This paper studies the relative contributions of the scale, structural and intensity effects on the total change of electricity consumption in 18 major states in India using the Laspeyres-based Parametric Divisia Method and the Simple Average Parametric Divisia Method in period-wise decomposition analysis. It also ranks states according to the proportionate values of the three effects with respect to the total change in electricity consumption. In this context, the paper proposes relying on the Divisia index rather than simple adding up methods. This analysis is based on secondary cross section data of the sectoral electricity consumption and sectoral output share for three benchmark years, 1990-91, 1995-96 and 2000-01. In the study, large variations among the states have been found in the identified effects for the early 1990s and post 1990s.
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200857Decomposition Analysis of Electricity Consumption: A State-wise AssessmentShrabani MukherjeeThe author is grateful to Krishnendu Ghosh Dastidar, Ram Prasad Sengupta and Debdulal Thakur for their enormous help and support. However, the usual disclaimers apply.Shrabani Mukherjee (shrabani0808@gmail.com) is a research scholar at the Centre for Economic Studies and Planning, Jawaharlal Nehru University, New Delhi.This paper studies the relative contributions of the scale, structural and intensity effects on the total change of electricity consumption in 18 major states in India using the Laspeyres-based Parametric Divisia Method and the Simple Average Parametric Divisia Method in period-wise decomposition analysis. It also ranks states according to the proportionate values of the three effects with respect to the total change in electricity consumption. In this context, the paper proposes relying on the Divisia index rather than simple adding up methods. This analysis is based on secondary cross section data of the sectoral electricity consumption and sectoral output share for three benchmark years, 1990-91, 1995-96 and 2000-01. In the study, large variations among the states have been found in the identified effects for the early 1990s and post 1990s. It is a well known fact that power generation and capacity expansion have always been essential for economic growth. In India, where the demand for electricity has outpaced sup-ply, due to a chronic shortage of electricity for a long period, a well-managed and sufficient supply of electrical power in a plan is far from perfect. Added to this, there exist large interstate vari-ations in the sectoral consumption of electricity. These interstate disparities in consumer category-wise sale of electricity arise mainly due to three factors, viz, variations in total production of the economy (scale effect), structural composition of the sectors (structural effect) and technical efficiency or, per unit of electric-ity use with respect to gross state domestic product (GSDP) (inten-sityeffect). This paper collates sufficient information regarding the relative change in scale of the economy, structural composi-tion of the sectors and technical competence in the increasing level of consumption of electrical power for the states. It thereby converges towards a sectoral disaggregation in the demand anal-ysis required to endorse efficiency and safety in the allocation of electricity to the states. The relative contributions of the scale effect, structural effect and intensity effect in the total change in electricity consumption have been estimated individually for the 18 major states using period-wise decomposition analysis.The power sector in India is characterised by the vertical in-tegration between generation, transmission and distribution. There are state electricity boards(SEBs) which possess, control and put on the market electricity from the clutch of generat-ing units within the state boundaries. Apart from SEBs, there are other state-owned utilities also known as the central sector utilities(CSUs). Transmission and distribution (T&D) losses are high in most of the states due to inadequate investments, defec-tive metering, tapping and unmetered supply. The all India T&D losses as a percentage of availability have increased from 34 per cent in 2001-02 to 38.3 per cent in 2002-031 (TEDDY, 2003-04). The disappointing and weakening financial health of SEBs has acted as a constraint not only for adding new capacity, improv-ing the T&D system and carrying out renovation and modernising programmes, but also for carrying out much needed reforms in electricity utilities.Presently (2003-04), India has an installed generating capacity of nearly 112 giga watts (GW). This includes thermal (coal, gas and liquid fuel) hydro, nuclear and wind power. Out of the total installed capacity, 90 per cent is owned by the public sector. The annual gross electricity generation in the utility is currently about 558 billion units (BU) with a net availability of 519 BU. The avail-ability of power was short of demand and as a result, the country experienced a shortage of 7.1 per cent in energy and 11.2 per cent in peak-period power (TEDDY 2003-04). Electricity sector suffered
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly58from serious under-investment (both public and private) in the Ninth Plan period (1997-2002), and a significant shortfall in the Tenth Five-Year Plan (2002-07). The decline in the private sector involvement in generation reflects the fact that the distribution segment of the power sector remains financially unviable. Un-der such circumstances, the recent National Electricity Policy of the government of India (GoI) aiming to meet the power demand fully by 2012 sounds far more ambitious.2 The installed generation capacity of the utilities in the country in March 2002 was 1,04,917.5 MW of which 59.33 per cent was owned by the states, 30.12 per cent by the centre and 10.55 per cent was owned by the private sector. The T&D losses increased from a level of 24.53 per cent in 1996-97 to 27.8 per cent in 2001-02.3 The actual power supply position as on March 2002,an assessment by the Central Electric-ity Authority(CEA), indicates a peak deficit of 12.6 per centand energy deficit of 7.5 per cent at the all India level as against a peak deficit of 18 per cent and energy deficit of 11.5per cent dur-ing 1996-97. The per capita electricity consumption of India was 355 kWh during 1999-2000 as against 334 kWh in 1996-97, whereas in China it was 719 kWh dur-ing 1997. A gross subsidy for domestic, agriculture and interstate sale has increased from a level of Rs 20,210 crore in 1996 to Rs 43,060.1 crore in 2001-02. The pattern of sales to various consumers has undergone significant changes in the last 10 years. The average per capita electricity consumption of the country as reported by the CEA has increased from 334 kWh in 1996-97 to 355 kWh in 1999-2000.4 The inefficiency in use of created capacities has been under-determined by the financial viability of electricity sector units of the states. Under this present structure, for the viability of the power sector reforms, we have to determine the actual require-ments of the states’ electricity sectors. The main objective of this paper is to estimate the disaggregated demand for electricity structure of the states using the additive decomposition analysis or the Divisia index method. 1 PreviousStudiesSince the end of 1970s, there has been extensive empirical re-search interest in energy consumption and economic growth, with neither conclusive results nor convincing explanations. It is often debated how the total quantity of energy resources pro-duced or consumed in a nation or region should be considered for use as a variable in the economic models. In economic models, the total consumption of energy resources is normally expressed either in terms of total heating value or in terms of its economic value (Divisia Indices or Expenditure).Beginning with Berndt(1985), many economists have pointed out the theoretical flaws inherent in aggregating different types of energy resources based on the heating value index and have suggested alternatives, such as Divisia energy aggregates which combine different resources based on their economic value. Zarnikau et al (1996) have traced the microeconomic foundations of the Divisia energy aggregate and discussed how the form value attributes of different energy resources are reflected in Divisia aggregation approaches. Hong(1983) has demonstrated how the Divisia and heating value indices lead to different conclusions regarding the trend in energy-output ratios for the US econo-my. Nguyen and Andrews (1989) have shown the superiority of Figure 1: General Framework of Decomposition MethodologyEnergyconsumptionapproachEnergyintensityapproachDecompositionmethodologyAdditive methodMultiplicative methodMultiplicative methodAdditive methodTable 1: Consumption of Electricity for the Year 1990-91 E0d E0c E 0a E 0I E 0r E 0 E 0- Ed States/Sector DomesticCommercialAgriculture Industry RailwayTraction Total Total- DomesticAndhra Pradesh 2,217 467 5,237 7275 305 16,144 13,927Assam 175 117 15894 NA1,554 1,379Bihar 464 277 1,544 2787 388 5,716 5,252Gujarat 1,519 5,069 6,588 249 15,857 14,338Haryana 982 165 2,749 1,700 NA6,772 5,790Himachal Pradesh 225 93 26 604 NA 1,767 1,542Jammu and Kashmir 378 236 147 360 NA 1,490 1112Karnataka 1,741 228 3,241 5,829 NA11,3559,614Kerala 1,620 585 226 3,003 NA5,739 4,119Madhya Pradesh 2,592 476 1,428 8,177 613 15,036 12,444Maharashtra 2,852 677 5,874 10,790345 28,583 25,731Meghalaya 3432169NA361327Orissa 670 138 230 2,839 2224,4753,805Punjab 1,591 380 5,616 5,165 NA14,05012,459Rajasthan 1,043 521 2,784 3,794 199,140 8,097Tamil Nadu 2,300 1,350 3,850 7,254 295 16,235 13,935Uttar Pradesh 2,555 1,474 7,267 7,295 842 20,185 17,630West Bengal 750 310 520 1725 410 5,981 5,231The data collected from “Consumer Category-wise Sale of Electricity” provided by the Annual Report of Working Group of State Electricity Board. NA signifies the missing data in corresponding column for corresponding state. The unit of the value is in Mkwh. Suffix 0 means the base year, here it is 1990-91. The last column shows the total consumption by the states minus consumption in domestic sector. Last column is used as the total consumption in this demand analysis.Table 2: Consumption of Electricity for the Year 1995-96 E 1d E 1c E 1a E 1i E 1r E 1 E 1- Ed States/Sector DomesticCommercialAgriculture Industry RailwayTraction Total Total- DomesticAndhra Pradesh 3,276 704 11,399 6,470 632 23,562 20,286Assam 429 160 44506 NA1,804 1,375Bihar 831 339 1,268 3381411 6,544 5,713Gujarat 2,176 601 1,01329,109 331 24,69522,519Haryana 1,637 258 3,9052,017 9087457,108Himachal Pradesh 387 112 12 968 NA 2,647 2,260Jammu and Kashmir 439 80 304 216 NA 1,728 1,289Karnataka 2,654 440 7,363 4,546 3115,98413,330Kerala 2,777799 322 532 NA7,415 4,638Madhya Pradesh 3,387 655 7,982 7902 1,083 22,957 19,570Maharashtra 4,424 979 13,33214,870821 41,61937,195Meghalaya 7740261NA510433Orissa 1,047 273 175 2,866 131 5,179 4,132Punjab 2,764 581 5,868 6,512 NA16,41213,648Rajasthan 1,961 685 4,343 5,127 180 13,703 11,742Tamil Nadu 3,924 1,711 6,631 9,817 342 24,610 20,686Uttar Pradesh 6,148 2,142 9,843 6,674 773 27,107 20,959West Bengal 1,612 691 1,232 2,124 466 8,951 7,339The data collected from “Consumer Category-wise Sale of Electricity” provided by the Annual Report of Working Group of State Electricity Board. NA signifies the missing data in corresponding column for corresponding state. The unit of the value is in MkWh. Suffix 1 means the final year of one period and base year of another period, here it is 1995-96. The last column shows the total consumption ofelectricity by the states minus consumption in domestic sector. Last column is used as the total consumption in this demand analysis.
