Cross subsidy removal in electricity pricing in India

Energy Policy 100 (2017) 181–190

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Cross subsidy removal in electricity pricing in India Ranajoy Bhattacharyya, Amrita Ganguly

⁎

crossmark

Indian Institute of Foreign Trade, 1583 Madurdaha Chowbagha Road, Ward No. 108, Borough XII, Kolkata 700107, India

A R T I C L E I N F O

A BS T RAC T

Keywords: Electricity pricing Cross subsidies Computable General Equilibrium Direct benefit transfer

In India electricity price for agriculture is cross subsidized by the industries. The Indian government has started a process through which the extent of cross subsidization is gradually being reduced. The idea is to replace the cross subsidization by 2030 and introduce a rate structure that will increase with the amount of electricity usage. This paper uses the Computable General Equilibrium framework to evaluate the ex-ante impact of these policy changes on the Indian economy. The paper finds that removal of cross subsidies will increase inflation particularly food inflation resulting in a decline in household incomes more so in rural areas. Replacing cross subsidies with a progressive rate structure will compensate for only a small part of the negative effects of the removal of cross subsidies. Four other policy options are also investigated targeting household incomes, food inflation and general inflation. Most of these options do not work as the required increase in budget deficit is unlikely to be bearable to the government. The only feasible option appears to be a direct price subsidy to agricultural sector: in this case food prices are held down, inflation is moderate and effect on household incomes is minimal.

1. Introduction The average industrial power tariffs in India have increased from INR 4.16/kWh in 2007 to INR 7.64/kWh in 2015. This occurred in part because agricultural power tariffs are cross-subsidized by industrial tariffs. Agricultural tariffs were INR 0.77/kWh in 2007, which amounted to about 18% of industrial tariffs. Although agricultural tariffs increased (to INR 1.83/kWh in 2015), they still amount to only 24% of industrial tariffs. On average, Indian industries pay about 12% more than the average cost of supplying power, while agricultural consumers pay about 55% lower.1 Consequently, although industries consume more power (365 MWh in 2015) than agriculture (147 million MWh in 2015)2, power generation and distribution companies struggle financially. The scenario reported above varies significantly depending on context. States, the central government, and private organizations play roles in India's power sector. There are generation units wholly owned by state governments and the central government, and a significant portion of the total power supply is managed by private organizations. Although power is provided by the central government body through the Power Grid Corporation of India Limited, the distribution units are owned by either state-owned or private corporations. In such a situation, the structure of power tariffs throughout India is not homogenous. The tariffs structures vary across states and across sectors. For the end consumer, the tariff is based on the category and state to which the consumer belongs. In 2015, the agricultural tariff varied between INR 1.19/kWh and INR 2.50/kWh while the industrial tariff varied between

INR 3.99/kWh and INR 10.02/kWh. For the DISCOMs, the tariff varies across states and the source of generation (conventional or renewable). The Electricity Act of 2003 created a framework for setting electricity tariffs and defined rules and regulations for all organizations in the power sector. The Act also entrusted the responsibility of approving tariffs to the Central Electricity Regulatory Commission (CERC) for units owned by the central government and units selling power to more than one state and to State Electricity Regulatory Commissions (SERCs) for units selling power within a single state. The exact method for determining power tariffs is largely left to SERCs and the CERC, which publish rates and guidelines for different entities to fix their tariffs from time to time. Variations in power tariffs between states are also significant. For instance, for state-owned units in the state of Maharashtra, power tariffs are determined by the Maharashtra Electricity Regulatory Commission (MERC). There are several power generators and distribution licensees in Maharashtra, such as Tata Power, with a licensed area that is part of central Mumbai. The tariff Tata Power places upon different consumers in this distribution area is regulated and approved by the MERC. Reliance Infrastructure, Brihanmumbai Electric Supply & Transport (BEST), and Maharashtra State Electricity Distribution Company (MahaVitaran) are the other players that distribute power to Mumbai. While Tata Power and Reliance Infrastructure are private organizations, BEST and MahaVitaran are state-owned companies. The tariffs charged by all these Distribution Companies (DISCOMs) is regulated by the MERC. In Gujarat, the Madhya Gujarat Vij Company Limited is a leading DISCOM that supplies power to

⁎

Corresponding author. E-mail addresses: [email protected] (R. Bhattacharyya), [email protected] (A. Ganguly). Except the northeastern states (with the exception of Assam), where industrial power tariffs are also significantly subsidized. 2 Source: Ministry of Agriculture, Govt of India (ON643) & (16361) and Central Electricity Authority, Ministry of Power, Govt of India 1

http://dx.doi.org/10.1016/j.enpol.2016.10.024 Received 26 May 2016; Received in revised form 12 September 2016; Accepted 18 October 2016 0301-4215/ © 2016 Elsevier Ltd. All rights reserved.

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

charge+capacity charge+UI (Unscheduled interchange). The fixed charge is same as that discussed above. The capacity charge is for making the power available to them and depends on the capacity of plant and the third one is UI. If every things goes well, power demand is equal to power supplied and the system is stable and frequency is 50 Hz. But practically this rarely happens. One or more state overdraws or one or more Generating Station under supplies. This can lead to deviation in frequency and system stability. If demand is more than supply, frequency dips from normal and vice versa. UI charges are incentive provided or penalties imposed on the generating stations. If the frequency is less than 50 Hz, implies demand is more than supply, then the Generating Station which supplies more power to the system than committed is given incentives. On the other hand, if frequency is above 50 Hz, implying supply is more than demand, incentives are provided to Generating Station for backing up the generating power. Hence it tries to maintain the system stability. For further details on ABT, please refer to National Power Training Institute, 2009. Here the discussion is on mainly the first type of tariff i.e. the tariff that the end consumers pay to the DISCOMS. Recently, the Indian government has shown interest in reducing the extent of cross subsidization of tariffs. Between 2007 and 2015, the agricultural power tariff was increased by 138%, compared to the 47% increase in the industrial power tariff. As a substitute for cross subsidies, the government proposed a low base rate for both agriculture and industry (National Tariff Policy, 2006). The rate then became progressive; in other words, it increases with the amount of electricity consumed. The Electricity Amendment Bill of 2014, which further amended the Electricity Act of 2003, aimed to fully remove cross-subsidization in the electricity sector in India, requiring each SERC to specify a timeline for reduction of crosssubsidies. Some states have proactively started removing cross-subsidies. For example, the state of Maharashtra, which has one of the highest industrial power tariffs in India, is seeking to reduce its industrial power tariff by around 30% by decreasing farmers' need for the state to supply cheap electricity. It plans to give solar water pumps to farmers and reduce their dependence on conventional energy5. For simulation purposes, the above scenario has considerably simplified in order to make the analysis tractable. It is assumed that different states and organizations have uniform mean rates6. The broad research question is as follows: What are the effects of substituting cross subsidization with a progressive rate structure? This broad research question was split into two separate questions: What is the possible effect of increase in electricity tariffs (i) without changing the rate structure (cross subsidization) and (ii) with change in the rate structure (progressive rate structure)? The answers to the two research questions were combined to obtain the overall results. The four industrial sectors that contribute the most to the manufacturing GDP of India were selected for analysis. As of 2011–12, these industrial sectors included chemicals and chemical products (12.2%), machinery and equipment (11.1%), basic metals (9.7%), and textiles (9.2%)7. The agriculture sector was also considered, as it contributed to almost 17% of the overall GDP of India in the same period8. Before running the simulations, first it is estimated whether these industries experienced constant or increasing returns to scale using firm-level data.

