This article was downloaded by: [Moskow State Univ Bibliote]On: 20 February 2014, At: 12:06Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Economic Systems ResearchPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/cesr20

Input–Output Anatomy of China'sEnergy Use Changes in the 1980sXiannuan Lin a & Karen R. Polenske ba Center for Energy and Environmental Studies andDepartment of Geography , Boston University , 675Commonwealth Avenue, Boston, MA, 02215, USAb Department of Urban Studies and Planning , MassachusettsInstitute of Technology , Cambridge, MA, 02139, USAPublished online: 28 Jul 2006.

To cite this article: Xiannuan Lin & Karen R. Polenske (1995) Input–Output Anatomy ofChina's Energy Use Changes in the 1980s, Economic Systems Research, 7:1, 67-84, DOI:10.1080/09535319500000011

To link to this article: http://dx.doi.org/10.1080/09535319500000011

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoeveras to the accuracy, completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions and views of theauthors, and are not the views of or endorsed by Taylor & Francis. The accuracyof the Content should not be relied upon and should be independently verifiedwith primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages, and otherliabilities whatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms& Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Economic Systems Research, Vol. 7, No. 1, 1995

Input-Output Anatomy of China's Energy Use Changes in the 1980s

XIANNUAN LIN & KAREN R. POLENSKE

(Received June 1993; revised August 1994)

ABSTRACT China significantly reduced the energy intensity of its economy in the 1980s. In this paper, we conduct a structural decomposition analysis to explain China's energy use changes between 1981 and 1987-the years for which we have input-output tables. We find that China's energy saving during this period came about primarily by changes in how to produce (production technology changes) rather than changes in what to consume (final demand shi$s). The driving force of the energy intensity decline was energy efficiency improvements, which were multiplied across the entire economy through inter-industy input-output linkages.

KEYWORDS: China, energy, structural decomposition analysis

1. Introduction

Energy consumption typically grows faster than final economic output in developing countries. As Lin (1991, 1994) and Polenske and Lin (1993) have indicated, this increased energy intensity occurs as a result of at least six major changes associated with development: industrialization; increases in the capital-to- labor ratio; substitution of commercial energy for traditional energy; the construction of modern infrastructure; motorization; urbanization.' Levine et al. (1991) found that, between 1972 and 1988, energy consumption for the developing countries as a whole grew by about 20% more than did the gross domestic product (GDP). Schipper and Meyers (1992) similarly reported that, for most of the past

X. Lin, Center for Energy and Environmental Studies and Department of Geography, Boston University, 675 Commonwealth Avenue, Boston, MA 02215, USA. K. R. Polenske, Department of Urban Studies and Planning, Massachusetts Institute of Technology, Cambridge, MA 02139, USA. The research reported in this paper was funded under grant 9224003-SBR from the National Science Foundation. We thank Jan Oosterhaven and two anonymous referees for their helpful comments and suggestions on an earlier version of this paper. We take full responsibility for the views and conclusions expressed in this paper, which do not necessarily reflect the views of the National Science Foundation.

0953-5314/95/010067-18 01995 The International Input-Output Association

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

68 X. Lin & K. R. Polenske

20-30 years, commercial energy consumption increased more rapidly than the GDP in developing countries. Imran and Barnes (1990) showed that, despite sharp increases in energy prices, the average energy intensity of developing countries continued to rise in the late 1970s and 1980s.

However, China's economic development during the 1980s did not follow this pattern (Lin, 1991; Polenske & Lin, 1993). Between 1980 and 1990, China's real GDP (in 1980 constant prices) grew by 136.2%, from 447.0 billion Renminbi (RMB) to 10562 billion RMB, but primary energy consumption increased by only 63.7% from 602.8 million tonnes of standard coal equivalent (tsce) to 987.0 million tsce (Lin, 1994). The energy intensity, in grams of standard coal equivalent per RMB (gsce/RMB) of the real GDP, declined by 30.7%, from 1348.4 to 934.5 gscel RMB.

In this paper, we discuss a structural decomposition analysis (SDA) that we conducted to examine how this drop in China's energy intensity occurred between 1981 and 1987-the years for which we have input-output tables. We view the amount of energy required in the economy as the cumulative product of millions of decisions-the consumer deciding what to consume, the government making budget decisions, the investor evaluating where to invest, the trade representative negotiating tariff structures, the manager- choosing a production technology, etc. These decisions affect what to consume (final demand) and how to produce (production technology), which ultimately determine the nation's aggregate demand for energy. By performing an SDA, we determine how much of the energy use changes in China's economy from 1981 to 1987 can be attributed to final demand shifts and/or to production technology changes.

2. Structural Decomposition Analysis

SDA involves analysis of economic changes by means of a set of comparative static adjustments of key parameters of input-output tables (Rose & Miernyk, 1989). It has been widely used in energy studies. For example, Strout (1966) analyzed how changes in technology and in the level and composition of final demand affected US energy use between 1939 and 1954. Reardon (1976) conducted an input-output analysis of US energy use changes for 1947-1958, 1958-1963 and 1963-1967. With an input-output framework, Park (1982) measured the direct, indirect and income- induced energy effects of a change in final demand, and estimated the effect of technological change on energy consumption.

