Memorandum by Dr Ian Castles, Asia Pacific
School of Economics and Government, Australian National University,
1. In a press statement issued in Milan
on 8 December 2003, the IPCC charged that:
In recent months some disinformation has been
spread questioning the scenarios used by the IPCC as developed
in its Special Report on Emissions Scenarios (SRES). Like all
reports published by the IPCC, this publication was based on an
assessment of peer reviewed literature available at the time of
the preparation of the report and subject to the review and acceptance
procedures followed by the IPCC. As the work of the IPCC proceeds
further any new literature that becomes available will be assessed.
2. The statement identified "so-called
`two independent commentators' Ian Castles and David Henderson"
as the source of the "disinformation" about the IPCC
scenarios. The IPCC has dismissed our critique of the scenarios
and has determined that the SRES provides "a credible and
sound set of projections, appropriate for use in AR4."
3. Some months earlier, the process of assessment,
review and acceptance of the SRES that was summarised in the IPCC
press statement was outlined in detail by 15 of the SRES authors
Over the period of more than three years, the
IPCC writing team developed a set of long-term emissions scenarios
and documented them in a Special Report on Emissions Scenarios
(SRES, 2000). The international and interdisciplinary writing
team that assessed the scenarios literature and developed the
set of 40 emissions scenarios consisted of 53 Authors and four
Review Editors. In addition to scientists, the interdisciplinary
writing team also included experts from NGOs and the private sector.
The work on the scenarios included an "open process"
that invited modelling groups and individuals to contribute to
the scenario development. Six leading modelling teams participated
in formulating the scenarios that included systems-engineering
as well as macro-economic models . . . The scenarios were extensively
reviewed both by 89 experts and by governments prior to IPCC approval
and publication in 2000. Based on these reviews, first the writing
team and then the IPCC Working Group III plenary session in Katmandu,
Nepal, March 2000, modified and subsequently approved the [SRES].
It was published by Cambridge University Press . . . and is also
available through the IPCC website (Nakicenovic et al,
2003, "IPCC SRES Revisited: A Response", Energy and
Environment, volume 14, nos 2 and 3: 188-89).
4. In an article recently published in the
quarterly Newsletter of the Royal Economic Society and submitted
in evidence to the Committee, Professor David Henderson has written:
In relation to our own subject interests, we
question the claims to authority and representative status that
the IPCC makes on its own behalf. We do not question the numbers
of those involved, their diligence, or the existence and observance
of formal review processes. But we think that when it comes to
the treatment of leading economic issues, the milieu is neither
fully competent nor adequately representative. We also hold that
building in peer review is no safeguard against dubious assumptions,
arguments and conclusions if the peers are all drawn from the
same restricted professional milieu.
5. Professor Henderson and I have given
many examples of "dubious assumptions, arguments and conclusions"
in the joint critique of economic aspects of the work of the IPCC
that we have put forward over the past two or three years.
6. In this submission I draw attention to
eight further errors in the SRES and/or in the responses of the
SRES Teams to our critique, in addition to those identified in
our previous papers. My purpose is to provide additional evidence
in support of the view that the IPCC should not be accepted as
the authoritative source of information on the economic aspects
of climate change.
7. My submission is made in an individual
capacity. I do not speak for any interest or organisation.
8. The relationship between economic development
and energy intensity is presented in the SRES as follows:
Although there are consistent differences among
regions and energy paths, the level of energy intensities in developing
countries today is generally comparable with the range of the
now-industrialized countries when they had the same level of per
capita GDP . . . Global energy intensities diverge across different
scenarios, as shown in the shaded wedge in Figure 2-10. The wedge
clearly illustrates a persistent inverse relationship between
economic development and energy intensity across the wide
range of scenarios in the database (SRES, s 2.4.10, pps 97-98,
9. In fact there is no persistent inverse
relationship between economic development and energy intensity
if economic development is measured, as it must be, in GDP(PPP).