central Gujarat, while Dakshin Gujarat Vij Company Limited supplies power to southern Gujarat. Both are state-owned DISCOMs and their tariffs are regulated by the Gujarat Electricity Regulatory Commission. Each SERC has its own priorities and sets tariffs and subsidies accordingly, although within the limits set by the National Tariff Policy of 2006. There are two tariff systems in India, one for the consumers which they pay to the DISCOMS and the other one is for the DISCOMS which they pay to the generators, which is discussed later on. The distribution utilities continued to face financial constraints due to a variety of reasons – power theft, inefficiencies, high overheads etc. One of the major reasons is the reluctance of state governments to revise tariffs periodically and the supply of electricity to certain category of consumers at free of charge or well below the cost of supply. As a result of this, distribution utilities were forced to borrow funds from the financial institutions in the past few years on short term basis to manage their operations. The level of such short term borrowings as well as the payment to power generation companies accumulated to un-sustainable level and financial institutions refused to further continue lending the distribution utilities (Planning Commission, Govt of India, 2014). The approximate accumulative losses of all DISCOM companies in India amount to over INR 3000 billion and they are bleeding for about INR 800 billion every year3. They incur losses of about Re 1 per unit of power. These losses occur due to a variety of reasons such as cross subsidies, 23% Aggregate Technical & Commercial (AT & C) losses as of 2014–154, more supply to the subsidized agriculture sector as agriculture is becoming more mechanized and industry is increasing its captive generation capacity. With removal of cross-subsidy, one of the factors contributing to the ill-health of DISCOMs shall be taken care of. However, the losses on account of the high AT & C losses will continue to remain a factor in impacting the financial; heath of the DISCOMs. 1.1. Details of the tariff systems in India There are two tariff systems, one for the consumer which they pay to the DISCOMS and the other one is for the DISCOMS which they pay to the generating stations. First the discussion is about the tariff of electricityfor the consumer i.e. the cost consumer pay to the DISCOMS. The total cost levied on the consumer is divided into 3 parts usually referred as 3 part tariff system. Total cost of electrical energy (INR)=fixed cost +semi fixed cost +variable cost

=(a + b*kW + c*kWh) Here, a=fixed cost independent of the maximum demand and actually energy consumed. This cost takes into account the cost of land, labour, interest on capital cost, depreciation etc. b=constant which when multiplied by maximum kW demand gives the semi fixed cost. This takes into account the size of power plant as maximum demand determines the size of power plant. These demand charges are intended to recover the cost of facilities (such as power transformers, wires and power plants) available to provide the maximum amount of electricity which customer may require at any time. c=a constant which when multiplied by actual energy consumed kWh gives the running cost. This takes into account the cost of fuel consumed in producing power. Thus the total amount paid by the consumer depends on its maximum demand, actual energy consumed plus some constant sum of money. The tariff system existent in India for the DISCOMS is regulated by CERC and SERCs. This tariff system is called availability based tariff (ABT). It is a tariff system which depends on the availability of power. It is a frequency based tariff mechanism which tends to make the power system more stable and reliable. This tariff mechanism also has of 3 parts: Fixed

2. Literature review Energy subsidies have resulted in excessive and inefficient energy use, contributing to price volatility and discouraging much-needed investment 5 http://articles.economictimes.indiatimes.com/2015-02-10/news/59005191_1_ power-tariff-solar-water-pumps-state-grid 6 Since the basis of the CGE model is the SAM and the SAM is constructed at a national level (and not at individual state level), hence a uniform power tariff is assumed at India level for the sake of this analysis. This power tariff is an average tariff based on weighted average tariffs of each state (and category of consumer) and the weights used are the power consumed by each state (and category of consumer). 7 https://data.gov.in/catalog/manufacturing-gdp-sector-and-employment-projections 8 According to the Central Statistics Office.

3 Interview with Ashok Khurana, Director General, Association of Power Producers at http://m.moneycontrol.com/news/economy/need-to-dealdiscom-losses-onstate-levelexpert_1506161.html 4 http://powermin.nic.in/upload/loksabhatable/pdf/LS19032015_Eng.pdf

