IPCC Issues: A Swag of Documents

Ian Castles and David Henderson

February 2003

Over the past six months we have voiced criticisms of work done under the auspices of the Intergovernmental Panel on Climate Change (IPCC). In particular, we have queried some of the methods and conclusions of the Special Report on Emissions Scenarios (SRES). We have now brought together the various pieces that we have written on this subject. They fall into three distinct and successive groups.

First, there are three letters to the IPCC Chairman, Dr Rajendra Pachauri---two from Castles (sent in August 2002) and one (sent in October) from Henderson. These focus on the main single ground of our criticism---namely, the use in the SRES of nominal exchange rates, rather than purchasing power parity rates, for expressing the GDP of different countries in a common unit of measurement. They appear below as Document 1.

Second, there are three contributions that we were able to make at an IPCC meeting. Although the IPCC has made no formal substantive response to the criticisms we have made, we were invited in mid-December to attend an Experts' Meeting of its Technical Group on Climate Impact Assessment, held in Amsterdam on 7-10 January 2003. Arrangements were made in Amsterdam for an extended technical discussion of the points we had raised in our letters to Dr Pachauri, and both of us were given the opportunity to make short presentations to the full meeting of the Group. The texts that formed the basis for these presentations also appear below. Castles presents a critique of the IPCC scenarios, while Henderson makes the case, with some concrete suggestions, for wider professional involvement in the economic aspects of the IPCC's work. Included with these two presentation texts is a note by Henderson, first drafted in Amsterdam, on a specific proposal made in the course of the technical discussions there. These three texts make up Document 2.

Third, following the Amsterdam meeting, we decided to develop our case further, incorporating points that we had not covered in the pieces listed above. Document 3 contains two papers, one by each of us. The paper by Castles presents the case for a review of the IPCC emissions scenarios, while the piece by Henderson raises some further queries and concerns about both the scenarios themselves and the IPCC process of which they form part.

For some readers, it might be preferable to read the papers numbered 2(1) and 3(1) in sequence, since both are by Castles, and similarly, to take 2(2) and 3(2) together, since both are by Henderson, rather than working through Documents 2 and 3 in order.

The contents of these documents are more diverse, and less systematically put together, than would have been the case if we had set out with the intention to provide a full critique. All the eight texts were prepared at short notice, in response to immediate opportunities and concerns. Critics of our critique have rightly made the point that what we are saying has not been subject to peer review. But given that the IPCC has now embarked on the preparation of its Fourth Assessment report (AR4), and will be meeting in Paris later this month, we think it may be useful to circulate what we have written in its original state---as a contribution, timely albeit rough and ready, to an important debate which is now under way.

Ian Castles
National Centre for Development Studies
Australian National University, Canberra

David Henderson
Westminster Business School, London.

5 February, 2003

Publication Note

The following document comprises all eight texts mentioned above, in the order of their presentation. For the aid of those readers who wish to take up the 'navigation' suggestion which was also mooted above, we provide a small contents listing with hyperlinks to the individual pieces. Links returning the reader to this index have been inserted at the conclusion of various pieces within the major document.

 Ian Castles

 David Henderson
Letter to Dr Pachauri (6 August 2002) Letter to Dr Pachauri (28 October 2002)
Letter to Dr Pachauri (29 August 2002) Presentation to TGCIA Expert Meeting in Amsterdam (10 January 2003)
Presentation to TGCIA Expert Meeting in Amsterdam (10 January 2003) Ideas for a Special Meeting (15 January 2003)
IPCC Emissions Scenarios: The Case for a Review SRES and IPCC: Further Concerns (February 2003)


Document 1: Letters to Dr Pachauri

1 (1): Letter of 6 August 2002 from Ian Castles to Dr Rajendra Pachauri


Dr Rajendra Pachauri,
Intergovernmental Panel on Climate Change

Dear Dr Pachauri,

1. Thank you for your interest in my criticisms of the IPCC Special Report on Emissions Scenarios (SRES), and for inviting me to put my concerns in writing. I am taking the liberty of sending copies of this letter to others who joined in discussions on climate change issues with you in Canberra on 23 and 24 July, and also to a 'mailing list' of my colleagues in the international statistical community, government agencies and universities who follow my ongoing correspondence about the use and abuse of statistics in public debates about globalisation, poverty and the distribution of incomes both within and between countries. I will of course forward any comments that you or the SRES authors may have on my criticisms to all of those to whom I am copying this letter.


2. As mentioned in our discussions, I was formerly the Australian Statistician (1986 to 1994) and head of the Australian Department of Finance (1979-86). I am a former President of the International Association of Official Statistics (IAOS), a section of the International Statistical Institute (ISI), and have been a consultant to several national statistical offices and international organisations on a range of statistical issues.

Statistics of Global Poverty and Inequality

3. Following the release of the UNDP's Human Development Report 1999 (HDR 1999), I made extensive statistical criticisms of the treatment in that report of trends in global poverty and inequality. At the request of the 2000 meeting of the UN Statistical Commission (UNSC), those criticisms were examined by a group of expert statisticians constituted as the Friends of the Chair of the Commission. The report of the group upheld my more serious criticisms. In particular, the Friends of the Chair of the UNSC held that HDR 1999 had made a 'material error' (i.e., one which left the reader with 'a fundamentally distorted view of the phenomenon being described') in relying on national accounting aggregates converted into $US at current exchange rates to compare living standards between countries.

4. The HDR Office of the UNDP accepted the report, and has made major improvements in statistical presentation and reporting in subsequent issues of the HDR.

5. Both the SRES and the Contribution of IPCC Working Group III (WG III) to the Third Assessment Report (Climate Change 2001: Mitigation) cite incorrect statements from pre-2000 issues of the HDR in support of claims about the international distribution of incomes. Some of these statements involve exactly the same 'material error' as that referred to in paragraph 3 above. Thus the WG III Contribution includes the following:

    The distributional dimension of global poverty was illustrated vividly by the Human Development Report 1989 (sic---the first HDR was published in 1990), in the form that came to be known as the champagne glass [reproduced as Figure 1.4 in the IPCC Report]. This representation of global income distribution shows that in 1988 the richest fifth of the world's population received 82.7% of the global income, which is nearly 60 times the share of the income received by the poorest fifth (1.4%). More recent statistics indicate that inequality has widened further since then and that in 1999 the richest quintile received 80 times the income earned by the poorest quintile (UNDP 1999) (section 1.3.1).

6. Another paragraph in the same section of the Contribution of WG III cites statements in the World Bank's World Development Report 2000 in support of similar claims about high and rising global inequality. The Australian Treasury has published a detailed critique of this analysis by the Bank, and has cited statements included in this analysis as examples of 'material errors in the use of statistics' ("Global poverty and inequality in the 20th century: turning the corner", 2001, pps. 24-26, 34-40, available in pdf on the Treasury website at www.treasury.gov.au).

7. The SRES cites the UNDP's HDR 1993 in support of the proposition that 'The poorest 20% of Bangladesh's population ... earn per capita incomes that are a factor of 700 lower than that of the 20% richest Swiss population' (section This comparison is invalid, because it is based on the assumption that the poor Bangladeshi family has converted the whole of its income into foreign currency, and spent it on goods and services at average world prices rather than Bangladeshi prices.

8. The same false assumption underlies the claim that 'When measured across the four SRES regions in 1990, income per capita differences are nearly 40 to 1 (between ASIA and OECD90)' (SRES, section The difference in average incomes between these two regions, properly measured, was less than 10 to 1 in 1990, and has since contracted significantly. Thus the projected growth of real average incomes in the ASIA region (i.e., developing Asia) between 1990 and 2100 that is assumed in the A1 scenarios (an increase by a factor of about 140 to 1) and the B1 scenarios (an increase by a factor of more than 70 to 1) would take incomes in that region to far higher levels than the present OECD90 average, and possibly to higher levels than the OECD90 average in 2100. (This latter comparison is more problematic, because of the difficulty of allowing for the 'Gerschenkron effect').

9. The rates of growth in global GDP presented in the SRES significantly understate the true increases in GDP that underlie the emissions projections. This is because the regions that are expected, by assumption, to achieve the highest rates of economic growth in the twenty-first century are significantly underweighted in the calculations of global GDP. An indication of the possible extent of understatement can be derived by comparing the IMF's estimate of the average annual rate of growth in world GDP between 1994 and 2003 at market exchange rates (2.8 per cent) with the corresponding estimate of this growth measured on the correct basis using purchasing power parities (3.6 per cent). In per capita terms, calculation of the growth rate on the discredited exchange rate-based method used in the SRES yields an average annual increase over this decade of only 1.4 per cent, compared with an increase of 2.2 per cent on the correct purchasing power-corrected basis (IMF, World Economic Outlook, May 2002, p. 157).