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200861their respective values in the year “0” which is taken as the base year. In the two simple average based methods, no base year is specified and the mean values of relevant variables between year “0” and year “t” are used in the analysis.4 AnalysisThe study mainly applies the Divisia index to compute the im-pact of basic economic and structural changes of the economy on the electrical energy used by the sectors in the states. Basically, the objective of the paper is to find the relative contributions of theproduction effect, structural effect and intensity effect on the change of electricity use by the major 18 states (Andhra Pradesh, Assam, Bihar, Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Karnataka, Kerala, Madhya Pradesh, Maharashtra, Meghalaya, Orissa, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh,West Bengal) and then try to compare the estimated ef-fects between the period early 1990s and late 1990s. Here the year 1995-96 over 1990-91 represents the first period and later period signifies by the year 2000-01 over 1995-96. Further, the paper attempts to rank the states according to their per-formance of relative contributions of said effects for the respective periods. The present study is based on both sec-ondary data and information. Secondary data have been collected from different reports and publications.It is a state-wise sectoral analysis. Con-strained by the availability of relevant data, our study is based on the following assumptions:There are foursectors – (1) commer-cial sector which is mainly referred to communication, hotel and trade serv-ices, banking and insurance and public administration; (2) agriculture sector; (3) industry sector which is referred to mining, manufacturing and construc-tion; (4) transport sector, which here only refers to railway traction.This is a two-period analysis – early 1990s and late 1990s. (1) The paper takes 1995-96 as the bench-mark year and calculated five-year change taking 1990-91 as the base year for the early 1990s; (2) Similarly, in late 1990s, it takes 2000-01 as benchmark year using 1995-96 as the base year.Since the study attempts to find out the actual contribution in final electric-ity change by the sectors, the demand-side analysis, the model has to leave the domestic sector because it is generally used in the expenditure-side estimation in theory. In order to get the decomposi-tion analysis for the electricity demand Table A(b): Structural Effect for the Period of the Early 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -557.09 -1129.53 -795.56 -594.93Assam -412.89 -690.42 -348.74 -180.16Bihar -3160.82-1880.28-3523.10-4045.97Gujarat -607.79 -2773.97 -1130.27 -1113.79Haryana -191.01-1295.33-274.81-229.11Himachal Pradesh 186.55 -598.65 244.87 129.74Jammu and Kashmir 0.00 -1112.00 0.00 300.00Karnataka 345.8948.62231.5449.82Kerala -407.37-651.88-166.8422.86Madhya Pradesh -732.22 -2429.29 -939.26 -615.49Maharashtra -1395.99-9280.10-2215.71-1701.27Meghalaya -14.12-235.75-12.16-6.92Orissa -329.56 -674.01 -349.41 -201.18Punjab -89.84-1386.53-87.87-43.69Rajasthan 315.84-384.40278.30-77.28Tamil Nadu -155.78 -1314.63 -175.94 -119.18Uttar Pradesh -565.88 -1260.85 -672.51 -430.33West Bengal -314.76 -2544.28 -311.75 -180.89Table A(c): Intensity Effect for the Period of the Early 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -7413.16 -5412.83 -7149.55 -39209.80Assam -661.92 -793.39 -498.04 -2119.70Bihar 3855.48 9310.84 4324.65 31306.96Gujarat -4903.67-5614.03-5772.03-26905.51Haryana -2084.95-2710.03-2369.33-1819.16Himachal Pradesh -531.19 -1192.10 -656.42 -2704.28Jammu and Kashmir 0.00 -1112.00 0.00 1191.64Karnataka -7174.36-4523.91-6487.27-24334.08Kerala -8385.12-3534.42-5344.17-24252.29Madhya Pradesh -1413.20 -435.93 3037.70 9641.85Maharashtra -5623.40-12092.75 -6347.37-81.08Meghalaya -68.98-274.93-69.61-327.75Orissa -2783.87-2226.29-2693.45-22127.82Punjab -6289.91-6078.37-6731.46-24797.26Rajasthan -3630.90-3505.51-4171.51-14060.75Tamil Nadu -6896.96 -6452.44 -8327.16 -48453.96Uttar Pradesh -8358.24 -7028.92 -8534.57 -72307.62West Bengal -808.56 -2934.53 -889.29 -22020.94an integral path is undefined, this work is confined in the cross section analysis. But for comparison we have taken two bench-mark periods. The period-wise analysis needs some assumptions about the value of the parameters. For instance, α =β =ϒ = 0 like Laspeyres’ indices,α =β =ϒ = 1 like Paache’s, and α =β =ϒ=0.5 like Marshall-Edworth indices. We consider four specific decom-position methods depending on the values of the parameters pro-posed by many researchers as follows: 3.1 Absolute Consumption Approach(1) Laspeyres-based Parametric Divisia Method (LAS-PDM 1):ΔE pt = E0 ln (Yt /Y0) ΔE st = ∑E j,0 ln ( Sj,t/Sj.0) ΔE it = ∑E j,0 ln (Ij,t /Ij,0)Hereα = β = δ= 0. (2) Simple Average Parametric Divisia Methods (AVE-PDM 1):ΔE pt = 0.5 (Et + E 0) ln (Yt /Y0) ΔE st = 0.5 ∑ (Ej,t + Ej,0)] ln (Sj,t/Sj.0) ΔE it =0.5∑(Ej,t –Ej,0) ln (Ij,t /Ij,0 )Hereα = β = δ= 0.5.This method proposed by Boyd et al (1988). 3.2 ElectricityIntensityApproach (3) Laspeyres-based Parametric Divisia Method (LAS-PDM 2):ΔE pt =∑Ij,0Yt S j.0 – E0 ΔE st = ∑Ij,0Y0S j.t – E0 ΔE it = ∑Ij,tY0 S j.0 – E0 Hereα = β = δ= 0.Park (1992) proposes this method.(4)Simple Average Parametric Divisia Methods (AVE-PDM 1):ΔE pt = 0.5 (It + I 0)(Y t – Y0)… ΔE st = 0.5∑(Ij,0Y0 + I j,tY t)(S j,t–Sj,o)ΔE it = 0.5∑(Yj,0+Yj,t)(Ij,t– Ij,0)Hereα = β = δ= 0.5.This method is proposed by Reitler et al (1987). Generally, the analyst fixes par-ametric values.Now with these above formulae the values of “production effect”, “struc-tural effect” and “intensity effect” in the change of electrical power consumption can be estimated for the major states in India for periods, early 1990s and late 1990s. In the two Laspeyres-based meth-ods, each effect is isolated by measuring a change in electricity consumption as-sociated with a change in corresponding variable between year “0” and “t” while holding all the other variables constant at
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly62of the states, firstly, the total electricity consumption data should be adjusted by subtracting the domestic sector’s consumption value from it as shown in Tables 1 and 2. Since electricity is a non-storable commodity, consumer categorywise sale of electricity data can be used as the consumption of the end-use sectors. This data for three periods, 1990-91, 1995-96 and 2000-01 are shown by Tables 1, 2 and 3 (p 58, p 59) respectively.From Tables 1 and 2 it can be said that in the period, the early 1990s, the industrial consumption of electricity has gone down in most of the states, whereas the opposite picture emerges for the agriculture sector. The data for transport sector are unavailable for many states in 1990-91 and 1995-96.From Tables 2 and 3 no uniform directed (increased or de-creased) result for each sector by all the states for the period in post-1990s can be drawn. That means the electricity consump-tion level in the agricultural sector has increased for some of the states and also gone down for some states. The same thing hap-pened in the industrial sector also. Figures 2 and 3 (p 59, p 60) show the five-year change in total electricity consumption by the states for two periods, early 1990s and post-1990s. In order to get comparability between electricity consumption growth and output growth by the states, the paper needs to look at the trend of percentage change in electricity consumption as well as the percentage change in theGSDP of the corresponding states for the same two periods, early 1990s and late 1990s. Fig-ures 2 and 3 show the trends of GSDP growth and percentage change in electricity consumption for early 1990s and late 1990s respectively.From Tables 1, 2 and 3 and Figures 2 and 3, the total elec-trical energy consumption with the output growth of the states can be compared by the traditional method of comparability for each state. But, here, the search is something more specific than simple average, ie, the individual impact of the change in total production of the state, structural composition of the sectors and per unit use of electricity with respect to GSDP to the final change in electricity consumption. It takes into account the re-evaluation of actual efficiency of the states in electrical power