182

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

ln pi=α +βlnli+γlnki+ δlnmi + εlnei

in the energy sector. Such negative impacts of energy subsidies have encouraged governments to take initiatives for their reform. Energy subsidy reforms can have negative effects on GDP (Abouleinein et al., 2009; Breisinger et al., 2011; McDonald and van Schoor, 2005; Lin and Ouyang, 2014; Fan et al., 2007) as well as affect the poverty and inequality indices of a country. For example, in Indonesia, it was seen that 25% reduction of fuel subsidies increases poverty by 0.259% and if this money were fully allocated to government spending, the poverty would decrease by 0.27%. 100% removal of fuel subsidies and the reallocation of 50% of the amount to government transfers and other subsidies could decrease the incidence of poverty by 0.277% (Dartanto, 2013). In Ghana, it was seen that almost 78% of fuel subsidies benefited the richest group, with about 3% of subsidy benefits reaching the poorest quintile, hence with subsidy reform the Government introduced national cash transfer program (Livelihood Empowerment Against Poverty-LEAP). This program when expanded to 150,000 households reduced poverty by 1.6% and when expanded to 500,000 poor households, reduced poverty by 2.3% (Cooke, et al., 2014). Most of these studies focus on effects of energy subsidy reforms on macroeconomic parameters or household incomes of a country. For this research work, the focus was more on electricity price subsidy reforms and their impacts on the industrial sector of a country. There are two central effects of a rise in the cost of electricity: cost push inflation and output shrinkage. Which effect dominates in a particular sector depends on the elasticity of demand in the sector. Either effect can dominate in a particular sector. For instance, Alvarez and Valencia (2015) found that output responded to a change in cost more than prices in Mexico. However, Coupal and Holland (2002) found that prices responded more than output in the US. The effect of a change in the cost of electricity has also been evaluated for other variables. For instance, Grave etal. (2015) examined the effect of increased electricity prices on exports and GDP in Germany and found that a 3.5% increase in electricity prices reduced German exports by 0.3% and Germany's GDP by 0.15%. Swain and Charnoz (2012) found that eliminating electricity subsidies in the agriculture sector in India may increase rural poverty by reducing farmers' disposable income and exacerbate food security by reducing agricultural yields. In the US, Sands et al. (2011) found that higher energy-related production costs lowered agricultural output, raised the prices of agricultural products, and reduced farms' income, regardless of the reason for the energy price increase. Another study by GGGI (2014) concluded that increases in electricity prices negatively impacted all the important sectors in South Africa and noted that many industries adopted energy efficiency measures to reduce electricity bills. However, we could not find any work analysing the impact of electricity tariff variation on industries and agriculture in India using the CGE model. The rest of the paper is arranged as follows: Section 3 describes how returns to scale were determined for each sector, while Section 4 reports the methodology of this study. Section 5 discusses the results of the study, Section 6 discusses on direct benefit transfer scheme and Section 7 concludes the study.

Returns to scale are determined by the equation µ=β+γ+δ+ε. In cases with constant returns to scale, µ=1 is used, while in cases with increasing returns to scale, µ > 1 is used. For the selected industrial sectors, data on output, labour, capital, materials, and energy were taken from the Prowess database. “Output” is defined as deflated sales adjusted for changes in the inventory and purchase of finished goods. “Purchase of finished goods” is defined as finished goods purchased from other manufacturers for resale. Thus, the purchase of finished goods was subtracted from sales to arrive at the firms' manufactured output. The database also provides information on the purchase of raw materials, which is used to determine parameters for materials. The database provides information on the wages and salaries of firms and does not provide any information on the number of people employed by firms. Hence, the available information is used to determine the number of employees of each firm, dividing the salaries and wages of the firm by the average wage of the industry to which the firm belongs. To arrive at an average wage, data from the Annual Survey of Industries (ASI) is used, dividing total emoluments by total persons engaged in the relevant industry. “Power and fuel expenses” is used as a proxy for electricity expenses, while “net fixed assets” is used as the capital input. All the relevant data were obtained from Prowess and the ASI and were used to determine returns to scale for 2007–08 and 2013–149. The heteroskedasticity-consistent estimates of µ for 2007–08 and 2013–14 are presented in Table 1. The table shows that two sectors, chemicals and chemical products and machinery and equipment, experience increasing returns to scale, while all the other sectors experience constant returns to scale. This is also consistent with the fact that the chemicals and chemical products and machinery sectors have high fixed costs, while the other sectors have relatively lower fixed costs. (Fig. 1). 4. Description of the methodology used for development of electricity SAM and structure of benchmark CGE model a) Electricity SAM The calculations are based on the SAM for India for the year 2007– 08 following Pradhan, Saluja and Sharma (2013). To construct the Electricity SAM, relevant sectors from the above SAM were aggregated into primary (agriculture sector consists of all agricultural products, minerals, primary products such as iron ores, crude petroleum and agro process activities), secondary (Manufacturing sector comprised mainly of all manufacturing activities such as cotton and textile, plastic, rubber and leather products, cement, different chemical products, etc. without crude oil, LPG, petrol and diesel) and tertiary (Service sectors such as education, health care services, public administration, bank and insurance, postal services etc.) sectors. Textiles, chemicals, basic metals and machinery and equipment have been taken as separate sectors. Electricity has been proportionately taken out from the sector – services. Thus, the SAM that has been used here has one energy sector (electricity) where subsidies have varied over the years and across sectors and six non-energy sectors (agriculture; manufacturing and services together, textiles, chemicals, basic metals and machinery and equipment) where there is no subsidy as well. Four types of agents in the economy have been considered, namely, (a) household, (b) firm, (c) government and (d) Rest of World (ROW). The three types of labour (unskilled, semi-skilled and skilled) were aggregated into one sector – labour. Households have been aggregated into two types – urban and rural household.

3. Description of the methodology for determining returns to scale in industry To determine whether industries demonstrate constant or increasing returns to scale, the standard Cobb–Douglas production function is used:

pi =aαliβ kiγ miδ eiε

(2)

(1)

Where, i=1, …….., N enumerates observations within a given industry. pi=output. li=labour input. ki=capital input. mi=material input. ei=energy input. Taking the log of both sides' yields:

9 For all the sectors considered here, material costs contribute most to overall costs, while energy costs contribute the least. For the chemical industry, capital costs contribute to about 45% of overall costs, indicating that the fixed costs are very high. Similarly, capital costs contribute to about 52% of the overall costs in the machinery sector, indicating that the fixed costs are very high. However, the capital costs in the other sectors (basic metals, textiles, and agriculture) are much lower, with agriculture being the lowest.

