International Comparisons of Sectoral Carbon Dioxide Emissions Using the Mine/Yours Method

The Energy Journal (May 2001)

Lee Schipper*

International Energy Agency

Scott Murtishaw

LawrenceBerkeley National Laboratory

Fridtjof Unander

International Energy Agency

Discerning which sources contribute most to differences in per capita carbon emissions and why presents a daunting task for analysts, since several underlying factors affect emissions from hundreds of end-uses. This paper provides details of an international comparison methodology and carries out the comparison on a number of International Energy Agency (IEA) Member countries. These calculations show where differences in the components of emissions lead to large gaps among countries. The data, from national sources, are the most extensive and disaggregated ever compiled for this kind of international analysis. Overall, activity differences account for the largest part of the gap in per capita emissions among IEA countries. If we normalize emissions to GDP, then transport activity levels, energy intensities, and utility carbon intensity share about equally in explaining the differences in the carbon/GDP ratios among countries. Most of the structural variations arise in the freight, services, and household sectors, sectors less sensitive to international competition than manufacturing.

* 2002- at World Resources Institute, 10 G St NEWashingtonDC20002. Please address all correspondence to Lee Schipper, ,

1

1Introduction

Since the 1992 Earth Summit at Rio de Janeiro, attention has focused on reduction of emissions of carbon dioxide and other greenhouse gases (GHG). With the signing of the Kyoto Protocol and adoption of targets for GHG reduction, interest has grown to understand how each country’s economic and human activity is linked to emissions (Schipper 1997). A number of related papers[1] have analyzed carbon emissions from several IEA Member countries and decomposed them over the period from the early 1970s to the mid 1990s. These time series decompositions are helpful for understanding why some countries’ economies have decarbonized – reduced the ratio of carbon emissions to GDP – at a greater rate than others’. However, as important for reaching international agreements, negotiators need to understand why differences in emissions occur between countries at a given time and what the implications of these differences are for carbon reduction goals.

This paper charts out for the first time findings from a cross-sectional study of energy-related carbon emissions in 14 IEA Member countries (which as a group we refer to as the IEA-14) in 1994, the last year for which complete disaggregated data were available for the United States.[2] This effort marks the first analysis to disaggregate the energy uses to the level of twenty-seven subsectors or end-uses, so there is no real comparison with previous work. The analysis at this stage is descriptive, not causal, but we point to important links with causal factors that should be investigated. While more sophisticated analyses are planned for forthcoming publications, the present step is necessary for exploring the basic magnitudes of emissions differences. For comparison, we normalize results to population but note at times how normalizing by GDP might affect results. In the final analysis we show how much emissions in any country differ from the average of the other countries because of any single component or sector’s characteristics in that country. Thus the calculation is not intended to show “how much X would emit if it were like Y”, but rather, “what characteristic of X makes its emissions differ from those of Z, the average of the 13 other countries?”

A key theme of this analysis is that a disaggregated approach across all sectors is needed to demonstrate how and where emissions differ among countries. Exploring the underlying factors that differentiate CO2 emissions is important for several reasons. First, some differences may be irreducible and, consequently, some features of a country’s economy may always lead to high (or low) emissions. Some of these “irreducible features” of an economy, such as its winter climate, its size and other geographical features, or its natural resource endowment, may force key parts of the emissions pattern away from the negotiating table. Second, some of the variations in emissions arise because of differences in energy supply or energy-use technologies, two aspects of the energy-economy link that climate policies may aim to change. Third, some parts of the differences in emissions among countries arise because of differences in other policy areas, like industrial policies, housing policies, and fiscal policies. Finally, by understanding how these underlying components of emissions differ among countries, policy-makers may begin to appreciate differences in the rates at which emissions can be lowered.

2Background: Data Sources and Methodology

Data sources for this study are described in the individual sectoral papers listed in footnote 1, as well as a long list provided in Schipper 1997. Most data come from official national sources in energy, transport, industry, and housing ministries and national statistical agencies. Key studies that make the breakdowns by end use are described in Schipper 1997. The time series study of Schipper et al. 1997 was the first to use these data, to which readers are directed for more discussion of this issue.

