Up Previous Next Title Page

1.4 Aggregation and linkages

1.4.1 Why we need condensed information: the example of GDP

Economic policy disposes, with the System of National Accounts and its main aggregate, the Gross Domestic Product, of powerful tools which help policy-makers to take informed decisions, and which enable the public to judge the performance of their elected representatives. GDP has an overwhelming influence on the political process. For example, it is hardly possible to read a newspaper not containing at least one article that is somehow related to government actions intended to increase economic growth (measured as GDP, of course). Even more striking is the way in which economic information is condensed on television: in autumn ´96 Bill Clinton and Bob Dole fought for two hours on TV for the votes of the American people. On a European TV programme, this event was condensed into the following statement: “ Clinton said he managed to increase GDP by x percent during his presidency ”. Two hours, or 7,200 seconds, of carefully elaborated arguments between high level politicians were substituted by a five-seconds commentary, a “compression ratio” of 1440 : 1. This example may be a drastic one, but it is symptomatic of the way we replace detailed information by simple indicators. The same “compression” is apparent, for example, when the economic performance of Europe and the US are being compared: a rather complex story is reduced to a look at the respective GDP growth rates, plus the difference of the unemployment rates (the second “big” economic indicator; No. 3 would be the inflation rate).
There are two main reasons for this use respectively mis-use of indicators:
  1. People trust objective figures more than a politician’s words. Experience tells the average voter that statistical yardsticks like GDP or the unemployment rate reflect economic reality more accurately than the eloquent statements of politicians during election campaigns.
  2. People are often not able or willing to assess a multitude of figures in their entirety. While economists will feel a need to look at a dozen relevant economic variables before they judge the state of the economy, the average citizen wants to “reduce the complexity”. Similarly, when busy decision-makers receive a one-hundred page report, they will ask for a two-page “executive summary”. Indicators are “executive summaries” of complex realities.
The number of indicators that the public uses for judging the government’s performance is small enough to be “aggregated” to an overall judgement reflecting a citizen’s personal priorities. For example, an unemployed person might attach more importance to the unemployment rate than a wealthy share-holder. Surprisingly, the inflation rate still plays a significant role in policy-making, although inflation is a purely nominal effect with little impact on real income or income distribution.
In this “mental aggregation process”, environmental indicators have almost no chance; even environmental experts cannot simultaneously judge the “overall performance” derived from sixty indicators, or even ten pressure indices. A non-expert would just exclude the environmental indicators from the evaluation. This is the principal reason why aggregation to indices is needed: to increase the visibility of the environment in the set of popular indicators used by the public. It should be kept in mind that ten indices, as currently foreseen in the pressure indices project, still would have little chance to compete with the three big economic indicators. However, further aggregation to one overall pressure index would require a thorough public discussion between scientists and the societal actors (government, economic sectors associations, environmental NGOs) on trade-off between
In any case, climbing the top level of the information iceberg, that is, further aggregation of economic and environmental performance through an overall welfare index, is neither easy nor desirable. Currently, nobody seems to be eager to aggregate GDP and the unemployment rate into one overall “economic performance index” (although the methodological problems are much less difficult than e.g. the ones of monetary valuation of environmental damages). Obviously, there is no demand for such an aggregation; the citizen prefers the “mental aggregation” based on own priorities. Likewise, environmental experts will always prefer the set of ten pressure indices (and to do the “mental aggregation” themselves), and specialized experts will insist to see the individual indicators separately. It should be recognized, however, that an overall pressure index would satisfy the information needs of people with all levels of expertise (because the disaggregated indicators would be available as well), while a restriction of the aggregation level to ten indices or even sixty indicators would exclude ordinary citizens from adequate access to information on environmental policy performance.
Figure 6: The Indicator-Politics Interface

1.4.2 Environmental Pressure Indices: The Dutch and Finnish examples

On a European scale, the process of establishing consensus weighting systems that would allow higher levels of aggregation is a slow one, given the environmental diversity of European nations (and the intellectual diversity of their indicator experts). On a national scale, however, two remarkable examples are available:

1.4.2.1 Policy Performance Indicators: the Dutch Environmental Pressure Index

(to be developed)

1.4.2.2 The Index of Environmental Friendliness: Statistics Finland

(to be developed)

1.4.3 Aggregation, weighting and the economics of politics

A system of environmental indicators, if intended for use by ordinary citizens, would have to compete with many well-established other information sources, for example
Obviously, there is no single environmental indicator that reaches the level of attention of e.g. GDP, the unemployment rate or even the Dow Jones index. There are two distinct reasons for this lack of representation of environmental issues.
The two reasons are not completely independent of each other, but it is worthwile to analyse them separately.