183

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

Table 1 Determination of value of µ for different sectors. Sector

Variable

2007–08

2013–14

Coefficient

t-Statistic

Coefficient

t-Statistic

Chemicals and chemical products

β γ δ ε µ

0.40 0.25 0.56 −0.07 1.21

4.74 3.19 5.27 −0.90 –

0.43 0.16 0.57 −0.01 1.2

5.64 1.90 7.41 −0.15 –

Machinery and equipment

β γ δ ε µ

0.40 0.11 0.63 −0.08 1.1

5.09 1.90 7.95 −2.02 –

0.50 0.03 0.56 0.01 1.1

5.44 1.32 8.40 0.24 –

Textile

β γ δ ε µ

0.19 0.01 0.82 0.08 1.0

2.15 0.17 6.36 1.25 –

0.30 0.06 0.73 −0.03 1.0

5.06 1.70 9.96 −0.42 –

Basic metals

β γ δ ε µ

0.21 0.06 0.77 0.08 1.0

2.96 1.88 8.89 1.50 –

0.30 0.01 0.75 0.05 1.0

3.56 0.38 8.53 1.55 –

Agriculture

β γ δ ε µ

0.60 0.23 0.39 0.12 1.0

5.47 1.56 3.56 1.04 –

0.42 0.25 0.59 −0.04 1.0

3.49 1.57 5.26 −0.37 –

Electricity has been proportionately taken out from the sector – services. Four types of agents in the economy have been considered, namely, (a) household, (b) firm, (c) government and (d) Rest of World (ROW). There are two types of households, namely, (a) Rural and (b) Urban. All other countries and regions are clubbed together into ROW.

For this research work, 2007–08 is used as the base year for the following reasons: (i) the latest SAM of India is available for 2007–08 and (ii) The National Tariff Policy was enacted in 2006 which proposed a progressive rate (increases with amount of electricity consumed) on a low base rate for both agriculture and industry as a substitute for cross subsidies (Table 2). b) Assumptions of the Benchmark CGE Model under perfect competition

4.2. Production and factor inputs

Our benchmark CGE model is based on perfect competition10 and constant returns to scale assumption both in commodity market and factor market. The model is based on the following assumptions.

Two basic factors of production have been considered, namely, labour and capital that take part in the production process within which substitution is possible through Cobb–Douglas production technology. Each production unit requires intermediate inputs following fixed-coefficient-type Leontief technology.

4.1. Sectors and agents

4.3. Prices

Following SAM for India for the year 2007–08 produced by Pradhan, Saluja and Sharma (2013), all sectors of the economy were grouped into seven aggregated non-energy sectors, namely, (i) primary (agriculture sector consists of all agricultural products, minerals, primary products such as iron ores, crude petroleum and agro process activities), (ii) secondary (Manufacturing sector comprised mainly of all manufacturing activities such as cotton and textile, plastic, rubber and leather products, cement, different chemical products, etc. without crude oil, LPG, petrol and diesel), (iii) tertiary (Service sectors such as education, health care services, public administration, bank and insurance, postal services etc.), (iv)textiles, (v) chemicals, (vi) basic metals and (vii) machinery and equipment sectors.

Product prices are determined from the equality of price and average cost. The average cost is comprised of basic factor cost, cost of intermediate inputs that includes cost of energy inputs. Increasing returns to scale is assumed through the presence of fixed cost in the production units (as discussed in the next sub-section). 4.4. Household Income and expenditure Households are rendering factor services in terms of labour and capital while in return they are receiving factor payments in the form of wages and rentals. Two types of household, rural and urban have been considered. A household spends its income for consumption purposes. It is assumed that the linear expenditure-system-type demand function for a household.

10 It is assumed over here that zero profit condition actually holds in the minimum point of average cost curve. This also indicates there is constant returns to scale in the long run and there is no excess capacity present in the market. In the electricity sector, there are many buyers and in this era of liberalization and globalization, quite handsome number of firms actually produce the same homogenous product - electricity. Hence perfect competition assumption maybe a very close approximation to the reality when there is no presence of increasing returns to scale, and zero profit condition holds at the minimum point of the long run average cost curve.

4.5. Government Income and expenditure Sources of income of the government are (a) direct, indirect and corporate taxes; (b) import tariff and (c) income from entrepreneurial activity. In the expenditure front it is assumed that government's 184

Energy Policy 100 (2017) 181–190

Frequency

R. Bhattacharyya, A. Ganguly

4.6. Investment and savings

Agriculture

40

It is considered that Neo-classical-type closure rule where investment is guided by saving. Total saving is comprised of (a) household saving, (b) government saving, (c) corporate saving and (d) foreign savings. Total saving is converted to total investment.

20 0

4.7. Armington function and trade

Fixed cost (INR million)

Chemicals and Chemical Products

20

Frequency

International trade in our model is guided by Armington function. Total availability of composite commodity in the domestic economy is composed of domestically produced variety of the good demanded by the domestic people and foreign variety of the same good. Both types of the varieties are combined together following a constant elasticity of substitution type preference function.

10

4.8. Production of output and transformation

0

Total supply of each domestic good produced using labour, capital and intermediate input is used up by export of that good and to meet up the domestic demand of domestic variety. Both export and domestic demands of the produced good are combined together following the constant elasticity of substitution (CES)-type transformation function.

Fixed cost (INR million)

4.9. Factor prices and equilibrium Two basic factors of production, namely, labour and capital have been considered. Total supply of the basic factors is fixed in value terms, and factor prices are flexible. Physical quantity of labour or capital may change in different simulation experiments following demand and supply equilibrium mechanism in the factor market. Demand for a factor is originated from the production of goods and services.

More

Fixed cost (INR million)

1000

900

800

700

600

500

400

300

0

200

50 100

Frequency

Texles 100

4.10. Equilibrium in commodity market

Frequency

Basic Metals 30

In the commodity market, total supply of the composite commodity is constituted by domestic variety as well as imported foreign variety corresponds to each good. Demand for the composite commodity is generated from household consumption, government consumption expenditure, total investment demand and demand for intermediate input. Composite commodity price is determined from the demand and supply of composite commodity.

20 10 0

4.11. GDP and welfare

Fixed cost (INR million)

Under perfect competition GDP has been computed adding all sectoral outputs. Social welfare has been of Cobb–Douglas type and depends on private household consumption, that is, elasticity of substitution between any two sectors' product is constant and takes the value unity.

Frequency

Machinery and Equipment 20 10

4.11.1. Database and calibration of model parameters After specifying Electricity CGE model, parameters of the model will have to be estimated from the benchmark dataset. Electricity SAM has been constructed by segregating the seven sectors and one energy sector. SAM of India for the year 2007–08 has been constructed based on the SAM constructed by Pradhan, Saluja and Sharma (2013). The Electricity SAM was used for the calibration of the CGE model. Subsidy rates and rate hikes are supplied exogenously.