Ostblom (1982) attributed the changes in the energy output ratio of the Swedish economy to changes in direct energy coefficients, changes in the output share of industrial sectors and changes in the composition of final demand. For the US in 1963 and 1980, Hannon (1983) compared the energy costs of providing goods and services. Proops (1984) decomposed changes in the energy output ratio into three factors: changes in energy intensities; changes in final demand; changes in the structure of inter-industry trading. For Danish manufacturing industries, Ploger (1985) assessed the effects of changes in the output mix and energy coefficients on energy consumption. Casler and Hannon (1989) examined the readjustment potential in the industrial energy efficiency and structure in the US. The Office of Technology Assessment (OTA, 1990) staff performed an SDA on US energy use changes between 1972 and 1988.

Recently, Rose and Chen (1991) made a major contribution to advance the state of the art of SDA. They extended the analysis to a two-tier KLEM (capital, labor,

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 69

energy and materials) flexible production function framework, which produced 11 separate sources of energy use changes and three 'interactive' effects. They also formally derived a system of estimation equations that are mutually exclusive and completely exhaustive. They applied their model to study energy demand changes in the US for the period 1972-82 (Rose & Chen, 1991) and those in Taiwan between 1971 and 1984 (Chen & Rose, 1990). They showed that, overall, the model can yield as much insight as more elaborate and data-intensive KLEM econometric models of production technologies.

In this paper, we develop an alternative formulation of SDA that is simpler than the Rose-Chen model. Even so, we generate a system of mutually exclusive and completely exhaustive estimation equations. Our formulation differs from the Rose-Chen model in two important aspects. First, we do not formulate the SDA with the KLEM flexible production framework. Instead, we decompose final demand changes, based on changes in the level and mix of spending on different product groups by different final users. We partition technological changes into energy and non-energy input changes in different sectors. Secondly, Rose and Chen incorporated energy into a standard monetary input-output table, by using output-to-fuel coefficients to convert energy output values from the standard input-output table into physical quantities. In contrast, we apply a 'hybrid' method adapted from the work of Bullard and Herendeen (1975) and Bullard et al. (1978), and replace all the energy rows in the monetary input-output table with physical energy flows. Miller and Blair (1985) demonstrated that the 'hybrid' formulation is generally superior to the output conversion approach, because it always conforms with energy conservation conditions.

2.1. Model Structure

We start our model forinulation from a standard static monetary input-output model. Mathematically, the structure of the input-output model is simple and can be expressed as

A X + Y = X (1)

where X is the vector of gross output, Y is the vector of final demand and A is the matrix of direct input coefficients, each of which shows the input required to produce one unit of gross output.

The solution of equation (1) gives

x = (I-A)-'Y (2)

where (I -A)-' is the matrix of total input requirements. For an energy input-output model, the monetary flows in the energy rows in

equation (2) are replaced with the physical flows of energy, to construct the energy flows accounting identity, which conforms with the energy balance condition (Miller & Blair, 1985)

where E, is the vector of intermediate energy consumption, Ed is the vector of final energy consumption and E is the vector of total domestic energy consumption.

The intermediate energy consumption is the energy used by production sectors as an input, i.e. the energy used in production activities. The direct energy consumption is the energy used or sold directly to final users, such as households

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

70 X. Lin & K. R. Polenske

and government agencies. It is not-in the time period under consideration-used as inputs by business firms to produce output.

Using equation (I), we can rewrite the energy balance condition of equation (3) as

where e is a diagonal matrix composed of ones and zeros. The ones appear in the column locations that correspond to energy sectors and all the other elements of the matrix are zeros. The matrix selects the energy rows from the input-output table.

We can obtain information about the amount of intermediate energy required in the economy, by combining and rearranging equations (2) and (4) to give

Then, to calculate the direct energy consumption Ed, we need to adjust the final energy consumption for energy exports, imports and inventory changes. Mathematically, we have

where E, is the vector of final energy consumption, E, is the vector of imported energy, E, is the vector of exported energy, E, is the vector of energy net inventory change and n is a matrix consisting of ones and zeros, with ones in the diagonal locations that correspond to those columns that are not imports, exports and net inventory changes, and zeros in all other elements of the matrix, i.e. it excludes energy imports, exports and net inventory changes from the calculation of direct energy consumption.

The total energy consumption E in the economy is the sum of the intermediate and direct energy consumption, i.e.

where F = e[(I-A)-'-I]. Equation (7) shows that the total energy consumption in an economy is determined by the total intermediate energy requirements F and direct energy requirements by final demand (Y). F in turn is a function of production technology, measured in terms of the technical coefficients matrix A, which includes both energy and non-energy inputs.

Therefore, energy use in the economy can change because of changes in final demand and/or because of changes in production technology. We apply equation (7) to describe the changes in China's energy consumption from 1981 to 1987 as

Here, (F87Y87-F81Yxl) represents changes in the intermediate energy use, which depend both on changes in production technology F and changes in final demand Y. The term e(Yx7-Ysl)n measures changes in the direct energy consumption, which is solely a function of final demand shifts.

To determine how much of the energy use change results from changes in what to consume (final demand shifts) and from changes in how to produce (production technology changes), we introduce a hypothetical economy with 1981 production technology (F,,) and 1987 final demands (Y,,). The energy consumption in this

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

hypothetical economy would be

E F ~ ~ Y S ~ = F8iYs7 + eYs7n

China's Energy Use Changes 71

where measures the amount of energy that would be consumed in China's economy if the 1981 production technology were used to deliver 1987 final demands. Using EF81y87 as a reference point, we can rewrite the energy use changes from 1981 to 1987 as

AE = E87 + E F ~ I Y ~ ~ - E F ~ I Y ~ ~ - Esl

=Fs l (Y~7-Y~l )+e(Ys7-Y81)n (finaldemandshift) (10)

+ (F87-F81)Y87 (production technology change)

The final demand shift indicates the energy impact of final demand changes while holding the production technology constant. The production technology change quantifies the energy effect of changes in the production technology, with a given final demand.

