Monthly Archives: February 2006

69 – Climate change

There have been several Pannell Discussions covering specific aspects of climate change. This piece sets down my views on the issue as a whole, with a focus on the appropriate policy responses.

How should we respond to climate change? We know that the Kyoto Protocol, even if fully implemented, would not mitigate climate change, but what should we do instead? More? Or less? Or just different? It is a complex question encompassing many issues. Here I will present my considered views, based on a lot of reading, and informed by my experiences as a builder and user of large complex computer models, an economist with expertise in decision making under uncertainty, and a researcher working across disciplines to integrate diverse technical and socioeconomic information to evaluate policy options.

The past and present

The question of what climate change there has already been is not crucial in determining how we should best act now. What matters for that is how much we can reduce future adverse climate impacts, and what it would cost to do so. Nevertheless, information about climate change that has already occurred might give clues about the extent to which climate change has been human-caused and, therefore, clues about future consequences of current actions.

It seems clear that average temperatures rose during the 20th century, but we need to ask whether this rise was unusual, or within the normal historical range. Answering this is hard because reliable records of temperature don’t go back far.

To examine the distant past, scientists have to rely on proxy variables, like tree-ring widths, which are at least partly related to temperature. Based on studies of this type, a group of scientists (often called the “hockey team”) has concluded that average global temperatures have been on a rising trend since around 1900, after being fairly stable for 900 years before that, giving a graph in the shape of an ice hockey stick. However this work has been heavily criticised by several groups. In particular, Steve McIntyre at has identified serious problems with their statistical methods, data and statistical inference. I find McIntyre’s arguments compelling. By comparison, the hockey team appears evasive, defensive, slap-dash, and inclined to play the man rather than the ball.

Other evidence (e.g. tree lines in cold regions, written historical records) indicates that there was a long “Medieval Warm Period”, at least in Europe, and that temperatures were not constantly cool for 900 years. On longer time scales, studies of ice cores indicate that temperatures over hundreds of thousands of years have undergone frequent fluctuations, larger than current changes. Overall, current temperatures are still well within the historical range of fluctuations.

This evidence indicates that we need some humility in our conclusions about human-induced climate change. There are theoretical reasons to expect climate change as a result of greenhouse gas emissions, but it is extremely difficult to judge how much of the observed changes are human-induced and how much are due to natural fluctuations. The confidence of many climate scientists on this question seems to me to go well beyond what can be justified by the evidence.

The future

Ideally, to make decisions about the merits of current actions to prevent climate change, we’d need to know what will happen to future climate if we don’t take preventative actions, what will happen if we do, and how much the resulting difference would matter in terms of human welfare and the environment.

Predictions of the future of climate are based on global climate models. I think our ability to model the long-term climate is weak, in the sense that confidence intervals around predictions for a century in the future are very wide. The models are pretty rough even for modelling current climate, and they vary widely in their predictions for the future, so I don’t have any confidence in their ability to represent global climate over 100 years under changed conditions (further elevated CO2). The models are certainly sophisticated and complex, but real climate is much more complex and, to some extent, chaotic.

Interestingly, the Intergovernmental Panel on Climate Change (IPCC) (2001, page 774) says: “In climate research and modeling, we should recognize that we are dealing with a coupled non-linear chaotic system, and therefore that long-term prediction of future climate states is not possible.” This impossibility cannot be overcome by improved models; chaotic systems are inherently unpredictable.

To predict the future of climate, even if we believed that existing global climate models were sufficient, we would need to be able to predict the inputs required to drive the models. Crucially, this means being able to predict technological change and economic activity. On the scale of a century, no meaningful predictions about these things are possible, only speculations. The range of possible futures is extremely wide. The IPCC deals with this by defining a range of scenarios, which is a reasonable approach in the circumstances, but their choices of scenarios have been criticised by some for being biased towards the pessimistic end of the distribution (i.e. resulting in the greatest climate change).

Overlaid on extreme uncertainty about (a) the technological and economic scenario, (b) greenhouse gas emissions for any scenario, and (c) climate for any pattern of emissions, is uncertainty about impacts. Even if we knew future temperatures with certainty, we have very limited capacity to predict impacts on human welfare and the environment. The aspects of long-term climate that really matter (the frequency and severity of storms and droughts) are not able to be predicted by the existing global climate models. Then there is further substantial uncertainty about what those impacts are worth in dollar terms, due in part to the weakness of economics in handling the distant future and non-financial values. McKibbin and Wilcoxen (2002, p.37-38) sum it up well: “Uncertainty is the single most important attribute of climate change as a policy problem.” The uncertainties are pervasive, large, and unlikely to be resolved.

Overall, I judge that our ability to predict the long-term benefits from current climate policy initiatives is very low, and is not likely to be appreciably improved. We have no choice but to make decisions without good information about their consequences. This should make us cautious investors.

What should we do about it?

There are two broad types of possible strategies: try to prevent or moderate climate change, or adapt to climate change if and when it occurs.

On prevention, the models tell us (a) that the changes in human economic activity required to reduce emissions enough to avoid climate change are extremely great, vastly greater than Kyoto requires, and (b) that there are very long time lags between any current changes in emissions and future effects on climate. These seem to be conclusions about which we can be confident.

Point (a) means that the cost of enforcing sufficiently drastic emission targets is high, so much so that the benefits per dollar spent are likely to be small. Point (b) means that the benefits to current political decision makers resulting even from an effective action would be small. Putting those two things together, my view is that very few countries would comply with emissions constraints that were tight enough to influence climate.

Overall, even if we accept that the observed climate change has been largely human-caused, the best available information indicates that benefits of taking large-scale preventative action are probably small, while the costs are large. On this basis, it seems that prevention may not be a worthwhile strategy.

There is a potential exception to this. Given the scale of emission reduction that is needed, emissions-trading schemes cannot possibly moderate climate change unless they stimulate massive changes in energy technologies. But if massive technological change is possible, it would (it seems to me) be vastly cheaper if governments invest directly in the required R&D. Unlike emissions trading, an R&D-driven program could be cheap enough to be worth doing, and would be much more likely to be effective because it would not be rejected by large emitters for economic reasons.

What about adapting to climate change as and when it occurs? This should a major part of our strategy. My expectation is that adapting to climate change is what we will actually end up doing, although we will also waste a lot of money on ineffectual preventative actions.

The experts

This synopsis is pretty different to the public messages of most climate experts and commentators on the environment. Many of them are prepared to ignore the massive uncertainties, and they seem to advise us to take action whatever the cost and without weighing up the likely benefits. I once heard a scientist on radio admit that the best-bet consequences of climate change are not very severe, but then he argued that we need to worry about the “unknown unknowns”, and spend big money to avoid them. What stupidity!

In my view, the judgements of many experts have been influenced by non-scientific factors, including some or all of the following:

  • passion for the subject and concern to help the planet;
  • extreme ‘group think’, so that dissent from the consensus view is marginalised;
  • invested interests in the public having an inflated view of the issue;
  • disciplinary narrowness resulting in simplistic judgements about policy.

To reduce these problems and ensure that much more balanced advice is provided, the international community needs to change the institutional arrangements around the IPCC.

David Pannell, The University of Western Australia

Further Reading

Intergovernmental Panel on Climate Change – Working Group I (2001). Climate Change 2001: The Scientific Basis. Cambridge University Press: Cambridge.

McKibbin, W.J. and Wilcoxen, P.J. (2002). Climate Change After Kyoto: Blueprint for a realistic Approach. Brookings Institute: Washington D.C.