0 Fixed cost (INR million)

Fig. 1. The distribution of firms in the sample according to power cost.

expenditure in any sector is exogenously determined, that is, determined in the government's budget and adjusted to benchmark SAM. The difference between government's income and expenditure is government's savings.

a) Modeling returns to scale (Introducing market imperfections in the benchmark CGE model- imperfect competition) The effect of electricity price changes on industries depends on the 185

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

Table 2 Electricity SAM of India for 2007–08 (INR billion). Sector

C1

C2

C3

C4

C5

C6

F1

LAB

CAP

RHH1

UHH1

PVT

PSE

GOV

IDT

INV

EXT

Total

C1 C2 C3 C4 C5 C6 F1 LAB CAP RHH1 UHH1 PVT PSE GOV IDT INV EXT Column Total

2388 1824 57 393 0 54 118 4720 3749 1 1 1 1 1 −326 1 198 93843

3443 13845 36353 −1738 −2431 −1545 670 20798 20025 1 1 1 1 1 1173 1 6686 1452

351 −560 610 239 2 59 85 383 502 1 1 1 1 1 −1 1 126 1452

71 −1349 14 1348 15 43 93 171 628 1 1 1 1 1 95 1 953 2016

1 −1383 2 46 1279 57 108 141 747 1 1 1 1 1 249 1 1394 2645

14 −4160 25 150 1261 2739 133 494 1246 1 1 1 1 1 421 1 3266 5580

2 917 4 11 4 106 103 588 92 1 1 1 1 1 26 1 1 1858

1 2 1 1 1 1 1 1 1 9989 12560 1 1 1 1 1 1 22564

1 2 1 1 1 1 1 1 1 8033 5125 2928 1142 987 1 4846 1 23072

4388 11747 966 237 1 456 68 1 1 1 1 1 1 820 247 6045 1 20594

2077 10931 577 203 1536 500 96 1 1 1 1 1 1 982 263 5396 1 20491

1 2 1 1 1 1 1 1 1 1 1 1 1 1921 1 2928 1 4864

1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1142 1 1158

100 −1786 55 734 5755 175 83 1 1 2189 1502 1921 1 1 114 −1982 1 8765

1 2 1 1 1 1 1 1 1 1 1 1 1 4050 1 1 1 4066

−6 12170.6 195 243 1 5674 1 1 1 1 1 1 1 1 735 1 1 19029

280 7293 716 587 1 986 1 −25 −180 378 1297 1 1 1 728 644 1 12429

47676 39521 39521 2065 7427 9256 1447 22559 23069 20601 20496 4863 1157 8771 4055 19028 12436

Note: C1: agriculture; C2: manufacturing and services; C3: textiles; C4: chemicals and chemical products; C5: basic metals; C6: machinery and equipment; F1: electricity; LAB: labour; CAP: capital; RHH1: rural household; UHH1: urban household; PVT: private enterprise; PSE: public enterprise; GOV: government; IDT: indirect taxes; INV: capital account; EXT: rest of the world

the industry. At the industry level, therefore, the effects of increased electricity price are symmetric with increasing and constant returns to scale, price rises, and output falls. To analyze demand, the standard Dixit and Stiglitz (1977) set up is used. In this set up, the price elasticity of demand is Epi=αi+(1 − αi )/ ni , where αi is the elasticity of substitution between any two varieties and ni is the number of varieties in the ith industry (Helpman and Krugman, 1985). Epi was computed in our model by comparing the pre- and post-shock situations. The values for the five industries considered in this study are as follows: −0.17 (textiles), −0.47 (chemicals and chemical products) −0.28 (basic metals), −0.41 (machinery and equipment), and −0.12 (agriculture).

industries' cost structures. In the CGE model specified here, the average cost function (including fixed costs) is as follows:

pdi=ai pi Z i+Zi2

∑ bj,i qj +FCi /Zi j

(3)

Where, pdi=Average total cost for ith sector. ai=composite factor input. pi=Composite factor price. bj, i=Intermediate input required. qi=composite good price. FCi=Fixed cost in ith sector. Zi=Output of ith sector. ai x pi= unit basic factor cost. ∑bj, i x qi=unit intermediate input cost for ith sector and jth intermediate good. FCi/Zi=average fixed cost for ith sector. Therefore, the average total cost is the sum of the unit basic factor cost, the unit intermediate input cost, and the average fixed cost. Unit basic factor cost includes both labour and capital costs, while capital cost excludes fixed costs. Electricity cost is an element within the quadratic term of the cost function. Therefore, an increase in the price of electricity or removal of subsidies causes the average cost function to shift up and to the right. The falling parts also become steeper, so the minimum points shift to the left. The effect of increased electricity price on a particular firm thus clearly depends on the position of the firm in the average cost function. Whether firms in a particular industry are located in the falling or minimum part of the average cost function is immaterial as far as price is concerned; in both cases, prices rise as firms exit the industry. In cases with perfect competition, the market demand curve shifts to the left, increasing market price and, hence, the price of electricity for each firm. In cases with monopolistically competitive firms, the residual demand curve shifts up and becomes steeper as the number of substitute products fall. However, the effect on firm output is asymmetric. When firms have constant returns to scale, industry outputs fall, while the opposite occurs for firms with increasing returns to scale.11 Although the share of the market served by each remaining firm rises with increasing returns, the aggregate industry output falls along the market demand curve due to the aggregate price rise. Thus, the decrease in supply due to a firm's exit from the industry must overcompensate for the increase in supply for every other firm in 11

5. Results 5.1. Removal of cross subsidization and introduction of a progressive rate structure The Table 3 shows that the most important effect of removal of cross subsidization is food inflation, which leads to a steep decline in the real income of households, especially in rural areas, where the proportion of food consumption is higher. Note, however, that general inflation in terms of the aggregate price index12 decreases, as the prices in most industries fall due to a decrease in the cost of production. This increases the competitiveness of Indian industrial produces in the international market, resulting in a corresponding rise in certain exports. However, agricultural exports would be adversely affected by removing cross subsidization. The main concern of removing cross subsidization, therefore, is decline in household incomes. Since this would affect politics in a democratic country like India, the government must mitigate the effect of such a change on households. As mentioned above, the government of India is planning to replace cross subsidization with a progressive rate structure. Under this structure, firms that use more electricity will be charged higher tariffs in addition to a base tariff. Since our data was obtained at the industry level rather than the firm level, it is not possible for us to n 12 The aggregate price index is calculated as follows. PIy= ∑i =1 wi, y xpi, y , where PI=aggregate price index for year y, pi,y=price of ith industry in year y, and

wi, y=

% contributiontoGDPbyindustryiinyeary TotalGDPinyeary

Inflation between the two years under consideration is

defined as the percentage difference between the CPI of the two years. The production index (Pi,y) is defined as the sum of the production of all industries in year y.

