CC BY
204Carbon Dioxide Emissions from Indian Manufacturing Industries: Role of Energy and Technology Intensity*
Santosh KUMAR SAHU
Lecturer, Madras School of Economics, India [email protected]
K. NARAYANAN, Professor
Department of Humanities and Social Sciences, Indian Institute of Technology, Mumbai, India [email protected]
Abstract. Industrial energy efficiency has emerged as one of the key issues in India. The increasing demand for energy that leads to growing challenge of climate change has led to major issues. This paper is an attempt to compute Carbon Dioxide (CO2) emission from fossil fuel consumption for firms in Indian manufacturing sector from 2000 to 2011 by adopting the IPCC Reference Approach. The contribution of this paper lies in estimating CO2 emission at the firm level and analyzing the factors that explain inter-firm variation in CO2 emission. The results indicate that there are differences in firm-level emission intensity and they, in turn, are systematically related to identifiable firm specific characteristics. This study found size, age, energy intensity and technology intensity as the major determinants of CO2 emission of Indian manufacturing firms. In addition, capital and labour intensity of the firms are also related to the firms' CO2 emission intensity. We conclude the short run policy implications should be aimed at encouraging firms to invest more in R&D and technology sourcing, and at long run firms should be able to adapt cleaner energy to reduce CO2 emission from the fuel consumption.
Аннотация. Вопрос энергоэффективности промышленности Индии в последнее время приобрел ключевое значение. Увеличение спроса на энергию, которое вызывает растущую проблему изменения климата, привело к серьезным последствиям. Эта статья является попыткой оценить выбросы углекислого газа (CO2) в индийской обрабатывающей промышленности с 2000 по 2011 год на основе методики МГЭИК. Результаты данного исследования свидетельствуют, что существуют различия в интенсивности выбросов на уровне предприятий, которые, в свою очередь, находятся в зависимости от идентифицируемых специфических характеристик предприятий. Исследование показало, что к числу основных факторов, определяющих выбросы CO2 индийских предприятий, относятся размер, время использования, энергоемкость и интенсивность технологий. Кроме того, объем капитала и интенсивность труда также связаны с интенсивностью выбросов CO2. Мы пришли к выводу, что в краткосрочной перспективе политика должна быть направлена на поощрение фирм инвестировать больше в НИОКР, а в долгосрочной перспективе компании должны переходить на более чистые энергетические технологии для снижения выбросов CO2 от потребления топлива.
Key words: CO2 emission, technology intensity, firm heterogeneity, panel data, Indian manufacturing.
1. introduction
Political efforts to address the concerns of climate change have developed greatly in the last twenty years. In 1992, at the Rio Summit, the United Nations Framework Convention on Climate Change (UNF-CCC) was established. Then, in 1997, despite its flaws, the Kyoto Protocol set targets to curb greenhouse gas emissions on a number of industrialized countries emis-
sions between 2008 and 2012. While natural scientists identified the relationship between greenhouse gas concentrations and climate change and highlighted many of the threats, social scientists, particularly economists, played a crucial role in developing strategies for mitigating climate change (Nordhaus, 1991; Cline, 1992; IPCC, 2007). Economists have been influential in arguing that the cost of mitigation may not be as great as many
* Зависимость уровня выбросов углекислого газа в индийской обрабатывающей промышленности от энергетических технологий.
expected (Porter, 1991; Fischer & Newell, 2008) and there may be substantial benefits (Stern, 2004; Sterner & Persson, 2008). They also proposed mechanisms for trading responsibilities and credits related to greenhouse gas (GHG) emission reductions, which have been central tools to agreement on targets related to Kyoto Protocol (Atkinson & Tietenberg, 1991; Stavins, 1995). At a national level, many governments have introduced taxes to discourage the consumption of high-carbon energy sources (Pearce, 1991; Newbery, 1992; Oates, 1995; Parry and Small, 2005; Nordhaus, 2007; Sterner, 2007). In other words, economists have become highly influential in the global efforts to achieve climate stability.
In a parallel line at macroeconomic perspective there is no consensus on the effect of international trade on the environment; and in particular on the effect of trade on global emissions. Neither theoretical nor empirical literature provides a clear-cut answer to the link between trade and CO2 emissions. Literature survey emphasizes on limited studies in explaining patterns of emission for industry or firm level. However, studies in eco-innovation are given much importance. Studies in eco-innovation can be broadly divided into two categories:
1) The first mainstream research deals with the drivers of eco-innovation strategies. The seminal work by Jaffe and Palmer (1997) studies environmental innovation (R&D and patents) at industry level, followed by Brunnermeier and Cohen (2003), employs panel data on manufacturing industries to provide new evidences on the determinants of environmental innovation measured by number of patents. Rennings et al. (2003) exploit OECD survey data in order to investigate whether environmental auditing schemes and pollution abatement innovation are correlated. Mazzanti and Zoboli (2008) present evidence for manufacturing sector at a district level, focusing on an extended set of drivers (environmental R&D, policy induced costs, industrial relations, and other innovations). Frondel et al. (2004) use an OECD survey data on manufacturing firms and focus on internal firm-based strategies, external policy variables and test the drivers for end-of-pipe measures or integrated cleaner production processes.
2) The second stream of research is focused on environmental innovation and employment effects. The main contributions in this stream include Rennings & Zwick (2001) and Pfeiffer & Rennings (1999).
In economic literature, it is also argued that innovation through technological advancement makes firm/ industries competitive and productive. In such reviews, arguments are also attempting to link the complementarity of energy to capital, where energy is considered as an instrument to capture the technological indicator. This line of study follows productivity framework and necessarily tries to estimate the relationship of energy
and non-energy inputs. Gap in literature lies in analyzing the patterns of firm level emission, and relate to firm characteristics. This study attempts to fill this gap and computes firm level CO2 emission. Given the absence of theoretical and empirical reviews in this line of thought, we employ the structure-conduct-performance paradigm in analyzing the inter-firm differences in CO2 emission. This paper assumes the choice of primary source of energy as one of the eco-innovation strategies of firms, and we would like to arrive at the inter-firm differences in the determinants of the negative externalities, specifically the CO2 emissions from the choice of fossil fuel consumption. The remainder of the paper is as follows. Section 2 discusses the review of literature; section 3 describes the estimation of CO2 emission; section 4 describes the econometric model and variable construction. Section 5 presents the descriptive analysis of the sample. Section 6 presents the empirical results, and section 7 concludes with policy implication.