consump-tion. Therefore, the paper postulates the disaggregated effects in total electricity used by the states using the said “consumption Table B(a): Production Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh 11039.05 12807.65 11359.30 11838.97Assam 771.89-130.31 741.86790.10Bihar 3634.95 4487.87 3990.30 4148.12Gujarat 9995.338924.97 11281.2011365.20Haryana 4200.034213.054640.544781.35Himachal Pradesh 1488.49 -150.11 1604.08 1683.13Jammu and Kashmir 728.57 -233.10 798.37 821.31Karnataka 8378.84 9881.60 8324.15 8900.99Kerala 2677.36-1693.743000.053073.71Madhya Pradesh 8163.11 7172.97 8625.28 8779.79Maharashtra 18321.1211903.7517553.20 18464.65Meghalaya 258.64-245.82269.69282.62Orissa 1298.69585.242009.641930.24Punjab 7812.69 9326.549147.34 9262.62Rajasthan 5632.79 4955.34 6312.55 6391.27Tamil Nadu 11569.18 11679.74 13303.49 13510.20Uttar Pradesh 11158.94 12134.57 10362.98 11076.51West Bengal 4704.79 1228.95 4692.29 5024.50Table B(b): Structural Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -1213.37 -1960.79 -1020.47 -603.44Assam 248.01-324.23226.6175.42Bihar 180.56-77.13148.7320.36Gujarat -3816.99-5355.49-4718.22-3560.59Haryana -615.27-1399.65-698.28-429.94Himachal Pradesh -146.78 -1304.22 -166.30 -100.30Jammu and Kashmir -91.79 -769.86 -110.68 -76.17Karnataka -1572.35-2399.19-1433.28-709.08Kerala 831.07-969.18 3346.75 1470.45Madhya Pradesh -1540.97 -3053.98 -2020.07 -1565.53Maharashtra -4370.27-10886.50-3954.35-2118.35Meghalaya 2.20-327.763.252.11Orissa -312.25-958.41-298.54-172.36Punjab -556.64 -1210.68 -669.46 -415.85Rajasthan -1490.42-2766.71-1762.66-1125.29Tamil Nadu -2620.58 -4520.61 -3046.16 -1954.55Uttar Pradesh -1187.40 -2622.69 -884.80 -319.34West Bengal -1461.58 -3443.46 -1533.20 -3664.55Table B(c): Intensity Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -8431.52 -7842.21 -8657.99 -35966.16Assam -919.08 -1172.41-777.76 -4158.23Bihar -2699.87-2322.89-2860.87-26445.51Gujarat -1116.82-1917.95-437.1523351.24Haryana -1758.07-2300.77-1901.52-9701.72Himachal Pradesh -299.70 -1430.04 -342.56 -1264.06Jammu and Kashmir -273.14 NA -127.46 -244.55Karnataka -7508.62-6434.18-6887.48-34691.77Kerala -575.50-3430.87-894.982092.63Madhya Pradesh -4642.19 -4791.93 -3989.66 -742.02Maharashtra -10537.63 -16037.43 -10526.24 -71020.04Meghalaya -15.93NA-16.67-47.47Orissa -869.18-1390.66-811.09141.95Punjab -3067.72NA-3548.06-15439.09Rajasthan -1288.79-2218.31-1014.64-2895.72Tamil Nadu -3312.87 -5000.61 -3648.54 -37784.29Uttar Pradesh -17833.69 -13014.60 -14459.84 -67258.93West Bengal -273.33 -372.94 -403.70 421198.74SPECIAL ISSUEBUDGET 2007-08April 7, 2007Fiscal Adjustment: Rhetoric and Reality – M Govinda RaoDeft Draughtsmanship sans Broader Vision – Pulin B NayakBudgetary Policy in the Context of Inflation – Prabhat PatnaikDividend Taxation Revisited – Amaresh BagchiEducation, Agriculture and Subsidies: Long on Words – Mala LalvaniImplications for Education – Anit Mukherjee No ‘New Deal’ for Farm Revival – S Mahendra DevFor copies write to: Circulation Manager Economic and Political Weekly,Hitkari House, 6th Floor, 284, Shahid Bhagatsingh Road, Mumbai 400 001.email: circulation@epw.org.in
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200863Table 4: Ranking of the States Corresponding to Their Decomposition EffectsMethods LAS_ PDM 1 AVE_PDM 1 LAS_ PDM 2 AVE_PDM 2States DEpt DEst DEit DEpt DEst DEit DEpt DEst DEit DEpt DEst DEitAndhraPradesh 9 9 127 9 107 4 5 13 4 12Assam 18 1 1 151 1 18 1 1 18181Bihar 17 18 2 18 11 17 16 17 2 1 2 18Gujarat 147 7 2 159 106 4 16154Haryana 5 131311 2 13 6 1112 6 1214Himachal Pradesh 12 2 9 8 5 16 14 10 11 11 6 10Jammu and Kashmir 16 5 3 17 4 18 17 18 16 5 14 9Karnataka 7 3 14 6 1611 5 2 9 9 8 15Kerala 2 1618 1 1014 2 1418 3 1 3MadhyaPradesh 15104 1613 7 12 7 3 17 9 6Maharashtra 8 116 4 8 6 11 9 8 15 7 13Meghalaya 6128101715151614837Orissa 1 17175 7 121 15174 165Punjab 3 8 16133 8 3 1215 2 1316Rajasthan 13 4 109 6 5 9 3 7 14118Tamil Nadu 10 6 11 3 14 4 8 5 6 12 10 11Uttar Pradesh 4 151514123 4 8 13 7 5 17West Bengal 11 14 5 12 18 2 13 13 10 10 17 2Table 5: Ranking of the States Corresponding to Their Decomposition EffectsMethods LAS_ PDM 1 AVE_PDM 1 LAS_ PDM 2 AVE_PDM 2States DEpt DEst DEit DEpt DEst DE it DEpt DEst DEit DEpt DEst DE itAndhra Pradesh 1 17 18 12 16 18 1 15 18 1 15 17Assam 16182 10183 2 4 2 16172Bihar 5 6 17 3 6 17 2 6 14 5 7 16Gujarat 11157 16156 8 111012165Haryana 7 1115 6 1115 4 1213 7 1115Himachal Pradesh 3 12 13 1 12 14 11 17 17 3 10 13Jammu and Kashmir 6 10 14 9 10 10 12 16 8 6 12 9Karnataka 17 2 1 4 2 1 18 2 1 17 2 1Kerala 8 4 107 3 131310168 3 6MadhyaPradesh 4 16161717163 1415 4 188Maharashtra 15 3 5 144 5 14 3 4 15 4 3Meghalaya 2 7857916187 2 611Orissa 13 8 6 188 7 10 7 9 13 8 7Punjab 12 9 12 8 9 12 5 8 6 11 9 12Rajasthan 9 149 15148 7 13119 1410TamilNadu 10 13111113116 912 10 1314Uttar Pradesh 14 5 4 135 4 15 5 5 14 5 4WestBengal 18 1 3 2 1 2 17 1 3 18 1 18DE pt, DE st, DE it signifies scale effect, structural effect and intensity effect, respectively.approach” and “intensity approach” of the decomposition analy-sis in a unified framework. The choice between the two methods would be based on the growth pattern of energy consumption andGSDP of the states. Generally,PDM 1 and PDM 2 would be preferred if the growth pattern is in logarithmic trend and linear trend, respectively. Since different states have different growth trends in both, the consumption of electricity and sectoral output share, we have calculated the period-wise decomposition analy-sis for all the states in bothPDM 1 and PDM 2 methods. 5 EmpiricalResultsThree major disaggregated effects, viz, “production effect”, “structural effect” and “intensity effect” have been estimated by using decomposition method, referred in our above dis-cussion for the major 18 states for said two periods – early 1990s and late 1990s as shown in Tables A(a) and B(b), respectively.Tables A and B (p 60, p 62) show that same trend exists in production effect for almost all the states by the said four different methods.It can be said that all the said methods (LAS-PDM 1, AVE-PDM 1, andAVE-PDM 2) except one (LAS-PDM 2) give more or less the same results for the states. Ac-cording to the LAS-PDM 2 method, Meghalaya had ex-tremely low amount of structural effect with respect to other values. LAS-PDM 1, LAS-PDM 2, AVE-PDM 1 give the same trend for all the states. Only the AVE-PDM 2 method gives fluctuating results for the intensity effects of the states. All the methods provide the same trend for produc-tion effect of most of the states in the second period. Andhra Pradesh, Himachal Pradesh, Jammu and Kashmir, Kerala and Maharashtra got the negative scale effect byLAS-PDM 2 method.Only theLAS-PDM 2 method among the others shows a different trend for the structural effect of all the states for the period late 1990s.In both the periods, the production effect is posi-tive, structural effect and intensity effect is negative for almost all the states. In the early 1990s,LAS-PDM 2 shows scale effect is negative for Bihar, Jammu and Kashmir and Meghalaya where as in the late 1990s that effect is negative for Assam, Jammu and Kash-mir, Kerala and Meghalaya. In the early 1990s,LAS-PDM 1,AVE-PDM 1 and AVE-PDM 2 show that Himachal Pradesh and Karnataka have a positive structural ef-fect, but in the late 1990s, this effect is positive for Assam and Bihar. In the early 1990s, all the methods show the intensity effect is positive for Bihar where-as in the late 1990s, West Bengal, Kerala and Orissa have a positive intensity effect. The residual terms are large, which means that there are some exogenous ef-fects in the total change of power consumption by the states, which have to be taken into account. The power sector reforms were initiated from June 1990. But the major policies and decisions for restructuring the electricity sector were implemented and start-ed working for most of the states after 1995. Therefore, the pe-riod late 1990s gives an immediate impact of the power sector reforms for the states. Since different states have different lev-els of income, size and growth, we rank the states according to their share of the effects in total change of electrical power con-sumption during the corresponding periods. Since the lower the value of the effects in change in final electricity consumption, the higher is the efficiency in use of scarce electrical power resources in an economy; we rank the states in a descending order of the estimated values of the effects. Therefore, a higher rank shows better quality in development of electricity sector of the states.5.1 Early 1990s and Late 1990sBeing able to disaggregate the total effects into three differ-ent components, now, the rank of the states has been made for
