessor upstream stages of production.
In this paper, we consider CO 2 (defined here to be the combined greenhouse gases measured in CO 2 equivalent) as a waste material associated with industrial production activities.

Graphical representation of connectedness of I-O sectors
Different sectors tend to be more connected in modern developed economies than in developing economies. This is because, in a modern economy, unproductive sectors will become more productive, with inputs from more productive sectors to survive. In addition, primary sectors and supplier sectors of manufacturing are connected to assembly sectors of manufacturing in a functional and efficient manner in supply chains. These functional connections are often missing in developing economies.    Intuitively speaking, Figures 3-5 show the degrees of connectedness between sectors in terms of economic transactions. For example, sectors whose transactions are mostly within themselves are depicted as single dots. On the other hand, if two different sectors have more transactions with each other, then those two sectors are connected by a box. Multiple sectors with transactions, such as sectors that define supply chains, are shown with larger boxes containing them. As expected, Figures 3 and 4 show that sectors of the Indian economy are not much connected to each other, though there are considerably more connectedness observed during 2003 than during 1995. This implies that there are increasingly more supply chain type relationships emerging in the Indian economy in recent years. The Japanese economy has developed well-defined supply chain based relationships among sectors in many industries [6]. This is clearly observed in Figure 5. We speculate from these figures, for example, that environmental management performance of upstream suppliers affect the performance of downstream firms much more in Japan than in India. 4 In the I-O analysis presented in Section 3.1, it is customary to include output which has economic value. 5 It is also customary to assume that industrial waste has no economic value in the form it is generated. For these reasons, industrial wastes are not included in our analysis in Section 3.1. 6 We treat waste materials separately here. Suppose we have estimated E1 j , the amount of waste generated per unit of output produced in sector j (j = 1,2,..,n). We denote by E the corresponding n × n diagonal matrix with E1 j in the jth diagonal position. Then the amounts of waste produced by the output of sector j along the successive stages of a supply chain are given as follows:

Estimation of wastes along a supply chain
Denote by w j the amount of waste generated in sector j, and denote by w an n × 1 vector consisting of w j (j = 1,2,..,n).
Then in the final stage, stage 0 (k = 0), of a supply chain, the demand is f, and the waste generated is which is the waste generated from assembly operations of final output f.
In the immediate predecessor upstream stage, stage 1 (k = 1), of the supply chain, the amount of waste generated (called indirect output for stage 1) is Similarly, we can derive the amount of waste generated along the upstream stages (k = 2, 3, …,) of the supply chain as follows: This is shown in the last row of Table 1. Production output along the stages of a supply chain

Downstream: final stage of a supply chain (final demand)
… + A k f Waste output generated along the stages of a supply chain Table 1. Production and waste output along the stages of a supply chain.

Output along a firm-specific supply chain and statistically obtained average output along the average supply chain
We do not have data on individual firm-specific supply chains that expand from downstream to upstream stages of production. However, element a ij of input-output matrix A = {a ij , i, j = 1,2,..,n} is in fact the statistically estimated average fraction of output of sector i that goes to sector j. This statistical method of obtaining matrix A (called the commodity flow method) thus allocates input X ij from the data of total output x j [27], that is, a ij connects downstream sector j to its immediate upstream sector i statistically. We have used this property of matrix A to obtain the average production and waste output along the stages of the average supply chain, x (k) and w (k) , k = 1,2,…, given downstream demand vector f. If we had data on production and waste output for the stages of all firm-specific supply chains for given f, then our I-O based estimates give the first-order approximation to the average output of the quantities for all such firm-specific supply chains. (The first-order approximation arises because of the linearity of a ij which defines the I-O matrix A.)

Input-output matrix
As we have noted in Section 3.1, our estimation methodology uses an n × n matrix A consisting of I-O technical coefficients a ij (i,j = 1,2,…,n), where n is the number of economic sectors being considered. Since 1973, estimated values for a ij (i,j = 1,2,…,n) are published every 5 years as I-O tables by the Government of India [9,28]. In this paper, we primarily use the

Waste and y-products surveys, and I-O matrix A
The environmental input-output table that we use here, based on greenhouse gas emission estimates (GOI, 2010), I-O table, material table, calorific table, combustion ratio table, and other data, was constructed by [9,12,29,30].