10. On the basis of estimates by Angus Maddison which are used in the SRES itself in other contexts, average real incomes in the United States increased by a factor of perhaps 5 to 1 in the nineteenth century, and average real incomes in Japan increased by a factor of almost 20 to 1 in the twentieth century. Thus the historical record gives no support to projections that in the course of the twenty-first century there could be increases in average incomes in the entire continent of Asia by a factor of 140 to 1---or even of 70 to 1, which is the assumption underlying the scenario yielding the lowest projected level of emissions. Yet it is upon such fantastic assumptions that the IPCC's projections of emissions, and therefore of temperatures, are predicated.

The B1 IMAGE Projections

11. The B1 IMAGE projection is of particular interest, because this is the marker projection that yields the lowest increase in temperature between 1990 and 2100---between 1.4 degrees C. and 2.5 degrees C. for the seven climate models, with a projected increase averaged over those seven models of 1.98 degrees C. from 1990 and of 1.82 degrees C. from 2000 (Climate Change 2001: The Scientific Basis, Appendix II, Table II. 4).

12. According to the SRES Terms of Reference (SRES, Appendix I), the process for developing the scenarios was to be: 'First, key input assumptions would be provided to modelers', and 'Second, modelers would be asked to construct emissions scenarios based on the input assumptions provided'.

13. In this case, the opposite process was followed. The B1 IMAGE modelers (1) assumed an extraordinarily rapid growth in annual global emissions of carbon dioxide from the burning of fossil fuels (an increase of 1.6 billion tonnes between 2000 and 2010, and a further increase of 1.5 billion tonnes between 2010 and 2020, compared with increases of only 0.8 billion tonnes in the 1980s and of 0.7 billion tonnes in the 1990s); (2) allocated the whole of the very large increases in these emissions in both decades to developing countries; and (3) used the model to estimate the levels of income, energy use and emissions of other greenhouse gases and aerosols that might be associated with the assumed levels of fossil carbon dioxide emissions.

14. In the current decade, for example, the results of the B1 IMAGE model are predicated upon assumed increases in emissions of exactly 0.8 billion tonnes both in the ASIA and the ALM (Africa, Latin America and the Middle East) regions. In other words, the modellers assumed that increases in emissions in each of the SRES developing regions (ASIA and ALM) would be greater in the current decade than the increase for the world as a whole between 1990 and 2000.

15. These assumptions are patently unrealistic, even for a 'high emissions scenario'. They translate into increases in per capita emissions of fossil carbon dioxide of 24% in ASIA and of 46% in ALM. On this basis, the output of the B1 IMAGE model suggests that GDP per head could rise by around 50% in both regions.

16. In the case of the ALM region, it is already certain that such a growth in incomes will not occur. In the IMF's World Economic Outlook May 2002, it is estimated that real GDP per head declined in this region in 2001 and that there will be a further decline in 2002 (p. 163). Even if the IMF's rather optimistic 'world medium-term baseline scenario' for the years 2003 to 2007 were to be realised (p. 224), it would require a further increase of no less than 40 per cent in real GDP in 'ALM' in the succeeding three years to achieve the increase in decadal GDP projected in the B1 IMAGE scenario. If the modelers had followed the procedure laid down in the SRES terms of reference, they would have 'fed in' a much lower rate of growth in GDP as a 'key output assumption'. They could then have concluded that fossil carbon emissions in this region will not increase by anything like 0.9 billion tonnes.

17. The projected growth in Asian GDP in the current decade may well be realised, but with a far slower growth in carbon dioxide emissions than is assumed in the B1 IMAGE projection. There is no obvious reason why the rapid decline in 'emissions intensity' in ASIA in the 1990s should not be maintained during the current decade. Again, if the procedure for scenario development laid down in the SRES terms of reference (paragraph 11 above) had been followed, a fast rate of decline in emissions intensity in this region would have been a 'key input assumption'. The output of the model would then have revealed that carbon emissions would not need to increase by anything like 0.8 billion tonnes.

18. In short, the B1 IMAGE projection, which belongs to the 'storyline and scenario family' that features 'rapid change in economic structures toward a service and information economy', 'reductions in material intensity', 'the introduction of clean and resource-efficient technologies' and 'global solutions to economic, social and environmental sustainability', is itself a 'high emissions scenario'---at least in the early decades of the century. It assumes that there will be a sharp reversal in the downward trend in global fossil fuel emissions per head that occurred in the last two decades of the twentieth century. This seems unlikely: under the 'reference' scenario in the World Energy Council Commission Report Energy for Tomorrow's World (1993), global carbon dioxide emissions per head were projected to decrease by 7% between 1990 and 2020. It is significant that the WEC study, unlike the SRES, was based on purchasing power parity estimates of GDP.


19. I believe that it is important that governments be advised as soon as possible that the economic projections used in the IPCC emissions scenarios are technically unsound, having been derived by converting national GDPs in nominal values into a common currency using exchange rates. This procedure is not permissible under the internationally-recognised System of National Accounts, and was recently rejected by an expert group in a report to the UN Statistical Commission. The practice of using exchange rate conversion is especially inappropriate in relation to projections of physical phenomena such as emissions of greenhouse gases and aerosols.

20. In the introduction to the first edition of his book Global Warming: The Complete Briefing (1994), Sir John Houghton, then Co-Chairman of the Scientific Assessment Working Group of the IPCC, said that 'scientists have a responsibility to communicate the best possible information about the likely magnitude of climate change, along with clear statements of the assumptions made and the level of uncertainty in the estimates.' As there was no clear statement of the assumptions underlying the projections of the likely magnitude of climate change in the IPCC's Third Assessment Report, I consider it vital that governments be advised that the lowest of the SRES projections assumed that GDP per capita would increase to more than 70 times its 1990 level in Asia (excluding Japan) and to nearly 30 times its 1990 level in Africa, Latin America and the Middle East, by the end of the century. Contrary to statements made in the SRES, these projections imply that real incomes in the whole of the developing world will be many times greater than those in the richest countries in the world today.

21. As I mentioned in our discussions, it would also be desirable to seek the involvement of national statistical offices and of the International Statistical Institute in the new emissions projections that I understand are to be prepared for the IPCC's Fourth Assessment Report.

23. I hope that these comments are helpful to you. I repeat my congratulations to you on your appointment as Chairman of the IPCC, and wish you well in your difficult but important task.


Ian Castles
National Centre for Development Studies
Australian National University

1 (2): Letter of 29 August 2002 from Ian Castles to Dr Rajendra Pachauri


Dr Rajendra Pachauri,
Intergovernmental Panel on Climate Change

Dear Dr Pachauri,

In my letter to you of 6 August, I said that I believed that it was "important that governments be advised as soon as possible that the economic projections used in the IPCC emissions scenarios are technically unsound, having been derived by converting national GDPs in nominal values into a common currency using exchange rates".

The pernicious consequences of using this false method of measuring output are apparent in the analysis of greenhouse issues in the World Development Report 2003, released by the World Bank last week.

For example, the Bank argues that "non-OECD countries use ... 3.8 times as much energy per dollar of GDP [as OECD countries], and claims that "This disparity suggests looking for ways that developing and transition countries can increase efficiency and reduce fuel costs---with reduced GHG emissions as a welcome side-benefit ..." The Bank goes on to wonder "why these apparent 'win-win' situations are so elusive", and decides that the answer lies in two types of institutional failure: "distortions in energy policy [which] benefit special interests", and the neglect by firms and households of profitable ways of saving energy "because it is simply too much trouble to pursue them" (p. 177).

There is a simpler answer to the question that the Bank poses. The assumption of a huge margin of difference in energy intensity between OECD and non-OECD countries which the Bank is seeking to explain is false. The ratio of use of energy per unit of GDP in non-OECD countries to that in OECD countries, calculated using PPPs rather than the spurious exchange-rate conversion basis favoured by the Bank (and the IPCC), is not 3.8:1 but 1.2:1.

On the same page of WDR 2003, the Bank wonders what will happen when people "aspire to the current lifestyle of a prosperous country", and puts forward some "simple arithmetic" to show why the Bank supposes this to be impossible:

"Among the prosperous countries, Norway has one of the lowest rates of CO2 emissions per capita from energy, owing in part to ample use of hydro-power. Yet if the global population of 2050 emitted CO2 on average at this rate, the total would be about 2.5 times current global emissions, which would greatly exceed the planet's absorptive capacity."