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly64the entire individual components by the four unambiguous methods for the two periods (shown by Tables 4 and 5, p 63). There are large variations in the position of the states between the periods, early 1990s and late 1990s. The variations are due to change in the production level, change in share of sectoral output in totalGSDP, change in per unit average tariff, change in techni-cal efficiency in transmission, distribution, final use of power by the states, etc, over the years 1990 to 2001. Now, for each state and for each method, the clear perception can be given about the state’s electricity exhausts whether for the increased produc-tion level or structural composition of the sectors or technical in-efficiency in power use. This method is more focused than the simple trend of percentage change in the electricity use. ConclusionsIn order to ensure the best use of scarce resources which is an imperative policy for energy sustainability, the information about the disaggregated effects of the total electricity consumption by the end use sector is the essential part of the state plan for the power sector, especially for electricity. From the above data, method and econometric analysis, it follows that there are significant variations in the electricity sector quality among the states. The consumer category-wise sale of electricity in In-dia hasincreased over the years not only for the change in pro-duction volume, but also a change in structural composition of the sectors, increasing technological efficiency and other changes inthe economy. Each of these changes is significant for every state with different dimensions. That means the de-mand for electricity arises unequally for the different factors in different states. This paper has traced the disaggregated effects (scale effect, structural composition effect and power intensity of GSDP effect) on the overall change of demand for electricity of the states ap-plying additive decomposition method. The basis of choosing among the four parametric Divisia methods applied should be the growth patterns of output and electricity consumption. Further, the choice of the parameter values depend on the assumptions re-lated to the chosen values that can best meet the objective of the study. In order to promote the developmental quality of the state electricity sector, the government should stress on the critical in-put among the effects (ie, scale, structure and intensity) used for quality measurement of the states at a point of time. Thereby, this study assumes significance in order to ensure the necessary and sufficient information about the disaggregated components of the demand for electricity of each state separately and thus facilitate understanding the individual requirements and needed modifications. The study reflects that decomposition analysis is crucially important to recognise the focused limitations of the states for a successful and sustainable integrated policy for re-gional development of the power sector in India.Notes1 TERI Energy Data and Directory Yearbook, 2003-04 (TEDDY 2003-04).2 Regulatory, TERI, 2003.3 Planning Commission, May 2002.4 Annual Report, May 2002.ReferencesAng, B W (1993): ‘Sector Disaggregation, Structural Change and Industrial Energy Consumption: An Approach to Analyse the Interrelationship’,En-ergy, Vol 18, No 10, pp 1032-44. – (1994): ‘Decomposition Methodology in Industri-al Energy Demand Analysis’,Energy International Journal, 20, 1081-95.Ang, B W and S Y Lee (1994): ‘Decomposition of Indus-trial Energy Consumption – Some Methodologi-cal and Application Issues’,Energy Economics, 16:83-92.Berndt, E (1985): ‘Aggregation Energy Efficiency and Productivity Measurement’,Annual Rev Energy, pp 225-73.Boyd, G A, D A Hanson, T Sterner (1988): ‘Decomposi-tion of Changes in Energy Intensity: A Compari-son of the Divisia Method and Other Methods’, Energy Economics, Vol 10, No 4, pp 309-12.Boyd G, J F McDonald, M Ross and D A Hanson (1987): ‘Separating the Changing Composition of US Manufacturing Production from Energy Efficien-cy Improvements: A Divisia Index Approach’,The Energy Journal, Vol 8, No 2, pp 77-96.Das, Anjana, Jyoti Parikh and Kirit Parikh (1999): ‘Power: The Critical Infrastructure’ in Kirit S Parikh (ed), ‘India Development Report 1999-2000’, New Delhi, p 120.Dhungel, Kamal Raj: ‘Income and Price Elasticity of Demand for Energy: A Macro-Level Empirical Analysis’, Pacific Asian Journal of Energy, 13(2) pp 73-84.Diewert, W E (1976): ‘Exact and Superlative Index Numbers’, Journal of Economics, 4, pp 115-45.Hankinson, G A and J M W Rhys (1992): ‘Electricity Consumption, Electricity Intensity and Industrial Structure’, Energy Economics, Vol 4, pp 265-70.Hankinson, G A and J M N Rhys (1983): ‘Electricity Consumption, Electricity and Industrial Struc-ture’,Energy Economics, 5, 146-52.Howarth, R B, L Schipper, P A Duerr and S Strom (1991): ‘Manufacturing Energy Use in Eight OECD Countries’,Energy Economics, 13, 135-142.Hong, N (1983): ‘Two Measures of Aggregate Energy Production Elasticities’,The Energy, 4(2): 172-27.Huntington, G H (1989): ‘The Impact of Sectoral Shifts in Industry on US Energy Demand’,Energy, Vol 14 No 6, pp 363-72.Li, J W (1990): ‘Structure Change and Energy Use’, Energy Economics, 12, 109-15.Liu, Q X, B W Ang and H L Ong (1992): ‘The Appli-cation of Divisia Index to the Decomposition of Changes in Industrial Energy Consumption’,The Energy Journal, Vol 13, No 4, pp 161-77. – (1992): ‘Inter Fuel Substitution and Decomposi-tion of Changes in Industrial Energy Consump-tion’,Energy, Vol 17, No7, pp 689-96. – (1992): ‘The Application of the Divisia Index to the Decomposition of Changes in Industrial En-ergy Consumption’,Energy Journal, 13, 161-77.Mishra, S K and V K Puri (2003): Indian Economy, Himalaya Publishing House, New Delhi.Munasingha, M and M Peter (1993): Energy Policy Analysis and Modelling, Cambridge University Press. National Account Statistics, CSO, Back Series (1960-61 to 2003-01), EPW Research Foundation.National Account Statistics, CSO, Government of India.Nguyen, S V and S H Andrews (1989): ‘The Effect of Energy Aggregation on Energy Elasticities: Some Evidence from US Manufacturing Data’,The Energy, 10: 149-55. Park, S H (1992), ‘Decomposition of Industrial Energy Consumption: An Alternative Method’, Energy Economics, Vol 14, No 4, pp 265-70.Paul, Shyamal, R N Bhattacharya (2004): ‘CO2 Emis-sion from Energy Use in India: A Decomposition Analysis,The Energy Policy, Vol 32, pp 585-93.Rangarajan, V (2005): ‘Determining T and D Losses in India: Their Impact on Distribution Privatisa-tion and Regulation’,Economic & Political Weekly, February 12.Reitler, W, M Rudolph and H Schaefer (1987): ‘Analy-sis of Factors Influencing Energy Consumption in Industry: A Revised Method’, Energy Economics, Vol 9, No 3, pp 145-48.Reitler, W, M Rudolph and M Schaefer (1987): ‘Analy-sis of the Factors Influencing Energy Consump-tion in Industry: A Revised Method,Energy Eco-nomics, 9, 145-48.Rejikumar, R (2005): ‘National Electricity Policy and Plan: A Critical Examination’,Economic & Politi-cal Weekly, May 14.Sengupta R P (1993): ‘Energy Modelling for India’, Planning Commission, Energy Division, India.TERI Energy Data & Directory Yearbook, various years.The Energy Research Institute (TERI), Regulatory, Is-sue 24, March 2005.The Energy Research Institute (TERI), Regulatory, Is-sue 25, March 2005.Torvanger, A (1991): ‘Manufacturing Sector Carbon Dioxide Emissions in Nine OECD Countries, 1973-87’,Energy Economics, 13, 168-86.Vogt, A (1978): ‘Divisia Indices on Different Paths’ in W Eichhorn, T Henn, O Opitz and R W Shephard (eds),Theory and Application of Economic Indices, Physicaverlag, Wurzburg, pp 297-305.Zarnikar, Jay (1999): ‘Defining Total Energy Use in Economic Studies: Does the Aggregation Ap-proach Matter?’,Energy Economics, 21, pp 485-4.Zarnikau J, S Guermouche and P Schmidt (1996): ‘Can Different Energy Resources be Added or Compared’,The Energy, 21(6) 483-91.