This is apparent from the estimates presented by Professor Angus
Maddison in Table 4b of his submission to the Committee. These
estimates show, for example, that in 2001 the level of energy
intensity in the United States was higher than in most developing
countries and regions, including China, India, Other Asia and
Latin Americaand that this was also true (with the single
exception of China) in 1973. If there were an inverse relationship
between economic development and energy intensity, one would expect
energy intensity in the United States to have been lower than
in developing countries in both years.
10. The myth that there is an inverse relationship
is widespread, and arises as a consequence of the belief that
GDP(MER) is "the preferred measure of economic growth"
(see, for example, Nakicenovic et al, 2003, "IPCC
SRES Revisited: A Response", Energy & Environment,
volume 14, nos 2 and 3: 187 and 199). For reasons that have been
explained in detail in our articles, estimates of GDP(MER) do
not measure relative volumes of output, and should not be used
as indicators of levels of economic development.
OF SRES WORLD
11. Appendix III to the SRESentitled
"Definition of SRES World Region (sic)"lists
207 countries and territories in the world, and shows how they
are allocated to the four SRES regions. All countries and territories
of any significance in the world are included in the IPCC list,
with one significant exception: the United Kingdom. It is surprising
that the British Government participants in the processes described
in the quotation in paragraph 3 above did not notice the absence
of their country from the list. The other 207 countries and territories
in the world are listed at http://www.grida.no/climate/ipcc/emission/149.htm.
GLOBAL GDP IN
12. There are many differences in estimates
of regional and global GDP in the base year of the projections
(1990), even among scenarios in the same modelling group. In the
case of the OECD90 region, for example, the estimated GDP(MER)
in 1990 is only US$15.3 trillion for the A2 MESSAGE scenario,
but US$16.4 trillion for the other eight MESSAGE scenarios. The
estimated GDP(PPP) in 1990 for the OECD90 region is US$14.067
trillion for all nine MESSAGE scenarios.
13. It is possible that the reason why the
estimate of GDP(MER) in 1990 for the OECD90 region in the A2 MESSAGE
scenario is about US$1 trillion less than the corresponding estimate
for the other MESSAGE scenarios is that the GDP(MER) for the United
Kingdom (which was of the order of US$1 trillion in 1990) was
inadvertently omitted (see paragraph 11 above).
14. The estimated GDP(MER) for the World
in 1990 is reported as US$20.9 trillion for all nine of the MESSAGE
scenarios. This corresponds to the total for the four SRES regions
for eight of these scenarios. However, the total of the four regions
in the A2 MESSAGE scenario is US$20.080 trillion, or about 4 per
cent less than the estimate given for the World for this scenario.
15. It is possible that this is because,
for the A2 MESSAGE scenario, the United Kingdom is included in
the estimate for the World but not in the estimate for the OECD90
region (see paragraphs 11 and 13 above).
16. The "Glossary of Terms" in
the SRES (Appendix IX) defines "scenario" as "a
plausible description of how the future may develop on a coherent
and internally consistent set of assumptions about key
relationships and driving forces" (SRES, p 594, emphasis
17. Some of the SRES scenarios are not internally
consistent because the composition of the four world regions is
defined differently for different parameters in the same scenario.
For example, Turkey is listed as part of the "OECD90"
region in Appendix III to the SRES, and is presumably included
in that region in all parameters in the MESSAGE scenarios, including
18. However, it is clear from the distribution
of the global population between regions in the B1 IMAGE scenario
that Turkey is not included in the OECD90 region in this scenario
so far as population estimates and projections are concerned:
the estimated population of this region in 1990 in B1 IMAGE was
859 million, compared with an estimated population in the same
year of 919 million in the B1 MESSAGE scenario. The population
of the ALM region (Africa, Latin America and the Middle East)
was correspondingly higher in the B1 IMAGE scenario than in B1
MESSAGE (and most other) scenarios.