The argument is grossly misleading for a number of reasons. But the key point that it illustrates is the Bank's failure to understand the basis of the IPCC emissions projections, the lowest of which assumes that developing countries will not only aspire to but will in fact achieve far higher living standards than those of the most prosperous countries today.

Pasted below is the text of an article which appears under my name in this morning's Canberra Times, under the heading "Greenhouse emissions calculations quite wrong". It puts the view that the IPCC should base its climate projections on realistic assessments of future greenhouse emissions, based in turn on realistic projections of the future of the world economy, rather than on the quantification of fantastic "storylines".

With best wishes,

Ian Castles

Text from Canberra Times:

In January last year the Intergovernmental Panel on Climate Change (IPCC) released its latest projections of prospective global warming. The key finding was that "globally averaged mean surface temperature is projected to increase by 1.4 to 5.8°C over the period 1990 to 2100".

The statement led to widespread alarm. Most commentators, including many scientists, interpreted the IPCC's new projected range as a forecast of massive rises in global temperatures, but the IPCC made projections, not predictions, by feeding hypothetical levels of future greenhouse emissions into climate models. The output of such models cannot be better than the input assumptions upon which they are based.

The simulated temperature increases in the IPCC's lowest emissions scenario ranged from 1.4 to 2.5°C. Some assumptions incorporated in this scenario were conservative, but it also assumed an extraordinarily high rate of economic growth in the developing world.

Specifically, the IPCC assumed that the volume of goods and services produced per head in 2100 would be more than 70 times 1990 levels in developing countries in Asia, and nearly 30 times 1990 levels in other developing countries. Far from marking the lower bound of likely outcomes, such astronomic increases are extremely improbable.

The reasoning that produced these assumptions was as follows. Productivity in the rich countries is likely to continue to increase.

In 1990, average incomes in these countries, on the exchange rate-converted basis used in the IPCC projections, were 40 times higher than in Asian developing countries and 12 times higher than the average of developing countries elsewhere. If this gap is to be substantially closed by 2100 on these assumptions, this century must be an era of unprecedented growth.

In fact, average incomes in developing countries are three or four times higher than the IPCC assumed. By adopting the long-discredited method of converting incomes into a common currency using current exchange rates, the IPCC modellers greatly overstated the size of the development gap, but there are two more fundamental objections to the modellers' argument.

First, living standards in the developing countries in 2100 will depend on their actual economic growth during the coming century. No significant country has ever achieved a 20-fold increase in output per head in a century, let alone the 30-fold or 70-fold increases projected by the IPCC for most of the world's population.

Secondly, and paradoxically, the IPCC's model-builders are hostile to wealth per se. They are obsessed by the belief that growth in productivity and affluence inevitably leads to unacceptable growth in greenhouse emissions. For example, they argue that "if governments support the development of rapid-growth sectors, the tendency may be to promote long-term economic growth, increase household income and consumption, and hence increase GHG emissions".

They even claim that "protectionist policies may ... reduce national economic efficiency, which dampens economic growth and tends to restrict growth in GHG emissions".

These concerns are misplaced. Economic growth maximises the output of goods and services for a minimum expenditure of scarce resources. Conversely, reductions in economic efficiency tend to increase the volume of resources required to produce a given volume of final output, and therefore raise the level of GHG emissions.

In Britain, the first developed economy, average carbon dioxide emissions exceeded 2.5 tonnes of carbon per head of the population in 1880, before the motor age began.

Now Britain produces at least five times the volume of goods and services per head as in 1880, but per capita emissions of carbon dioxide have not increased at all.

According to economic historian Angus Maddison, average incomes in China are now higher than in Britain in 1880, but China's carbon emissions are only 0.6 tonnes of carbon per head---less than a quarter of the levels in late-Victorian Britain.

And China's emissions per unit of output are less than half their levels of twenty years ago.

Global carbon dioxide emissions per head from the burning of fossil fuels reached a peak of over 1.2 tonnes per head of population in 1979. They have since declined by nearly 10 per cent.

It is not true that the per capita emissions of rich countries will necessarily increase as they become still richer. No country in western Europe today emits the 3.2 tonnes of carbon per head that Britons emitted in 1913, and per capita emissions in the United States, Canada, Germany, France, the Netherlands, Belgium and Sweden are now lower than the peak levels reached in the 1970s or earlier.

None of the high-income countries of the Organisation for Economic Co-operation and Development now emits the volume of carbon per head that the failing Communist regime in East Germany was emitting in the late 1980s, and poverty-stricken Communist North Korea emits more carbon dioxide per head than South Korea (and most other OECD countries).

It is true that per capita emissions in most developing countries will increase as the world's poor get richer, but this will be happening in a world in which emissions in many rich countries will continue to decline.

Sadly, there is a serious risk that poverty will escalate in many of the poorest countries, especially in sub-Saharan Africa. The real problem is that the people of these countries may remain very poor, not the impact on the world's climate if they and the rest of the developing world become very rich.

The IPCC should base its climate projections on realistic assessments of future greenhouse emissions, not on the quantification of improbable 'storylines' that assume that all of the world's problems except climate change will be magically overcome.

Ian Castles
National Centre for Development Studies
Australian National University

1 (3): Letter of 28 October 2002 from David Henderson to Dr Pachauri


Dear Dr Pachauri,

The former head of the Australian Bureau of Statistics, Ian Castles, who is an old friend of mine, wrote to you on 6 August, and later on 29 August, to raise a number of issues concerning the projections that are set out in the 2001 Special Report on Emissions Scenarios. In replying, you have said that you will be making in due course a full response to him.

I would like to take further two of the arguments that Castles has made.

FIRST, there is the issue of whether comparative cross-country GDP data should be given at market exchange rates or in purchasing power parity (PPP) terms: in the scenario projections, the former procedure was followed. 

Ian Castles has made the case, here as in other contexts, for choosing the PPP basis instead. In this he speaks for the general body of economists and economic statisticians who are working with these data; and as you know, the OECD, the IMF and other agencies moved some years ago to the use of PPP-based weighting systems for their projections. But reading Ian's letters to you, I was puzzled to know how and why the choice of market exchange rates would affect the SRES scenario projections of total GDP and GDP per head for the 'Annex 2' countries over the period 1990-2100, projections which he views as unrealistically high. Why (I asked myself) should GDP projections for one group of countries alone be affected by their relative position, in relation to the rest of the world, in the base year from which the projections start out?

Having now looked at the scenarios, I think I have found an answer to that question. 

The answer lies in the fact that the scenario projections start from an assumption that the 'Annex 2' countries---broadly, the developing countries---will progressively and substantially gain ground over time, in terms of GDP per head, with respect to the 'Annex 1' group (which comprises the core OECD countries and the economies in transition). Each of the scenarios takes as a point of departure an estimate of the extent of this catching up, or convergence, over the whole period from 1990 to 2100.

With this procedure, the choice between market exchange rates and PPP rates in the base year can make a substantial difference. This is because with the former, as opposed to the latter, there is significantly more ground to be made up: the initial divergence, the 'gap', is greater. Hence projected GDP in the poorer countries has to grow faster in order to achieve the postulated degree of convergence in later periods---with corresponding implications, other things being equal, for energy use and for CO2 emissions.

To illustrate this, one can compare the 1990-2100 projections in the B1 IMAGE scenario, which is one of the 'markers', with how these projections would have looked if the same working assumptions had been made but with Angus Maddison's PPP-based 1990 data as the point of departure. (Here I use Maddison's fine book, The World Economy: A Millennial Perspective, though with appropriate regrouping of countries).

The B1 IMAGE scenario projects for the Annex 1 countries an increase in GDP per head, between 1990 and 2100, by a factor of just over 5. It further assumes that by 2100 the ratio of per capita GDP in those countries to that of the Annex 2 countries will have fallen to just over 1.8. In the 1990 base year this ratio (using market exchange rates) is put at 16.7. In order to move from this initial ratio of 16.7 to the postulated 1.8 in 2100, given the projected growth in the Annex 1 group, the total GDP of the Annex 2 countries is projected to rise, between 1990 and 2100, by a factor of just under 65.