1.4.3.1 The absence of clear messages and the role of the media

With the exception of perhaps noise, odours and litter, most of our environmental problems can not be directly assessed by people through their own senses. People believe that Climate Change is a big problem because the media have told them that scientists have told them that it is a big problem. Ideally the media should give unbiased descriptions of the dangers which we impose on this planet through our actions (and it would be unwise to assume that environmental journalists are not well-informed). However, we are living in a market economy. The media have to sell information with a high value for their clients. The "news value" of an information is composed of several elements:
The latter two elements may be self-enforcing, because "existing knowledge" means the media have already talked about it. It is often difficult to launch new themes, but once e.g. the TV has started talking about a subject, the newspaper reader expects more details. "Importance" is judged by the number of media which dealt with the issue. The initial snowball effect is re-enforced, leading to characteristic "waves":
Figure 7: How issues reach critical attention levels through indicators

Themes are discovered, quickly reach the headlines, and disappear shortly after their summits (because by definition "news" have to be new). This pattern applies to many kinds of news, e.g. about civil wars, earthquakes, crime. For a number of days, the news are headlines covered by all media. Some days later, the media attention drops significantly. However, the message "there is a war in Yugoslavia" is now established in the minds of people. The reporting has broken through a "critical threshold" which might be called the "lasting attention level". Themes which have reached the headlines for some days are retained in peoples' memories, others will be soon forgotten. The importance of themes therefore follows a "bitwise" logic: either they are important, or they are not. The scale of importance is non-linear. A war that causes half a million deaths does not get a hundred times more media coverage than a war with "only" 5000 casualties, especially if it is far away. This "logarithmic" behaviour of perceived importance of events is further enforced by the fact that waves inevitably have "valleys", in spite of overlaps. These phases of low media profiles e.g. for environment are "news holes”, market niches for journalists, to be filled with whatever new theme appears. For example, intensive reporting on Climate Change at the UNCED Conference or the Kyoto will later on create a vacuum that will be filled, in the absence of really important themes, by "pesticides in spinach" or the dangers of electric fields from mobile telephones. The expert knows that Climate Change will cause millions of deaths in the long run, while electrical fields are relatively harmless, but the not-so-well-informed media client will perceive the two issues as equally important, because her/his favourite newspaper dedicated comparable space to both themes.
Since there are too many environmental issues to be followed, none of them can permanently overcome the "critical threshold level". Thus, environmental issues have a handicap compared to the well-organized economic issues. Or, expressed the other way round: economic performance would not have the same weight in politics if the Statistical Services would refuse to produce GDP, and instead would publish a disaggregated set of economic indicators for e.g. the seventeen divisions of the NACE...

1.4.3.2 Headline indicators: “Reducing the complexity”?

The visibility problem is, of course, a great concern especially for environmental politicians. They would like to raise the profile of “the” environment with a few “headline” indicators that would catch the attention of the public in the same way as e.g. the unemployment rate. While this is perhaps a key reason for the existence of the Pressure Indices project, it is important to make a sharp distinction between:

... like NOx or CO 2 emissions, total material use, water use per capita etc., that is, selecting a few important indicators, and highlighting them. For example, one could pick the ten pressure indicators most frequently mentioned by the Scientific Advisory Groups (SAG), and give them a high political weight (the example on the left alludes at the difficulties of providing sufficiently robust data e.g. for the Loss of Biodiversity and Waste indicators). Of course, this would also mean that no attention would be paid to the remaining 50 issues in the list.
or, alternatively,