2. REVIEW OF LITERATURE
This section of the paper attempts to look at the mainstream research similar to the objectives of this work. To start with eco-innovation, the study carried out by Rennings and Zwick (2001) is based on a sample of eco-innovative firms for five European Union (EU) countries in manufacturing and service sectors. The result of the study indicates that in most of the firms employment does not change as a consequence of eco-innovations. The econometric results show that, apart from some product innovations, eco-innovation typologies do not influence the level of employment, though as expected, according to their evidence environmentally oriented innovations seem to lead to a skill-based effect. Also end-of-pipe innovations are related to a higher probability of job losses, while innovations in recycling have a positive effect on employment. Employment effects may thus be unevenly distributed with strong negative effects from environmental strategies/policies on low skills intensive industries and potentially positive effects on other industries. It could also be argued that product and process eco-innovation strategies may bring about (potentially negative) net effects on employment, attributable to a destruction of the low skilled labour force and a creation of high skilled positions (R&D).
There is a complementary stream of literature that has focused on the various static and dynamic relationships between eco-innovation, environmental performances and firm performances. Konar and Cohen (2001) investigated the effect on firms' market performance of tangible and intangible assets, including two environmental performance-related elements as explanatory factors. Cohen et al. (1997) also analyzed the relation-
ship between environmental and financial performances. Overall, authors found that investing in a green portfolio did not incur a penalty and even produced positive returns. Gray and Shadbegian (1993) used total factor productivity and growth rates of firms over 1979-1990 as performance indicators to test the impact of environmental regulations and pollution abatement expenditures. They found that 1$ more expenditure on abatement is associated with more than 1$ worth of productivity losses. Analysis on variations over time or growth rates, the relationship between abatement costs and productivity was found insignificant. Greenstone (2001) estimated the effects of environmental regulations, using data from 175 million observations of firms in 1967-87, US censuses of manufacturers. According to the study environmental regulations negatively affect growth in employment, output and capital shipments.
The EU-based study by Ziegler et al. (2008) focused on the effects of environmental strategies on the stock performances of corporations using standard cross section/ panel approaches and event studies that analyze whether there are exogenous unexpected policy effects on the short-term performance of environmentally minded firms. This study was criticized for their intrinsic very short-term focus. Based on official datasets they conclude that the evidence focusing on stock market performance is limited since the majority of firms especially in Italy are of medium or small size and do not appear in stock market data. Innovation dynamics are close to productivity trends, which in the end are the main engines of firm performance. Doonan et al. (2005) examined the role of communities to create incentives for local industrial facilities to reduce pollution. They found that firms face both internal and external pressures to improve their environmental performance. Using primary data collected for 750 Canadian pulp and paper industries during 1992, they found that the government policies are much of a barrier for the industries. However, financial and consumer markets are not most important barriers. They found that education status of employee is one of the important determinants of environmental performance. The regulatory intervention is also found as the major determinant of environmental performance of the pulp and paper industries. In case of the Indian manufacturing industries firm level energy intensity and their determinants have been studied majorly by Goldar (2011) and Sahu & Narayanan (2011). Both the studies use Indian manufacturing data from CMIE PROWESS online database and follow structure-conduct-performance theory of the firm and analyzed the determinants of energy intensity at firm level. In both of the studies energy intensity is considered as a proxy for energy efficiency of firm. However, what is relevant to our study is the main hypothesis that increasing environmental efficiency by environmental innovations
strengthens competitiveness and the firm heterogeneity. The above discussion on the existing review of literature concludes that environmental performance has direct or indirect relationship with the firm performance, in terms of employment or in terms of productivity growth. However, none of them have linked the negative externality such as CO2 emission as a byproduct of the firm to firm heterogeneity. Hence, the motive of the paper is to focus on emission (CO2) intensity with the firm heterogeneity.
3. THE CONCEPTUAL FRAMEWORK
One of the objectives of this work is to estimate the CO2 emission at firm level. Further, we econometrically model the factors explaining determinants of inter-firm differences in the CO2 emission. We begin explaining the construction of the firm level CO2 emission for the sample of firms in Indian manufacturing industries. From figure 1, we can see that share of CO2 emission of the manufacturing industries (at aggregate level) is higher as compared to other emissions (LEAP)1. From the figure we can also observe that industrial value added is increasing from 2000 to 2010 with fluctuations. The industrial value added is downward sloping till 2005, the share of CO2 emission shows an increasing trend for the entire period. LEAP captures this data from the output point of view. Data at the aggregate level is available, but firm level emission information is not reported. One of the ways to capture the firm level emission is to compute the emission from the input use that is from the fossil fuel used by the firms. This is an indirect measure based on a scientific approach, closely related to the emission generated from firm according to the IPCC.
The estimation of emission from the fossil fuel consumption is based on the IPCC reference approach that refers as a top-down approach using aggregate information of fossil fuel consumed, to calculate the emissions of CO2 from combustion. However, the study has few data limitations such as quality of coal used. This is not considered mainly because the calculation is carried out for the first time at firm level in Indian manufacturing firms using PROWESS data base. Data is collected from the Center for Monitoring Indian Economy database PROWESS 4.0. This data is a combination of the annual audited balance sheet (that gives information of the firm characteristics) and energy consumption at firm level. Therefore, firms that don't report energy consumption are dropped from the active data sheet. Also, since we are adopting the IPCC reference approach, we have considered only fossil fuels consumed by the firms.
