Global Science Report is a weekly feature from the Center for the Study of Science, where we highlight one or two important new items in the scientific literature or the popular media. For broader and more technical perspectives, consult our monthly “Current Wisdom.”

A new paper shows that climate models are getting worse at replicating a collection of known climate changes as incentivized efforts to improve them have them universally veering off course.

Anyone who is familiar with John Allison’s book The Financial Crisis and the Free Market Cure knows that incentives can drive otherwise “independent” decisions in a common direction, with sometimes disastrous results. Allison documents how a collection of government incentives (intentionally and unintentionally) drove decisions in the wider financial markets towards overinvesting in residential real estate. The resulting massive misallocation of funds and ultimate bubble burst sent us into the Great Recession, from which we have yet to recover.

Obviously, that was not the intended outcome of the federal policies, but as Allison writes “Intentions that are called ‘good’ often do not produce favorable outcomes.” Allison argues that a free market, one that is free from centralized incentives, and one in which truly independent decisions are being made, is less susceptible to a universal failure and that when failures do occur, they are not as severe and they are more quickly recovered from. Had the financial markets been operating without federal regulations and incentives, not only would the Great Recession not have occurred (or would have been minor), but that our country would be in a much healthier financial state with an overall higher standard of living for everyone.

Not only can (and do) targeted incentives lead financial markets astray, they also operate the same way in the field of science.

In either case, the ultimate effect is to steer the outcome away from its most efficient pathway and instead send it veering towards dangerous territory that is marked by a decline in our overall well-being.

This is nowhere more evident than in the field of climate science, as a new paper by the University of Wisconsin-Milwaukee’s Kyle Swanson clearly illuminates.

In his work “Emerging selection bias in large climate change simulations,” Swanson finds that the new generation of climate models has become worse at matching recent climate change than the generation of models which they supplant.

According to Swanson:

[T]he current generation (CMIP5) model ensemble mean performs worse at capturing the observed latitudinal structure of warming than the earlier generation (CMIP3) model ensemble. This is despite a marked reduction in the inter-ensemble spread going from CMIP3 to CMIP5, which by itself indicates higher confidence in the consensus solution. In other words, CMIP5 simulations viewed in aggregate appear to provide a more precise, but less accurate picture of actual climate warming compared to CMIP3.

The climate model collective is becoming more precise, but more precisely wrong.

Swanson suggests that the reason for the deteriorating overall performance of a collection of somewhat “independent” models is that the model developers are focusing too intently on trying to improve a single aspect of climate model performance, that being a better replication of the observed climate changes which have been taking place in the Northern Hemisphere high latitudes. Previous versions of climate models generally tend to underestimate the magnitude of the loss of Arctic sea ice in recent decades and the resulting feedbacks to other aspects of the regional climate.

There is increasing incentive for model developers to get this aspect of climate change right.

It comes from the high profile attention (from climate activists and the media) that is heaped upon observed changes in the Arctic. Just think of all the stories that you have heard about the loss of Arctic sea ice, and the threat to polar bears, seals, native peoples, etc. When it comes to a poster child for global warming, there is little better than some furry dewy-eyed arctic creature precariously perched upon the last bit of ice in an otherwise ice-free Arctic ocean. If climate models are under-predicting the extent of climate changes there, their potential as a tool to force climate change mitigation efforts (e.g., a carbon tax) is being under-realized. The sooner this situation is rectified, the sooner we can get on with saving mankind from itself. Or so it goes.

But does doing a better job in the Arctic lead to more valuable models?

Swanson isn’t so sure:

While the observed Arctic warming is spectacular and important, it is unclear why it is more important from the perspective of the evolution of the overall climate system than the relatively modest warming in the tropics and southern hemisphere. It is unclear whether the CMIP5 simulations are even getting the reason for the actual Arctic warming correct, as they are inconsistent with the strong Arctic warming but only modest warming in the Northern Hemisphere midlatitudes and tropics that best describes the recent evolution of the actual climate system.

So what we are left with is a giant mess. Newer models overall perform worse than older models, and in the specific region where the new models are designed to replicate the observed climate the best—the Arctic—it isn’t even clear that they are getting things right for the right reason.

This situation is not much different than the situation leading up to the Great Recession.

In that case, government incentives reduced the independence of decision-making in financial markets. With climate models, environmental incentives have reduced the independent decision-making in model development.

Swanson describes the situation like this:

[T]he proverbial ‘marketplace of ideas’ about how climate change has and will continue to occur has shrunk. Instead of 38 unique models, each responding differently to increased anthropogenic forcing, in the CMIP5 project climate simulation is evolving towards a state where there are 38 ‘unique’ models that all respond the same.

Incentives have driven a misallocation of resources towards producing an outcome which is getting ever further from the optimal one. Consequences may follow which lead to a decrease in human well-being (for example, restricting the development of inexpensive and reliable sources of energy).

Swanson provides some suggestions as to how to get climate models back on the right course:

Whether through re-examination of the radiative forcings that underlie climate change, the dynamical variability of the models, the sensitivity of the models to imposed radiative forcings or the heat uptake of model oceans, a healthy dose of diversity must somehow be reintroduced into climate simulation enterprise. [references removed]

Allison also recognizes the need to have more diversity in financial decision-making to avoid looming future disasters from deficits in social security and Medicare, annual operating deficits and unfunded government pension liabilities. In his book, he lays out the steps that must be taken to get things going again in the proper direction.

Righting the ship will not be easy. Allison sums up the situation why, which in form, applies equally well to financial markets as well as climate (and all) science:

People cannot think unless they are free to think. Government rules and regulations literally prevent thought and prevent experimentation. A free market is a massive experiment in competing ideas, the most productive of which win out. Most of the experiments fail, but even failed experiments lead to a better understanding. When intellectual elitists stop the experiments because they are smarter than the rest of us and know what is in the “public good’ the learning stops—witness the Soviet Union. By now, the elitists should know better. Often the use public good as an excuse for their own lust for power. Those of us who have had to face government bureaucrats often see the lust for power as the true motivation and the “public good” as the bureaucrat’s rationalization.

References:

Allison, J. a., 2013. The Financial Crisis and the Free Market Cure: Why Pure Capitalism s the World Economy’s Only Hope. McGraw-Hill, New York, 278pp.

Swanson, K., 2013. Emerging lection bias in large-scale climate change simulations. Geophysical Research Letters, doi:10.1002/grl.50562