However, EF81Y87 is not the only reference point that can be used to separate energy use changes into the final demand shift and production technology change components. An alternative formulation is to use &87yX1 as a reference point, in which case equation (10) becomes

AE = Es7 + E F ~ ~ Y x ~ -EFWYS~ -Esi

= FS7(Y87 - Ysl) + e(Y8, - Ysl)n (final demand shift) (11)

+ (F87 -Fsi)Ysl (production technology change)

It is obvious that, except by pure chance or under some strict mathematical conditions, equations (10) and (11) will attribute energy use changes to final demand shifts and production technology changes differently. This ambiguity stems from the problem of indexing, i.e. weights from one year usually do not give the same answer as weights from another year, and there is no single 'correct' answer (Strout, 1966; Carter, 1970).

In fact, the two equations are designed to answer different questions, as pointed out by Strout (1966). In equation (lo), we ask 'how much more (or less) energy would have been required in 1987 if the 1981 production technology had still been used to satisfy 1987 final demands?' In equation ( l l ) , we try to find out 'how much less (or more) energy would be used in 1981 if the 1987 production technology had been available to deliver 1981 final demands'. Because, in this study, we are mainly interested in the energy impact of using different technologies to deliver 1987 and not 1981 final demands, we choose EF81Y87 rather than EF87Y81 as the reference point.2

2.1.1. Components of final demand shifts. We further decompose the final demand shift component along three dimensions. First, we identify the energy use changes associated with changes in the level, distribution and pattern of final demand. The level of final demand refers to the overall level of total demand (i.e. the GDP), which equals the sum of all final outputs or expenditures. The distribution of demand refers to the distribution of the total demand along individual final demand sectors, such as personal consumption, government expenditures, capital investment, exports and imports. The pattern of demand refers to the mix of goods and services within each individual final demand category.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

72 X. Lin & K. R. Polenske

In matrix notation, final demand is the product of its level, distribution and pattern components, i.e.

Y = MDL (12)

where M is the matrix of the spending mix of individual final demand categories, D is a diagonal matrix with the categorical composition of total demand on the diagonal and L is a diagonal matrix with the overall total demand level on the diagonal.

We use equation (12) to quantify the energy effects of final demand level, distribution and pattern changes. Mathematically: we have

AEY = F81(Y87-Y81)+e(Y87-Y81)n

= Fsl [Ms7Ds7L87 - M ~ I D ~ ~ L ~ I ] + e[Ms7D87L87 - M s ~ D s L L ~ ~ ~ ~

= F81M81Dsl(L87 -L81) + eM81D81(L87 - L ~ l ) n (level effect) (13)

+FglMsl(Ds7 -D81)L87 + eMsl(DX7 -Dsl)Lg7n (distribution effect)

+ Fsl(M87 - M81)D87L87 + e(Mg7 - M81)D87L87n (pattern effect)

Secondly, we calculate the amounts of energy use changes that originate in individual final demand sectors, such as personal consumption, investment, exports and imports. Mathematically, this is very simple, because final demand in the input-output system is additive, giving

where A E ~ , is the change in energy use resulting from changes in final demand sector h.

Thirdly, we determine how changes in the purchase of an individual product or product group affect the energy consumption, using

AE, , = ~ ~ ~ ~ ( 9 8 ~ -981) + e ~ ( 9 8 7 -981)n (1%

where AE,,, is the matrix of energy use changes associated with each product k by fuel type, Ps7 and 98, are diagonal matrices of the total final demand vectors in 1987 and 1981, and K is a matrix consisting of ones and zeros, with ones in the row locations that correspond to products (k) and zeros in all other elements of the matrix.

From equation (15), we also estimate how much of the energy use changes resulting from final demand shifts comes directly from purchases of energy products and how much comes indirectly from purchases of non-energy products.

2.1.2. Components of production technology changes. The production technology change component in equation (10) measures the energy use change associated with changes in the total intermediate energy requirements of final goods and services. It can be rewritten as

= e(G87-G8l)Y87

where G87 = (1-As7)l and G81 = (I -A8l)-l.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 73

It follows sequentially that

Inserting equation (17) into equation (16) results in

We split the production technology into two portions. The energy portion represents the direct use of energy inputs, such as coal, oil and electricity, by sector. It measures the energy requirement per unit of output. The non-energy portion contains all the other inputs by production sector, such as plastics, steel and chemical fertilizers.

We use equation (18) to separate the effect of changes in direct energy requirements and direct non-energy requirements of energy use, by partitioning and writing the changes in the technical coefficients (Ax7 -Asl) as

where AE represents the energy rows of the technical coefficient matrix and AN represents the non-energy rows. Equation (18) then becomes

= eGx7(As7,~ - Asi,~)GslYs7 (changes in energy inputs) (20)

+ eGs7(Ax7,~ - A 8 1 , ~ ) ~ 8 1 ~ 8 7 (changes in non-energy inputs)

This tells us that the change in intermediate energy demand can be caused not only by changes in direct energy inputs (AE) but also by changes in direct non- energy or material inputs (AN). Furthermore, the changes in direct input requirements will be multiplied across the economy, through inter-industry input- output linkages, which are quantified by the total input requirements matrix, G.

