One man's carbon is another man's bread: understanding differences in the structure of carbon emissions

I. Introduction to Indicators

Energy indicators measure the performance of energy use or emissions in much the same way stock indices measure economic performance. The indicators relate energy or emissions – broken down by activity, end-use, or output – to a measure of that activity, end-use, or output. In this sense, the IEA energy indicators rely on two other indicators – monetary and physical – to measure two kinds of activity, economic and human. The two types of activity are distinct because the latter is not registered directly in national accounts or other economic statistics. This is important because an increasing share of emissions arises from household energy use and private transportation.

Indicators are not data. They are derived from basic data on the structure of economic and human activity, combined with measurements or estimates of the energy used for those activities. Using standard coefficients, quantities of final energy use are converted to quantities of carbon emissions. By relating energy use or emissions to activity, we normalise energy use and arrive at intensities or intensive quantities. Intensive quantities are more comparable over time or among countries than are extensive or un-normalised quantities. This is important, because in some contexts, total energy use or total carbon emissions are less interesting than either of those quantities normalised to a key parameter like population or GDP.

II. Motivation: A Closer Look Beyond Energy

The fundamental problem that motivates the IEA effort is that the most widespread indicator of energy use — the ratio of energy use to GDP — does not really measure anything.

Since the denominator represents many diverse activities, the ratio cannot really measure efficiency. Since the numerator aggregates many fuels and stirs electricity into the mix as well, even the notion of "energy" is confused. Moreover, the mix of activities generally varies from country to country and over time. In response we have decided to disaggregate energy uses and activities and to calculate intensities where numerators and denominators match as closely as possible.

The most desirable level of disaggregation depends on the questions that have to be answered. For the energy technologist, the housing expert or the industrial engineer, dozens of different energy uses have to be separated out so each key technology and energy intensity can be identified. Many policy experts need to know how various energy technologies, or energy-saving programmes have affected energy use. Since these are almost invariably aimed at specific energy uses, energy uses must be disaggregated.

Energy Intensities

Measuring "efficiency" is far more difficult than it seems to be because we rarely observe the physical quantities that define an "efficiency" in the engineering sense and we rarely measure or estimate the economic inputs and outputs that define economic efficiency. To avoid this confusion we introduced energy intensities, defined as energy use per unit of activity or output for a large number of economic and human activities. These intensities can be aggregated under certain conditions, but should not be confused with the ratio of energy use to GDP, which unfortunately is still used widely to measure "efficiency".

Intensities reflect behaviour, choice, capacity or system utilisation and other factors besides just engineering ones. Observers can of course debate whether larger or smaller cars are "better" from the standpoint of emissions but we have learned that in this debate it is better to separate the normative "better/worse" from the objective "more/less", "higher/lower" or "rising/falling".

Data Availability in the IEA

The IEA used many official data sources unofficially to produce its first book of energy and CO₂ indicators, "Indicators of Energy Use and Efficiency". Since then, the IEA role has been to work with Member countries to assure comparability and compatibility of data, to deal with issues of data reliability and continuity and to help countries with less experience develop the more disaggregated views from available data. Gradually we hope to improve our efforts by working with the OECD and others to develop questionnaires to be sent to Member governments. This is already done for industrial statistics.

The Importance of the Structure of Economic and Human Activities

The IEA has not limited its efforts to measuring energy uses and energy intensities. Measuring the underlying structure of activities for which energy is used is equally as important. The evolution of this structure of the economy and human activities can itself cause changes in energy consumption that mimic or offset changes caused by changes in energy intensities. Governments need to know with some degree of accuracy how each component has affected energy use. Energy policy analysts want to know which changes in energy use are directly caused by, or related to, energy policies, energy prices, and energy technologies, and which are largely attributable to the other factors which we called "structural" components above. Thus energy policies may have affected how much energy is used to heat a square metre of an average home. Housing and fiscal policies, on the other hand, may explain why homes are of a given size. With some skill in indexing or other mathematical devices, we can produce measures of the structure of activities or output.

Moving From Energy to Environment: Carbon Emissions

In the 1990s energy has been seen less than in the past as a scarce commodity. We recognise that our supply of energy is not running out. As a result, measures of energy use alone tell us little about the economy or about issues relating to sustainability. Using energy data can cause undesirable side effects, or externalities. One of these problems concerns carbon emissions from fossil fuels (the dominant source of energy in most countries), which are associated with the threat of climate change. Since climate change is certain to be a worse problem for future generations than for our own, energy use today is inexorably linked to sustainability. If we need indicators of energy use as they relate to sustainability, we must have good indicators of how energy use (and fossil fuel use) is linked to economic and human activity. Our indicators serve this purpose.