In other words, the zero profit curve shifts up (see Krugman, 1979).

186

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

Table 3 Effect of increased electricity tariffs in different scenarios in India (percentage change between 2007 and 2014). Sector

With cross subsidies

Removal of cross subsidies

Progressive rate structure – I1

Progressive rate structure – 22

Power tariff

Agriculture Manufacturing and services Textiles Basic metals Machinery and equipment Chemicals and chemical products

138% 50%

380% −11%

377% −16% −7% −3% 1% 5%

357% −16% −7% −3% 1% 5%

Production

Agriculture Manufacturing and services Textiles Chemicals and chemical products Basic metals Machinery and equipment

−0.015% −0.0008% −0.02% −5% −0.07% −3%

−0.02% −0.01% −0.03% 1% −0.1% 2%

−0.02% −0.01% −0.03% 1% −0.1% 2%

−0.02% −0.01% −0.03% −4% −0.1% −3%

Domestic price

Agriculture Manufacturing and services Textiles Chemicals and chemical products Basic metals Machinery and equipment

18% 9% 20% 13% 16% 24%

73% 0% −6% −4% −5% −5%

32% 1% −1% 7% 2% 1%

29% 1% −1% 7% 2% 1%

Aggregate price index (inflation)

17%

6%

7% 6%

Income

Rural household income Urban household income Income index (weighted average income)

−1% −2% −1%

−8% −3% −6%

−2% −2% −2%

−2% −1.60% −2%

Export

Agriculture Manufacturing and services Textiles Chemicals and chemical products Basic metals Machinery and equipment

−23% −10% −5% −31% −10% −13%

−40% 4% 3% 13% 5% 9%

−37% 6% 0% −6% 3% −1%

−33% 6% 0% −6% 3% −1%

Note: 1) The assumed rate hikes for the five sectors are as follows: 5% for agriculture, 10% for textiles, 15% for basic metals, 29% for machinery and equipment, and 25% for chemicals and chemical products. 2) The assumed rate hikes for the five sectors are as follows: 0.5% for agriculture, 10% for textiles, 15% for basic metals, 29% for machinery and equipment, and 25% for chemicals and chemical products.

some sensitivity, but industries with constant returns will show almost no change. Therefore, misallocation of resources due to cross subsidization is minimal, perhaps because electricity costs are a small part of firms' overall costs.

impose a strict progressive rate structure for electricity tariffs during simulations. Instead, the distribution of the sample of firms used in Section 2 according to their electricity usage (Fig. 1) has been looked at. Sectors with a large concentration of firms with low electricity usage will have lower average tariffs than sectors with a high concentration of firms with high electricity usage. The base tariff has not yet been specified. The simulations that were conducted in the last two columns of Table 3 assume that the base tariff will be the cost of power generation in India, which was INR 3.50 per unit in 2014– 15. Given the distribution of firms, the last column reports the assumed rate hikes for the five sectors: 5% for agriculture, 10% for textiles, 15% for basic metals, 29% for machinery and equipment, and 25% for chemicals and chemical products. In the other column, the rate hike is reported to be 0.5% for the agricultural sector and the same as the last column for the other sectors. Since general inflation is expected to fall drastically, a progressive rate structure will partially mitigate the effects of removing cross subsidies on households' income. When the structure is implemented, all sectors will experience inflation, although food inflation could be significantly lowered, depending on the increase in tariffs applied to the agricultural sector. As a result, rural households could suffer significantly less. However, this combined policy change would be costly, even with as little as a 0.5% markup for agriculture products. Removal of cross subsidies for power tariffs thus creates three specific problems that the government must deal with: a decrease in household income, food inflation, and economic inflation. Interestingly, the effect of removing cross subsidies on industries' output is insignificant; industries with increasing returns will show

5.2. Possible ways to make the policy change less impactful In this section four alternative policies are considered that are targeted to address some of the problems created by removal of cross subsidization. These policy options were tested with the assumption that the government's existing policy is similar to the situation in the last column of Table 4. Two of these policies are specifically designed to ensure that household income remains unaffected after a policy change regarding power tariffs. One of the remaining policies targets food inflation, and the other targets general inflation. Since in our model the change in household income is due to inflation, the policy that targets general inflation would also ensure that household income remains unchanged. To keep household incomes constant two alternative sources of providing them subsidy are considered: pure government grants and a combination of corporate tax hikes and government grants. Unlike the first option, which would double the fiscal deficit (and thus is an unlikely choice for the government), the second option would not have a significant negative effect on national finances though corporate taxation has to increase by a staggering 15%. However, the first option would significantly lower the general inflation level. Note the similarity between this policy and the progressive rate structure, which would raise tariffs for the agricultural sector by 5% (Table 4) but reduce household incomes. Since doubling of fiscal deficits or increasing 187

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

Table 4 Alternative ways to ensure consumer safety (percentage change between 2007 and 2014). Parameter

Production

Domestic price

Sector

Targeting food inflation

Targeting general inflation

Adjustment in government deficit

Corporate taxation

Price subsidy on agriculture

−0.002% −0.001% −0.003% −4%

−0.002% −0.001% −0.003% −4%

−0.002% −0.001% −0.003% −4%

0% 0% 0% −2%

−0.01% −3%

−0.01% −3%

−0.01% −3%

0% −1%

Agriculture Manufacturing and services Textiles Chemicals and chemical products Basic metals Machinery and equipment

32% 1% −1% 7%

35% 9% 5% 17%

0.7% 15% 6% 19%

0% 1% 0% 0%

2% 1% 7%

12% 14% 15%

14% 16% 13%

0% 0% 0%

Rural household income Urban household income

0% 0% −95%

0% 0% −12% 15%

−0.6% −1% −30%

0% 0% −147%

Agriculture Manufacturing and services Textiles Chemicals and chemical products Basic metal Machinery and equipment