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200857Decomposition Analysis of Electricity Consumption: A State-wise AssessmentShrabani MukherjeeThe author is grateful to Krishnendu Ghosh Dastidar, Ram Prasad Sengupta and Debdulal Thakur for their enormous help and support. However, the usual disclaimers apply.Shrabani Mukherjee (shrabani0808@gmail.com) is a research scholar at the Centre for Economic Studies and Planning, Jawaharlal Nehru University, New Delhi.This paper studies the relative contributions of the scale, structural and intensity effects on the total change of electricity consumption in 18 major states in India using the Laspeyres-based Parametric Divisia Method and the Simple Average Parametric Divisia Method in period-wise decomposition analysis. It also ranks states according to the proportionate values of the three effects with respect to the total change in electricity consumption. In this context, the paper proposes relying on the Divisia index rather than simple adding up methods. This analysis is based on secondary cross section data of the sectoral electricity consumption and sectoral output share for three benchmark years, 1990-91, 1995-96 and 2000-01. In the study, large variations among the states have been found in the identified effects for the early 1990s and post 1990s. It is a well known fact that power generation and capacity expansion have always been essential for economic growth. In India, where the demand for electricity has outpaced sup-ply, due to a chronic shortage of electricity for a long period, a well-managed and sufficient supply of electrical power in a plan is far from perfect. Added to this, there exist large interstate vari-ations in the sectoral consumption of electricity. These interstate disparities in consumer category-wise sale of electricity arise mainly due to three factors, viz, variations in total production of the economy (scale effect), structural composition of the sectors (structural effect) and technical efficiency or, per unit of electric-ity use with respect to gross state domestic product (GSDP) (inten-sityeffect). This paper collates sufficient information regarding the relative change in scale of the economy, structural composi-tion of the sectors and technical competence in the increasing level of consumption of electrical power for the states. It thereby converges towards a sectoral disaggregation in the demand anal-ysis required to endorse efficiency and safety in the allocation of electricity to the states. The relative contributions of the scale effect, structural effect and intensity effect in the total change in electricity consumption have been estimated individually for the 18 major states using period-wise decomposition analysis.The power sector in India is characterised by the vertical in-tegration between generation, transmission and distribution. There are state electricity boards(SEBs) which possess, control and put on the market electricity from the clutch of generat-ing units within the state boundaries. Apart from SEBs, there are other state-owned utilities also known as the central sector utilities(CSUs). Transmission and distribution (T&D) losses are high in most of the states due to inadequate investments, defec-tive metering, tapping and unmetered supply. The all India T&D losses as a percentage of availability have increased from 34 per cent in 2001-02 to 38.3 per cent in 2002-031 (TEDDY, 2003-04). The disappointing and weakening financial health of SEBs has acted as a constraint not only for adding new capacity, improv-ing the T&D system and carrying out renovation and modernising programmes, but also for carrying out much needed reforms in electricity utilities.Presently (2003-04), India has an installed generating capacity of nearly 112 giga watts (GW). This includes thermal (coal, gas and liquid fuel) hydro, nuclear and wind power. Out of the total installed capacity, 90 per cent is owned by the public sector. The annual gross electricity generation in the utility is currently about 558 billion units (BU) with a net availability of 519 BU. The avail-ability of power was short of demand and as a result, the country experienced a shortage of 7.1 per cent in energy and 11.2 per cent in peak-period power (TEDDY 2003-04). Electricity sector suffered 
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly58from serious under-investment (both public and private) in the Ninth Plan period (1997-2002), and a significant shortfall in the Tenth Five-Year Plan (2002-07). The decline in the private sector involvement in generation reflects the fact that the distribution segment of the power sector remains financially unviable. Un-der such circumstances, the recent National Electricity Policy of the government of India (GoI) aiming to meet the power demand fully by 2012 sounds far more ambitious.2 The installed generation capacity of the utilities in the country in March 2002 was 1,04,917.5 MW of which 59.33 per cent was owned by the states, 30.12 per cent by the centre and 10.55 per cent was owned by the private sector. The T&D losses increased from a level of 24.53 per cent in 1996-97 to 27.8 per cent in 2001-02.3 The actual power supply position as on March 2002,an assessment by the Central Electric-ity Authority(CEA), indicates a peak deficit of 12.6 per centand energy deficit of 7.5 per cent at the all India level as against a peak deficit of 18 per cent and energy deficit of 11.5per cent dur-ing 1996-97. The per capita electricity consumption of India was 355 kWh during 1999-2000 as against 334 kWh in 1996-97, whereas in China it was 719 kWh dur-ing 1997. A gross subsidy for domestic, agriculture and interstate sale has increased from a level of Rs 20,210 crore in 1996 to Rs 43,060.1 crore in 2001-02. The pattern of sales to various consumers has undergone significant changes in the last 10 years. The average per capita electricity consumption of the country as reported by the CEA has increased from 334 kWh in 1996-97 to 355 kWh in 1999-2000.4 The inefficiency in use of created capacities has been under-determined by the financial viability of electricity sector units of the states. Under this present structure, for the viability of the power sector reforms, we have to determine the actual require-ments of the states’ electricity sectors. The main objective of this paper is to estimate the disaggregated demand for electricity structure of the states using the additive decomposition analysis or the Divisia index method. 1 PreviousStudiesSince the end of 1970s, there has been extensive empirical re-search interest in energy consumption and economic growth, with neither conclusive results nor convincing explanations. It is often debated how the total quantity of energy resources pro-duced or consumed in a nation or region should be considered for use as a variable in the economic models. In economic models, the total consumption of energy resources is normally expressed either in terms of total heating value or in terms of its economic value (Divisia Indices or Expenditure).Beginning with Berndt(1985), many economists have pointed out the theoretical flaws inherent in aggregating different types of energy resources based on the heating value index and have suggested alternatives, such as Divisia energy aggregates which combine different resources based on their economic value. Zarnikau et al (1996) have traced the microeconomic foundations of the Divisia energy aggregate and discussed how the form value attributes of different energy resources are reflected in Divisia aggregation approaches. Hong(1983) has demonstrated how the Divisia and heating value indices lead to different conclusions regarding the trend in energy-output ratios for the US econo-my. Nguyen and Andrews (1989) have shown the superiority of Figure 1: General Framework of Decomposition MethodologyEnergyconsumptionapproachEnergyintensityapproachDecompositionmethodologyAdditive methodMultiplicative methodMultiplicative methodAdditive methodTable 1: Consumption of Electricity for the Year 1990-91 E0d E0c E 0a E 0I E 0r E 0 E 0- Ed States/Sector DomesticCommercialAgriculture Industry RailwayTraction Total Total- DomesticAndhra Pradesh 2,217 467 5,237 7275 305 16,144 13,927Assam 175 117 15894 NA1,554 1,379Bihar 464 277 1,544 2787 388 5,716 5,252Gujarat 1,519 5,069 6,588 249 15,857 14,338Haryana 982 165 2,749 1,700 NA6,772 5,790Himachal Pradesh 225 93 26 604 NA 1,767 1,542Jammu and Kashmir 378 236 147 360 NA 1,490 1112Karnataka 1,741 228 3,241 5,829 NA11,3559,614Kerala 1,620 585 226 3,003 NA5,739 4,119Madhya Pradesh 2,592 476 1,428 8,177 613 15,036 12,444Maharashtra 2,852 677 5,874 10,790345 28,583 25,731Meghalaya 3432169NA361327Orissa 670 138 230 2,839 2224,4753,805Punjab 1,591 380 5,616 5,165 NA14,05012,459Rajasthan 1,043 521 2,784 3,794 199,140 8,097Tamil Nadu 2,300 1,350 3,850 7,254 295 16,235 13,935Uttar Pradesh 2,555 1,474 7,267 7,295 842 20,185 17,630West Bengal 750 310 520 1725 410 5,981 5,231The data collected from “Consumer Category-wise Sale of Electricity” provided by the Annual Report of Working Group of State Electricity Board. NA signifies the missing data in corresponding column for corresponding state. The unit of the value is in Mkwh. Suffix 0 means the base year, here it is 1990-91. The last column shows the total consumption by the states minus consumption in domestic sector. Last column is used as the total consumption in this demand analysis.Table 2: Consumption of Electricity for the Year 1995-96 E 1d E 1c E 1a E 1i E 1r E 1 E 1- Ed States/Sector DomesticCommercialAgriculture Industry RailwayTraction Total Total- DomesticAndhra Pradesh 3,276 704 11,399 6,470 632 23,562 20,286Assam 429 160 44506 NA1,804 1,375Bihar 831 339 1,268 3381411 6,544 5,713Gujarat 2,176 601 1,01329,109 331 24,69522,519Haryana 1,637 258 3,9052,017 9087457,108Himachal Pradesh 387 112 12 968 NA 2,647 2,260Jammu and Kashmir 439 80 304 216 NA 1,728 1,289Karnataka 2,654 440 7,363 4,546 3115,98413,330Kerala 2,777799 322 532 NA7,415 4,638Madhya Pradesh 3,387 655 7,982 7902 1,083 22,957 19,570Maharashtra 4,424 979 13,33214,870821 41,61937,195Meghalaya 7740261NA510433Orissa 1,047 273 175 2,866 131 5,179 4,132Punjab 2,764 581 5,868 6,512 NA16,41213,648Rajasthan 1,961 685 4,343 5,127 180 13,703 11,742Tamil Nadu 3,924 1,711 6,631 9,817 342 24,610 20,686Uttar Pradesh 6,148 2,142 9,843 6,674 773 27,107 20,959West Bengal 1,612 691 1,232 2,124 466 8,951 7,339The data collected from “Consumer Category-wise Sale of Electricity” provided by the Annual Report of Working Group of State Electricity Board. NA signifies the missing data in corresponding column for corresponding state. The unit of the value is in MkWh. Suffix 1 means the final year of one period and base year of another period, here it is 1995-96. The last column shows the total consumption ofelectricity by the states minus consumption in domestic sector. Last column is used as the total consumption in this demand analysis.