Calculating the amounts of waste materials
Using application of the input-output analysis described above, we used the estimated quantities of CO 2 for each of the 130 Indian I-O sectors, which we use in our regression analyses. We also used value-added estimations for each of these I-O sectors.
Using I-O analysis, we estimated the amounts of CO 2 generated per unit output for each of the 130 I-O sectors.
We are interested in studying the behavior of CO 2 emissions in firms' decision processes. In this paper, we denote by CO 2 emissions the total emissions in carbon dioxide equivalents of all greenhouse gases. CO 2 has certain characteristics in common in terms of their implications for firms' own economic incentives and government regulations. 7 For example, CO 2 emissions, like some other nontoxic wastes, are harmless to human health. On the other hand, CO 2 emissions, like some other waste materials, may also mean firms' excessive use of costly inputs (fossil fuels in case of CO 2 ). (Note that CO 2 emissions and fossil energy use are highly correlated [32,33].

Waste output along supply chains: example of an auto industry
One topic of research interest is to evaluate the relationships that might exist between downstream and upstream firms in terms of their waste behavior. Input-output analysis identifies statistically average economic relationships that exist between upstream and downstream firms. It is then possible to use input-output analysis also to find the average amounts of wastes that are generated by upstream firms in supply chains in response to production activities for the final products of downstream firms. Table 1 illustrates how our production of output and waste takes place along a supply chain starting from the final downstream demand. By tracing backward, final assembly plant 7 In addition to direct regulations of various types pertinent to toxic wastes and CO 2 , indirect regulations often dictate firms' management of some of nontoxic wastes as well. For example, firms' nontoxic wastes are sometimes indirectly regulated in terms of the amounts of such waste materials that firms are allowed to bring to landfills and other waste processing facilities. Some nontoxic wastes have commercial value as well.

Example of an auto industry example
Economic Performance, Greenhouse Gas Emissions, Environmental Management, and Supply Chains in India: A Comparison with Japan http://dx.doi.org/10.5772/62535 receives inputs from suppliers in upstream stage 1, who in turn receive their inputs from suppliers in upstream stage 2. As we have shown, I-O analysis allows us to estimate inputs between two successive stages of production along a supply chain.

A numerical example, India and Japan
This example illustrates the supply chain effects in the propagation of waste (CO 2 ) generation along supply chains in India and Japan.   Tables 2 and 3 show how much CO 2 emissions occur along the auto supply chains in producing passenger cars with certain characteristics: median size cars in India and cars with 2000 cc engines in Japan.
We see from Table 2 that firms along the auto supply chain in Japan generate 5.26577 tons of CO 2 emissions, but only 2% of this amount is generated by the final assembler firms. The remaining 98% of CO 2 emissions are generated by suppliers and other upstream firms in the supply chain. In comparison, the corresponding figures for India are: 4.485602 tons of total CO 2 emissions per car are generated in total, of which 3% is generated by the final assembler firms and the rest (97%) of the emissions are generated by suppliers ( Table 3). This similarity in the patterns of CO 2 emissions along auto supply chains between India and Japan suggests that production technology of autos is reasonably standardized, perhaps due to the fact that many auto plants in India are owned and operated to a large extent by Western automakers. Another noteworthy point is that total CO 2 emissions per car produced is somewhat lower in India than in Japan. This difference occurs in part because of the sizes of passenger cars considered here that are different between India and Japan, and also in part because of the difference between India and Japan in the amounts of CO 2 emissions induced by imported car parts. The use of more imported parts implies lower levels of domestic CO 2 emissions, which is the case for India. For passenger car production, this ratio is 0.05846 for India and 0.02316 for Japan.
Based on the results given in Tables 2 and 3, we conclude that government environmental regulations about greenhouse gas emissions need to include not only the final auto producers but also many upstream suppliers, in order to be effective.
We noted that our results in Tables 2 and 3 are consistent with the possibility that downstream firms might be able to upload the processing of CO 2 in particular to their upstream suppliers, while processing relatively large amounts of nonenergy-intensive tasks themselves in-house. This could easily happen in practice, since processing energy-intensive tasks is generally expensive.
We also note that this hierarchical structure of processing of the waste materials emitted by firms in assembly-based industries is likely to be typical. This is because of the nature of the types of assembly-based industries, which are most efficiently done by streamlining their supply chains so that assembly operations come last. In addition, assembly firms are generally more powerful than suppliers in their supply chains and hence have the most bargaining power.
Detailed processes of generation of CO 2 emissions by upstream and downstream firms are presented in Tables