The method presented here grew out of time series decomposition methods examining changes in energy use and carbon emissions (Greening et al. 1996; Schipper et al. 1997; Greening et al. 1997). In these decompositions, changes in total energy use for each sector are disaggregated into activity, structure, and energy intensity terms. Depending on the sector, activity is measured either as value added, passenger-kilometers (pkm), tonne-kilometers (tkm), population, or built area. “Structure” divides activity further into industry branches, transportation modes, or measures of residential end-use activity. Energy intensity is simply a measure of how much energy is consumed per unit of this subsectoral activity. For carbon emissions decompositions we include two additional terms: one to account for changes in end-user fuel mix (final fuel mix) and another for changes in the utility carbon intensity.[3] Changes in energy use or carbon emissions are attributed to the underlying decomposition factors to determine the direction and magnitude that these effects have had over time. The Mine/Yours method, however, uses the same terms to illuminate the causes of differences in emissions during any given year.

The starting point for this analysis is the “GASIF” identity:

G  A * Si * Ii * Fij

In simple terms, the comparison starts with the GASIF identity for a single country, where G equals the emissions from a given sector factorized into the components described above. Sectoral activity is represented by scalar A while vectors Si and Ii indicate the activity share (structure) and energy intensity of subsector i. The i may represent nearly a dozen end-uses in the household sector, seven subsectors in manufacturing, and up to five modes in either sector of transport. The carbon factors (carbon emitted per unit of energy) of each fuel j in each subsector are represented by the F term. The first two terms, activity and structure, comprise what we refer to as energy services. This indicates the total demand for energy needed to support a given level of activity. The last two terms determine the carbon intensity of each subsector i. In other words, the product of the I and F terms indicates how much carbon is released for every unit of energy services delivered. Because there may be several fuels providing energy to a given subsector or end use, j can range to 6 or 7 according to sector and subsector.

To evaluate the importance of each term in differentiating emissions, we measure the ratio of Gm/y,c, where we have substituted the Mine/Yours (m/y) average value of GASIF component c for that of the country in question, to Ga, the country’s actual sectoral emissions. The resulting difference between actual and hypothetical emissions shows how much the component in question drives the difference in emissions between that country and the average. For example, substituting the average utility carbon intensity (in tonnes of carbon per terajoule (TJ) of final electricity) for that of a country with low utility carbon emissions raises that country’s emissions in proportion to both total electricity consumption and the difference between its own and the average utility carbon intensity. When a comparison is made between one country and the group average, that country’s own share in the average is left out. This approach prevents large countries like the U.S. and Japan from exerting a strong pull on the averages to which they are compared. The “Mine/Yours” comparisons are carried out for sectoral activity, for sectoral structure, for the combined substitution of activity and structure (also called “energy services”), and for each of the three terms related to carbon intensity: energy intensity, final fuel mix, and utility carbon intensity. This procedure is repeated for each sector, it being understood there may be different numbers of subsectors or fuels, but always one energy intensity per subsector.

The calculations for the activity, structure, and intensity terms differ among the sectors since the measures of activity and level of data disaggregation vary. Detailed descriptions of the methodology are provided in the sectoral results sections below. However, in general terms, the activity substitution calculates the resultant emissions in each sector that would occur if the country in question had the same per capita activity level as the average of the other countries, holding the other components constant. The structure substitution uses the country’s own activity level but adjusts the subsectoral shares of activity to reflect the average shares. Finally, the energy intensity comparison replaces each country’s subsectoral energy intensities with the average subsectoral energy intensities to determine the role of energy intensity in differentiating emissions. This approach offers a crucial advantage to more aggregated energy intensity comparisons: summing the results due to subsectoral energy intensity differences allows us to control for important structural differences within each sector.[4]

The same approaches to calculating final fuel mix and utility carbon intensity effects are used for every sector (see Table 1). The final fuel mix method adjusts the country’s own fuel mix to reflect the average shares of fuels by sector or subsector (where applicable) at the same energy use, calculating the new carbon emissions that result from this change. The carbon coefficients for coal, oil, and gas are constant[5], but district heat and electricity coefficients must be determined for each country based on that country’s fuel inputs and generation efficiency.[6] Each country’s own district heat and electricity carbon coefficients are used for the final fuel mix calculation, while the average utility coefficients are used in the utility carbon intensity formula. An important assumption made in this work is that emissions from purchased heat and electricity are apportioned based on each country’s national average carbon coefficient. In other words, no distinctions are made based on regional, intra-national differences in utility carbon intensity. The relative importance of each term in differentiating emissions across all countries is discussed in Section 4.

Table 1. Mine/Yours Final Fuel and Utility Carbon Intensity Calculations by Sector

Final Fuel Mix / Utility Carbon Intensity
Calculate emissions with own final energy at the others’ share of each fuel, with the resulting emissions from electricity and district heat taken at own utility carbon intensity. / Calculate emissions with the own share of each final fuel but the others’ utility carbon intensity for producing electricity and district heat.