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200861their respective values in the year “0” which is taken as the base year. In the two simple average based methods, no base year is specified and the mean values of relevant variables between year “0” and year “t” are used in the analysis.4 AnalysisThe study mainly applies the Divisia index to compute the im-pact of basic economic and structural changes of the economy on the electrical energy used by the sectors in the states. Basically, the objective of the paper is to find the relative contributions of theproduction effect, structural effect and intensity effect on the change of electricity use by the major 18 states (Andhra Pradesh, Assam, Bihar, Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Karnataka, Kerala, Madhya Pradesh, Maharashtra, Meghalaya, Orissa, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh,West Bengal) and then try to compare the estimated ef-fects between the period early 1990s and late 1990s. Here the year 1995-96 over 1990-91 represents the first period and later period signifies by the year 2000-01 over 1995-96. Further, the paper attempts to rank the states according to their per-formance of relative contributions of said effects for the respective periods. The present study is based on both sec-ondary data and information. Secondary data have been collected from different reports and publications.It is a state-wise sectoral analysis. Con-strained by the availability of relevant data, our study is based on the following assumptions:There are foursectors – (1) commer-cial sector which is mainly referred to communication, hotel and trade serv-ices, banking and insurance and public administration; (2) agriculture sector; (3) industry sector which is referred to mining, manufacturing and construc-tion; (4) transport sector, which here only refers to railway traction.This is a two-period analysis – early 1990s and late 1990s. (1) The paper takes 1995-96 as the bench-mark year and calculated five-year change taking 1990-91 as the base year for the early 1990s; (2) Similarly, in late 1990s, it takes 2000-01 as benchmark year using 1995-96 as the base year.Since the study attempts to find out the actual contribution in final electric-ity change by the sectors, the demand-side analysis, the model has to leave the domestic sector because it is generally used in the expenditure-side estimation in theory. In order to get the decomposi-tion analysis for the electricity demand Table A(b): Structural Effect for the Period of the Early 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -557.09 -1129.53 -795.56 -594.93Assam -412.89 -690.42 -348.74 -180.16Bihar -3160.82-1880.28-3523.10-4045.97Gujarat -607.79 -2773.97 -1130.27 -1113.79Haryana -191.01-1295.33-274.81-229.11Himachal Pradesh 186.55 -598.65 244.87 129.74Jammu and Kashmir 0.00 -1112.00 0.00 300.00Karnataka 345.8948.62231.5449.82Kerala -407.37-651.88-166.8422.86Madhya Pradesh -732.22 -2429.29 -939.26 -615.49Maharashtra -1395.99-9280.10-2215.71-1701.27Meghalaya -14.12-235.75-12.16-6.92Orissa -329.56 -674.01 -349.41 -201.18Punjab -89.84-1386.53-87.87-43.69Rajasthan 315.84-384.40278.30-77.28Tamil Nadu -155.78 -1314.63 -175.94 -119.18Uttar Pradesh -565.88 -1260.85 -672.51 -430.33West Bengal -314.76 -2544.28 -311.75 -180.89Table A(c): Intensity Effect for the Period of the Early 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -7413.16 -5412.83 -7149.55 -39209.80Assam -661.92 -793.39 -498.04 -2119.70Bihar 3855.48 9310.84 4324.65 31306.96Gujarat -4903.67-5614.03-5772.03-26905.51Haryana -2084.95-2710.03-2369.33-1819.16Himachal Pradesh -531.19 -1192.10 -656.42 -2704.28Jammu and Kashmir 0.00 -1112.00 0.00 1191.64Karnataka -7174.36-4523.91-6487.27-24334.08Kerala -8385.12-3534.42-5344.17-24252.29Madhya Pradesh -1413.20 -435.93 3037.70 9641.85Maharashtra -5623.40-12092.75 -6347.37-81.08Meghalaya -68.98-274.93-69.61-327.75Orissa -2783.87-2226.29-2693.45-22127.82Punjab -6289.91-6078.37-6731.46-24797.26Rajasthan -3630.90-3505.51-4171.51-14060.75Tamil Nadu -6896.96 -6452.44 -8327.16 -48453.96Uttar Pradesh -8358.24 -7028.92 -8534.57 -72307.62West Bengal -808.56 -2934.53 -889.29 -22020.94an integral path is undefined, this work is confined in the cross section analysis. But for comparison we have taken two bench-mark periods. The period-wise analysis needs some assumptions about the value of the parameters. For instance, α =β =ϒ = 0 like Laspeyres’ indices,α =β =ϒ = 1 like Paache’s, and α =β =ϒ=0.5 like Marshall-Edworth indices. We consider four specific decom-position methods depending on the values of the parameters pro-posed by many researchers as follows: 3.1 Absolute Consumption Approach(1) Laspeyres-based Parametric Divisia Method (LAS-PDM 1):ΔE pt = E0 ln (Yt /Y0) ΔE st = ∑E j,0 ln ( Sj,t/Sj.0) ΔE it = ∑E j,0 ln (Ij,t /Ij,0)Hereα = β = δ= 0. (2) Simple Average Parametric Divisia Methods (AVE-PDM 1):ΔE pt = 0.5 (Et + E 0) ln (Yt /Y0) ΔE st = 0.5 ∑ (Ej,t + Ej,0)] ln (Sj,t/Sj.0) ΔE it =0.5∑(Ej,t –Ej,0) ln (Ij,t /Ij,0 )Hereα = β = δ= 0.5.This method proposed by Boyd et al (1988). 3.2 ElectricityIntensityApproach (3) Laspeyres-based Parametric Divisia Method (LAS-PDM 2):ΔE pt =∑Ij,0Yt S j.0 – E0 ΔE st = ∑Ij,0Y0S j.t – E0 ΔE it = ∑Ij,tY0 S j.0 – E0 Hereα = β = δ= 0.Park (1992) proposes this method.(4)Simple Average Parametric Divisia Methods (AVE-PDM 1):ΔE pt = 0.5 (It + I 0)(Y t – Y0)… ΔE st = 0.5∑(Ij,0Y0 + I j,tY t)(S j,t–Sj,o)ΔE it = 0.5∑(Yj,0+Yj,t)(Ij,t– Ij,0)Hereα = β = δ= 0.5.This method is proposed by Reitler et al (1987). Generally, the analyst fixes par-ametric values.Now with these above formulae the values of “production effect”, “struc-tural effect” and “intensity effect” in the change of electrical power consumption can be estimated for the major states in India for periods, early 1990s and late 1990s. In the two Laspeyres-based meth-ods, each effect is isolated by measuring a change in electricity consumption as-sociated with a change in corresponding variable between year “0” and “t” while holding all the other variables constant at 
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly62of the states, firstly, the total electricity consumption data should be adjusted by subtracting the domestic sector’s consumption value from it as shown in Tables 1 and 2. Since electricity is a non-storable commodity, consumer categorywise sale of electricity data can be used as the consumption of the end-use sectors. This data for three periods, 1990-91, 1995-96 and 2000-01 are shown by Tables 1, 2 and 3 (p 58, p 59) respectively.From Tables 1 and 2 it can be said that in the period, the early 1990s, the industrial consumption of electricity has gone down in most of the states, whereas the opposite picture emerges for the agriculture sector. The data for transport sector are unavailable for many states in 1990-91 and 1995-96.From Tables 2 and 3 no uniform directed (increased or de-creased) result for each sector by all the states for the period in post-1990s can be drawn. That means the electricity consump-tion level in the agricultural sector has increased for some of the states and also gone down for some states. The same thing hap-pened in the industrial sector also. Figures 2 and 3 (p 59, p 60) show the five-year change in total electricity consumption by the states for two periods, early 1990s and post-1990s. In order to get comparability between electricity consumption growth and output growth by the states, the paper needs to look at the trend of percentage change in electricity consumption as well as the percentage change in theGSDP of the corresponding states for the same two periods, early 1990s and late 1990s. Fig-ures 2 and 3 show the trends of GSDP growth and percentage change in electricity consumption for early 1990s and late 1990s respectively.From Tables 1, 2 and 3 and Figures 2 and 3, the total elec-trical energy consumption with the output growth of the states can be compared by the traditional method of comparability for each state. But, here, the search is something more specific than simple average, ie, the individual impact of the change in total production of the state, structural composition of the sectors and per unit use of electricity with respect to GSDP to the final change in electricity consumption. It takes into account the re-evaluation of actual efficiency of the states in electrical power consump-tion. Therefore, the