Centre International de Recherche

sur l’Environnement et le Développement

Nos tutelles

CNRS Ecole des Ponts CIRAD EHESS AgroParisTech

Nos partenaires



Accueil > Rubrique de services > Archive Equipe > Stéphane Hallegatte

Comments on the Stern Review and its assessment of the economic cost of climate change

Stéphane Hallegatte

publié le , mis à jour le

La Stern Review représente un remarquable effort pour rassembler et structurer les connaissances sur les conséquences du changement climatique. Toutefois, l’évaluation du coût économique du changement climatique qu’elle propose est extrêmement contestable, pour des raisons qui sont décrites ci-dessous.

Toutefois, il est important de noter que, pour la majorité des scientifiques, la justification d’une action pour réduire les émissions de gaz à effet de serre ne vient pas d’une évaluation agrégée précise du coût futur du changement climatique, puisque ces évaluations sont largement sujettes à caution. Cette justification vient de l’incertitude sur l’évolution future du climat, qui fait planer un risque important - identifié bien que non quantifiable à l’heure actuelle - sur nos économies, nos modes de vie, et les écosystèmes qui nous entourent.

The publication of the Stern Review on the economics of climate change have generated an impressive media coverage, especially on its numerical assessment of the future economic cost of climate change. Considering the uncertainties that are widely recognized in this field, however, I think that it might be useful to provide a few insights on this assessment. Indeed, if I feel sympathetic with the policy conclusion of the report ("acting without delay is rational"), I am very concerned about the methodologies used by the Stern Review to assess climate change economic costs.

First, this is the only part of the Stern Review that is not a review but an independent modeling exercise. It is unclear why they decided to carry out their own analysis in this section, but this difference should be noted, since this assessment is not based on peer-review literature like other chapters of the Review.

The problem with this analysis is that their economic assessment is based on a methodology and a set of hypotheses, which are very difficult to support. It is noteworthy, however, that the information one can find in the Review is not sufficient to carry out a detailed review process ; the following comments could possibly change in view of additional information about their analysis. In view of the available information, the main concerns that can be raised about this assessment are :

(i) Analyses are carried out until 2200. However, the emission scenario from 2100 to 2200 is not described in details. If it is a simple extrapolation of the A2 scenario, it would reach unrealistic emission levels in 2200. They also selected only one of the published pdf for climate sensitivity, which predicts a rather large warming. As a consequence, the probability of a warming lower than 2.5°C in 2100 is close to zero (A2 scenario). The model used to predict temperature up to 2200 is not described and, unfortunately, no distribution of warming in 2200 is plotted to check realism. Important information, therefore, is missing to assess the report.

(ii) Considering the large warming that is considered (my guess would be more than 12°C in 2200), the use of simple functions like those of the PAGE2002 model (like T^n, calibrated around a warming of +2°C) can be dangerously misleading. No detailed analysis of the economic consequences of very high warming has been carried out so far, making it difficult to calibrate a model like PAGE2002. Additionally, there is no information about how non-market costs are modeled.

(iii) The treatment of adaptation is unclear, in spite of its importance when looking at very long time horizons like 2200 : it is quite unrealistic to imagine that damages will increase regularly and continuously over two centuries without any effective adaptation action. Recent literature has shown that adaptive capacity (and constraints to adaptation strategies) will play an essential role in determining how large climate change cost will be.

(iii) Most importantly, the Review is very good in chapter 3 to 5 to describe mechanisms through which climate change will impact human societies and economies (e.g., extreme events as an obstacle to economic development). Unfortunately, few of these mechanisms are accounted for in the modeling exercise from chapter 6 (as far as I can say). Adding these additional (adverse) mechanisms to the Stern’s estimates would probably lead to huge numbers...

Finally, the most important point is that this modeling exercise uses the same methodology than Nordhaus, Mendelsohn, Tol and others (who found only small costs), but reaches large numbers because of different underlying hypotheses.

I have strongly criticized this methodology, because it does not account for most of the mechanisms that can lead to significant damages (imperfect foresight and uncertainty about future climate ; difficulty in climate change detection and attribution ; technical, financial and institutional constraints to adaptation ; loss of basic needs ; conflicts between social categories, economic activities or countries, etc.). The Stern Review does not make any progress in this direction, but use extreme hypotheses in the same unsuitable methodological framework. I do not think it is a fruitful approach. If future damages are to be of the orders of magnitude suggested by the Stern Review, I do not think that the method they use is able to predict them.

I think that published and well-accepted science is enough to justify action against climate change. Trying to demonstrate this conclusion using questionable methodologies and questionable assumptions may be counter-productive, because it is very easy to show that the chosen assumptions are excessively weak.

All this is particularly unfortunate because the rest of the report on impacts is an excellent information source, well written and well constructed. The Chapter 6 seems to me to be by far the weakest point, on which the whole report will probably be attacked.