1 The Long-range Energy Alternatives Planning system (LEAP) is a widely-used software tool for energy policy analysis and climate change mitigation assessment developed at the Stockholm Environment Institute (SEI).
Year
Share of CO2 emission in total emission from industries
Industrial Value Added
figure 1. Share of CO2 emission and industrial value added in Indian manufacturing industries Source: Author's estimate from LEAP Data Base, www.energycommunity.org/LEAP.
The IPCC reference approach of estimating emissions from fossil fuels is as follows:
7/ - ^
C02 = z
V
(acf x cff x ccf )x x10-3 -ec
f
44
xcoff x —
y
12
(1)
where, acf = apparent consumption fuel, cff = conversion factor for the fuel to energy units (TJ) on net caloric value basis, ccf = carbon content (ton C/TJ i.e. to kg C/GJ), ecf = excluded carbon defined as carbon in feed-stocks and non-energy use excluded from fuel combustion emissions (Gg C), coff = carbon oxidation factor defined as fraction of carbon oxidized (usually the value is 1, reflecting complete oxidation). Lower values used only to account for carbon retained indefinitely in soot, and (44/12) is the molecular weight ratio of CO2 to carbon (C). Further, following Chen et al. (2010) we construct the firm level emission from equation (1) as:
C = Z Ct = Z Eht X NCV, X CEFt X COFt x (44/12)
i=1 i=1
(2)
Where, Ct = flow of carbon dioxide with unit of 10,000 tons, NCVt = net calorific value provided by IEA energy statistics for India, 2011, CEFt = carbon oxidization factor provided by 2006 National Greenhouse Gas Inventories in IPCC, COFi is the carbon oxidization factor set to be 1 in this study. Therefore, based on equation (2) in manufacturing industries the calculated CO2 emission coefficient for coal is 2.0483 (kg CO2/ kg coal), for oil 3.272 (kg CO/ kg oil) and for natural gas 2.819 (kg CO/m3 natural gas).
4. ESTIMATION METHOD AND vARIABLE CONSTRuCTION
Following Copeland and Taylor (1995) and assuming each firm produces two outputs: a manufactured good (x) and emission (e), the testable implication of the study follows a log linear relation of the following type:
ln- = \fh x
(3)
y
Where, ln— = Natural log of firm level emission intensity and fh is a vector representing firm characteris-
x
tics. We use an unbalanced panel data for the estimation of equation (3). Following similar framework as in Goldar (2011) and Sahu & Narayanan (2011) for the Indian manufacturing industries, the general form of equation (3) is estimated with the following econometric specification:
e 22
ln - = a,-, + PiC iit + p2 liU + $3eiU + P2 st + P2 su + P2 aSit + p2a&2 + P2 tit + P2 rdit + $2mneit + Pi, (4)
X it
where ci: capital intensity, li: labour intensity, ei: energy intensity, s: firm size, s2: square of firm size, ag: age of the firm, ag2: square of age of firm, t: technology import intensity, rd: research and development intensity and mne: multinational affiliation.
Different empirical works that study reasons for energy (in)efficiencies pay attention to the market share or value added to the industry output and find the evidence that it can make a contribution to the explanation of inefficiencies as the factor of market power (Hrovatin and Urib, 2002). It is worth mentioning, that fossil energy resources are characterized by the considerable undesirable outcome (such as CO2 emissions) and still their share in total energy generation is dominant, while the role of renewable energy sources is comparatively low, though extended recently. We have selected the following variables which influence the emission intensity of firms. Output is deflated net sales adjusted for change in inventory and purchase of finished goods. In PROWESS database the purchase of finished goods is defined as finished goods purchased from other manufacturers for resale. Hence, we subtracted purchase of finished goods from sales to arrive at the firms manufactured output. A positive increase in inventory is added to sales to arrive at output and a decrease subtracted.
Capital is measured as defined in Srivastava (1996) for the measurement of capital stock, which revalues the capital given at historical cost to a base year. Actual investment for the present period is estimated by taking the difference between Gross Fixed Asset (GFA) for current year and that of last year. The real investment value is expressed in the base price of 1993-1994=100. This enables us to use the perpetual inventory method to construct capital stock. In estimating the capital stock we first revalue the GFA at historical cost to a particular base year value. We have used GFA, after deflating it with the wholesale price index for machinery and machine tools, as plant and machinery accounts for 71 percent of the GFA (RBI Bulletin, 1990). Firms can gain a technological advancement not only through their own innovation but also through purchases of new capital or intermediate goods from other sectors. Capital intensity is measured in terms of deflated GFA as a proportion of output.
The PROWESS database provides information on wages and salaries of the firm and provides no information on the number of employees. Therefore, we need to use this information to arrive at the number of persons engaged in each firm. Number of persons engaged in a firm is arrived at by dividing the salaries and wages at the firm level by the average wage rate of the industry (at the three digit level) to which firm belongs. Hence, Number of persons engaged per firm =Salaries and Wages/
Average Wage Rate. To arrive at the average wage rate we make use of the Annual Survey of Industries (ASI) data on Total Emoluments as well as Total Persons Engaged for the relevant industry. And Average Wage Rate = Total Emoluments/Total persons engaged.
In most of the productivity studies of four factors of production, energy consumption is considered as one of the indicators for innovation. This implies that in cost minimization a firm can shift from one source to the other sources of energy. Hence, it will be of interest to check the relationship between energy efficiency and emission at the firm level. Size of the firm is the proxy for several effects as observed by Bernard and Jensen (2004). Size of firm is one of the components of firm heterogeneity. Because of scale economics bigger firms might use the efficient fuel and emit less. In the present study, firm's size is measured by the natural log of total sales. There could be a non-linear relationship between emission intensity and firm's size. Age of the firm is calculated as the deference between years of the study to year of the incorporation of the firm as reported in the PROWESS database.