In addition, we carry out a more detailed analysis of the energy and non-energy inputs, by examining production technology changes in individual sectors or sector groups-in this case, agriculture, energy, non-energy industrial sector, construc- tion, transportation and commerce. We assess their relative contributions to intermediate energy demand changes. Mathematically, we have

AET = C BE', i

= C [eG87(~i87,E-Aj81,E)G81Ys7 (changes in energy inputs) i

(21)

+ eGs7(Ai7,, -A$l,~)~XlY871 (changes in non-energy inputs)

where AE$ is the change in energy use resulting from production technology changes in sector j.

We summarize the hierarchical structure of the estimation equations we use in our overall analysis in Table 1.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

74 X. Lin G.' K. R. Polenske

Table 1. Structural decomposition of energy use changes

Factor Equation

Final demand shift F81(Y~7-Y81) + e(Ys7-Ys1)n Level effect F ~ I M ~ I D ~ ~ ( L s ~ - L ~ ~ ) + eM81D81(L~7 -L81)n Distribution effect FSIMB~(DS~-DSI)L~~+~MS~(D~~-D~I)L~~~ Pattern effect F ~ I ( M ~ ~ - M S I ) D ~ ~ L S ~ + e(M87-M81)D87L87n

For demand source h F s I ( ~ ~ ~ ~ - Y ~ I ) + ~ ( Y ! ~ - Y ~ ~ ) ~

For product group k F S I K ( P S ~ - Y ~ I ) + ~ K ( P s ~ - P s ~ ) ~

Production technology change Energy inputs Non-energy inputs

For individual sector j Energy inputs Non-energy inputs

Actual energy use changes E87-E81 = (Fa7Ya7 -F81Y81)-e(Ysi.-Ysl)n

2.2. Model Implementation

To implement the model and conduct the SDA of energy demand changes, we require three key data components for both 1981 and 1987: input-output tables, price indexes and energy flow data.

2.2.1. Input-output tables. The input-output tables used in our SDA modelling are commodity-by-commodity tables for China in 1981 and 1987, compiled using the system of national accounts (SNA) conventions. The tables were constructed by the Institute of Systems Science (ISS), Chinese Academy of Sciences, based on the Input-Output Table of China, 1981 and The Input-Output Table of China, 1987 (SSB, 1985, 1990). We group final demand sources into eight sectors:

(1) rural personal consumption; (2) urban personal consumption; (3) social consumption; (4) capital investment; (5) inventory changes; (6) exports; (7) imports; (8) other.

We also subdivide the production activities of China's economy into 18 industrial groups and present energy-intensive sectors at a more disaggregated level than that of the rest of the economy, as shown in Table 2.

Ideally, we would like to conduct the SDA at a finer industrial classification to increase the degree of homogeneity within each industrial group and to have considerable detail about changes in the production technology and spending mix. However, the energy consumption data compatible with such an industrial classification are not available and are not of a reasonable quality. Although limited, the basic 18-sector classification is revealing and the presentation of energy-intensive sectors at a rather disaggregated level enables us to capture most major energy-related production technology changes and final demand shifts.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 75

Table 2. Production sectors in the SDA model

Code Sector

Agriculture Coal Petroleum Natural gas Electricity Iron and steel Non-ferrous metals Chemical fertilizers Heavy chemicals Cement Construction materials Heavy machinery Light industry Construction

15 Freight transport and telecommunication 16 Commerce 17 Passenger transport 18 Services

We must stress, however, that the results of the SDA are not invariant with respect to the industrial classification. A change in industrial disaggregation will not change the total levels of intermediate output and final demand, but it will affect the production technology and spending mix. Therefore, the size of the final demand level effects will remain about the same. In contrast, the final demand mix effects and technology change components will become greater or smaller, depending on whether the products or inputs that are growing are grouped together to reinforce each other's growth or are grouped with products or inputs that are declining, so that the changes tend to cancel each other. This also depends on whether greater detail is devoted to static or to changing elements (Carter, 1970). A priori, there is no systematic relationship between the fineness of the industrial classification and the relative size of the final demand mix and production technology components.

2.2.2. Price indices. The analysis of change in energy use patterns over time requires that each year's input-output tables be based on the same set of prices, because a million RMB's worth of output in 1981 had a much different value or physical unit from a million RMB's worth of output in 1987.4 We use 1981 as the base year (so no price changes were necessary for the 1981 table) and adjust the 1987 table to 1981 'prices, using the price indices from the ISS. The ISS staff estimated the price indices for each of the 18 sectors, based on various real output numbers and price indices reported by statistical agencies, many of which were unaccessible to researchers at large. We use the indices to convert the output in the 1987 table into 1981 constant prices for all industries within their respective sectors.

2.2.3. Energy f i w data. The primary source for our energy flow data was the energy flow matrices developed by the ISS staff for their 1981 and 1987 input- output tables. The matrices show the flows of nine different energy products that

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

76 X. Lin & K. R. Polenske

are consumed by all 18 production sectors and eight categories of final demand in our input-output model. The nine energy products are raw coal, coke, crude oil, fuel oil, gasoline, kerosene, diesel, natural gas and electricity, which accounted for all the primary energy consumption and about 95% of final energy use in the 1980s. We convert all the energy products into standard coal, based on their average net calorific values or lower heating values, and aggregate them into four categories, i.e. coal, petroleum, natural gas and electricity, to be consistent with the 18-sector industrial classification of the input-output tables. We then follow the approach used by analysts at the University of Illinois (Bullard & Herendeen, 1975; Bullard et al., 1978) and the US Office of Technology Assessment (OTA, 1990), and allocate energy consumption to the production and final demand sectors that completely used up energy in 1981 or 1987.