III. The Devil is in the Decomposing: Seeing Beyond the Surface

A comparison of carbon emissions from a variety of sectors and end-uses in 1994, normalised to GDP tempts us to draw conclusions. But much more analysis must take place before we can either understand the differences among countries or explain them. We need a way of disentangling all the factors that lie behind a particular level of carbon emissions (or any other pollutant problem).

In our model we show how the indicator approach can be used to break down changes in carbon emissions into different underlying elements. Demand for energy services is generated from sectoral activity levels, such as value added and person-km and the structure within each sector, such as industry mix and the mix of transport modes. The driving factors behind the activity and structure developments are GDP, population, income distribution and prices, as well as geographic factors like climate. Energy intensity measured at end-use level represents the delivered, or final, energy needed per unit of activity. By including supply-side losses for each energy carrier and multiplying all fuels by their emission factor we calculate the emissions resulting from each of the activities in the various sectors.

From this approach, observed changes in the end-use of energy can be separated into changes in activity changes in structure and changes in energy intensities. Changes in carbon emissions can be further decomposed into changes in fuel mix and in supply-conversion efficiency. Hence, changes related to improved end-use energy efficiency (reductions in energy intensity) can be isolated from changes deriving from changes in other factors.

The result is an index that measures the relative impact of the varying term (energy intensity, for example) on carbon emissions. The results can literally be interpreted as "how emissions would have evolved if only one term had changed over time and all other terms had remained the same". The benefits of this approach are that the indices are simple to obtain and the results are easy to interpret, since everything is expressed relative to a single base year.

It is important to separate the components of the carbon intensity effect (which is related to energy efficiency and fuel choices) from those related to demand by people and enterprises for energy service, since they change for different reasons and in response to different stimuli. Polices and measures aimed at reducing emissions are most often directed towards the first category. Demand for energy service is related to welfare and economic development, e.g. industrial production, travel activity, appliance ownership, etc., and seldom the target for energy and environmental polices. The disaggregated approach presented here allows for a better understanding of how the various components have shaped and will shape energy and emission developments. This can help determine where policies can be most effective.

Let us now consider in a more general sense the factors behind differences in carbon emissions, some of which are relatively independent of GDP. These factors drive some of the enormous differences in emissions per unit of GDP, even among countries with roughly similar GDP and economic structure:

• Heating degree-days vary from 900 in Australia and 1 800 in Japan to over 4 500 in Canada and Finland;
• Automobile use per capita varies from a high of 13 500 km/year in the US to a low of 3 900 km in Japan. Trucking tonne-km vary by a factor of two, and the ratio of freight hauled/GDP varies by more than a factor of two, clearly a function of geography as well as other considerations;
• Home size varies from over 155 square metres/home (or roughly 60 square metres/capita in the US) to less than 90 square metres/home, and closer to 30 square metres/capita in Japan;
• Per capita manufacturing output varies by two to one. The manufacturing share of GDP varies from under 20% in Australia or Denmark to nearly 30% in other countries, and the share of the energy-intensive raw materials sectors varies from over 30% in Australia, the Netherlands, Finland and Sweden, to only 15% in Denmark. Similar considerations in the service sector reveal an almost two-to-one ratio of built area to population or to service-sector GDP.
• Among important driving factors, road-fuel prices vary by a factor of three, home-heating fuel prices by a factor of two, electricity prices in any sector by as much as a factor of three. GDP per capita varies by 1.75 to 1 across the countries considered, all compared using purchasing power parity.

Some of these factors are "natural", such as climate or geography. Others, such as those affecting the size of homes and buildings, depend on tax, housing and urban policies. Is any of these driving factors likely to relax its apparent grip on patterns of energy use and therefore carbon emissions? We cannot say, but we estimate that these kinds of differences and the policies underlying them contribute more than half of the differences in the ratio of carbon emissions to GDP, with the primary fuel mix, the final fuel mix, and energy intensities contributing the rest.

It is also very important to avoid confrontation among countries in the name of comparison. To some extent low energy intensities are associated with low costs. Low costs, in turn, encourage high levels of activity or structural concentrations. Long driving distances in the US, high per capita production of aluminium in Norway and high indoor temperatures in heating-efficient Sweden are related to low fuel prices or to very high efficiencies.

The key issue for policy makers is how to repeat the achievements of the past. For countries where less carbon savings have occurred, the bad news of the past may be good news for the future. Thus there is almost no carbon left to save in the home heating sector in Norway, but five times more per square metre and per degree-day in western Germany.