Technology import intensity is defined as the expenses on import of capital goods and royalty and technical fees payments in foreign currency, to net sales of the firm. Higher the technology import it is assumed that firm might be emitting less as technology advancement of the firm might enable the firm to be energy efficient and emit less. R&D intensity is also one of the innovation strategies that might help firms in emitting less. Here, we define R&D intensity as the ratio of R&D expenditure to net sales. There is empirical evidence that foreign-owned firms tend to be more efficient in energy conservation (Faruq and Yi, 2010) and, at the same time, there is also evidence in Zelenyuk and Zhe-ka (2006) that reveals a negative correlation between foreign ownership and firm's environmental efficiency level. We have created a dummy to capture the multinational affiliation (mne), where firm belonging to foreign affiliation takes a value 1 and the domestic firms takes a value of 0.
5. TRENDS AND PATTERNS OF CO2 EMISSION: DESCRIPTIVE ANALYSIS
This section of the study depicts the descriptive analysis of the sample. We have estimated CO2 emission based on equation (2), from the fossil fuel consumption of sample of firms in Indian manufacturing from 2000 to 2011. Figure 2 presents the aggregate mean annual CO2 emission of the sample. From the figure we can observe that the aggregate CO2 emissions of the sample of Indian manufacturing firms are fluctuating over the period with an increasing trend from 2005 to 2011.
12000 10000 J 8000
5« 5«
S
w 6000
cO
u
4000 2000 0
vo vo vo
K>
o o o
k> o o
K>
o o
K>
K>
o o
K> K> K>
o o o
o o o
-Ê- <Ji CT\
K> O
o
K>
o o 00
K>
o o
vo
K>
o
K>
o
Year
■ CO Emission
---Poly (CO2 Emission)
Figure 2. Average annual CO2 emission (over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy.
S
W
cT
u
0,7 0,6 0,5 0,4 0,3 0,2 0,1 0
vo vo vo
k> o o o
K>
o o
K>
o o
K>
K>
o o
(Jj
K>
o o
K>
o o
K>
o o
Cn
K>
o o
K>
o o 00
K>
o o
vo
K>
o
K>
o
Year
' CO2 Emission Intensity
---Poly (CO2 Emission Intensity)
Figure 3. Average annual CO2 emission intensity (over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy.
Emission intensity is considered as better measurement as compared to emission in absolute form as it is normalized with the output of the firm. Emission intensity is drawn in Figure 3. The standard deviation between the average emission and emission intensity, however, are different. We can observe that for emission intensity, the standard deviations across years are fluctuating as compared to the standard deviation
of aggregate emission for the sample in Figure 2. If we compare between the distributions of both the series we can observe that emission of the sample are more stable as compared to emission intensity. However, the trend is quite similar for both the distributions. One of the interesting findings of this comparison is that the trend in average annual emission is flatter then the emission intensity over the years, which is due to the
45 40 35 30 25 20 15
8 10
5
o o o o o o o o o o
O i—' K< <Jj -ti-
to K<
o o
o o
<J1 CT\
K<
o o
K<
o o 00
K<
o o
vo
K<
o
K<
o
Year
•CO2 Emission Intensity
Tech Intensity
10 9 8 7 6 5 4 3 2 1
Figure 4. Comparison of CO2 emission and technology intensity (over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy.
M
o
"S s ■s
o
s
structure and production of the firms. Hence, when we normalize the emission with output we can get a different movement as compared to the absolute level of emission due to fossil fuel consumption.
From this discussion, we can assume that increase in emission of firms is related to the output, which is definitely related to the technology in use for the production process. To check this we plot technology intensity of firm with the emission intensity for the same period. Figure 4 presents this exercise. From this figure we can observe that from 2000 to 2008 the emission intensity is fluctuating while technology intensity was increasing. However, from 2009 even technology intensity continued to increase, emission intensity has started declining. For few years such as 2001, 2003 and 2008 emission intensity is higher and for the rest emission intensity has been less.
One of the indicators of firm heterogeneity is multinational affiliation of firms. In Figure 5 we have plotted technology import, R&D and the emission intensity of firms. As a result of technology import and investment on research & development a firm might lead to reduce emission. For a growing country such as India, we need to also look at the differences in emission for the multinational affiliated and the domestic firms. From Figure 5 we can see that foreign firms are higher technology intensive and higher R&D intensive. However, the domestic firms are emitting high. Emission intensity is not widely differenced, but domestic firms emit more than the foreign firms. Twelve years' data of the sample states that the CO2 emission is higher for the domestic firms (0.32) as compared to the foreign firms (0.21).
Firms use different sources of fossil fuel as primary sources of energy hence, emissions from each of the energy type are supposed to be different. Therefore, we have attempted to see the difference in emission from different fossil fuel use. In addition we would also like to compare technology intensity with emission with different fossil fuel consumption. Figure 6 gives the comparison of energy, technology, R&D and emission intensity classified by different sources of primary energy consumption. From the figure we can see that energy intensity is higher for firms using coal and oil, however, firms using natural gas are energy-efficient. In case of technology intensity we can observe that firms using coal as primary source of energy are importing higher technology as compared to firms using natural gas and oil. However, oil-consuming firms are higher R&D intensive as compared to the other two classifications. Emission intensity is similar for firms using coal and oil, where coal-using firms are found higher emission intensive as compared to oil-using firms, but the firms using natural gas are least in emission intensity.
As literature suggests, the determinants of firm heterogeneity are size and age of the firm. We have classified the sample based on age and size distribution of firms and relate with firm level emission. Figure 7 presents group of firms classified based on size and age. In case of age of the firms we have created four classifications. The classification G-1 represents firms' age between 1-10 years old, G-2 represents 11-25, G-3 represents 26-50, and G-4 represents firms older than 51 years. Similarly, for size of the firms: G-1 represents 10th percentile of the sample, G-2 represents 25th percentile, G-3 represents 50th percentile and G-4 represents high-
0
0
0,60 0,50 0,40 0,30 0,20 0,10 0,00
Technology Intensity R&D Intensity Emission Intensity ■ Foreign Firms ■ Domestic Firms
Figure 5. Emission, technology and R & D intensity of domestic and foreign firms (average over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy.