The electricity sector poses a special difficulty for our SDA modelling. With the 18-sector input-output tables, we are unable to separate primary electricity (almost all of which comes from hydropower) from secondary electricity generated in thermal power stations. This is problematic because, to avoid double counting both primary energy (such as coal) used to generate secondary energy (such as electricity) and the consumption of secondary energy, we should include only primary electricity in calculating the total energy consumption for the whole economy. We solve this problem by introducing a hypothetical hydropower sector into the input-output model. The hydropower sector sells all its output to the electricity sector for power generation and obtains all its inputs from the earth. We estimate the standard coal equivalent of hydropower based on the amount of fossil fuel that would be required to generate the equivalent amount of hydroelectricity.

This way of incorporating hydropower enables us to trace both primary energy requirements and electricity end-uses.

2.3. Strengths and Limitations of the Model

SDA has at least four major strengths for use in energy analyses. First, it integrates energy data with an input-output account and provides a unified framework for describing the relationships between energy, other factor inputs and other final products; consequently, this gives a framework for the relationship between energy and the economy. Secondly, the framework includes all business sectors and covers the entire energy production and consumption cycle. Thirdly, unlike the traditional static input-output model, which assumes fixed technical coefficients, the SDA model allows for input substitution and technological changes. Finally, the model describes the economy as a system of interdependent activities, and enables analysts to trace inter-industry linkages and account for both direct and indirect energy uses; for example, energy consumption in a radio-assembly line would be a direct use of energy, while the use of plastics in making a radio is an indirect use of energy, because energy is required to produce plastics.

We realize that our SDA also has some important limitations. First, we assume a linear production function or constant returns to scale, so failing to account for energy intensity changes caused by changes in the scale of production. Secondly, we disaggregate intermediate inputs to only 18 sectors, because of a lack of data. This level of disaggregation may be too high to capture some important changes in specific industries and specific production processes. Thirdly, SDA is fundamentally a top-down macro-economic model and provides no information on how energy technologies and energy use practices change at the micro-level.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 77

Therefore, analysts must complement SDA with other types of analysis, such as econometric analyses, case studies and institutional analyses. There are many methods and models available for examining energy use changes, each with strengths and weaknesses. Thus, the question is not which method or model is the best one, but which one sheds the most light on which problem. We believe that one key to a good energy policy analysis is to match approaches to the task at hand and to mix different methods and models as needed.

3. Major Results from SDA Modelling

With these limitations in mind, we present in Tables 3 and 4 the main results of our SDA modelling. The tables display, in terms of amount (Table 3) and rate of

Table 3. SDA of primary energy use changes in China from 1981 to 1987 (million

Source Coal Petroleum Natural gas Hydropower Total

Actual change 219.0 29.5 1.5 13.0 263.1

Final demand shift 347.5 101.9 15.0 23.3 487.7 Level effect 376.1 101.6 14.3 22.5 514.5 Distribution effect -27.0 - 2.5 -3.3 - 2.7 -35.5 Pattern effect - 1.5 2.7 4.0 3.6 8.8

Demand source Rural consumption Urban consumption Social consumption Capital investment Inventory change Exports Imports Other

Product group Energy 40.5 4.4 1.6 2.0 48.4 Agriculture 42.0 15.7 3.9 3.8 65.3 Heavy industry 4.7 -0.8 - 0.8 0.8 3.9 Light industry 110.6 37.8 5.4 8.8 162.6 Construction 119.3 28.6 3.9 6.1 158.0 Transportation 9.8 6.9 0.3 0.4 17.4 Services 20.6 9.3 0.8 1.5 32.1

Technology change - 128.5 - 72.4 - 13.5 - 10.3 - 224.7 Energy inputs - 167.1 - 83.6 - 14.2 -13.2 - 278.1 Non-energy inputs 38.6 11.2 0.7 2.9 53.5

Sector Agriculture 5.6 -5.0 0.2 - 0.4 0.4 Energy 11.6 - 14.4 - 1.9 - 3.5 - 8.2 Heavy industry - 115.5 - 37.3 - 9.7 - 7.0 - 169.5 Light industry -52.7 - 14.8 - 2.4 - 2.0 -71.9 Construction 28.0 3.9 0.0 1.5 33.4 Transport -11.3 - 3.1 -0.1 0.3 - 14.1 Service 5.9 - 1.7 0.3 0.7 5.2

"Numbers may not add to totals or subtotals, as a result of rounding.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

78 X. Lin & K. R. Polenske

Table 4. SDA on the growth rate of primary energy consumption in China's economy from 1981 to 1987 (per cent of 1981 total energy consumptiona)

Demand source Coal Petroleum Natural gas Hydropower Total

9.0 48.8 43.1 Actual change

Final demand shift Level effect Distribution effect Pattern effect

Demand source Rural consumption Urban consumption Social consumption Capital investment Inventory change Exports Imports Other

Product group Energy Agriculture Heavy industry Light industry Construction Transportation Services

Technology change Energy inputs Non-energy inputs

Sector Agriculture Energy Heavy industry Light industry Construction Transport Service

"Each percentage figure is calculated by dividing the per cent of energy use changes from 1981 to 1987 by the 1981 total coal, petroleum, natural gas, hydropower and all primary energy consumption.

change (Table 4), by fuel type, the sources of the energy use changes in China's economy.