Price index Income

Targeting household income

Govt savings Corporate tax

In Section 5, four such policies have been evaluated from which a direct subsidy in the form of direct cash transfer for agricultural products or food subsidy appears to be the best option, as the government deficit is affected less than by other policies and household incomes are only marginally affected. However it should be admitted that even this option is difficult to be implemented. If at all the direct benefit will have to be given to the people by using the, so called, Aadhaar platform. In India the Aadhaar card is a proof of identity and address and designed basically as the Social Security card for US citizens. However like in US where the institutional set up for issuing and using such a card for fund transfers is in place, in India such an institutional set up does not exist yet. Critical success factors for this scheme include effective issuance of Aadhaar cards to all beneficiaries, access to banking facilities at affordable costs and mobile connectivity. At present, 975 million individuals (about 75% of the population13) hold an Aadhaar card. A large number of the remaining 25% of the people are exactly the ones that need to be compensated by the direct benefit scheme. Further a large number of these people do not have a bank account. Since this is mandatory to make the direct transfer feasible, further effort needs to be put in to encourage people to open bank accounts. To provide access to affordable banking, the government introduced the Pradhan Mantri Jan Dhan Yojana scheme in 2014. Under this scheme, 220 million bank accounts were opened in India of which about 134 million accounts were rural accounts.14 Rural connectivity turned out to be the single biggest challenge in implementing this plan. Banks have also been engaging business correspondents who go to areas where it does not make economic sense for banks to open a branch. Regulations governing the remuneration of business correspondents need to be reviewed to ensure that commission rates are sufficient to encourage them to remain active. Also greater focus would be needed to improve the beneficiary database and develop the business correspondent infrastructure. India has over 1 billion mobile phone users which is around 85% of the population15. This could also

corporate taxes by 15% are unlikely to be feasible for any government, these options cannot be considered to be possible candidates for active consideration by the government. A more direct way to target household income is to lower inflation. Of all the sources of general inflation, food inflation affects poor households the most significantly. Thus, the next alternative policy that was investigated targeted food inflation. To lower food inflation, the government must subsidize the price of agricultural products. A 30% subsidy would almost entirely eliminate food inflation. However, inflation in other sectors may also decrease household incomes and increases the fiscal deficit by 30%. On the other hand, if general inflation is reduced to zero by appropriately subsidizing products from all sectors, the fiscal deficit would increase by almost 150%. Removal of cross subsidies in India's electricity prices is going to be costly for the Indian government, either in terms of dissatisfaction among households, especially in rural areas, or in terms of revenue. Of the four policies that were investigated, a direct subsidy for agricultural products appears to be the best option, as the government deficit is affected less than by other policies and household incomes are only marginally affected. Note that this option amounts to a restructuring of the rationing (fair price shop) system. The rationing system or public distribution system in India, distributes subsidized food and non-food items to the poor. However, there have been many instances where dealers replace supplies received from Government of India with inferior stock and sell the supplies in black market and several other malpractices and illegal diversions of commodities. To avoid black markets, the rationing system can be replaced by a system that directly benefits households, like the liquefied petroleum gas subsidies in India. In such a system, consumers buy products at the market price and subsidies are credited to their bank accounts at a later date. 6. Is direct cash transfer feasible? From the above discussion it is clear that an important effect of reduction of cross-subsidies in power pricing would be a hike in agricultural or food products which would lead to a decline in household incomes and a rise in inflation. Any removal or reduction in cross subsidy of electricity tariff should therefore be accompanied by policies which mitigate such negative effects of cross-subsidy reforms.

13 http://www.financialexpress.com/economy/economic-survey-2016-jan-dhan-seesonly-46-penetration-needs-a-leg-up/216687/ 14 http://www.business-standard.com/article/economy-policy/why-jan-dhan-yojanais-gaining-currency-in-uttar-pradesh-west-bengal-116052600920_1.html 15 http://www.forbes.com/sites/saritharai/2016/01/06/india-just-crossed-1-billionmobile-subscribers-milestone-and-the-excitements-just-beginning/#53d2adc45ac2

188

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

of Average Cost of Supply by FY 2010-11. However, SERCs were unable to adhere to the Tariff Policy. This paper examined why this was the case. The cost of removing cross-subsidies is daunting in terms of price inflation and household consumption, especially in rural areas. Additionally, there are no obvious ways in which the government (central or state) can mitigate the financial burden placed on households due to a change in tariffs. Two alternative methods of financing the policy change were considered: the government and the corporate sector. In the first case, the increase in the fiscal deficit is so high that it is not feasible. In the second case, corporate taxes would increase by 15%, which is too high for the government to consider. Therefore, abolition of cross subsidies in India is not possible without placing a considerable financial burden on households. An extension of the above research work would be to include more sectors so that sectoral impacts can be properly evaluated. Secondly the CGE model used was static. Dynamic CGE models are now available on purchase. An extension of the above results to the dynamic CGE version is urgently called for. Thirdly, coal was totally left out of the analysis. But it well known that coal is one of the major fuels used in India in the power sector. Also subsidies in coal have been reduced over time by introduction of the coal cess (Clean Energy Cess). What was the effect of this cess – this is an important question to answer and can be looked into during future research.

be used for enhancing the mobile banking and mobile money space in India. Till all that is achieved, the direct transfer mechanism will not be effective. Achieving all these is a tall order. And since we have shown that without adequate policy backup removal of cross subsidies will be extremely costly for the households, immediate removal of cross subsidies does not seem sensible at this point of time. 7. Conclusion This paper demonstrates why the implementation of structural adjustment programs is expected to be slow in a parliamentary democracy like India. Electricity tariffs in India are related to consumers' capacity to pay, keeping in mind the socio-economic considerations of different social sectors. Industrial consumers pay tariffs up to 40% higher than the average cost of supply of power in many states, although the cost to serve these consumers is lower than the average cost of supply due to lower technical and commercial losses. On the other hand, agricultural consumers in some states pay tariffs that cover only about 50% of the average cost of power. The Electricity Act of 2003 declared that tariffs must progressively reflect the cost of supply. This was reinforced by the National Electricity Policy and Tariff Policy, which changed tariffs for all categories of consumers to +/−20% of the average cost of supply by 2010–11. In India, as with all forms of subsidies, removing cross-subsidies on electricity also has been proven to be difficult. The review of performance of the states shows that SERCs were unable to adhere to the Tariff Policy guidelines to bring down cross subsidies to within +/−20%

Appendix See Table A1.