0,00
-
- 0,223 0,196 0,222
- 0,06 0,06 0,07
0,11 0,10 0,09
0,08 0,07 0,08
Coal
1 Energy Intensity R&D Intensity
Natural Gas
Oil
1 Technology Import Intensity 1 Firm Level Emission
Figure 6. CO2 emission, energy, technology and R & D intensity between three fuel sources (average over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy
er than the 50th percentile. Emission intensity of G-2 firms is the least as compared to the other classifications. Older firms are emitting the highest. For size of the firms, bigger firms are emitting higher compared to the smaller firms. The cross-tabulation might not give the concluding result between firm size and age of the firm to emission intensity; hence the econometric analysis will help us in determining the relation for policy formulation.
As of now we have analyzed the sample at aggregate level. However, as we know the sample consists of different types of firms at industrial classification. There are firms which use energy more and emit more. However, energy and emission intensity can be different based on the output level of such industrial classes. Hence, further the sample is classified in two digit industrial classification based on the NIC-2008 GoI classification. Indicators such as energy and emis-
sion intensity are further calculated for the set of these classifications. Table 1 and Figure 8 present the result. From the table and the figure we can observe that coke and refined petroleum industries are energy intensive in the set of 17 classes of two digit industries, whereas industries related to printing and reproduction are found to be energy efficient in the same classification.
Five energy intensive industries from the sample are (1) coke and refined petroleum, (2) paper and paper, (3) textiles, (4) wearing apparel and (5) fabricated metal products industries. In the similar classification five energy efficient industries are (1) printing and reproduction, (2) beverages, (3) food, (4) wood products and (5) computer and electronic industries. For the emission intensity we have also tried the similar classification and found that coke and refined petroleum industries are the emission intensive ones, and food industries are the least emitting industries for the sample. The
G-2 G-3 G-4
Classification of Age of the Firm
1 Emission based on Age Emission based on Size Mean Emission
Figure 7. CO2 emission for group of firms classified by age and size (average over study period) Source: Author's estimates from PROWESS, Center for Monitoring Indian Economy.
Table 1. Comparison of energy intensity and CO2 emission intensity across industries over study period
NIC Industry Type Energy Intensity CO2 Emission Intensity
18 Printing and reproduction 0.116 4.195
11 Beverages 14.381 6.443
10 Food 19.059 1.586
16 Wood products 29.519 21.002
90 Computer and electronic 32.557 6.754
12 Tobacco 44.398 8.051
15 Leather 46.320 33.301
60 Manufacture of basic metals 48.742 7.490
22 Rubber and plastics 56.561 40.630
20 Chemicals 60.122 4.399
50 Non-metallic mineral products 61.970 10.320
40 Pharmaceuticals, medicinal chemical 69.943 14.709
80 Fabricated metal products 86.847 9.532
14 Wearing apparel 109.536 8.174
13 Textiles 159.866 28.508
17 Paper 165.525 22.240
30 Coke and refined petroleum 252.124 55.362
Full Sample 104.702 22.977
Source: PROWESS, Center for Monitoring Indian Economy.
top five emission intensive industries in the sample are (1) coke and refined petroleum, (2) rubber and plastics, (3) leather, (4) textiles and (5) paper industries. The top five emission efficient industries are (1) food, (2) printing and reproduction, (3) chemicals, (4) beverages and (5) computer and electronic industries.
A comparison of energy and emission intensity of the three digit classification of industries is presented in Figure 8. From the top five industries in case of efficient and intensive firms for energy and emission
intensity it is not clear whether industries which are higher energy intensive also higher in emission intensity. Hence, we have attempted the Spearmans' rank correlation analysis for three digit industrial classification for energy, emission intensity and technology intensity of the industries. The technology intensity is classified in two major classifications: (1) technology import intensity and (2) research and development intensity. From the analysis of the rank correlation we conclude that energy and emission intensity are highly correlat-
Energy Intensity CO2 Emission Intensity
Figure 8. Comparison of energy intensity and CO2 emission intensity across industries Source: PROWESS, Center for Monitoring Indian Economy.
ed at 1% of statistical significance. The relationship is found to be positive hence, energy intensive firms are found to be higher emitting. In addition we can also observe from Table 2 that research and development intensity in negatively related and statistically significant to energy intensity of industries.
Above discussion has tried to link emission intensity with technology intensity and firm heterogeneity. We began from the macro point of view in relating industrial value added to CO2 emission of industries. From the LEAP database we confirm that CO2 emission is one of the major emissions from the industrial sector of the Indian manufacturing industries. However, this emission information is from the output generated of the firms as bad output. As data on CO2 emission at firm level is not available we have constructed the emission at firm level from the IPCC reference case approach. Further, we also classified emission intensity to different firm characteristics, specifically MNE affiliation, firm size, age etc. Further, at three digit industrial classification we have computed the emission and energy intensity to find out the intensive and efficient industries in the sample. Also, this section tried to establish relationship of emission, energy intensity with technology intensity using rank correlation. However, the question of why firms differ in CO2 emission was not established. The next section tries to attempt this question using an econometric approach at firm level.
6. THE ECONOMETRIC RESULTS
The theoretical background of the estimated equation (4) is given earlier in section 3. However, before at-
Table 2. Rank correlation between various intensities.