19. Yet Turkey is not excluded from the
OECD 90 region for purposes of the GDP estimates and projections
in the B1 IMAGE scenario: the GDP numbers for this region are
virtually identical to those for the B1 MESSAGE scenarios in all
years. The population estimates for the OECD 90 and ALM regions
in the B1 IMAGE scenarios are therefore inconsistent with the
corresponding GDP estimates for the entire period of the projection.
The population estimates for these two regions in 1990 and 2000
must also be inconsistent with the corresponding emissions numbers,
which are standardised for all scenarios and follow the SRES Definition
of World Regions throughout.
GDP(PPP) AND GDP(MER) GROWTH
20. Under the heading "Comparing apples
and oranges", the SRES Team allege that:
. . . in their critique Mr Castles and Mr Henderson
systematically confuse MER and PPP GDP growth measures as reported
in SRES and elsewhere, both in terms of comparing the scenarios
to each other as well as in comparing the scenarios to both the
historical and prospective (scenario) literature. In short, they
systematically construct a case out of comparing apples and oranges.
For instance, throughout their critique an inappropriate comparison
is made between the historical growth rates reported in the formidable
statistical work of Angus Maddison, that reports GDP in PPP, with
the MER growth rates reported in the SRES report, whereas a valid
comparison would have considered the SRES PPP scenarios instead
(Nakicenovic et al, 2003, op cit: 196-97).
21. As pointed out in Castles and Henderson,
2003, "Economics, Emissions Scenarios and the Work of the
IPCC, Energy and Environment, volume 14, no 4: 422-23,
the so-called PPP-based figures in the MESSAGE scenarios are not
genuine measures of GDP. The implied growth rates cannot therefore
be compared with the historical growth rates reported by Angus
Maddisonsee William Nordhaus, 2005, "Alternative Measures
of Output in Global Economic-Environmental Models: Purchasing
Power Parity or Market Exchange Rates?", Prepared for IPCC
Expert Meeting on Emission Scenarios, US-EPA Washington, DC, 12-14
January 2005, revision dated 12-14 January 2005:
A final issue concerns how to construct panels
of countries with PPP outputs or expenditures . . . The most
common practice is to combine the cross section of a base year
(which is a PPP measure) with extrapolations forward and backward
for each country using that country's prices and outputs . . .
This procedure is used by the World Bank, the OECD, and Angus
Maddison (p 11, emphasis added).
THE 1990-2000 DECADE
22. In another example under the heading
"Comparing apples and oranges", the SRES Team claim
Likewise, in his critique of the MESSAGE scenarios,
Mr. Henderson . . . wonders why the 1990 to 2000 GDP growth varies
between 20.6 and 35.4 per cent . . ., simply ignoring that the
former (lower) number refers to PPP, while the latter (higher)
number refers to MER . . .
23. This is not correct. Both numbers refer
to the reported growth in GDP(MER) in the MESSAGE scenarios. The
lower number refers to the A2 scenario, which projects an increase
in GDP(MER) from $20.9 trillion in 1990 to $25.2 trillion in 2000
(SRES, p 481), and the higher number refers to the B2 scenario,
which projects a corresponding increase from $20.9 trillion to
$28.3 trillion (SRES, p. 561). These increases in absolute levels
of GDP represent growth rates of 20.6 per cent and 35.4 per centie,
the numbers quoted by Professor Henderson and repeated by the
24. The corresponding GDP(PPP) projections
from the MESSAGE scenarios are readily available from the same
tables in the SRES, and imply an almost equally wide range in
the rates of growthfrom 21.4 per cent for the A2 scenario
to 35.2 per cent for the B2 scenario.