paper postulates the disaggregated effects in total electricity used by the states using the said “consumption Table B(a): Production Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh 11039.05 12807.65 11359.30 11838.97Assam 771.89-130.31 741.86790.10Bihar 3634.95 4487.87 3990.30 4148.12Gujarat 9995.338924.97 11281.2011365.20Haryana 4200.034213.054640.544781.35Himachal Pradesh 1488.49 -150.11 1604.08 1683.13Jammu and Kashmir 728.57 -233.10 798.37 821.31Karnataka 8378.84 9881.60 8324.15 8900.99Kerala 2677.36-1693.743000.053073.71Madhya Pradesh 8163.11 7172.97 8625.28 8779.79Maharashtra 18321.1211903.7517553.20 18464.65Meghalaya 258.64-245.82269.69282.62Orissa 1298.69585.242009.641930.24Punjab 7812.69 9326.549147.34 9262.62Rajasthan 5632.79 4955.34 6312.55 6391.27Tamil Nadu 11569.18 11679.74 13303.49 13510.20Uttar Pradesh 11158.94 12134.57 10362.98 11076.51West Bengal 4704.79 1228.95 4692.29 5024.50Table B(b): Structural Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -1213.37 -1960.79 -1020.47 -603.44Assam 248.01-324.23226.6175.42Bihar 180.56-77.13148.7320.36Gujarat -3816.99-5355.49-4718.22-3560.59Haryana -615.27-1399.65-698.28-429.94Himachal Pradesh -146.78 -1304.22 -166.30 -100.30Jammu and Kashmir -91.79 -769.86 -110.68 -76.17Karnataka -1572.35-2399.19-1433.28-709.08Kerala 831.07-969.18 3346.75 1470.45Madhya Pradesh -1540.97 -3053.98 -2020.07 -1565.53Maharashtra -4370.27-10886.50-3954.35-2118.35Meghalaya 2.20-327.763.252.11Orissa -312.25-958.41-298.54-172.36Punjab -556.64 -1210.68 -669.46 -415.85Rajasthan -1490.42-2766.71-1762.66-1125.29Tamil Nadu -2620.58 -4520.61 -3046.16 -1954.55Uttar Pradesh -1187.40 -2622.69 -884.80 -319.34West Bengal -1461.58 -3443.46 -1533.20 -3664.55Table B(c): Intensity Effect for the Period of the Late 1990sStates LAS-PDM1LAS-PDM2AVE-PDM1AVE-PDM2Andhra Pradesh -8431.52 -7842.21 -8657.99 -35966.16Assam -919.08 -1172.41-777.76 -4158.23Bihar -2699.87-2322.89-2860.87-26445.51Gujarat -1116.82-1917.95-437.1523351.24Haryana -1758.07-2300.77-1901.52-9701.72Himachal Pradesh -299.70 -1430.04 -342.56 -1264.06Jammu and Kashmir -273.14 NA -127.46 -244.55Karnataka -7508.62-6434.18-6887.48-34691.77Kerala -575.50-3430.87-894.982092.63Madhya Pradesh -4642.19 -4791.93 -3989.66 -742.02Maharashtra -10537.63 -16037.43 -10526.24 -71020.04Meghalaya -15.93NA-16.67-47.47Orissa -869.18-1390.66-811.09141.95Punjab -3067.72NA-3548.06-15439.09Rajasthan -1288.79-2218.31-1014.64-2895.72Tamil Nadu -3312.87 -5000.61 -3648.54 -37784.29Uttar Pradesh -17833.69 -13014.60 -14459.84 -67258.93West Bengal -273.33 -372.94 -403.70 421198.74SPECIAL ISSUEBUDGET 2007-08April 7, 2007Fiscal Adjustment: Rhetoric and Reality – M Govinda RaoDeft Draughtsmanship sans Broader Vision – Pulin B NayakBudgetary Policy in the Context of Inflation – Prabhat PatnaikDividend Taxation Revisited – Amaresh BagchiEducation, Agriculture and Subsidies: Long on Words – Mala LalvaniImplications for Education – Anit Mukherjee No ‘New Deal’ for Farm Revival – S Mahendra DevFor copies write to: Circulation Manager Economic and Political Weekly,Hitkari House, 6th Floor, 284, Shahid Bhagatsingh Road, Mumbai 400 001.email: circulation@epw.org.in 
SPECIAL ARTICLEEconomic & Political Weekly january 19, 200863Table 4: Ranking of the States Corresponding to Their Decomposition EffectsMethods LAS_ PDM 1 AVE_PDM 1 LAS_ PDM 2 AVE_PDM 2States DEpt DEst DEit DEpt DEst DEit DEpt DEst DEit DEpt DEst DEitAndhraPradesh 9 9 127 9 107 4 5 13 4 12Assam 18 1 1 151 1 18 1 1 18181Bihar 17 18 2 18 11 17 16 17 2 1 2 18Gujarat 147 7 2 159 106 4 16154Haryana 5 131311 2 13 6 1112 6 1214Himachal Pradesh 12 2 9 8 5 16 14 10 11 11 6 10Jammu and Kashmir 16 5 3 17 4 18 17 18 16 5 14 9Karnataka 7 3 14 6 1611 5 2 9 9 8 15Kerala 2 1618 1 1014 2 1418 3 1 3MadhyaPradesh 15104 1613 7 12 7 3 17 9 6Maharashtra 8 116 4 8 6 11 9 8 15 7 13Meghalaya 6128101715151614837Orissa 1 17175 7 121 15174 165Punjab 3 8 16133 8 3 1215 2 1316Rajasthan 13 4 109 6 5 9 3 7 14118Tamil Nadu 10 6 11 3 14 4 8 5 6 12 10 11Uttar Pradesh 4 151514123 4 8 13 7 5 17West Bengal 11 14 5 12 18 2 13 13 10 10 17 2Table 5: Ranking of the States Corresponding to Their Decomposition EffectsMethods LAS_ PDM 1 AVE_PDM 1 LAS_ PDM 2 AVE_PDM 2States DEpt DEst DEit DEpt DEst DE it DEpt DEst DEit DEpt DEst DE itAndhra Pradesh 1 17 18 12 16 18 1 15 18 1 15 17Assam 16182 10183 2 4 2 16172Bihar 5 6 17 3 6 17 2 6 14 5 7 16Gujarat 11157 16156 8 111012165Haryana 7 1115 6 1115 4 1213 7 1115Himachal Pradesh 3 12 13 1 12 14 11 17 17 3 10 13Jammu and Kashmir 6 10 14 9 10 10 12 16 8 6 12 9Karnataka 17 2 1 4 2 1 18 2 1 17 2 1Kerala 8 4 107 3 131310168 3 6MadhyaPradesh 4 16161717163 1415 4 188Maharashtra 15 3 5 144 5 14 3 4 15 4 3Meghalaya 2 7857916187 2 611Orissa 13 8 6 188 7 10 7 9 13 8 7Punjab 12 9 12 8 9 12 5 8 6 11 9 12Rajasthan 9 149 15148 7 13119 1410TamilNadu 10 13111113116 912 10 1314Uttar Pradesh 14 5 4 135 4 15 5 5 14 5 4WestBengal 18 1 3 2 1 2 17 1 3 18 1 18DE pt, DE st, DE it signifies scale effect, structural effect and intensity effect, respectively.approach” and “intensity approach” of the decomposition analy-sis in a unified framework. The choice between the two methods would be based on the growth pattern of energy consumption andGSDP of the states. Generally,PDM 1 and PDM 2 would be preferred if the growth pattern is in logarithmic trend and linear trend, respectively. Since different states have different growth trends in both, the consumption of electricity and sectoral output share, we have calculated the period-wise decomposition analy-sis for all the states in bothPDM 1 and PDM 2 methods. 5 EmpiricalResultsThree major disaggregated effects, viz, “production effect”, “structural effect” and “intensity effect” have been estimated by using decomposition method, referred in our above dis-cussion for the major 18 states for said two periods – early 1990s and late 1990s as shown in Tables A(a) and B(b), respectively.Tables A and B (p 60, p 62) show that same trend exists in production effect for almost all the states by the said four different methods.It can be said that all the said methods (LAS-PDM 1, AVE-PDM 1, andAVE-PDM 2) except one (LAS-PDM 2) give more or less the same results for the states. Ac-cording to the LAS-PDM 2 method, Meghalaya had ex-tremely low amount of structural effect with respect to other values. LAS-PDM 1, LAS-PDM 2, AVE-PDM 1 give the same trend for all the states. Only the AVE-PDM 2 method gives fluctuating results for the intensity effects of the states. All the methods provide the same trend for produc-tion effect of most of the states in the second period. Andhra Pradesh, Himachal Pradesh, Jammu and Kashmir, Kerala and Maharashtra got the negative scale effect byLAS-PDM 2 method.Only theLAS-PDM 2 method among the others shows a different trend for the structural effect of all the states for the period late 1990s.In both the periods, the production effect is posi-tive, structural effect and intensity effect is negative for almost all the states. In the early 1990s,LAS-PDM 2 shows scale effect is negative for Bihar, Jammu and Kashmir and Meghalaya where as in the late 1990s that effect is negative for Assam, Jammu and Kash-mir, Kerala and Meghalaya. In the early 1990s,LAS-PDM 1,AVE-PDM 1 and AVE-PDM 2 show that Himachal Pradesh and Karnataka have a positive structural ef-fect, but in the late 1990s, this effect is positive for Assam and Bihar. In the early 1990s, all the methods show the intensity effect is positive for Bihar where-as in the late 1990s, West Bengal, Kerala and Orissa have a positive intensity effect. The residual terms are large, which means that there are some exogenous ef-fects in the total change of power consumption by the states, which have to be taken into account. The power sector reforms were initiated from June 1990. But the major policies and decisions for restructuring the electricity sector were implemented and start-ed working for most of the states after 1995. Therefore, the pe-riod late 1990s gives an immediate impact of the power sector reforms for the states. Since different states have different lev-els of income, size and growth, we rank the states according to their share of the effects in total change of electrical power con-sumption during the corresponding periods. Since the lower the value of the effects in change in final electricity consumption, the higher is the efficiency in use of scarce electrical power resources in an economy; we rank the states in a descending order of the estimated values of the effects. Therefore, a higher rank shows better quality in development of electricity sector of the states.5.1 Early 1990s and Late 1990sBeing able to disaggregate the total effects into three differ-ent components, now, the rank of the states has been made for 