Energy Intensity co2 Emission Intensity Technology Import Intensity R&D Intensity
Energy intensity 1.000
co2 emission Intensity 0.633""" 1.000
Technology import intensity -0.377 0.082 1.000
R&D intensity -0.499"" -0.226 0.193 1.000
Source: PROWESS, Center for Monitoring Indian Economy.
tempting equation (4) let us compute the correlation matrix for select variables. From the correlation coefficients (Table 3) it is seen that energy intensity is statistically significant and positively related to emission intensity, where size of the firm is negatively related and statistically significant to emission intensity. Capital intensity, age of the firm, technology import intensity and R&D intensity are also found to be positively related to firm level emission. Equation (4) is estimated first using pooled OLS and further using panel data econometrics that are fixed and random effects models. Based on the coefficients of Hausman statistics, the fixed effects estimates are selected over the random effects estimates. Methodologically, result of fixed effects is robust and efficient compared to the pooled OLS estimates. Hence, we have interpreted the results of the fixed effects model in Table 3. Equation (4) is a
Table 3. Correlation matrix
Capital Intensity Labour Intensity Energy Intensity Size of the firm Age of the firm Technology Import Intensity R&D Intensity Firm Level CO2 Emission Intensity
Capital Intensity 1.000
Labour Intensity 0.846 1.000
Energy Intensity 0.004 -0.016 1.000
Size of the firm 0.345 0.298 -0.235 1.000
Age of the firm 0.104 0.138 0.020 0.172 1.000
Technology Import Intensity 0.590 0.479 0.039 0.387 0.219 1.000
R&D Intensity 0.092 0.155 0.107 0.127 0.040 0.109 1.000
Firm Level CO2 Emission Intensity 0.024 -0.015 0.582 -0.232 0.077 0.117 0.007 1.000
Source: PROWESS, Center for Monitoring Indian Economy.
semi log model and the definition of firm's size is also natural log of net sales, hence the econometric specification turns out to be a double log model. Hence, the coefficients of the model are hence elasticities. The detailed results of the pooled OLS and random effects are presented in Appendix Table A1, and the result of the fixed effects model is described in Table 4.
Two parameters are considered for technology intensity, namely technology import intensity and R&D intensity. We can observe from the result of the fixed effects estimates that technology import intensity is statistically significant at 1% and carries a negative sign with emission intensity implying that firms importing higher technologies are emitting less. Hence, higher the import of technologies of firms, lower the emission intensity of the firms. Because we have constructed emission from the input use of firms in terms of energy choice, the result indicates only one explanation of the relationship that is technology intensive firms mostly use cleaner source of energy such as natural gas. From the cross tabulation it is also clear that natural gas using firms are emitting less. Hence, higher technology intensive firms might be using natural gas in the energy mix instead of other two inputs. From the result it is clear that research and development intensity is statistically significant and negatively related to emission intensity. This means that firms investing more in research and development are emitting less. This result is also similar to the earlier discussions on the selection of energy choice of firm. Relationship of technology import and research & development intensity with emission intensity clearly states that they are negatively related, and firms depending more on technology import and investing more in R&D are emission efficient.
The result of technology intensity can also be discussed as firms import technology for upgrade of plant and machinery or develop the output process by using
such technologies. In addition through higher R&D intensity firms learn from the technology imported and hence both parameters help the firm in achieving higher emission efficiency. Therefore, the role of technology intensity for firms is an important indicator in reducing emission. Linking these results with eco-innovation studies such as Konar and Cohen (2001) we can assume that technology import and the R&D investments are eco-innovation strategies of the firms in reducing emission intensity.
Since the sample consists of very small as well as very large firms, we have tried to establish non-linear relationship between firm's size and CO2 emission intensity. The result indicates a positive and negative coefficient for size and size square variable. This implies that very large and very small firms are emitting less and the medium sized firms are emitting more. Similar exercise is also carried out for the age of the firms. We also found a nonlinear relationship for emission and age of firm indicating an inverted 'U' shaped relationship as the coefficients are statistically significant at 1% and carry positive and negative signs. That in turn indicates that both older and younger firms are emitting less, whereas medium-aged firms are emitting more. These results are akin to the literature on environmental Kuznets curve for both developed and developing countries. It means that as size and age of the firm increase, CO2 emission also increases. However, with increasing innovation, technologies awareness and building capabilities of firms, emission level starts declining beyond a threshold point. In other words it may be easier for both older and younger firms as well as bigger and smaller firms to either adapt to or shift to cleaner energy sources compared to the medium-sized and aged firms to adapt or shift from the existing energy sources.
Further, capital intensity has a negative relation with emission intensity of the firm, and is highly
Table 4. Determinants of CO2 emission intensity
Independent Variables Coefficient Standard Error t value
Capital Intensity -0.003 0.001 -2.200**
Labour Intensity -0.005 0.003 -2.520***
Energy Intensity 1.293 0.164 4.870***
Firm Size 0.043 0.154 2.280***
Firm Size2 -0.132 0.048 -2.730***
Firm Age 0.013 0.004 3.280***
Firm Age2 -0.004 0.003 -2.010***
Technology Import Intensity -0.539 0.274 -1.970**
R&D Intensity -0.016 0.104 -2.160**
MNE Dummy -0.042 0.168 -0.250
Constant -1.305 0.141 -9.290
R2 (overall) 0.289
R2 (within) 0.288
R2 (between) 0.294
(u_i=0) F(2324, 621) 8.290***
F(7,621) 20.65***
Number of observations 2275
Source: PROWESS, Center for Monitoring Indian Economy.
significant at 1%. This means firms with the larger capital are emitting less compared to firms with the smaller capital. If we compare the results of age and size of firm to emission we can see that older and bigger firms are emitting less and capital-intensive firms are also emitting less. Hence, we can now assume that older and bigger firms might be higher capital intensive firms. Hence, being capital intensive, older and larger emit less as compared to the less capital intensive firms. According to Narayanan (1998), accumulation of technological capabilities through learning by doing is facilitated by the skilled manpower employed in a firm. The calculation of the labour intensity is quite similar to Narayanan (1998), hence labour intensity can also refer as a proxy for skill manpower. The result of the labour intensity is statistically significant at 1% and negatively related to emission intensity. Therefore, labour intensive firms are emitting less as compared to the less labour intensive sample firms. Rennings and Zwick (2001) found that employment is not a major reason for eco-innovation of firms; in contrary we found that higher labour intensive firms are less emitting.