The indicators also point out how countries differ in fundamental ways. Finland and Canada have more than five times as many heating degree-days as Australia, and two-and-a-half times as many as Japan. Surely this kind of difference must be taken into account in accounting for differences in emissions. Australia is blessed with low-cost coal and bauxite, which leads to high carbon emissions for primary aluminium production. Australia, the US, and Sweden have more than two times the ratio of domestic freight tonne-km to GDP than most European countries and Japan. This leads to higher emissions from road, rail, and inland shipping. Americans have the largest homes, which raises emissions from space heating. Are these structural differences rigidly fixed in a way that will always make some countries high emitters? Decomposition using indicators raises such questions, but unfortunately cannot answer them.

C. Why Only Intensities? What about Structure?

Clearly, overall sustainability depends on reducing overall externalities from economic and human activity, particularly those that will affect future generations, such as carbon emissions. But few authorities have advocated reducing overall economic activity as a way of reducing those externalities. Instead, the primary focus has been on reducing the amount of an externality per unit of activity, that is, on reducing its intensity. Hence we find that the intensities convey useful information because they measure the link between emissions and activity. We call the goal of reducing the intensity of a particular externality "flexing the link".

At the same time we are aware of the vital role that the overall scale of an economy and its structure together play in determining the level of carbon emissions. A tendency to focus only on normalised indicators hides an important part of the development process. Lower-income countries are growing rapidly, usually much too fast for improved energy efficiency and fuel switching to offset the rise in carbon emissions. Compounding this problem is that such growth often fires increases in home and transportation-related energy uses, which are carbon-intensive. This spurt of activity might even increase the ratio of emissions to GDP, which gives a misleading impression of the development that is really occurring. Only with a careful, disaggregated approach can these natural pauses in the long-term decarbonisation of an economy be explained.

IV. Policy Implications for IEA Countries: You Cannot Hide from Indicators

One important though sometimes discomforting benefit of a disaggregated approach to measuring energy use is that the analyst can see developments invisible at the aggregate level. If restraining carbon emissions from fossil fuel use is important to sustainability, then it is important to know whether technologies actually work; whether raising fuel prices work; whether non-price policies work. It is very hard to say how these kinds of stimuli affect energy use or emissions. But without knowing how energy or emissions levels have changed, the arguments about "why" are meaningless.

Driving Forces: Pointing the Way to What Matters

Displaying indicators in novel ways reveals the importance of certain causal factors. For example, showing the carbon emissions from freight sectors of six IEA countries against each country’s respective GDP in material industry reminds us to take into account the strength of the historical coupling (in this case, freight levels vs. GDP) in thinking about changes in the future. And it reminds us that, all else being equal, emissions from freight are likely to rise with GDP. The economy will not stand still while we devise ways of lowering emissions.

Why are these such important pieces of the puzzle? The answer is that reducing carbon emissions, relative to GDP or even in an absolute sense, depends on the kinds of changes described above continuing for many years or decades. Understanding how prices and incomes shape emissions is also a step to solving the puzzle. Monitoring those changes, and adjusting the stimuli when the reduction in emissions slows, will be important. The IEA approach, using indicators, provides important lighting to make this task a bit easier.

V. Separating the Meat from the Carbon: Indicators on the Road from Buenos Aires

Energy and carbon indicators are nothing but tools to disaggregate and measure the link between emissions, the economy, and human activity. This is the only way to understand how carbon can be shed from the meat in any growing economy with minimum or no economic loss. Unfortunately, there is very little real agreement on what to measure and how to measure it today. There is no mechanism, for example, for "adjusting" emissions for a very cold or warm year. And there are few negotiators or their advisors able to make comparisons of differences in emissions patterns as the basis for differentiated emissions-reductions targets. The main reason is not lack of data per se, although this is an enormous problem for most developing countries. Rather, the real hindrance is belated recognition of the importance of transparency of the energy-economy-carbon links. Aggregation may reduce errors, but it also blinds observers and negotiators alike to the differences between the meat and the carbon. Thus there is an urgent need to be able make this distinction, made even more urgent as negotiations over carbon reductions become more binding and the year 2010 approaches. Whether supported by governments, by multilateral agencies like the World Bank, the FCCC or IEA, or other recognized and neutral international organisations, data collection and analysis, expert training, and a consensus-building process are the key steps to separating the meat from the carbon.

Note! The opinions expressed in this paper are those of the authors and NOT the International Energy Agency.