Suppose now that we make the same assumptions about the gap in 2100 (the 1.8 ratio) and the growth in per capita GDP in the Annex 1 countries (the factor of 5), but start in 1990 from Maddison's PPP-based estimates for that year. With these, the 1990 ratio of GDP per head in the Annex 1 countries to that of the Annex 2 countries is 6.3 only, as compared with 16.7. To bring this figure down to 1.8 by 2100 would require (if my arithmetic is correct) an increase in the total GDP of the Annex 2 countries, over the period 1990 to 2100, by a factor of only 24.5, as compared with just under 65.

The use of PPP data as the 1990 starting point of the scenarios would therefore have brought with it significantly lower projections for GDP growth in the Annex 2 countries, and hence lower figures for their CO2 emissions. This in turn might have affected the leading and much-quoted conclusion of the whole IPCC Third Assessment, namely, that the prospective extent of global warming, as between 1990 and 2100, could be put in a range of from 1.4 to 5.8 degrees C.

It seems to me that for this reason alone---though there are a number of other substantial questions that could be raised in relation to it---there is a good case for reviewing the whole scenario exercise. This brings me to my other point.

SECOND, Ian Castles has suggested that in the next IPCC Assessment national and international statistical agencies should be brought in and represented. I agree with him: it is high time that these agencies involved themselves in the process. But I would take the argument further. I think that the central economic departments of state---treasuries, ministries of finance, ministries of economics, and organisations such as the US Council of Economic Advisers---should likewise be taking an active part. Their expertise is pertinent, and the economic stakes are high enough to require their attention. They should not remain on the sidelines.

I am sending copies of this letter, on a personal basis, to a number of friends and acquaintances across the world with whom I have been discussing these and related issues, a few of whom are connected either with national statistical offices or with central economic departments of state. I am also sending it to two persons whom I have not met, because of their current roles and responsibilities. One of these is Henry Derwent, who led the British delegation to the recent COP8 meeting in Delhi, and the other is Professor Glenn Hubbard, who is Chairman of the US Council of Economic Advisers and of the OECD's Economic Policy Committee.

Last, I attach a summary CV by way of orientation.

With best wishes for your exacting role and tasks.

David Henderson
(Westminster Business School, London)

Document 2: TGCIA Presentations and Memo

2 (1): Text used by Ian Castles in making a presentation to the IPCC TGCIA Expert Meeting in Amsterdam on 10 January 2003.

My thanks go first of all to the IPCC for inviting me to this meeting to elaborate on concerns about the SRES which I expressed in messages I sent to Dr. Pachauri, at his invitation, some months ago. I am most grateful to the IPCC for meeting the costs of my travel from Australia, and for giving serious consideration to the issues that David Henderson and I have raised.

Although I have spent most of my life working for the Australian Governments, I need to make clear that I am not now affiliated with any government agency, nor indeed with any other organisation involved in climate change matters.

Secondly, I want to thank Dr. Richard Moss for the great efforts he has made to accommodate this presentation within what was already a very full program for the meeting, and also for finding many hours during his extremely busy program to serve as convenor for the productive discussions which David Henderson and I have had with SRES authors and other experts during the course of this week.

My final thanks go to Professor Nakicenovic and other participants for their courteous and frank approach to these discussions. We did succeed in removing some of the differences between us. And, for the rest, we now have a better understanding of what our differences are.

In recent months I have been told, many times, that I have not understood the purpose of the SRES scenarios---that I do not know why and how they were prepared, nor how they are being used in research. I have certainly learned more about these matters in recent months, and especially in recent days. And I have to say that the more I learn about the processes and the outcome, the more uneasy I have become about the way that the scenarios have been used by the IPCC and are being used by the research community.

Let me agree at once that the SRES process has been an open one. At the outset, an open and widely advertised invitation was issued to modellers to take part in the exercise; later, initial scenario results were posted on the internet and revised in the light of comments received; and in the final stages, following standard IPCC procedures, the draft SRES as well as the scenarios were subject to a series of expert and government reviews.

But whatever the merits of the process, serious ambiguities and contradictions remain. On the one hand, the SRES authors state repeatedly that the scenarios 'are neither predictions nor forecasts'; yet, on the other, the blurb on the back cover states explicitly that the report 'describes new scenarios of the future, and PREDICTS greenhouse gas emissions associated with such developments' (emphasis added). Perhaps the statements can be reconciled, along the lines that the scenarios themselves are not predictions but that their quantification in terms of model results are statements about the future which are conditional upon the realisation of the stated assumptions about driving forces. This seems to me to be splitting hairs, but in any case the key question about the future turns upon the reasonableness and consistency of those assumptions.

Where did the assumptions come from? The Terms of Reference for the SRES required that modelling teams would 'be provided with information on the input assumptions and other necessary information', and that the writing team, for its part, 'would ensure that the range of results reflects the underlying uncertainty' (p. 324). And the report itself states that 'Quantitative storyline targets recommended for use in all scenarios within a given family included, in particular, population and GDP growth assumptions' (p. 174).

It would seem to follow that the various storyline targets relating to GDP should have reflected the full range of underlying uncertainty relating to the future growth in world output. But they did not do so. The SRES writing team analysed hundreds of scenarios from 'different literature sources and other scenario evaluation activities', which spanned 'a wide range of assumptions about ... levels of economic development ... and other factors' (p. 79) and found that, for 2050, the gross world product (GWP) in most of the scenarios clustered around a rather narrow range equivalent to 5 times the 1990 level (p. 94). In fact, less than 5% of 166 scenarios analysed assumed that GWP would reach 6.5 times the 1990 level by 2050 (Figure 2-6a on p. 94).

Yet the storylines for the A1 and B1 scenario families---the groups that include the scenarios with the highest and the lowest projections of emissions---assume higher levels of GWP in 2050 than more than 95% of the scenarios in the open literature. In fact the storyline for the A1 scenario family, which includes 17 of the 40 SRES scenarios, assumes a higher level of GWP in 2050 than any of the 166 scenarios in the database. For technical reasons that are explained in my letters to Dr. Pachauri, the levels of future GWP which are in fact assumed in the A1 and B1 scenarios are even higher than the projected levels reported in the SRES.

The SRES identifies GDP as one of the main 'primary driving forces' of future greenhouse gas trajectories, but uses an erroneous method of aggregating the GDPs of individual countries and regions. It is surprising that this unacceptable method was used in the report, because the UN System of National Accounts that was formally adopted by governments and leading international agencies in 1993 is explicit that 'When the objective is to compare the volumes of goods and services produced or consumed per head, data in national currencies must be converted into a common currency by means of purchasing power parities and not exchange rates'. This recommendation was soon reflected, insofar as practice had not already changed, in the work of the IMF, the OECD and other international agencies.

An exception, until recently, was the United Nations Development Programme (UNDP), in its annual and widely-quoted Human Development Report (HDR). But the UNDP has now changed its ways, following a report by an expert group appointed by the UN Statistical Commission, which said that the HDR's practice of presenting international comparisons of GDP per head using market exchange rates was a 'material error'---that is, an error 'which left the reader with a fundamentally distorted view of the phenomenon being described'.

It is regrettable that 'material errors' of this kind, taken from still-unreconstructed volumes of the HDR, are presented both in the SRES and in Chapter 3 of the Report of Working Group III of the IPCC. One might ask how many of the multitude of authors and reviewers of both documents, and the anonymous officials involved in the review process for these two reports, were aware of the existence of the SNA and the fact that its recommendations on the use of purchasing power parities had been endorsed by UN member governments and the leading international economic agencies.

As someone who is strongly committed to the application of sound scientific method in the social sciences no less than in the natural sciences, I was concerned to learn from Dr. Parry's presentation on Wednesday that the IPCC Data Dissemination Centre now provides projections (or are they predictions?) of GDP converted at exchange rates for all countries at 5-year intervals to the year 2100, for the IPCC marker scenarios. I would question whether such an exercise can be of any value. If it is to be undertaken, the resulting projections should certainly be accompanied by a strong 'health warning', which should make clear that the numbers do not reflect relative volumes of output, and therefore cannot properly be used in international comparisons of living standards or GHG emissions.

There is a serious risk that the dissemination of this material will be worse than useless: it will encourage researchers to base their work on faulty data and to reach unsound conclusions---as indeed the SRES authors themselves do in the charts purportedly reporting historical data on 'energy intensity' in various countries and the related discussion in the report (p. 97, 125). The unprofessional use of statistics to make exaggerated statements about differences in income levels between regions and countries of the world (p. 197) places at risk the status of the IPCC as an objective and policy-neutral body.