MNE affiliation of firms is not found to be statically significant, but looking at the descriptive statistics on the relationship of MNE, R&D and technology import intensity we can find that foreign firms are investing more in technology import and R&D compared to domestic firms. Even in case of the emission we can see that there is difference between the domestic and foreign firms. Therefore, we assume that the presence of foreign affiliation might be captured either in the technology import or in the research and development ex-
penses of firms in the model. Further, energy intensity of firm is found to be positively related and statistically significant with the emission intensity. This implies energy intensive firms are also emission intensive. This result is akin to the rank correlation coefficient between energy and emission intensity. As the CO2 emission is from the fossil fuel consumption this result is accepted.
7. SUMMARY AND POLICY IMPLICATIONS
The climate change, green house gases, and emissions are matters of increasing concern not only for developed countries but also for the developing as well as the underdeveloped countries. In addition, concerns have been also reinvigorated by the global and local environmental problems caused by the ever-increasing use of fossil fuels, and so it is clearly an enormous challenge to fuel economic growth in an environmentally sustainable way. India, being one of the largest and rapidly growing developing countries, needs a special focus on the issue of emission. Analysis of the emission from the industries of Indian economy should not only be at the aggregate level and national level. Specific interest must be given to the sub-sectors as well. This work is an attempt to compute CO2 emission of sample firms in Indian manufacturing from 2000 to 2011 by adopting the IPCC reference approach. The results indicate that there are significant differences in firm-level emission intensity and they, in turn, are systematically related to identifiable firm specific characteristics. This study found size, age, energy intensity and technology intensity as the major determinants of CO2 emission inten-
sity of Indian manufacturing firms. In addition, capital and labour intensity of the firms are also related to the firms' emission intensity.
Indian manufacturing industries play a significant role in the country's economic growth. However, this sector has to upgrade the technologies and should achieve energy as well as emission efficiency. In addition, specific policy measures should be formulated to encourage medium-sized and older firms to upgrade technology and invest in technology import and research and development pertaining to eco-innovation to reduce CO2 emission. In addition, by reducing fossil fuel consumption and adopting cleaner and green energy firms will be able to become both energy- and emission-efficient. Summarizing the findings: R&D, technology sourcing, fuel switching should be given due attention for green growth. The contribution of this paper lies in estimating CO2 emission at the firm level and analyzing the factors that explain inter-firm variation in CO2 emission.
REFERENCES
Arimura, T., A. Hibiki, and H. Katayama (2008), "Is a voluntary approach an effective environmental policy instrument? A case for environmental management systems", Journal of Environmental Economics and Management, (55), 281-295. Atkinson, S. E., and T. Tietenberg (1991), "Market failure in incentive-based regulation: the case of emissions trading", Journal of Environmental Economics and Management, (21), 17-31. Bernard, A. B., and B. Jensen (2004), "Why some firms export?", The
Review of Economics and Statistics, 86 (2), 561-569. Brunnermeier, S., and M. Cohen (2003), "The determinants of environmental innovation in US manufacturing industries", Journal of Environmental Economics and Management, 45, 278-293. Chen, S., and Amelia U. Santos-Paulino (2010), "Energy consumption and carbon emission-based productivity change and industrialization in postreform China", Working Paper No. 78, World Institute for Development Economics Research, available at http://www.econstor.eu/handle/10419/54075 Cline, W. R. (1992), "The Economics of Global Warming', Institute for
International Economics. Washington, DC. Cohen, M., S. Fenn, and S. Konar (1997), "Environmental and financial performance: are they related?", Investor Responsibility Research Center, available at http://www.vanderbilt.edu/vcems/ papers/irrc.pdf.
Cole, M. A., R. Elliott, and K. Shimamoto (2005), "Industrial characteristics, environmental regulations and air pollution: an analysis of the UK manufacturing sector", Journal of Environmental Economics and Management, 50 (1), 121-143. Collins, A., and R. Harris (2005), "The impact of foreign ownership and efficiency on pollution abatement expenditures by chemical plants: some UK evidence", Scottish Journal of Political Economy, 52 (5), 746-768.
Copeland, B. R., and M.S. Taylor (1995), "Trade and transboundary pollution", American Economic Review, (85), 716-737.
Dales, J. (1968), "Pollution, Property, Prices", University Press, Toronto.
Doonan, J., P. Lanoie, and B. Laplante (2005), "Determinants of environment performance in Canadian pulp and paper industries: an assessment from inside the industry", Ecological Economics, 55, 73-84.
Faruq, H. A., and Yi. David T. (2010), "The determinants of technical efficiency of manufacturing firms in Ghana", Global Economy Journal, 10 (3), pp. 1-21.
Fischer, C., and R.G. Newell (2008), "Environmental and technology policies for climate mitigation", Journal of Environmental Economics and Management, 55 (2), pp. 142-162.
Foulon, J., P. Lanoie, and B. Laplante (2002), "Incentives for pollution control: regulation or information?", Journal of Environmental Economics and Management, 44, 169-187.
Frondel, M., J. Horbach, K. Rennings, and T. Requate (2004), "Environmental policy tools and firm-level management practices: empirical evidence for Germany", CAU Economics Working Paper No. 2, Department of Economics, Kiel, available at http://www. oecd.org/germany/31685533.pdf.
Goldar, B. (2011), "Energy intensity of Indian manufacturing firms: effect of energy prices, technology and firm characteristics", Science Technology Society,16 (3), pp. 351-372.
Gray, W., and R. Shadbegian (1993), "Environmental regulation and manufacturing productivity at the plant level", NBER Working Paper No. 4321, available at http://www.nber.org/ papers/w4321.pdf.
Greenstone, M. (2001), "The impacts of environmental regulations on industrial activity: evidence from the 1970 and 1977 clean air act amendments and the census of manufacturers", NBER Working Paper No. 8484, available at http://www.nber.org/papers/ w8484.pdf.