Several caveats are in order. First, we repeat that this analysis is not aimed at showing how much country X would emit if it were like the others’, only how much various components contribute to differences between X and the others. Second, while the overall analysis quantifies the sources of carbon emissions by energy end-use, these results alone do not indicate how much carbon a country could “shed” through various measures; that must be done through an analysis of economic and political costs. Third, not all of the emissions we identify are linear with their respective structural components: larger homes do not necessarily consume more heating fuel than small homes in proportion only to area.[7] Fourth, while the differences between own and others’ emissions can be roughly captured by the product of the Mine/Yours to actual ratios for each GASIF term, there are significant unexplained residuals. This occurs largely because each substitution is made individually, not progressively.

One final caveat must be added. The substitutions in this calculation assume that all of the Mine/Yours terms are separable. In reality they are not entirely separable, and we point to examples where a high value for a structural factor appears related to a low value for another factor, sometimes with offsetting effects on overall emissions. Above all, low energy costs, which may be associated with energy intensities that fell due to energy-saving investments, may stimulate activity levels, while high intensities may inhibit growth in activity levels. This kind of “rebound effect” is probably small for most sectors, but could be large for energy-intensive sectors like non-ferrous metals, heavy chemicals, or air travel (Schipper and Grubb 2000).

3Mine/Yours Results by Sector

Figure 1 shows the starting point for this analysis, emissions per capita from all 14 countries in 1994. Note the wide range of per capita emissions, with U.S. emissions nearly four times those of Norway. Differences within the travel, freight, and manufacturing sectors vary by about the same factor, while per capita emissions from residential and services energy use differ by a factor of about 15. A marker showing total emissions per GDP is also included to highlight the importance of national income in determining emissions. This shows that the US is by no means the most carbon intensive economy, measured against GDP. The following sections examine all these sectors using the Mine/Yours method to elucidate the underlying causes of these wide ranges in emissions.

Figure 1. IEA-14 per Capita Carbon Emissions by Sector, with Total Emissions per GDP

In the following sections we use a special method to illustrate the results of the substitutions.

3.1Manufacturing

The manufacturing sector is the largest source of energy-related carbon emissions in most of the IEA-14 countries and accounted for about 29% of 1994 emissions in the IEA-14 overall. Per capita manufacturing value added and aggregate carbon intensities (total manufacturing emissions divided by total manufacturing value added) vary widely among these countries. A detailed analysis of energy use shows that aggregate level comparisons hide important differences (Unander et al. 1999). The Mine/Yours calculations described below help to reveal the extent to which structural variations contribute to the differences in per capita carbon emissions from this sector.

3.1.1Methodology

The basic unit of accounting for the energy services provided in the manufacturing sector is value added.[8] This provides a common denominator for comparisons across industry branches and among thousands of final products. Energy intensity is defined as the final energy consumed by each industry branch per dollar’s worth of value added generated. Table 2 describes how the energy consumption and value added data are used to estimate the hypothetical level of carbon emissions that arise under various Mine/Yours scenarios.

Table 2. Mine/Yours Calculations for the Manufacturing (Mfg) Sector

Activity / Others’ mfg output/capita own mfg output/capita x own emissions
Structure / Sum for each branch: Others’ share of mfg value-added own mfg value added 
own carbon intensity
Energy Intensity / Sum for each branch: Others’ branch energy intensity own branch energy intensity own branch emissions

3.1.2Manufacturing Mine/Yours Results

Figure 2 depicts the contributions of activity and structural characteristics toward raising or lowering actual emissions relative to the average of the other countries. The first bar shows actual per capita carbon emissions for each country relative to the other countries’ average, which is indexed to 100. The second and third bars show the ratios of each country’s actual emissions to the hypothetical emissions at others’ structure and activity respectively, while the final bar combines structure and activity effects into an energy services term. In other words, these bars portray the relative contributions of the Mine/Yours terms to raising or lowering per capita emissions compared to the others’ average. For example, the U.K. emits about 30% fewer emissions per capita from this sector than the others do, and this is due to both a less than average carbon-intensive structure (which “saves” about 8% of manufacturing emissions, since the structure bar is at 92) and low manufacturing value added per capita (which “saves” about 24% relative to the others’). If each of these partial components, including the carbon intensity terms shown in Figure 3, is multiplied together, the result is, to first order (i.e., except for a residual) the difference between a country's actual emissions and the average).