I’ve done some writing about uncertainty and the role it plays in climate change analysis. (See: here, here.) I continue to think that it’s one of the most widely misunderstood aspects of the mess we’re in. Insofar as uncertainty enters climate discussions, it’s usually in dopey arguments over whether “the science is settled.” In fact, it’s true both that the basic science is settled and that we face enormous uncertainties about climate impacts and their cost. We need to start dealing with those uncertainties in a more sophisticated way.

One place uncertainty is not well-represented is in the economic models used to determine what’s called the “social cost of carbon” — that is, how much it’s worth to reduce carbon emissions.

Efforts to remedy that are underway. One came across my radar recently (via the ever-vigilant RL Miller). It’s called “The Social Cost of Stochastic and Irreversible Climate Change.”

Wait! It’s not boring! Okay, maybe a little boring. But important!

Grist thanks its sponsors. Become one.

The problem is that the typical study in this area — like a recent one [PDF] conducted by the U.S. Government Interagency Working Group on Social Cost of Carbon — uses what’s called an Integrated Assessment Model (IAM), which “assumes that the economy and climate systems evolve deterministically.” But those systems do not, in fact, evolve that way. Rather, they are what’s called “stochastic” systems — there’s an element of randomness that makes them intrinsically unpredictable, especially at policy-relevant time scales. More accurate models must incorporate stochastic change.

Previous models have done some of this stochastic stuff. What these researchers (out of Stanford and the University of Zurich) add is a more realistic estimation of “social risk preferences,” a measure of how risk averse we are with regard to future generations. There’s some technical stuff in there about measuring risk aversion, but I will spare you the details.

So anyway, these researchers “analyzed the optimal level and dynamic properties of the carbon tax in the face of stochastic and irreversible climate change and its interaction with economic factors, including business cycle fluctuations and preferences about risk.” Sexy, huh? (The work involved mixing elements of a DSGE model with an IAM model, but really, you don’t need to know that.)

The part I find most interesting is the attempt to incorporate stochastic climate change effects. Most previous studies have assumed that “damages are a function of contemporaneous temperature” — i.e., that climate damages rise in a steady, linear fashion along with temperature.

Grist thanks its sponsors. Become one.

But this leaves out the much-ballyhooed “tipping points,” positive feedback loops (like melting of the Siberian permafrost) that become self-sustaining. “While the likelihood of tipping points may be a function of contemporaneous temperature,” the researchers note, “their effects are long lasting and might be independent of future temperatures.” In other words, once a tipping point is crossed, the trajectory of climate damages may change, sharply, and stay changed for some indeterminate amount of time. Realistic models must incorporate the possibility of irreversible changes of indeterminate duration. (“Irreversible Changes of Indeterminate Duration” is also, coincidentally, the name of my new math rock band.)

The researchers boast:

In contrast to other approaches in the literature … we are endowed with an annual-frequency, full-dimensional, stochastic IAM with intrinsic uncertainty about annual economic productivity and stochastic climate components.

Suck it, other nerds!

Anyway, what’s the take-home message here?

When you incorporate the threat of tipping points into a model, the social cost of carbon goes up. (In nerdspeak, inclusion of stochastic change “induces significant and immediate increases in the social cost of carbon, even for low-probability and low-impact tipping events.”)

When you incorporate more realistic levels of social risk aversion, the social cost of carbon goes up.

When you incorporate uncertainty about the duration and severity of post-tipping point damages, the social cost of carbon goes up.

Long story short: Grappling with the uncertainties involved in the economy and the climate system, with some consideration of the welfare of future generations, leads us to a much higher social cost of carbon, and thus a much higher price on carbon.

Uncertainty about climate change is not an excuse for inaction. It’s a reason for more aggressive action.