The SRES Summary for Policymakers claims that 'the ... 40 SRES scenarios together encompass ... the current knowledge of and uncertainties that arise from scenario driving forces such as demographic, social and economic ... that drive the models, as described in the storylines' (p. 3). In my view, this statement is not consistent with the fact that most of the scenarios, including those that yield the lowest levels of future emissions, assume higher levels of global output in the decades ahead than does the bulk of the open literature.

In his very interesting paper in the session 'The SRES scenarios and application in climate research' on Wednesday afternoon, Dr. Tom Wigley identified the A2G IMAGE and the B1T MESSAGE scenarios as 'the extreme scenarios in terms of the 2100 forcing pattern'. He noted that the A2 IMAGE scenario projected a burden of 780 parts per million (ppm) of CO2 in 2100, whereas the B1T MESSAGE scenario projected a burden of 480 ppm. And he described these extremes as capturing 'the total range of possible variation'.

But do they? The B1 family of scenarios, including the B1T MESSAGE scenario, is described in the SRES itself as assuming a future world with 'high levels of economic activity' (p. 182) and a 'high' rate of economic growth to a 'high' level of per capita income in 2100 (Table 6-2a, p. 317). In the decade immediately ahead, the projected rate of growth in world output in the B1T MESSAGE scenario is far higher than that predicted by the leading international institutions. For the developing regions, the B1T MESSAGE projections of output growth are higher than the highest of three scenarios considered by the World Bank in assessing progress towards the achievement of the Millennium Development Goals. The projection of fossil CO2 emissions in B1T MESSAGE for 2010 is not particularly low---it is, in fact, around the median for the 40 SRES scenarios (8.3 GtC) and represents an average rate of increase in per capita emissions of 1.2% annually between 1999 and 2010 (compared with an average annual rate of DECREASE in per capita emissions between 1990 and 1999 of 0.8%). Far from marking the lower bound of prospective emissions at the end of this decade, the B1T MESSAGE projection of emissions of the main greenhouse gas seems unlikely to be realised.

It is true that the projections in this scenario rest upon optimistic assumptions about the rate of technological advance and its application in the second half of the century---and it is mainly for this reason that B1T MESSAGE marks the lower extreme of the SRES scenarios in terms of the 2100 forcing pattern. But it is not obvious that the technological advances assumed in this scenario could only be achieved if developing countries achieve unprecedented rates of growth in income and output in the decades immediately ahead. This argues for the early development of some variants to the SRES models, which combine the long-term characteristics of these models with a more realistic appraisal of short- and medium-term prospects for the developing countries. Some of these variants would probably yield a model with a lower forcing pattern in 2100 than that of B1T MESSAGE, although it is impossible to be more specific until the detailed modelling work is done.

I now pass to David Henderson. He will make some further points relating to the scenario work, and offer some ideas for strengthening it through a broader professional involvement and participation.

2 (2): Text used by David Henderson in making a presentation to the IPCC TGCIA Expert Meeting in Amsterdam on 10 January 2003.

Like Ian Castles, I am pleased to have had the opportunity to take an active part in these proceedings. I too would like to thank Dr Pachauri for the invitation to come to Amsterdam as a full participant in the Experts' Meeting of the TGCIA, Dr Moss for the time and effort that he has devoted to chairing the technical discussions that Ian and I have taken part in here, and Professor Nakicenovic and the other participants in those discussions for their courteous and frank approach

I think that these discussions that Ian and I have been involved in, with SRES authors and others under Richard Moss's able chairmanship, can be aptly described in the traditional diplomatic communique formula: we had a full and frank exchange of views in a cordial atmosphere. Substantial differences remain, as Ian has noted, but the meetings were helpful and constructive. They also gave rise to a promising suggestion for taking things further at the next stage, which I will mention at the end of my remarks.

Before doing so, let me acknowledge, what you will in any case have noticed, that both Ian and I are well beyond the stage of being salarymen: neither of us has any official status, and some of you may have drawn unflattering parallels with Jack Lemmon and Walter Matthau. But the issues that we have raised concerning the IPCC scenario work, as also some further issues that we might have raised or taken further had this been possible, do not just reflect the eccentric intuitions of two elderly retirees: we are making central points which many if not most of our interested professional colleagues would broadly endorse. Yet it needed outsiders to raise these points that bear on the IPCC process, and bring them to the attention of participants in that process.

Why this need for outsiders? The SRES lists 53 authors and 89 reviewers, and to these 142 persons one might add those involved in the sections of the WGIII Report where similar topics are treated in a similar way. Yet the issues that we have raised were not picked up by, or at least did not clearly emerge from, these serried ranks of IPCC participants. In the IPCC process, as elsewhere, a lot of weight is given to peer reviewing. But what has happened shows that multiplying peer reviewers is no safeguard against dubious procedures and conclusions if the peers are all drawn from the same restricted professional milieu.

The moral of this episode, as I see it, is that the scenario exercise, and the IPCC's economic work more generally, should be broadened. You need to establish a more extended, and more fully representative, economic milieu. In particular, you need to ensure wider participation, or at any rate greater awareness of what you are doing, with respect both to institutions and to professional ways of thinking.

As to institutions, what this chiefly involves is (1) national and international statistical offices, and (2) the central economic departments in member governments---i.e., treasuries, ministries of finance, ministries of economics, and agencies such as the US Council of Economic Advisers. You need a wider involvement of official responsibilities and expertise.

As to people as distinct from institutions, the main need is for closer involvement of economic historians and historically-minded economists. The historical element in the IPCC's economic work is greatly in need of strengthening, the sooner the better.

Alas---be warned!---broadening and reinforcing the economic programme, and professional participation within it, will not necessarily give the carbon cycle and global climate change modellers a much firmer basis for their work: have no illusions about what economists can say hope to say with confidence about the future. But what I propose would bring with it two improvements.

  • It would make the IPCC less vulnerable, less exposed to charges of bias, superficiality, or omission.
  • It could help to bring new ideas and insights to bear.

Under both headings, it would serve to make this element in the IPCC process more professionally watertight, which right now it is not.

I do not make these suggestions in a spirit of reproach. The IPCC could no doubt have tried harder to involve these missing groups of professionals; but equally---perhaps more equally, so to speak---people within these groups could themselves have shown greater interest in, and awareness of, IPCC-related issues, and greater readiness to participate. In any case, what chiefly matters is the future. What to do now?

On this, I would like to make two points. The first is longer-term, and the second more immediate.

Should the scenario exercise be rethought? Yes, chiefly for the reasons that we have given but also for some others that we could have developed had time permitted. But does a radical revision need to be put in place, or attempted, specifically for AR4? I think the answer to that second question is No. The work that has already been done on emissions scenarios, with suitable amendments to take account of considerations of the kind to which Ian has just referred, can provide an adequate basis for sensitivity analysis in the next assessment. Rethinking can have a longer time-horizon, which it will probably need.

However, the process of rethinking should begin now; and in our technical discussion group we emerged with a proposal for immediate action. In particular, the suggestion was made to hold a specially convened experts' meeting to review the kinds of issues that we had been debating within the group; and one or two of us have taken this idea further through memoranda containing specific proposals for giving effect to it. In my own memo I make the suggestion that such a meeting could be jointly sponsored by the IPCC and my former organisation, the OECD. So I will conclude by expressing the hope that some of us will meet again in Paris in the spring!

2 (3): Ideas for a Special Meeting

[The memorandum that appears below sets out a suggestion by David Henderson as to the agenda, venue and attendance for a meeting that could be held with the aim of reviewing the kinds of projections of economic change that enter into the scenario work of the IPCC. The memo builds on a proposal for such a meeting that was made in the course of technical discussions held in connection with an Experts Meeting of the IPCC's TGCIA, which took place in Amsterdam on 7-10 January 2003. This text makes some corrections to an earlier draft that was put together in Amsterdam, and incorporates some improvements in wording suggested there by Professor Nakicenovic.]


A proposal has been made, which I support, for convening a special meeting to consider ways of viewing and assessing long-term economic changes and development, in the context of the work and mandate of the IPCC. I believe that one object in holding such a meeting would be to secure the interest, participation and advice of professionals who have not so far been much involved in IPPC work. This would contribute to a wider understanding of what has been done within the programme, and could help to generate new ideas for its future conduct.

I have two groups of persons especially in mind for closer involvement in the IPCC process, starting with this proposed meeting. On the official side, there are representatives of national statistical offices and finance ministries. In the groves of academe, the main target groups are economic historians and historically-minded economists. With this in mind, here are some first suggestions as to venue, subject-matter and attendance.

