Climate scenarios are malarkey

Past data cannot be reliably used to infer the future effects of climate risk. A wide range of people, from senior regulators to junior researchers and respected commentators, have all pointed out that our experience of climate change in the coming decades will bear no resemblance to that which we have experienced over the past half century. The implication is that statistical modeling of historical data will not prepare us for the future. We instead require speculative, narrative-based methods like scenario analysis to properly understand the financial risks of climate change.

The previous paragraph is complete malarkey. In this post I want to highlight why this is so and explain why the prevalence of this view is crippling our ability to truly understand financial climate risk.

The first point to note is that this way of thinking is anti-scientific. Name one other field of human endeavor where data is, at the outset, deemed to be irrelevant. Theoretical physics might come close, but the most esteemed theories remain those that have been confirmed (or at least not rejected) by diligent experimentation and observation. Even the most book-centric theoretician would be able to describe the set of observations that would lead to the repudiation of their ideas.  

In every other field that I can think of, even the most dynamic, rapidly changing ones, human knowledge is advanced through inference. Scientists will hold a set of established theories but these will constantly be challenged by testing their salience against the available evidence. Established wisdom is thus dynamic in nature, constantly changing as new data and new methods of observation emerge.

In contrast, a system based on untestable assertions about an uncertain future can never be effectively challenged. I can run a scenario analysis and, irrespective of the results I present, you will not be able to refute my findings. We could wait until a set of circumstances emerge that approximate the proposed scenario, allowing us to check whether the assertions made at the outset hold up.  This is a bit like waiting to observe a Lamarckian speciation, something that hasn’t happened since the theory was first proposed at the outset of the 19th century.  (Lamarckianism was a precursor to Darwinian natural selection in which giraffes have long necks because they stretch for high food in trees, passing the newfound skills on to their offspring.)

To refute an assertion you must appeal to evidence.  There’s no other way.

For those who espouse the view that future climate risk cannot be inferred from historical data, I pose the following challenge:  Demonstrate that your assertion is valid.  Is it true that past predictions have failed?  Can it be shown that models based on historical data are missing critical features that make the predictions from the models fundamentally wrong?  As long as this challenge remains unmet, the assertion must be rejected on its face.  

The reality is that past data provides our only useful guide to future risks.  On climate science generally, physical scientists have, for decades, used highly empirical methods to make astoundingly accurate long term predictions of how carbon emissions cause warming.  They have successfully predicted many of the consequences of global warming.  Their work continues to improve as data becomes richer.  We will often see news stories where scientists have been able to dispel widely held views.  Unfortunately these observations seem always to point to ever more dire climate-related outcomes.

On the financial side, meanwhile, the data are far more benign.  Empirical methods would suggest that investors, especially in rich countries, and major banks are not overly impacted by changes in the climate and may even benefit from the circumstances that cause global warming.  Past predictions based on rigorous statistical methods point to a robust financial sector – and indeed the sector seems to be very robust to climate risk.  So far the predictions made by rigorous methods are holding up well. 

If you believe that this view is too rosy, you need to find concrete evidence of the emerging threat.  Perhaps you could identify the pockets of the banking sector that are more acutely affected by warming or maybe you could search for evidence of a bend in the relationship between financial outcomes and temperature increases.  Presenting scenarios showing outcomes like this does not constitute evidence.

And maybe financial robustness will one day end.  If this day comes, or when this day comes, evidence will emerge that will allow us to develop rigorous risk management tools.  

Until this happens my only scientific advice is to keep looking.