Hrovatin, N., and S. Urib (2002), "The determinants of firm performance after ownership transformation in Slovenia", Communist and Post-Communist Studies, 35, 169-190.
IPCC (2007), "Climate Change 2007 - Mitigation of Climate Change: Working Group III Contribution to the Fourth Assessment Report of the IPCC", Cambridge University Press, Cambridge, UK
Jaffe, A., and K. Palmer (1997), "Environmental regulation and innovation: a panel data study", The Review of Economics and Statistics, 79, (4), 610-619.
Konar, S., and M. Cohen (2001), "Does the market value environmental performance?", The Review of Economics and Statistics,83, (2), 281-289.
Kumar, N., and N.S. Siddharthan (1994), "Technology, firms size and export behavior in developing countries: the case of Indian enterprises", Journal of Development Studies 31 (2), 289-309.
Mazzanti, M., and R. Zoboli (2008), "Complementarities, firm strategy and environmental innovations", Environmental Sciences, 5 (1), 17-40.
Newbery, D. M. (1992), "Should carbon taxes be additional to other transport fuel taxes?", The Energy Journal, 13 (2), 47-60.
Nordhaus, W. D. (1991), "To slow or not to slow: the economics of the greenhouse effect", The Economic Journal, 101 (4), 920-937.
Nordhaus, W. D. (2007), "To tax or not to tax: the case for a carbon tax: alternative approaches to slowing global warming", Review of Environmental Economics and Policy, 1, 26-44.
Nordhausm W.D. (1977), "Economic growth and the climate: the carbon dioxide problem", American Economic Review, 67 (1), 341-346.
Oates, W. E. (1995), "Green taxes: Can we protect the environment and improve the tax system at the same time?", Southern Economic Journal, 61, 915-922.
Parry, I. W. H., and K.A. Small (2005), "Does Britain or the United States have the right gasoline tax?", American Economic Review, 95 (4), 1276-1289.
Pearce, D. W. (1991), "The role of carbon taxes in adjusting to global warming", Economic Journal, 101, 938-948.
Pfeiffer, F., and K. Rennings (1999), "Employment impacts of cleaner production on the firm level: empirical evidence from a survey in five European countries", ZEW Discussion paper No. 46, Centre for European Economic Research, Mannheim, available at ftp://ftp.zew. de/pub/zew-docs/dp/dp3299.pdf.
Porter, M. E. (1991), "America's green strategy", Scientific American, 264 (4),168.
Rennings, K., A. Ziegler, K. Ankele, E. Hoffmann, and J. Nill (2003), "The influence of the EU environmental management and auditing schemes on environmental innovations and competitiveness in Germany: an analysis on the basis of case studies and a large scale survey", Discussion Paper 03-14, ZEW, Mannheim, available at http://econstor.eu/bitstream/10419/24817/1/dp0314.pdf.
Rennings, K., and T. Zwick (2001), "The employment impact of cleaner production on the firm level: empirical evidence from a survey in five European countries, Discussion Paper, Centre for European Economic Research (ZEW), Mannheim, available at ftp:// ftp.zew.de/pub/zew-docs/dp/dp0146.pdf.
Sahu, S. K., and K. Narayanan (2011), "Determinants of energy intensity in Indian manufacturing industries: a firm level analysis", Eurasian Journal of Business and Economics, 4 (8), 13-30.
Srivastava, V. (1996), "Liberalisation, productivity and competition: a panel study of Indian manufacturing', Oxford University Press, New Delhi.
Stavins, R. (1995), "Transaction costs and tradeable permits", Journal of Environmental Economics and Management, 29, 133-148.
Stern, D. I. (2004), "The rise and fall of the environmental Kuznets curve", World Development, 32 (8), 1419-1439.
Sterner, T. (2007), "Fuel taxes: an important instrument for climate policy", Energy Policy, 35 (6), pp. 3194-3202.
Sterner, T. and U. M. Persson, (2008), "An even sterner review: introducing relative prices into the discounting debate", Review of Environmental Economics and Policy, 2 (1), 61-76.
Zelenyuk, V., and V. Zheka (2006), "Corporate governance and firm's efficiency: the case of transitional country Ukraine", Journal of Productivity Analysis, 25, 143-168.
Ziegler, A., M. Schroder, and K. Rennings (2008), "The effect of environmental and social performance on the stock performance of European corporations", Environmental and Resource Economics, 40, 609-680.
Appendix A
Table A1. Estimates of pooled OLS and random effects models
Variables Coef RSE t value Coef SE z value
Pooled OLS Random Effect
Capital Intensity 0.002 0.001 2.320 0.003 0.001 2.700
Labour Intensity -0.002 0.001 -1.710 -0.002 0.001 -2.630
Energy Intensity 5.096 0.863 5.900 3.409 0.120 6.430
Firm Size 1.219 0.113 5.830 0.881 0.074 5.870
Firm Size2 -0.174 0.038 -4.550 -0.085 0.023 -3.640
Firm Age 0.011 0.004 2.890 0.013 0.004 3.280
Firm Age2 -0.003 0.002 -1.790 -0.002 0.001 -2.010
Technology Import Intensity -0.896 0.378 -2.370 -0.848 0.103 -8.240
R&D Intensity -0.071 0.123 -1.580 -0.023 0.034 -0.680
MNE -0.038 0.180 -0.210 -0.042 0.168 -0.250
Constant -2.850 0.238 -11.960 -2.473 0.198 -12.460
F (10, 2942) 101.130 -
R2 (overall) 0.429 0.338
R2 (within) - 0.423
R2 (between) - 0.411
Root MSE 1.076 -
Wald chi2 - 1516
Number of observations: 2275
Note: Coef: Coefficient, RSE: Robust Standard Error, SE: Standard Error. Source: PROWESS, Center for Monitoring Indian Economy.
