The standard (expected utility) approach to assessing the cost of climate change is to
(i) derive a probability distribution for possible rates of climate change under some given projections,
(ii) attach a cost (or benefit) number to each possible outcome, expressed in utility terms,
(iii) calculate the expected utility gain (or loss)
(iv) express the calculated number as a percentage change in some income aggregate (usually GDP)
In this post, I’m going to look at step (ii). In most respects, the Stern review has adopted assumptions that favour strong action to mitigate climate change – relatively optimistic regarding the costs of stabilising CO2 levels, and relatively pessimistic regarding projections of changes in climate. But the cost calculations are conservative, probably because the previously published estimates of Mendelsohn, Nordhaus and Tol have been way too low.
There are two big problem areas. First, the published estimates assume very low adaptation costs – in fact, a lot of the Mendelsohn-Nordhaus estimates assume that adaptation costs are zero. But adaptation costs are already significant – we need only look at the large sums of money being spent in response to changes in rainfall patterns in south-eastern and south-western Australia. Of course, we can’t tell for sure that these changes are permanent climate shifts rather than bad draws from a stable distribution. But that doesn’t help us – we have to bear the costs now and take the chance that rainfall will return to normal, and our money will turn out to have been wasted. Mendelsohn and Nordhaus have got this issue wrong, and Stern largely follows them.
But the treatment of non-market costs is far worse. Even modest changes in climate will have huge impacts on ecosystems. Stern quotes estimates that a 2 degree change in climate (which is pretty much inevitable) will result in the extinction of between 15 and 40 per cent of all species. The business-as-usual scenarios are far worse Nordhaus and Boyer give trivially low estimates of the cost of such a catastrophe $2.5 billion per year for the United States, and $1 billion for Europe. The latter sum also includes damage to cultural sites such as possible flooding in Venice. The implied cost of ecological disaster is of the order of 0.02 per cent of GDP or about 20 minutes of output per worker per year – a tiny fraction of the time that people spend watching wildlife documentaries, visiting national parks and so on. The number is so absurdly wrong, I find it hard to believe Nordhaus and Boyer gave it any thought at all.
The problem is that this is a field that advocates of inaction have had to themselves. Most economists who have thought about the question have looked at the projections of species extinction, concluded that the costs are easily large enough to justify action, but impossible to quantify. The only people who have produced numbers have been those whose prior beliefs are in the far lower tail.
Stern pushes the non-market cost estimates a bit higher than those of Mendelsohn, Nordhuas and Tol, but they are still way below 1 per cent of GDP for 2 degrees of warming. In my submission to the review, I presented a number of lines of evidence suggesting that people would be willing to pay between 1 and 5 per cent of GDP to prevent large-scale species extinction. Looking again, I think my own numbers have been pulled down by the anchoring effect of the absurdly low starting point that’s been presented in the older literature. Over the course of a century, I’m confident people in developed countries would give up 10 per cent of their (rapidly growing) income or more rather than wipe out most of the species with which we share the planet.
In most respects, Stern has adopted estimates that favour action – relatively low for the costs of stabilisation, and relatively high for the projected rate of climate change under BAU. But the downward bias of the median cost has a bigger impact than the combined effect of these (more widely-discussed) biases going in the opposite direction.