SPECIAL ARTICLEjanuary 19, 2008 Economic & Political Weekly64the entire individual components by the four unambiguous methods for the two periods (shown by Tables 4 and 5, p 63). There are large variations in the position of the states between the periods, early 1990s and late 1990s. The variations are due to change in the production level, change in share of sectoral output in totalGSDP, change in per unit average tariff, change in techni-cal efficiency in transmission, distribution, final use of power by the states, etc, over the years 1990 to 2001. Now, for each state and for each method, the clear perception can be given about the state’s electricity exhausts whether for the increased produc-tion level or structural composition of the sectors or technical in-efficiency in power use. This method is more focused than the simple trend of percentage change in the electricity use. ConclusionsIn order to ensure the best use of scarce resources which is an imperative policy for energy sustainability, the information about the disaggregated effects of the total electricity consumption by the end use sector is the essential part of the state plan for the power sector, especially for electricity. From the above data, method and econometric analysis, it follows that there are significant variations in the electricity sector quality among the states. The consumer category-wise sale of electricity in In-dia hasincreased over the years not only for the change in pro-duction volume, but also a change in structural composition of the sectors, increasing technological efficiency and other changes inthe economy. Each of these changes is significant for every state with different dimensions. That means the de-mand for electricity arises unequally for the different factors in different states. This paper has traced the disaggregated effects (scale effect, structural composition effect and power intensity of GSDP effect) on the overall change of demand for electricity of the states ap-plying additive decomposition method. The basis of choosing among the four parametric Divisia methods applied should be the growth patterns of output and electricity consumption. Further, the choice of the parameter values depend on the assumptions re-lated to the chosen values that can best meet the objective of the study. In order to promote the developmental quality of the state electricity sector, the government should stress on the critical in-put among the effects (ie, scale, structure and intensity) used for quality measurement of the states at a point of time. Thereby, this study assumes significance in order to ensure the necessary and sufficient information about the disaggregated components of the demand for electricity of each state separately and thus facilitate understanding the individual requirements and needed modifications. The study reflects that decomposition analysis is crucially important to recognise the focused limitations of the states for a successful and sustainable integrated policy for re-gional development of the power sector in India.Notes1 TERI Energy Data and Directory Yearbook, 2003-04 (TEDDY 2003-04).2 Regulatory, TERI, 2003.3 Planning Commission, May 2002.4 Annual Report, May 2002.ReferencesAng, B W (1993): ‘Sector Disaggregation, Structural Change and Industrial Energy Consumption: An Approach to Analyse the Interrelationship’,En-ergy, Vol 18, No 10, pp 1032-44. – (1994): ‘Decomposition Methodology in Industri-al Energy Demand Analysis’,Energy International Journal, 20, 1081-95.Ang, B W and S Y Lee (1994): ‘Decomposition of Indus-trial Energy Consumption – Some Methodologi-cal and Application Issues’,Energy Economics, 16:83-92.Berndt, E (1985): ‘Aggregation Energy Efficiency and Productivity Measurement’,Annual Rev Energy, pp 225-73.Boyd, G A, D A Hanson, T Sterner (1988): ‘Decomposi-tion of Changes in Energy Intensity: A Compari-son of the Divisia Method and Other Methods’, Energy Economics, Vol 10, No 4, pp 309-12.Boyd G, J F McDonald, M Ross and D A Hanson (1987): ‘Separating the Changing Composition of US Manufacturing Production from Energy Efficien-cy Improvements: A Divisia Index Approach’,The Energy Journal, Vol 8, No 2, pp 77-96.Das, Anjana, Jyoti Parikh and Kirit Parikh (1999): ‘Power: The Critical Infrastructure’ in Kirit S Parikh (ed), ‘India Development Report 1999-2000’, New Delhi, p 120.Dhungel, Kamal Raj: ‘Income and Price Elasticity of Demand for Energy: A Macro-Level Empirical Analysis’, Pacific Asian Journal of Energy, 13(2) pp 73-84.Diewert, W E (1976): ‘Exact and Superlative Index Numbers’, Journal of Economics, 4, pp 115-45.Hankinson, G A and J M W Rhys (1992): ‘Electricity Consumption, Electricity Intensity and Industrial Structure’, Energy Economics, Vol 4, pp 265-70.Hankinson, G A and J M N Rhys (1983): ‘Electricity Consumption, Electricity and Industrial Struc-ture’,Energy Economics, 5, 146-52.Howarth, R B, L Schipper, P A Duerr and S Strom (1991): ‘Manufacturing Energy Use in Eight OECD Countries’,Energy Economics, 13, 135-142.Hong, N (1983): ‘Two Measures of Aggregate Energy Production Elasticities’,The Energy, 4(2): 172-27.Huntington, G H (1989): ‘The Impact of Sectoral Shifts in Industry on US Energy Demand’,Energy, Vol 14 No 6, pp 363-72.Li, J W (1990): ‘Structure Change and Energy Use’, Energy Economics, 12, 109-15.Liu, Q X, B W Ang and H L Ong (1992): ‘The Appli-cation of Divisia Index to the Decomposition of Changes in Industrial Energy Consumption’,The Energy Journal, Vol 13, No 4, pp 161-77. – (1992): ‘Inter Fuel Substitution and Decomposi-tion of Changes in Industrial Energy Consump-tion’,Energy, Vol 17, No7, pp 689-96. – (1992): ‘The Application of the Divisia Index to the Decomposition of Changes in Industrial En-ergy Consumption’,Energy Journal, 13, 161-77.Mishra, S K and V K Puri (2003): Indian Economy, Himalaya Publishing House, New Delhi.Munasingha, M and M Peter (1993): Energy Policy Analysis and Modelling, Cambridge University Press. National Account Statistics, CSO, Back Series (1960-61 to 2003-01), EPW Research Foundation.National Account Statistics, CSO, Government of India.Nguyen, S V and S H Andrews (1989): ‘The Effect of Energy Aggregation on Energy Elasticities: Some Evidence from US Manufacturing Data’,The Energy, 10: 149-55. Park, S H (1992), ‘Decomposition of Industrial Energy Consumption: An Alternative Method’, Energy Economics, Vol 14, No 4, pp 265-70.Paul, Shyamal, R N Bhattacharya (2004): ‘CO2 Emis-sion from Energy Use in India: A Decomposition Analysis,The Energy Policy, Vol 32, pp 585-93.Rangarajan, V (2005): ‘Determining T and D Losses in India: Their Impact on Distribution Privatisa-tion and Regulation’,Economic & Political Weekly, February 12.Reitler, W, M Rudolph and H Schaefer (1987): ‘Analy-sis of Factors Influencing Energy Consumption in Industry: A Revised Method’, Energy Economics, Vol 9, No 3, pp 145-48.Reitler, W, M Rudolph and M Schaefer (1987): ‘Analy-sis of the Factors Influencing Energy Consump-tion in Industry: A Revised Method,Energy Eco-nomics, 9, 145-48.Rejikumar, R (2005): ‘National Electricity Policy and Plan: A Critical Examination’,Economic & Politi-cal Weekly, May 14.Sengupta R P (1993): ‘Energy Modelling for India’, Planning Commission, Energy Division, India.TERI Energy Data & Directory Yearbook, various years.The Energy Research Institute (TERI), Regulatory, Is-sue 24, March 2005.The Energy Research Institute (TERI), Regulatory, Is-sue 25, March 2005.Torvanger, A (1991): ‘Manufacturing Sector Carbon Dioxide Emissions in Nine OECD Countries, 1973-87’,Energy Economics, 13, 168-86.Vogt, A (1978): ‘Divisia Indices on Different Paths’ in W Eichhorn, T Henn, O Opitz and R W Shephard (eds),Theory and Application of Economic Indices, Physicaverlag, Wurzburg, pp 297-305.Zarnikar, Jay (1999): ‘Defining Total Energy Use in Economic Studies: Does the Aggregation Ap-proach Matter?’,Energy Economics, 21, pp 485-4.Zarnikau J, S Guermouche and P Schmidt (1996): ‘Can Different Energy Resources be Added or Compared’,The Energy, 21(6) 483-91. 