Table A1 Calibrated values of the parameters. Parameter

Description

Agriculture

Manufacturing and services

Textiles

Chemicals and chemical products

Basic metals

Machinery and equipment

Electricity

βi (Labour)

Share parameter in production functionProduc Share parameter in production functionProduc Production function shift parameter Composite factor requirement Government consumption share Import tariff rate Indirect tax rate Scale parameter in Armington function Share parameter of imported good Share parameter of domestic good Elasticity of substitution in Armington function Scale parameter in transformation function Share parameter of export Share parameter of domestic good(Transformation) Substitution elasticity in transformation Description Direct tax rate. Parameter for govt. transfer. Propensity to save for households.

0.583

0.536

0.459

0.232

0.173

0.306

0.877

0.417

0.464

0.541

0.768

0.827

0.694

0.123

1.972

1.995

1.993

1.72

1.586

1.851

1.453

0.626 0.016

0.444 −0.281

0.515 0.009

0.759 0.116

0.881 0.907

0.91 0.028

0.369 0.013

0.3 −0.004 1.034

0.3 0.002 0.698

0.3 −1.14E−4 59.26

0.3 0.014 2.364

0.3 0.038 6.241

0.3 0.034 3.007

0.3 0.002 0.872

2.46E−4

0.16

0.212

0.852

0.562

0.937

0.007

1.0

0.84

0.788

0.148

0.438

0.063

0.993

0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.466

0.465

0.479

0.478

0.446

0.446

0.462

2 2.159

2 2.159

2 2.159

2 2.159

2 2.159

2 2.159

2 2.159

1.5

1.5

1.5

1.5

1.5

1.5

1.5

Rural 0.041 0.345 0.299

Urban 0.051 0.237 0.281

βi (Capital)

bj ayi mui taumi tindi γi

δmi δdi

ηi φi xiei xidi

ρi Parameter taudb gtb sspb

189

Energy Policy 100 (2017) 181–190

R. Bhattacharyya, A. Ganguly

an International Comparison, German Ministry of Economic Affairs and Energy. 〈http://www.ecofys.com/files/files/ecofys-fraunhoferisi-2015-electricity-costs-ofenergy-intensive-industries.pdf〉 Helpman, E., Krugman, P.R., 1985. Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and the International Economy. MIT Press, Cambridge. Krugman, P.R., 1979. Increasing returns, monopolistic competition, and international trade. J. Int. Econ. 9 (4), 469–479. Lin, B., Ouyang, X., 2014. A revisit of fossil-fuel subsidies in China: challenges and opportunities for energy price reform. Energy Convers. Manag. 82, 124–134. McDonald, S., Schoor, M.V., 2005. A Computable General Equilibrium (CGE) Analysis of the Impact of an Oil Price Increase in South Africa, Research in Agricultural and Applied Economics, Working Paper 2005: 1. Retrieved from 〈http://purl.umn.edu/ 15633〉 National Power Training Institute, 2009. Positives of Availability based Tariff for Bulk Power Generation available at 〈http://www.npti.in/Download/Distribution/YMPL/ third%20party%20papers-technical%20info/ABT.pdf〉 Pradhan, Saluja and Sharma, 2013. A Social Accounting Matrix for India 2007–08, IEG Working Paper No. 326. Planning Commission, Government of India (2014). Annual Report 2013–14 on the working of State Power Utilities & Electricity Departments, 17–18. Sands, R., et al., 2011. Impacts of Higher Energy Prices on Agriculture and Rural Economies. United States Department Of Agriculture. Economic Research Report Number 123. Swain, A., Charnoz, O., 2012. In Pursuit of Energy Efficiency in India's Agriculture: fighting ‘Free Power’ or Working with it? Agence Française de Développement, Working Paper 126.〈http://www.afd.fr/webdav/shared/PUBLICATIONS/ RECHERCHE/Scientifiques/Documents-de-travail/126-document-travail-VA.pdf〉.

References Abouleinein, S., El-Laithy, H., Kheir-El-Din, H., 2009. The Impact of Phasing Out Subsidies of Petroleum Energy Products in Egypt. Egyptian Center Economic Studies, Working Paper No. 145. Alvarez, J., Valencia, F., 2015. Made in Mexico: energy reform and manufacturing growth. IMF Working Paper Western Hemisphere Department, WP/15/45. 〈https:// www.imf.org/external/pubs/ft/wp/2015/wp1545.pdf〉 Breisinger, C., Engelke, W., Ecker, O., 2011. Petroleum Subsidies in Yemen: leveraging Reform for Development. World Bank Policy Research Working Paper 5577, World Bank, Washington D.C. Cooke, E., Hague, S., Cockburn, J., Lahga, J., Tiberti, L., 2014. Estimating the Impact on Poverty of Ghana's Fuel Subsidy Reform and a Mitigating Response, Policy Paper, UNICEF. Retrieved from 〈http://www.unicef.org/ghana/1389903137_PEP_ UNICEF_Ghana_WP_201_public.pdf〉 Coupal, R.H., Holland, D., 2002. Economic impact of electric power industry deregulation on the state of Washington: a general equilibrium analysis.. J. Agric. Resour. Econ. 27 (1), 244–260. Dartanto, T., 2013. Reducing fuel subsidies and the implication on fiscal balance and poverty in Indonesia: a simulation analysis.. Energy Policy 58, 117–134. Dixit, A.K., Stiglitz, J.E., 1977. Monopolistic competition and optimum product diversity. Am. Econ. Rev. 67, 297–308. Fan, Y., Jiao, J.L., Liang, Q., Han, Z., Wei, Y., 2007. The impact of rising international crude oil price on China's economy: an empirical analysis with CGE model.. Int. J. Glob. Energy Issues 27 (4). Grave, K., Hazrat, M., Boeve, S., von Blücher, F., Bourgault, C., Breitschopf, B., Friedrichsen, N., Arens, M., Aydemir, A., Pudlik, M., Duscha, V., Ordonez, J., Lutz, C., Großmann, A., Flaute, M., 2015. Electricity Costs of Energy Intensive Industries—

190