25. These very wide ranges in the projected
growth rates for a decade that was already over when the SRES
was published should have been considered in the Report. The SRES
Team's false charge that Professor Henderson had confused the
MER and PPP measures suggests that they did not realise the implications
of their own results: they even claimed that "It is an added
bonus that they (the projected GDP increases from 1990 to 2000)
turned out to be quite robust in forecasting actual short-term
historical development", and that "This is a tribute
to the enormous efforts that went into calibration of the base
year in the six models and the short-term dynamics such as the
capital vintage structures" (p 203). In fact, the projections
for the 1990s do not show this. On the contrary, as Professor
Henderson correctly observes,
. . . the SRES treatment of developments over
the 1990s seems curiously detached from what actually happened
. . . Neither the spread of these scenario figures nor their relationship
to actual events is the subject of comment . . . Consideration
of the reasons why some model results diverged sharply from actual
developments over the decadein respect of energy consumption
and CO2 emissions, as well as GDPcould well have been given
in the Report, both as part of the assessment of past trends and
in commenting on the properties and performances of different
models (Castles and Henderson, 2003, "The IPCC Emission Scenarios:
An Economic-Statistical Critique", Energy and Environment,
volume 14, 2 and 3: 183).
MESSAGE PPP SCENARIOS
26. In their second response to Castles
and Henderson (Grubler et al, 2004, "Emissions Scenarios:
A final response", Energy and Environment, volume
15, no 1), the SRES Team state that "three of us (Grubler,
Nakicenovic and Rogner) have developed jointly with other colleagues
the first set of PPP-based long-term emissions scenarios in a
joint study between IIASA and the World Energy Council (Nakicenovic
et al 1998) which was quoted by C & H as an example
of how PPP can be used in such scenarios (p 16, emphasis
27. This is untrue. We stated clearly and
emphatically that "The MESSAGE GDP series expressed in PPP
terms is not such a measureit is mislabelled" (Castles
and Henderson, 2003, op cit: 423). This comment applies
equally to the GDP(PPP) series in the MESSAGE SRES scenarios and
to those in the IIASA-WEC study (Nakicenovic, Grubler and McDonald,
eds, 1998, Global Energy Perspectives).The PPP-based series
which we quoted as examples of how PPP can be used in long-term
emissions scenarios are listed in the sentence beginning "Among
the PPP-based studies . . ." at the foot of p 418 in our
second paper in E&E. These do not include GEP.
28. The fact that the GDP(PPP) series in
the GEP scenarios are not genuine GDP series was recognised by
two of Professor Nakicenovic's colleagues at IIASA (Leo Schrattenholzer
and Yoshiyuke Fujie) in Schrattenholzer et al, "A
longer-term outlook on future energy systems", 2000, in International
Journal of Global Energy Issues, volume 14, nos 1-4: 348ff.
In this paper, Schrattenholzer et al attempted a comparison
between average annual changes in the global energy intensity
of GDP in the IIASA-WEC A2 scenario in Nakicenovic et al,
1998, op cit with those in the POLES scenario. After noting
that there are significant differences in the profile of changes
in the global energy intensity between the two scenarios, Schrattenholzer
et al make the following comment:
Note, however, that the comparison is made using
the units with which GDP is measured in the two models. IIASA-WEC
uses the more conventional market exchange rates to convert GDP
from different currencies into US dollars. In contrast POLES uses
the concept of purchasing power parity (PPP) for the same purpose.
Using POLES GDP growth rates per world region on base-year GDP
values expressed in the conventional (market exchange rate) way,
gives different values of primary energy intensities and its changes
29. If Schrattenholzer et al had
believed that the MESSAGE PPP projections were genuine GDP numbers,
there is no reason why they could not have compared them directly
with the POLES projections which, as they note, are expressed
in PPP. Instead, these experts were obliged to convert the POLES
projections into a spurious GDP(MER) series, in order to compare
apples with apples. It is unfortunate that the sound GDP(PPP)
projections reported in World Energy Council, 1993, Energy
for Tomorrow's World were replaced by two unsound sets of
projections (MER and the so-called PPP-based numbers in the MESSAGE
scenarios) after the IIASA-WEC partnership was formed in 1993.
1 March 2005