Between 1981 and 1987, China's total primary energy consumption increased by 263 million tsce or 43%. Final demand shifts-the increases in the level of economic activities and shifts in spending mix towards more energy-intensive products-were the main factors that pushed the energy use upward. All else being equal, these shifts would increase the energy consumption by 487 million tsce or 80%. This upward pressure on energy demand, however, was dampened by changes in production technology, which reduced the energy requirements per unit of goods and services. Holding all other factors constant, production technology changes would decrease the energy use by 224 million tsce or 37%.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 79

3.1. Energy Effects of Final Demand Shifts

We can view the energy impact of final demand shifts from three different dimensions:

(1) the final demand level, distribution (across final users) and pattern (i.e. spending pattern of final users);

(2) the sources of final demand, such as personal consumption, social consumption, capital investment and international trade;

(3) the types of final good and service purchased.

The three dimensions intercept one another and are different aspects of the same final demand shifts from 1981 to 1987. However, each provides a unique insight into the relationship between final demand and energy consumption in China's economy.

3.1.1. Level, distribution and pattern of final demand. Almost all the energy use increases resulting from final demand shifts came from the increase in the overall level of spending, which (other things being equal) would cause China's total energy consumption to increase by 515 million tsce (Table 3) or 84% (Table 4). Changes in the spending mix of the individual demand sectors led to an additional 9 million tsce or 1% growth in energy use, while changes in final demand distribution cut the energy consumption growth between 1981 and 1987 by 36 million tsce, i.e. just under 6%. China has over one-fifth of the world's population and one of the largest and fastest growing economies in the world. The sheer size of population and economic growth are two of the most important factors driving China's energy consumption growth.

3.1.2. Sources of final demand. A combination of the expansion in capital investment, the increase in personal consumption and the rise in exports was the main force behind the energy use increase associated with final demand shifts. Holding all other factors constant, these three elements combined would result in an increase of 665 million tsce in China's energy consumption. This strong upward pressure on energy use was dampened by the rapid growth in imports, which saved China 233 million tsce of energy. In general, changes in spending from individual demand sectors had similar impacts on different types of energy. As we can see from Table 4, with a few exceptions, the percentage changes associated with each factor were in the same direction and had a similar magnitude across fuel types.

3.1.3. Different product groups. We can attribute only 48 of the 488 million tsce (10%) of the demand-related energy use increases to the purchases of energy by final customers, with the other 440 million tsce (90%) being related to the purchases of non-energy products (Table 3). The expenditures on light industry products and construction projects were the most important factor that augmented China's energy consumption from 1981 to 1987. Again, the change in the demand for each product group had a similar impact on different types of energy, as shown in Table 4. A higher spending on construction projects, for example, would lead to a 27% increase in coal, 24% increase in oil and 23% increases in both natural gas and hydropower. A major exception was heavy industry, which caused an increase in the consumption of coal and hydropower but a decrease in the use of petroleum and natural gas.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

80 X. Lin & K. R. Polenske

3.2. Energy Effects of Production Technology Changes

Production technology changes act to offset the increased energy use associated with final demand shifts. Compared with the energy requirements of using 1981 production technology to satisfy the 1987 final demand, the adoption of 1987 production technology saved about 224.7 million tsce of primary energy (Table 3), which was 36.8% of the 1981 total primary energy consumption (Table 4). All these energy savings came from the improvements in energy efficiency-reductions in direct energy input coefficients-which reduced primary energy use by 278.1 million tsce or 45.6%. However, the changes in the non-energy portion of production technology increased China's energy consumption and reduced energy savings from the efficiency improvement by 53.5 million tsce or 8.8%.

The changes in non-energy inputs have energy impacts because, in terms of value, the direct use of energy inputs in production represents less than 10% of all inputs in China's economy. A large percentage of the energy requirements of providing final products comes indirectly from the remaining 90% of the inputs, which require a significant amount of energy to produce. In the construction sector, for example, the direct energy use in the sector was only 85 gsce per RMB of output in 1981 but 1724 gsce of energy were used indirectly, because many construction materials, such as steel, glass and cement, embody a large quantity of energy (Lin, 1994). Therefore, changes in non-energy inputs can have an important energy consequence.

Of the total 224.7 million tsce of energy savings from 1981 to 1987 that resulted from production technology changes, 128.5 million tsce (57.2%) was in the form of coal, 72.4 million tsce (32.2%) was petroleum, 13.5 million tsce (6.0%) was natural gas and 10.3 million tsce (4.6%) was in the form of hydropower. The contributions of petroleum and natural gas to the savings were disproportionally large, given that they accounted for only about 19.8% and 2.8%, respectively, of the total primary energy consumption in 1981. As shown in Table 4, the energy savings rate was 60.0% for petroleum and 79.5% for natural gas, compared with 28.8% for coal and 38.6% for hydropower.

In terms of sectoral contributions, the production technology changes in the heavy and light industries were the most important sources of energy savings and, all other things being equal, they reduce China's energy consumption by 169.5 million tsce and 71.9 million tsce respectively (Table 3). The rest of the energy savings came from technological changes in the transportation (14.1 million tsce) and energy sectors (8.2 million tsce). On balance, changes in the energy and non- energy portions of production technology in the construction, energy and service sectors increased rather than decreased the energy consumption of China's economy.