Venue. Given the objective of drawing in treasuries and official statistical agencies, in addition to the government departments and agencies that are already involved with the IPCC, a good place to hold the meeting would be the OECD where all these ministries are accustomed to meet and where they each have their own Secretariat back-up. Hence the meeting could be a joint IPCC-OECD venture, held in Paris. (If the OECD's restricted membership raises a problem, we could think in terms of tripartite sponsorship, through involving the World Bank).

Subject matter. The title could be, "Long Term Economic Scenarios in the Context of the Work of the IPCC". Given (say) a day and a half for the meeting, the final morning, or session, should look ahead to what could be done for the future. The meeting should try to come up with clear proposals for action.

There should be a background paper, distributed well in advance and primarily for information, on the past economic projections and the SRES. The list of subject headings (and possible speakers) for the meeting could be, or include:

  • reviewing and building on the past: a historical perspective: Nicholas Crafts
  • projecting growth in poor countries and rich: Shankar Acharya
  • the interactions of economic and technical change: Paul David
  • the basis for international comparisons: Ian Castles
  • the uses and limits of model-based scenarios: William Nordhaus
  • Ways ahead: methods and ways of thinking: Angus Maddison
  • Ways ahead: participants and procedures: an IPCC speaker.

(The fifth heading reflects a personal view of mine, that the use of scenarios should not be viewed as the only possible route to take).

Attendance. Besides individuals, the OECD (Environment and Statistics Directorates, Economics Department, and possibly other units), and the IPCC (including leading SRES authors), we should I think try to involve the IEA, the World Bank, the UNDP, the UN Statistical Commission, and the IMF if they proved to be keen and ready to contribute. One way of securing wide representation of OECD member governments while keeping numbers manageable would be to get nominations of individuals from the relevant OECD committees.

David Henderson
15 January 2003

3 (1): IPCC Emissions Scenarios: The Case for a Review

Ian Castles

In letters sent to the IPCC Chairman, Dr Pachauri, in August 2002, I criticised the use in the IPCC Special Report on Emissions Scenarios (SRES) of nominal exchange rates, rather than purchasing power parity rates, for expressing the GDP of different countries in a common unit of measurement. My letters acknowledged that there are large differences in income per capita between countries and regions of the world but argued that, as a result of the use of an invalid method of comparison, the authors of the report had greatly overstated the scale of these differences at the beginning of the projection period. They had thereby been led to overstate the growth in average incomes in developing countries that would be required to achieve the much more even distribution in global income that is envisaged in most of the scenarios. Prospective levels of emissions in these countries had probably been overstated as a result.

The IPCC has made no formal response to these arguments. However, the IPCC's Data Dissemination Centre (DDC) at the Centre for International Earth Science Information Network (CIESIN) at Columbia University, which manages the dissemination of socioeconomic data and applications relating to the SRES under the guidance of the IPCC's Technical Group on Climate Impact Assessment (TGCIA), now provides the following explanation on its website (http://sres.ciesin.columbia.edu/tgcia/):

The SRES report, in the majority of cases, expressed economic growth using GDP estimates converted into a common currency of US dollars using market exchange rates... This was done in full recognition of the fact that the preferred measure of wealth and poverty is to adjust GDP using purchasing power parity (PPP) estimates---a practice initiated by the World Bank in 1996... PPP estimates more accurately capture the command over resources in poorer nations. As such, they would be the preferred income measure for environmental impact and vulnerability studies as well... The reason the SRES report adopted market-based GDP is because most greenhouse gas emissions models in the peer-reviewed literature, including the models used in SRES, are run based on market GDP. Since the terms of reference of the SRES required that it review and reflect the emissions scenario literature, SRES GDP projections are also mostly market-based. In any event, the disaggregated GDP data supplied to CIESIN from the SRES marker models was in market exchange rates, so we have restricted our country level numbers to market exchange rates as well' ('Country level GDP Downscaled Projections. (p. 1).

The statement that the use of market exchange rates to measure relative output and incomes in the SRES was 'done in full recognition of the fact... that PPP estimates more accurately capture the command over resources in poorer nations' is puzzling. It appears to imply that claims in the report about the dimensions of current disparities, which are based entirely on comparisons made at market exchange rates, were known to be misleading at the time that they were made.

The statement that the practice of adjusting GDP using purchasing power parities was initiated by the World Bank in 1996 is inconsistent with the reference in SRES to the fact that the UNDP's Human Development Index, produced since 1990, includes as one of its components 'income as measured by real GDP per capita at PPP to represent command over resources to enjoy a decent standard of living' (SRES, p. 115). This statement correctly reflects the UNDP's approach, as outlined in the Human Development Report 1990:

The GNP figures typically used for international comparisons do not adequately account for the distorting effects of official exchange rates. To overcome these inadequacies, we use here the purchasing-power-adjusted GDP estimates developed in the International Comparison Project, a collaborative effort of the UN Statistical Office, the World Bank, EUROSTAT, OECD, ECE and ESCAP, now being expanded by USAID (p. 13).

The explanation that 'the terms of reference of the SRES required that it review and reflect the emissions scenario literature' is also puzzling, because the rate of growth in global output projected in most of the SRES scenarios (including those that yield the lowest projections of emissions) is substantially higher than that assumed in almost all of the scenarios in the literature. (This point was documented in the text of my presentation to the TGCIA Experts Meeting on 10 January and is further elaborated below).

The DDC now provides projections of GDP converted at exchange rates for all countries at 5-year intervals to the year 2100, for each of the four IPCC marker scenarios. In my presentation to the TGCIA Experts Meeting on 10 January, I questioned whether these estimates were of any value, since they would encourage researchers to base their work on faulty data.

The explanation accompanying the new disaggregated estimates states that 'One initial finding of the downscaling exercise is that the regional growth rate methodology is unacceptable for some countries with high initial incomes that also happen to lie within very high SRES GDP growth rate regions'. It acknowledges that the resulting projections for the Republic of Korea and for eight smaller countries imply 'unacceptably high incomes in 2100', but says that the data for these countries cannot be excluded from the spreadsheets because this would introduce large regional discrepancies within SRES.

In fact the disaggregated estimates now available on the IPCC's DDC reveal that the SRES projections of GDP in 2100 are 'unacceptably high' not only for the nine countries in respect of which this is acknowledged in the explanatory notes, but for many other developing countries and countries in transition. The problem is attributable in part to the regional growth rate methodology, but the more important cause is the extremely high growth in the income of poor countries that is projected in the A1 and B1 scenarios (which is attributable in turn to the overstatement of income disparities at the beginning of the projection period).

The dimensions of the problem can be illustrated by the case of South Africa. In 2000, this country's GDP per head, converted from nominal values using exchange rates, was only 12% of the US level. By 2050, the A1 marker scenario projects that the per capita income of South Africans on this basis will have reached more than four times the US level in 2000, and about twice the level that the US will have reached in 2050. And by 2100, this scenario projects that the per capita income of South Africans will be approaching twenty times the US level in 2000, and more than four times the US level at the end of the 21st century.

In the case of the B1 marker scenario (and other scenarios in the B1 family, one of which yields the lowest levels of emissions in the course of the century), the projected levels of average income in both countries in 2100 are somewhat lower than in the A1 marker scenario, but the level of affluence of South Africans exceeds that of Americans by an even wider margin than in the A1 projections. The total output of goods and services in South Africa in 2100, according to these downscaled A1 scenario projections, will be comparable to that of the entire world in 1990.

Other countries whose average income levels in 2100 are projected to be higher than those of the US in that year under the DDC downscaling exercise for the B1 scenario include Germany, Italy, France and Japan among the OECD90 countries; the Russian Federation and the Baltic States (Estonia, Latvia and Lithuania) among the countries in transition; the Republic of [South] Korea, the Democratic People's republic of [North] Korea, Malaysia, Singapore and Hong Kong among Asian countries; and South Africa, Libya, Algeria, Tunisia, Saudi Arabia, Israel, Turkey and Argentina among the 'Africa, Latin America and the Middle East' group of countries.

Under the A1 scenario, the US is projected to have slipped even further down the ladder of relative affluence by 2100. The disaggregated projections on the CIESIN website imply that the countries that will by then have reached higher levels of average real incomes than the US under this scenario include most of those listed in the preceding paragraph together with (among others) Thailand and the United Kingdom.