Relative contributions of direct and indirect CO 2 emissions to the total sectorial emissions
It is intuitively clear that final output (called output from downstream sectors), whether assembled manufactured products, or output from primary sectors such as mining and agriculture, uses much output produced in their predecessor sectors including suppliers (upstream sectors). It is then likely that the total emissions attributable to any final product (e.g., a passenger car) consist of significant amounts of indirect emissions from upstream sectors and direct emissions which are emitted from the final car assembly stage in the downstream part of the supply chains. Figures 6 and 7 show the breakdown of direct and indirect emissions for 16 sectors. Industries 9-13 with asterisks are thought to be assemblybased manufacturing industries.
We see in these figures that CO 2 emissions are skewed toward upstream firms in manufacturing supply chains. This is particularly evident for Japan (Figure 7). Figure 7 also shows that proportions of toxic wastes show a similar pattern.  List of industries used in Figures 6 and 7

Sustainable Supply Chain Management
Are the patterns of CO 2 emissions across upstream and downstream economic sectors that we observed in Figures 6 and 7 consistent with downstream firms' profit maximization behavior?
We are interested in testing the following hypothesis: H1: Downstream firms' performance (measured by their value added) is affected by their upstream firms' CO 2 emissions as well as their own.
In general, we expect upstream firms' generation of toxic wastes such as CO 2 to be a negative factor in firms' value added, but generation of nontoxic wastes may not be, since most nontoxic wastes have commercial value. We first focus on the impacts of downstream firms' immediate predecessor upstream firms on downstream firm performance, because the impacts, if any, of downstream firms' environmental management policies such as green procurement can extend most effectively to their immediate predecessor upstream suppliers. We have also run regressions with log of value-added as the dependent variable. We obtained estimation results which are qualitatively the same. Further, we experienced considerable multicollinearity when indirect emissions from both first and all stages entered regressions. Therefore, we only report regressions with either one of the indirect emission variables here. These regression results were calculated by the authors. Results for Japan in columns E and F are also reported in [6]. We test this hypothesis empirically by estimating the following regressions using a sample of economic sectors corresponding to Indian input-output sectors for which usable data are Economic Performance, Greenhouse Gas Emissions, Environmental Management, and Supply Chains in India: A Comparison with Japan http://dx.doi.org/10.5772/62535 available. The data used includes value added and the amounts of CO 2 emissions generated during direct and indirect stages of production for each of the input-output sectors in the sample. (Descriptive statistics for these variables for India and Japan are presented in Appendix 2.) In our specification, we regress value added on the amounts of CO 2 generated directly by downstream firms as well as the amounts of CO 2 generated indirectly by their upstream producers. Our OLS regression results for India are given in columns A and B of Table 6. Columns C and D show the corresponding results for Japan. 8 Even though CO 2 is not thought to be one of the industrial wastes in a traditional sense, the amounts of CO 2 emissions represent the levels of firms' inputs of fossil fuels. As such, like some other toxic wastes, firms have economic incentives, even without government regulations, to reduce such emissions of CO 2 , since the cost of energy can be a significant portion of firms' production costs. Furthermore, from policy perspectives, some policies introduced by the governments of developed countries have been promoting energy-efficient production processes for many years (e.g., beginning in the late 1970s, after the second oil crisis in Japan). And also, in recent years, CO 2 emission quota policies of various sorts are being introduced in Japanese, EU, and other nations' industries.
From Column C of Table 6, we see that 1 ton of direct waste output of CO 2 contributes to −0.0018 of firms' value added per yen of firms' output. On the other hand, contribution to firms' value added of the indirect waste output of CO 2 from their immediate upstream predecessor suppliers is −0.0115, which is numerically much larger and statistically more significant than our direct waste output. We conclude that firms face significant financial losses, measured by value added, when direct and indirect generation of CO 2 occurs in their own production processes. Generation of CO 2 emissions by firms' immediate upstream predecessor suppliers seems to have much larger negative effects on their value added than their own direct waste output. This suggests that downstream firms may have economic incentives to reduce waste output by their immediate predecessor upstream suppliers.
Comparing columns A (India) and C (Japan), we see similar patterns on how CO 2 emissions along supply chains affect final sectors' value added. As far as final sectors' direct emissions are concerned, direct CO 2 emissions have no impact on value added for India, since its coefficient (−0.0043) is statistically not significant. On the other hand, their immediate predecessor CO 2 emissions negatively affect final sectors' value added (with statistically significant coefficient (−0.0107). But direct emission coefficients in Column B are positive and statistically significant (0.0649), suggesting that the more fossil energy is used by the final sector, the more productive (in terms of value added) final sectors become. This might indicate inefficient use of fossil energy, but this is not clear, since the same coefficient in column A is statistically insignificant. 9 In all cases, indirect emissions from all supplier stages combined are statistically significant and negative. From these results, we tentatively conclude that, for India, environmental management policies encouraging suppliers in supply chains to reduce their CO 2 emissions will likely improve final sector firms' performance measured in terms of value added. These results for India are consistent with but are not as strong as the policy conclusions obtained for Japan [6].