It should be noted that the sectoral distribution indicates energy use changes that originate from energy production technology changes in each sector. It includes not only energy use changes within the sector but also those in other sectors that supply energy inputs to the sector and/or use the sector's output as an input. Through inter-industry input-output linkages, the energy efficiency improvements of any one sector will be multiplied across the entire economy, by reducing the indirect energy requirements of those sectors which use the industry's product as inputs. For example, a large percentage of the total energy requirements of the agricultural sector came from the use of chemical fertilizers. Thus, the higher energy efficiency in the chemicals industry reduces not only its own energy

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 81

intensity but also that of the agricultural sector. This repercussion is especially important for energy-intensive, basic materials industries, such as chemicals, metallurgy and building materials.

Many researchers who have studied China's energy intensity changes in the 1980s distinguish between energy savings from structural changes or sectoral shifts and those resulting from energy efficiency improvements. We can rearrange and add different components of the SDA to conform with this convention. The structural change factor equals the sum of the energy use changes associated with non-energy input changes and those associated with final demand distribution and pattern effects. The energy efficiency factor is equivalent to changes in energy inputs. When measured this way, we find that all the energy savings in China in 1987, relative to 1981, can be attributed to energy efficiency improvements. The structural changes from 1981 to 1987 would result in a slight increase rather than a decrease in China's energy intensity.

3.3. Policy Implications

The results from our calculations have at least two important policy implications- one related to final demand changes and one related to technology changes. Overall, increases in the GDP from 1981 to 1987 contributed to an 80% increase in primary energy consumption. Thus, one way in which Chinese policy-makers can avoid rapid increases in energy consumption is to adopt policies that promote (retard) growth of less (more) energy-intensive components of final demand. For example, they can design policies to encourage the importation of energy-intensive goods, such as steel, machinery and building materials, or they can determine types of policy to continue the change in the patterns of final expenditures to patterns that are less energy intensive.

Adopting policies that will encourage industries to change to more energy- efficient technologies is another way for policy-makers to affect energy consumption. One of the most feasible policy options appears to be to introduce new technology at a more rapid rate than in the past. More than half of the reduction in energy intensity caused by production technology changes between 1981-1987 in China occurred as a result of changes in technology in the heavy industry sector. From our field trips, we believe that there are still significant potential energy savings to be realized in China, by the purchase of new equipment and the building of new plants. Before determining whether or not this is so, we would need to review data on the vintage of the current equipment and plants. Two of the main problems with stressing this policy option are the constant shortage of capital and the current bias in China towards importing the new equipment technology. Although imports help relieve pressures on domestic capital-producing industries, such as electric generators, where demand far exceeds supply, they require scarce foreign exchange and, at the same time, in certain cases, they hinder the domestic industry from developing.

4. Conclusion

Our SDA of energy use changes in China's economy shows that China's energy savings from 1981 to 1987 were caused primarily by changes in how to produce (production technology changes) rather than by changes in what to consume (final demand shifts). The driving force of energy intensity decline was energy efficiency

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

82 X. Lin i3 K. R. Polenske

improvements-the reductions in direct energy input coefficients in most production sectors-which were multiplied across the entire economy, through inter-industry input-output linkages.

It is beyond the scope of this paper to examine detailed reasons for China's energy efficiency improvements.' However, we can identify three macro-economic factors that appear to be primarily responsible for the energy efficiency increases in China's economy between 1981 and 1987: (1) energy conservation programmes; (2) improvements in macro-economnic performance; (3) increases in energy prices.

First, the improvements in energy efficiency were a result of China's energy conservation programmes. There has been a major shift in energy policy since 1979, from conventional complete devotion to increasing supply to a policy of placing equal emphasis on supply expansion and energy conservation, with priority given to conservation in the short term. The government adopted a large number of administrative, financial and economic measures in the 1980s to reduce energy waste and promote energy efficiency in the industrial sectors, especially energy- intensive heavy industries. Overall, these measures were highly successful and resulted in large energy savings, partly because China had been using energy very inefficiently until the 1970s. Therefore, there was great potential for energy saving, much of which could be realized without a major capital investment and simply by restructuring energy use or reducing apparent waste.

Secondly, the improvements in energy efficiency were also a by-product of China's rapid economic growth and part of the overall trend towards higher productivity in the 1980s. This trend was a result of China's economic reform programme, which reduced central planning and increased the role of the market mechanism and incentive structures in the economy. China's rapid economic growth in the 1980s was accompanied by an expansion in production capacities and the addition of new equipment and facilities, which usually embodied better technology and had higher energy efficiencies than old existing capital equipment. More importantly, many measures that were aimed at increasing productivity and profitability also helped to save energy. For example, worker training to improve equipment operation enhances both productivity and energy efficiency. The introduction of a modern, large-scale blast furnace results in an increase in production capacity, as well as a decrease in the coke rate. When business firms shifted from making low value-added products to making high value-added products, they not only increase profits but also usually reduce the energy per unit of output.