... like the ten Pressure Indices from Air Pollution to Water Pollution & Water Resources . These indices would contain the full, but aggregated information of all sixty pressure indicators currently selected for publication.
The crucial difference is representativeness, or the question whether a given indicator covers sufficiently what it is supposed to measure [3]. We will see later ( Chapter 1.5.4.4, Core indicators) that some 60-100 physical indicators (i.e. 6-10 per policy field) are needed to give a reasonable coverage of all relevant pressures on the environment. While the general public might be pleased with simple indicators, for the expert it would look like a desperate attempt to “reduce the complexity” of the policy field “environment”. However, indicators cannot reduce complexity, since reality will not change just because there are simplifying figures around. All we can hope for is that indicators or indices may serve to make the complex reality more transparent, thus enabling decision-makers to better deal with it. Pressure Indices with a transparent weighting process are an ideal tool for doing that; in contrast, a handful of more or less arbitrarily selected indicators would have the opposite effect of giving highly biased messages to the decision-makers, such as “ you can solve all problems of the policy field Water Pollution & Water Resources by reducing water use ”. The principal risk is that, through such “headlines”, distortions are being introduced into a policy-making process that up till now is based on an equilibrium of weak forces, in particular scientists acting in various fora as policy advisors. By introducing headline indicators, the weak voice of the scientists would be quickly swept away by the power of the media. The effect on the overall efficiency of policy-making, on research budgets etc. would be disastrous. Aggregated indices, in contrast, give science many voices in a balanced way and are therefore more neutral.
Most of the popular economic headline “indicators” are technically speaking highly aggregated indices, like GDP, the Consumer Price Index, or the Dow Jones, composed of items that have a common unit but point to different policy issues. Only the unemployment rate is a “physical” indicator in the sense used in the Pressure Indices project, i.e. it measures only one policy issue. This is possible since the labour market problem can reasonably be described by the number of people not having a job; which makes it a much less complex policy field than e.g. the economy as a whole.
Nonetheless, non-aggregated headline indicators may be useful for occasionally waking up the public; but the resulting effect is temporary and should not be confused with the effect achieved by “real” indicators like GDP or the unemployment. They derive their enormous influence on politics from the fact that they a) fully represent an important policy area and b) are being published “stubbornly” year by year with exactly the same methodology; with the result that the media are waiting for them, and that politicians cannot escape the indicator even if in a particular year they are not pleased with the message that the indicator is telling to the public.
Headline indicators seem also particularly attractive to those who want to play the role of society’s watchdog. While it is necessary to have such a function in a democratic system, it is doubtful whether picking a handful of “misbehaving” indicators and using them as a stick is an intelligent strategy to make politicians move. People might get the impression that environmentalists always have something to complain, and thus the originally sharp weapon may lose its bite after a while. A dozen “highlights” produce a lot of light, but the eye adapts quickly to the brightness. More promising is to confront one or two “highlights” with a range of “normal” indicators. For example, the Consumer Price Index (CPI) is often not worth a headline, but journalists regularly report it in the form “consumer prices rose by 2%; the rise was mainly determined by government-administrated prices which jumped up by 5%” , thus attacking a single item by comparing it to the average. This would work analogously with Pressure Indices ( “greenhouse gases went down by 1%, but there is an alarming trend for xyz which went up 5%” ), and would have more effect than publishing 5 alarming headlines without any (less alarming) reference to compare with. In other words: one needs several boring indicators to produce a single highlight...

1.4.3.3 Does aggregation lead to a “loss of information”?

An often-heard argument against aggregation is “you lose information when you aggregate”. The validity of this argument depends crucially on the audience.
Let us go back to the previous example of the Consumper Price Index (CPI):
“Consumer prices rose by 2%; the rise was mainly determined by government-administrated prices which jumped up by 5%”.
This newspaper headline has entirely different impacts on the reader:
For the analysis of aggregation in the context of environmental pressure or Sustainable Development indices, the CPI example demonstrates two important lessons:
Since most, if not all, of the literature on environmental indices is being written by experts, it is not surprising that the argument “aggregation leads to information loss” is still alive. Is it possible to combine the citizen’s right to be adequately informed, and to overcome the expert community’s resistance against aggregation?
The only chance to get the experts’ support for aggregation will be to make accessible both the index and its components . When using the Internet, this is particularly easy: a simple mouse click on the index can make the user jump to the detailed tables s/he asks for. The following two figures, showing two hypothetical examples of an environmental Pressure Index, demonstrate the principle:
Figure 8 : Transparency and aggregation: an illustrative example of a concentric circles presentation of pressure indices

At the center of this presentation, the total “Environmental Pressure Index” (EPI) would display a “serious” trend, expressed through the red colour. Clicking into the center circle might let the non-expert jump to additional general information, such as a trend line of overall pressure for the last four years, or a comparison of her/his country against the EU average.
However, those who want more detailed information about the components, for example a waste expert, or an ordinary citizen arguing with her/his children about the benefits of separate waste collection, could click into the “Waste” pie of the above figure and jump to the next web page:
Figure 9 : Illustrative example: Pressure Index "Waste”

Of course, this would not have to be the end of the story. With every mouse click, the user could get access to more detailed information; for example, by clicking into the “Waste incinerated” pie, s/he would get a detailed explanation why this component of the waste problem is being presented in a friendly green colour (low-emission technologies, ...). Additional information such as the “policy field coverage”, showing to what extent the “problem pressure” in the policy field “waste” can be explained with only six indicators, would help the environmental expert to interpret the message.
There is no limit to the detail that could be provided through such a web-based system. The old medium “paper” was not adequate for doing this, because newspapers (the information source of the ordinary citizen and voter) simply do not have the necessary space to perform this; and thick State-of-the-Environment reports are too clumsy and thus often remain in the shelves even of the dedicated expert.
(Click below to see a test version of this feature; currently only waste is implemented; use the “Back” button of your browser to continue with the text part)
EPI test