However, the disaggregated A1 scenario projections imply that average incomes in the US will be nearly twice as high as in Australia, which shows the lowest rate of incomes growth of all major OECD countries under both the A1 and the B1 scenarios. Among the developing and transition countries which are assumed to enjoy higher average incomes than Australia in 2100, according to the A1 projections published on the CIESIN website, are China, Indonesia, Thailand, the Republic of Korea and Malaysia in Asia; South Africa, Zimbabwe, Algeria, Tunisia, Libya, Gabon, Namibia, Swaziland, Botswana and Mauritius in Africa; Brazil, Mexico, Venezuela, Argentina, Chile, Uruguay and Panama in Latin America; Iran, Saudi Arabia, Israel and many smaller countries in the Middle East; the Russian Federation, Ukraine, Belarus, Poland, Hungary, Romania, Bulgaria and the Czech and Slovak Republics among transition countries; and Papua-New Guinea, Fiji, New Caledonia, Vanuatu and Samoa in the Pacific.

Many of these countries are assumed to have higher average incomes than Australia in 2100 under the B1 scenarios also: this is true, for example, of Zimbabwe. Although no one can say with certainty that per capita incomes in Zimbabwe will not exceed those in Australia in 2100, it is difficult to believe that the SRES authors would consciously have made such an assumption---and it is unlikely that the SRES would have been accepted by reviewers and governments if the full implications of the GDP assumptions had been exposed.

What are the implications for the SRES projections of emissions of these very high projected rates of growth in economic activity? It is not possible to be precise without undertaking a major reworking of the scenarios. But there is no obvious reason for supposing that the overstatement of prospective growth rates and output levels in developing countries would NOT have led to a significant overstatement of projected emissions.

The SRES Summary for Policymakers (SPM) claims that the 40 SRES scenarios 'together encompass the current range of uncertainties of future GHG emissions arising from different characteristics of these models, in addition to the current knowledge of and uncertainties that arise from scenario driving forces such as demographic, social and economic, and broad technological developments that drive the models, as described in the storylines ...' (p. 3).

The implication of this statement, and of similar statements in the body of the SRES, is that the preparation of new scenarios is unnecessary, because the resulting projections of emissions would differ only marginally from the existing set and (in particular) would not result in a significant reduction in the projected levels of emissions in the scenarios with the lowest emissions profiles.

The scenario that yields the lowest cumulative total of CO2 emissions through to the end of this century is the B1T MESSAGE scenario, and this is also the scenario which was identified by Dr. Tom Wigley, in his presentation to the TGCIA meeting on 8 January, as representing the low extreme of possible variation 'in terms of the 2100 forcing pattern.'

But there are a number of reasons for believing that the B1T MESSAGE scenario does not by any means establish a reasonable lower bound: lower emissions levels can be projected on the basis of assumptions that are fully defensible. The following list provides some indication of the range and extent of modifications to B1T MESSAGE that would be necessary if this scenario is to be used as a measure of the lower bound of 'the total range of possible variation':

1) The B1T MESSAGE scenario assumes a far higher rate of growth in output per head in developing countries than does the bulk of the literature in the scenario database (see SRES, p. 79 ff): The literature review showed that the median average annual rate of GDP per capita growth in developing countries was about 2.5% in each of the periods 1990-2020, 2020-2050 and 2050-2100 (Figure 3-10 on p. 119). The rates of growth in these periods that were assumed in B1T MESSAGE were, respectively, 4.2%, 4.6% and 2.6% (derived from tables on pps. 529-30). A rough calculation suggests that, as a result of the far higher growth rates assumed in the first half of the century, the cumulative level of output in the developing countries for the century as a whole exceeds the median level in the literature by a factor of about 3.

2) So far as the first half of the century is concerned, the B1T MESSAGE scenario also assumes a substantially higher rate of growth in output than is envisaged in the B1 storyline itself: The storyline postulates that, by 2050, global per capita output will be about US$13,000 (p. 182), which is similar to the level of $12,755 assumed in the preliminary B1 marker scenario (Table VI-4 on p. 368, line 3). However, the final B1 IMAGE and the B1T MESSAGE scenarios assume that average global GDP in 2050 will be $15,600, over 20% higher than in the preliminary marker scenario (pps. 506, 526).

3) The higher growth in output that B1T MESSAGE assumes in the first half of the century, over and above that postulated in the B1 storyline, is entirely attributable to faster growth in developing countries: For 2050, projected per capita GDP in the ASIA region was 27% higher, and in the ALM region 57% higher, in B1T MESSAGE than in the preliminary B1 Marker (SRES, pps. 371-72, 529-30). An indication of the impact of this difference on projected emissions is that cumulative CO2 emissions in these two developing regions from 1990 to 2050 in B1T MESSAGE are 78 GtC, or almost 30%, greater than in the original quantification of the B1 storyline in the preliminary marker for this scenario (pps. 371-72, 529-30).

4) The B1T MESSAGE scenario assumes a higher rate of global population growth in the first half of the century than many demographers now believe to be likely: James Hansen's 'An Open Letter on Global Warming' quotes demographer Joel Cohen's guess that 'Assuming business as usual ..., I would not be surprised by any population in the range 7-9 billion [in 2050]' (Natural Science, 26 October 2000, pps. 9-10). On this basis the B1 scenario population projections may be well above the lower end of the range of possible outcomes. It is possible that further evidence in support of this proposition will emerge from population projections that are soon to be released by the UN Population Division.

5) The global level of CO2 emissions in 2000 was well below the standardized base for that year reported for all [40] SRES scenarios, including B1T MESSAGE: For fossil CO2 emissions, the standardized increase for the decade 1990 to 2000, calculated in the way explained in Box 5-1 (p. 243), was 0.91 GtC, or 15%. The most widely quoted estimate of the actual increase for the nine-year period 1990-99 (that published by the US Department of Energy-sponsored Carbon Dioxide Information Analysis Centre) is 0.35 GtC, or 6%. On average, therefore, the four unadjusted marker scenarios appear to have overstated actual growth in fossil CO2 emissions in the 1990s by a factor of about 2: a surprisingly wide margin having regard to the fact that trends in emissions for the greater part of the decade were already known at the time that the projections were produced. Whatever the reasons for this, the effective base for all of the SRES projections of CO2 emissions (i.e., the standardized figure for the year 2000) is too high, by a margin of at least 5%.

6) The decrease in methane emissions in industrialised countries between 1990 and 2000 was much greater than projected in the standardized scenarios, including B1T MESSAGE: For the OECD90 region, the standardized scenarios show a fractional increase in CH4 emissions between 1990 and 2000. Data submitted to the UNFCCC by OECD countries show an aggregate decrease of about 8% during this period. For the REF region, the standardized scenarios show a decrease of 17% between 1990 and 2000. Partial information on the UNFCCC database suggests that there was a decrease of more than 30% in this period.

7) Contrary to some assessments, the B1T MESSAGE scenario projects a significant growth in global emissions of methane during the next half century: The scenario reports a projected increase of 44% in global CH4 emissions between 2000 and 2050, compared with an increase of only 11% in the B1 IMAGE marker scenario which uses the same (high) projections of economic growth. In 'Proceedings of a Workshop On Air Pollution as a Climate Forcing, Edited by James E. Hansen' (provided to the TGCIA Expert Meeting), it is stated that there has been significant progress in reducing methane emissions in the US and the European Union, and that 'Such efforts, if international, will clearly stabilise methane abundances, avoiding the large increases projected in some of the SRES scenarios' (p. 2).

8) Again contrary to some assessments, the B1T MESSAGE scenarios projects significant growth in global emissions of CO: The scenario reports a projected increase of 66% in these emissions between 2000 and 2050, compared with a DECREASE of 46% for the B1 IMAGE marker scenario. According to the Workshop Proceedings cited in 7) above, 'Control of CO emissions ... is a clear indirect reduction in CH4 ' (p. 2).

It is stated in the SPM (p. 9) that 'The SRES scenarios cover most of the range of carbon dioxide..., other GHGs and sulfur emissions found in the recent literature and SRES scenario database'; and that 'The SRES scenarios extend the IS92 range [of cumulative emissions between 1990 and 2100] towards higher [carbon] emissions..., but not towards lower emissions. The lower bound for both scenario sets is approximately 770 GtC.'

The obvious question that arises is why the SRES scenarios did NOT extend the IS92 range towards lower emissions. The B1T MESSAGE scenario that yields the lowest cumulative level of emissions in the SRES set assumes a higher level of global average income, both in 2050 and 2100, than do ANY of the six IS92 scenarios (SRES Technical Summary, Table TS1, p. 33). Presumably a scenario that incorporated more realistic assumptions about the prospective growth in incomes in developing and transition countries, but which was otherwise similar to the B1T MESSAGE scenario, would have yielded significantly lower levels of cumulative emissions than B1T MESSAGE. Why was such a scenario not developed as part of the SRES set?

