Extremely unlikely weather still happens

Disregarding exceptionally unlikely events when formulating climate policy is extremely unwise, according to an analysis presented in the Australian Journal of Agricultural and Resource Economics.

Economist John Quiggin from the University of Queensland argues that a misunderstanding regarding the language of probabilities leads policy-makers to disregard the likelihood of very rare weather events. 

This, he says, is foolish, because it overlooks the fact that exceptionally unusual occurrences indicate a system at the peak of its sensitivities and vulnerabilities. It is events at “the upper tail of sensitivity distribution” that will ultimately determine the optimal price for carbon dioxide emissions.

Calculating the emissions value based only on climate changes that are probable or moderately likely will result in a figure of only about half the correct price, he warns.

Quiggin maintains that the issue arises because policy-makers fail to understand the way in which the Intergovernmental Panel on Climate Change (IPCC) deploys the idea of probabilities. In its Fifth Assessment Report in 2013 the IPCC uses certain terms to convey quite precise estimates of things happening.

The term “virtually certain”, for instance, indicates something that has a 99 to 100% probability of occurring. “Very likely” is code for a 90 to 100% probability, “likely” puts us in the realm of 66 to 100% probability and so forth.

At the tail end of this spectrum come events that are “extremely unlikely”, with a zero to five per cent probability, and “exceptionally unlikely”, which denotes a probability of less than one per cent.

“This terminology is useful in communicating scientific uncertainty,” he writes. 

“However, it can be highly misleading in policy evaluation and risk analysis. It might be supposed that an outcome that is regarded as ‘very unlikely’ can be safely disregarded.”

This is not the case, he says, because it misrepresents the real nature of low probability events – and that is, paradoxically enough, that they happen rather often.

“Many of the risks with which we are concerned in everyday decision-making are realised with probabilities in the range zero to five per cent or zero to one per cent,” he explains.

“Most notably, the annual risk of death from any cause in developed countries such as the United States is of the order of one per cent, but it would be absurd to disregard the risk of death in making decisions about, for example, smoking or driving behaviour.”

Ignoring the likelihood of extreme events caused by climate change is dangerous, given that their effects are almost certain to be major. Low-probability risks, he says, “often play a much larger role than their probability alone might indicate”.