1.4.3.4 On the decreasing overall importance of the environment in politics

Independently of the role of the current lack of organised and structured environmental information, the weight of environmental indicators, pressure indices or even one overall pressure index will also depend on the importance the public attaches to “the” environment. Fortunately, this is measurable; there have been plenty of surveys (e.g. in the Eurobarometer series) asking the citizen to judge which are the most important political problems to solve, so we can assess the trend observed since environment became an issue on the political agenda. One might observe a rough correlation to the important economic issues:
The explanation is probably that environment is considered a “luxury”; if people feel secure (high GDP growth, low unemployment), they are willing to sacrifice something for saving the planet. If, however, they fear to lose their jobs, the environment quickly loses weight. This effect might be re-enforced by the still prevailing belief that a higher level of environmental protection is bad for jobs and competitiveness.

1.4.4 Why we need linked indicators: Sustainable Development and scenarios

Pressure indicators and indices are a “common language” that helps the actors of the societal debate on the environment to avoid misunderstandings, to clarify their positions, and to identify the priority areas for political action. However, they do not offer direct solutions. Of course, it helps to know that, for example,
but the obvious question would be how these indicators are linked:
and how a mix of policy instruments should look like that would be most suitable to address all these issues simultaneously and in a way that is compatible with the “economic half” of the Sustainable Development concept.
It should be noted that such questions will not be answered by statistical figures:
Figure 10: Linking Environmental and Socio-Economic Indicators: from SD indicators to Scenarios

1.4.5 Valuation: the problem volume-problem share model

All honest policy-making is somehow aimed at improving the “welfare” of people. However, the definition of “welfare”, and the importance attached to its components, depends strongly on personal beliefs and values. While everybody will agree nowadays that the quality of the environment is an important element of welfare, they will not agree how much weight should be given to it, and how much of the efforts of society should be dedicated to which issue.
Figure 11: Decision hierarchies: a three-cluster model of welfare politics

Below the level of the three top clusters, relative comparisons can be made easily either on the basis of natural science (e.g. by comparing the toxicities of mercury and lead), or with the help of social science-based weighting methods (such as expert assessments following a Budget Allocation Process or Analytical Hierarchy Process).
In contrast, comparisons between the three top clusters, e.g. between Economy and Environment, as they are necessary for any attempt to “monetize” environmental damages, are a real challenge, because the “values” of the clusters are highly subjective. Different societal actors have different overall attitudes towards the environment, that is, their judgement of the overall “problem pressure” may be orders of magnitude apart. For example, people close to the economic actors (whether they are trade unionists or share holders) tend to think that competitiveness, economic growth and employment come first, and then the environment; those who work actively with environmental NGOs would agree that competitiveness and employment are important but still would put “environment” on top of their list of priorities.
Figure 12: The Problem Volume-Problem Share Model

Environmentalists and their counterparts e.g. in industry and agriculture will never agree on the overall importance of "the" environment, that is, the “volume” of the problem. However, they may find it easier to agree on the relative importance of the main components. This explains why monetary valuation of environmental damages is so far from a consensus and almost impossible, while aggregation to indices might be fairly non-controversial. [7]
The “problem cakes” above represent the third layer of the “three cluster model” presented in the figure before (Climate Change/Toxics/Waste). There are, of course, “problem cakes” further down the pyramid, for example the policy field “Dispersion of Toxics” on level 2 of the decision hierarchy:
Figure 13: The Problem Volume Model for Dispersion of Toxic Substances

A toxicologist who has always worked with environmental NGOs will believe that the dispersion of toxic substances is a very big problem. Her/his counterpart in industry will (while recognizing that toxics can be a problem) be convinced that everything is under control. While the two experts cannot agree on the “absolute” importance of toxics, they will have no difficulty to agree on the relative contributions of various key substances to the problem.
The shares displayed above are not purely illustrative; they are based on the number of times the respective indicators have been chosen as “core” indicators by the SAG for the policy field Dispersion of Toxic Substances.
It can be expected that those scientists specialised in heavy metals have again their specific “problem cake” defining the relative importance they would attach mercury, lead and other heavy metals; and that they have fairly different opinions on how important heavy metal emissions are in general.
[3] see also Figure 14 , “Influence of the number of indicators on their usage in policy-making”
[4] see Data, figures, indicators: some basic definitions
[5] Official title of a document better known as the European Union’s Fifth Environmental Action Programme
[6] Meadows et al., The New Limits to Growth, 1992
[7] A study among 600 German scientists conducted in 1990 at the Forschungsstelle für Gesellschaftliche Entwicklungen (FGE), University of Mannheim, confirms the validity of this model. Summary results available on request.
Up Previous Next Title Page