In my view, the 8 reasons given above, taken together, establish a strong case for reworking the B1T MESSAGE scenario, as part of a more general reexamination of the methods and results of the SRES.

3 (2): SRES and IPCC: Further Concerns

David Henderson


In this note I raise some additional queries and concerns about the SRES emissions scenarios, as also the IPCC process of which they form part. The note follows on from, and complements, what is said in the text that formed the basis for my presentation to the TGCIA in Amsterdam, and which forms Document 2 (2) above.

My further concerns fall under four headings:

  • The neglect of the past.
  • The treatment of the period from 1990 to 2000.
  • Questionable presumptions.
  • Overstating consensus.

SRES: the neglect of history

It is a surprising feature of the SRES that in a document surveying the long-term future, which contains over 300 pages of main text and presents 40 different scenarios prepared by six different modelling groups, there is no chapter which systematically reviews the evidence of the past. The starting point for any such quantitative future-oriented inquiry should be a clear and careful survey of earlier developments and trends, going right up to the present day. Such a survey should be both factual and analytical. It should contain facts and figures, tables and diagrams, and a commentary designed to bring out the main features of past changes, to review the causal influences that may have been at work, and to consider how far the evidence thus presented can be used to throw light on future possibilities. It should serve not only as a basis for reflecting on the future, but also as an authoritative source of public information about the past.

SRES: treatment of the period 1990-2000

The starting point or base year for the SRES was 1990, and all the scenarios generated model-based results for the decade 1990-2000 as well as for the 21st century. But by the time the SRES was published, in 2000, the decade of the 1990s was itself becoming past history. Even allowing for the lag between events and the published data that relate to them, and between the completion of a MS and its final publication, it would have been possible for the scenario work to take account of what had actually happened over most of the decade of the 90s. Indeed, these recent developments could and should have found a place in an opening chapter reviewing the past.

As it is, the SRES treatment of developments over the 1990s seems curiously detached from what actually happened. Consider for example the growth of world GDP. For publication in 2000, the SRES could probably have drawn on the IMF data that were available by April 1999. These would have yielded firm estimates for the change in world GDP over the period 1990-97, a still provisional figure for growth between 1997 and 1998, and the Fund projections for changes over the two following years. Combining these would have yielded a preliminary but soundly based estimate for growth over the whole decade of the 1990s: it would in fact have shown prospective world GDP in 2000 as 36.5 per cent above the 1990 figure. (The final outcome now appears as 39.4 per cent, since the IMF's April 1999 projections for the final two years of the decade proved to be too low).

Actual developments in the 1990s, as distinct from model-generated data, appear not to figure in the SRES. Instead, a wide variety of derived figures for world GDP in 2000 is offered, without reference to the recorded course of change since 1990. For example, across the various MESSAGE scenarios, produced by the International Institute for Advanced Systems Analysis, the figure shown for the increase in world GDP, as between 1990 and 2000, varies between 20.6 per cent and 35.4 per cent, as compared with the 36.5 per cent referred to above which was available in early 1999. Neither the spread of these scenario figures nor their relationship to actual events over the decade is the subject of comment, and the same applies to other scenario families in the SRES. Consideration of the reasons why some model results diverged substantially from actual developments over the decade---in respect of energy consumption and CO2 emissions, as well as GDP---could well have been given in the Report, both as part of the assessment of past trends and in commenting on the properties and performance of different models.

IPCC and SRES: questionable presumptions

In some IPCC work, as in other writings on environmental issues, there is a tendency to portray the growth of GDP and advances in technology as separable and countervailing influences on 'sustainability'. Such a tendency is to be seen most clearly in the continued use of the dubious 'IPAT' identity:

Impact = Population x Affluence x Technology

In this formulation, which is given a respectful airing in the SRES (p. 105), a picture is presented in which increases in GDP bring with them, inexorably, the menace of greater Impact: in this respect, they are to be viewed as sources of concern. On the positive side, by contrast, there is the influence of new and improved Technology, which appears as an independent mitigating factor

This is a confused way of viewing ends, means and interrelations. It casts economic progress as a problem, rather than as a goal and an achievement; it leaves out of account the fact that gains in material welfare have gone together with, and indeed have partly comprised, a range of substantial environmental improvements; and it mistakenly presents 'Technology' as a separate influence, generated outside the economic process. It is as though talented men (and occasionally, women) in white coats, acting in their professional capacity from motives unconnected with the pursuit of material gain and profit, can fortunately be made available, given sound government policies, to rescue humanity and the planet from the consequences of economic growth. But 'Technology' and 'Affluence' are doubly inseparable. Not only is technological change a prime source of economic growth, but it is itself largely driven by the perception of profit-directed opportunity; and indeed, technological advances which people show themselves unwilling to pay the costs of are prima facie not worth introducing.

Under this and (as I think) some other headings, the IPCC milieu tends to take for granted, and lend support to, ways of viewing economic events, relationships and goals which are by no means established and agreed. These questionable presumptions, though they enter into the IPCC process, have their origins outside it: some have been endorsed by member governments and international agencies. There is here a general problem, which involves not just the IPCC but governments and the academic world as well---a problem of determining whether and how far there is broad agreement that can serve as a basis for policy, and if so, what form that agreement can be said to take.

IPCC: overstating consensus

Since its establishment in 1988, the IPCC has come a long way. It has brought to completion and publication a series of extremely detailed and broadly agreed reports, covering a wide range of extraordinarily complex issues and spanning several forms of expertise; it has secured for these reports and their conclusions, as also for its procedures, the broad approval of its many and diverse member governments; it has informed the thinking of those governments and prompted decisions by them; and it has now established both a provisional set of conclusions and a well defined process for continuing its work on the issues of climate change. Its participants and supporters might understandably claim that it has created a world-wide consensus on the nature of the problems, the kinds of actions to be taken on them, and the conduct of future inquiries into them.

I think there is a risk of overstating the extent to which consensus exists, or should necessarily be aimed at given the present state of knowledge and range of informed opinion. Despite the IPCC's achievements, it would be wrong to draw the conclusion that its status, ways of thinking and procedures should now go unchallenged. It may be that the natural wish of member governments for the IPCC to establish and present to them a consensus view has been pushed too far.

In its economic aspects, partly for reasons that Ian Castles and I have sketched out in these three documents, the work and preconceptions of the IPCC are open to question, in ways that have gone largely unnoticed by the many participants in the process, official and unofficial. As I have noted in Document 2(2), peer review provides no real safeguard against such failures of perception if the peers are drawn from the same milieu. To be made more professionally watertight, the IPCC has to extend its economic and statistical milieu, by involving a wider range of interests, viewpoints and expertise. Mere numbers are not enough.

Whether and how far the same argument may apply to other aspects of the IPCC's work, where other disciplines are involved, is not for me to judge; but the Panel and its member governments should perhaps consider the possibility that here too the now existing milieu, if it is fair to use the term, could usefully be broadened.

As to the member governments themselves, my impression (it is no more) is that the extent of the consensus that has been arrived at, and the readiness to endorse presumptions that others would question, has been made possible by the fact that, not surprisingly, the officials concerned have been largely drawn from the ministries, departments and agencies that are concerned with environmental issues. They too form a milieu. Its members bring to bear their own specialised knowledge and expertise, but within it economic aspects are not always well perceived or professionally handled. Hence my argument, as put in Document 2 (2), that other departments of state, with economic and statistical knowledge and expertise, should be ready and able to play a part in the IPCC process.

It may be that the IPCC and the professional groups associated with it, official and unofficial, are in danger of becoming victims of their own success. In some important respects, the networks and procedures that have been created appear as more limited, and less free from bias and presuppositions, than is generally realised. Broader professional involvement could admittedly bring with it new disagreements and new elements of complexity and uncertainty, thus putting consensus in question. But it would strengthen the basis of knowledge and understanding which the IPCC has rightly viewed as essential to its work and reputation.

Westminster Business School
4 February 2003

Lavoisier the Man
Bio and Image
Click above for latest SOHO sunspot images.
Click here for David Archibald on solar cycles.
Where is that pesky greenhouse signature?
Click here for David Evans's article.

Website designed and powered by Fergco Pty Ltd.
Copyright in the materials on this site resides with The Lavoisier Group Inc.