Finally, the improvements in energy efficiency were sometimes an enterprise's rational response to energy price increases. The Chinese government raised planned energy prices substantially in the 1980s. The government also introduced a dual-price system into the energy sector, which allowed energy products to be exchanged at two different prices: a state-set price, for the amount produced under central planning, and a higher free-market price, for above-plan output. These energy price increases provided some incentives for enterprises to reduce energy waste and improve the efficiency of energy utilization. However, these incentives were limited, because, despite the energy price increases, energy expenditures comprised a very small percentage of the total production cost in most sectors in the 1980s and were not that important in the overall scheme of production.

These three macro-economic factors contributed greatly to the energy efficiency improvements in China in the 1980s. When the 1992 input-output table becomes available later in 1995, we will be able to determine whether or not the increased

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

China's Energy Use Changes 83

energy efficiency trend is being maintained in the 1990s. We are also conducting individual case studies of key energy-intensive industries, such as iron and steel and electric power, to determine the types of change occurring at the micro- economic level.

Notes

1. Throughout this paper, we use the term 'energy' to refer to commercial energy, thus excluding traditional (mostly biomass) fuels. If biomass is included, we will note this explicitly.

2. It is possible, of course, to combine the two formulations and use their average as a measure of the energy impact of final demand and production technology changes. This means using [(FS7+Fsl)/ 2][(Ys7+Y8,)/2] as the reference point, which is often confusing and difficult to interpret.

3. Similar to equation (lo), the choice of the reference point and the order in which the demand level, distribution and pattern are varied are somewhat arbitrary.

4. The constant-price model does have one important drawback, however. It means that the price-an important factor in energy use-will be held constant, which does not allow analysts to examine the impacts of price changes on energy use, indicating the need to complement the SDA with other types of analysis, such as econometric analyses or studies of energy pricing policy changes.

5. See Lin (1994) for an in-depth examination of factors behind the improvements in China's energy efficiency in the 1980s.

References

Bullard, C. W. & Herendeen, R. A. (1975) The energy cost of goods and services, Energy Policy, 3, pp. 268-278.

Bullard, C. W. et al. (1978) Net energy analysis handbook for combining process and input-output analysis, Resources and Energy, 1, pp. 267-313.

Carter, A. P. (1970) Structural Change in the American Economy (Cambridge, MA, Harvard University Press).

Casler, S. & Hannon, B. (1989) Readjustment potentials in industrial energy efficiency and structure, Journal of Environmental Economics and Management, 17, pp. 93-108.

Chen, C. & Rose, A. (1990) A structural-decomposition analysis of changes in energy demand in Taiwan: 1971-1984, The Energy Journal, 11, pp. 127-146.

Hannon, B. (1983) Analysis of the energy cost of economic activities: 1963 to 1980, ERG Document 316 (Urbana, IL, Energy Research Group, University of Illinois at Urbana-Champaign).

Imran, M. & Barnes, P. (1990) Energy demand in the developing countries, World Bank Staff Commodity Working Paper Number 23 (Washington, DC, The World Bank).

Levine, M. D. et al. (1991) Energy Efficiency, Developing Nations, and Eastern Europe. A Report to the US Working Group on Global Energy Efficiency.

Lin, X. (1991) Declining energy intensity in China's industrial sector, The Journal of Energy and Development, 16, pp. 195-218.

Lin, X. (1994) China's energy-use changes from 1981 to 1987: a structural-decomposition analysis, PhD Dissertation (Cambridge, MA, Massachusetts Institute of Technology).

Miller, R. E. & Blair, P. D. (1985) Input-Output Analysis: Foundations and Extensions (Englewood Cliffs, NJ, Prentice-Hall).

Office of Technology Assessment (1990) Energy use and the U.S. economy, Report OTA-BP-E-57 (Washington, DC, US Government Printing Office).

Ostblom, G. (1982) Energy use and structural changes, Energy Policy, 19, pp. 21-28. Park, S. (1982) An input-output framework for analyzing energy consumption, Energy Economics, 4, pp.

105-110. Ploger, E. (1985) The effects of structural changes on Danish energy consumption, in: A. Smyshlyaev

(ed.), Input-Output Modeling (New York, Springer), pp. 211-220. Polenske, K. R. & Lin, X. (1993) Conserving energy to reduce carbon-dioxide emissions in China,

Structural Change and Economic Dynamics, 4, pp. 249-265. Proops, J. L. R. (1984) Modeling the energy-output ratio, Energy Economics, 6, pp. 47-51. Reardon, W. A. (1976) An Input-Output Analysis of Energy Use Changes From 1947 to 1958, 1958 to 1963,

and 1963 to 1967 (Richland, WA, Battele Pacific Northwest Labs).

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014

84 X. Lin & K. R. Polenske

Rose, A. & Miernyk, W. H. (1989) Input-output analysis: the first fifty years, Economic Systems Research, 1, pp. 229-271.

Rose, A. & Chen, C. Y. (1991) Sources of change in energy use in the US economy, 1972-1982, Resources and Energy, 13, pp. 1-21.

Schipper, L. & Meyers, S. (1992) Energy Efficiency and Human Activity: Past Trends, Future Prospects (Cambridge, Cambridge University Press).

State Statistical Bureau of China (1985) China Input-Output Table, I981 (Beijing, China Statistical Press) (in Chinese). "

State Statistical Bureau of China (1990) China Energy Statistical Yearbook, 1989 (Beijing, China Energy Statistical Press) (in Chinese).

Strout, A. M. (1966) Technological change and United States energy consumption, 1939-1954, PhD Dissertation (University of Chicago).

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

12:

06 2

0 Fe

brua

ry 2

014