Concluding remarks
Recent advances in supply chain based management methods have made it possible for many firms to organize their production and other business activities as part of the supply chains they belong to. Efficiency gains are realized in terms of reduced inventories, reduced lead times for new product development, and shorter delivery lags, among many other benefits. Our results suggest that including certain supply chain level environmental management schemes, such as "how to manage toxic and nontoxic wastes, as well as CO 2 emission for a supply chain as a whole," in such supply chain management methods might improve not only downstream firms' economic performance but also advanced economies' environmental performance significantly.
Consideration of such schemes may underlie some firms' proposals for green procurement policies. In many sectors of an advanced economy, as supply chain management becomes more sophisticated in pursuing economic efficiency, larger downstream firms tend to become more dominant as the primary driver of management decisions associated with their supply chains.
(Note, however, that this phenomenon is not limited to assembly-based manufacturing industries. In retail industries, Walmart and the like have become the primary decision makers for their entire global supply chains.) It is possible that, as a national economy develops and increases its sophistication in logistic capabilities, organic connections between upstream and downstream firms become more prevalent, as we see in Japan. This might make it easier for some downstream firms to adopt green procurement policies.
Another factor that might be important to consider in supply chain based environmental policies is firm ownership structures. Ownership structures of firms involved in supply chains are complex but tend to share some systematic patterns. Dominant downstream firms generally influence business decisions of their upstream suppliers via some forms of partial ownership and/or certain guaranteed purchase agreements. Dominant firms do not necessarily extend their partial ownership to all other firms in their supply chains, but, nevertheless, dominant firms often have significant influence over smaller upstream firms through various sorts of business relationships.
Current public policies on waste management in Japan focus on firms and/or establishments. Because of the reasons stated above, this is not appropriate for an advanced economy in which many firm decisions are made at their supply chain levels in interrelated ways. Our empirical results present limited evidence, for both India and Japan, that downstream firms' economic performance is affected not only by their own environmental policies but also by the environmental behavior of their upstream suppliers. Some profit-maximizing firms may see