Integrated assessment models: What do they tell us about climate change policy?
"Very little," according to Robert Pindyck in a new working paper.
Integrated assessment models (IAMs to practitioners) stitch together projections from climate models, energy sector models, agronomic crop models, models of other sectors of the economy, and partial or general equilibrium models that account for price and interactions with the broader economy to derive a more comprehensive evaluation of costs and benefits from climate change.
Pindyck is understandably frustrated with the false sense of precision these models can impart. As he explains, a few reasonable tweaks of any of these models can give very different estimates about the social cost of carbon---the price we should pay, but typically don't, for emitting CO2.
Pindyck raises some good criticisms about IAMs, or at least says out loud a lot of things that many economists have quietly said to each other. I'm glad he's bringing our varying assumptions and wildly varying cost-of-carbon estimates out into the open for all to see. Perhaps it will push us to make our modeling efforts a little more useful, or at least more transparent.
He's right to pick on false precision. But I wonder: has anyone really been fooled? My sense is no. One positive thing about these modeling efforts is that they allow us to see which assumptions are most critical. They are nice (black?) boxes for testing out the sensitivity of X on overall climate impacts. This might help us frame more reasonable discussion about the possibilities and what we should do. It might also help researchers focus future empirical efforts.
The extreme sensitivity of results to seemingly innocuous assumptions also shows how uncertain the impacts of climate change really are. Indeed, not long ago Pindyck published a paper in JEEM with results that are extremely sensitive to his assumption that the world will end in 500 to 1000 years (an assumption that could be more transparent--see his footnote #13), among others.
So let's take our IAMs with salt, and encourage developers of the models to be as transparent as possible about their assumptions and how and why their models differ from each other. But let's also not forget that they have a place in this business, albeit perhaps a bit less than IAM builders might have you believe.
Schneider and Schneider and Lane also have nice critiques of IAMs.
Integrated assessment models (IAMs to practitioners) stitch together projections from climate models, energy sector models, agronomic crop models, models of other sectors of the economy, and partial or general equilibrium models that account for price and interactions with the broader economy to derive a more comprehensive evaluation of costs and benefits from climate change.
Pindyck raises some good criticisms about IAMs, or at least says out loud a lot of things that many economists have quietly said to each other. I'm glad he's bringing our varying assumptions and wildly varying cost-of-carbon estimates out into the open for all to see. Perhaps it will push us to make our modeling efforts a little more useful, or at least more transparent.
He's right to pick on false precision. But I wonder: has anyone really been fooled? My sense is no. One positive thing about these modeling efforts is that they allow us to see which assumptions are most critical. They are nice (black?) boxes for testing out the sensitivity of X on overall climate impacts. This might help us frame more reasonable discussion about the possibilities and what we should do. It might also help researchers focus future empirical efforts.
The extreme sensitivity of results to seemingly innocuous assumptions also shows how uncertain the impacts of climate change really are. Indeed, not long ago Pindyck published a paper in JEEM with results that are extremely sensitive to his assumption that the world will end in 500 to 1000 years (an assumption that could be more transparent--see his footnote #13), among others.
So let's take our IAMs with salt, and encourage developers of the models to be as transparent as possible about their assumptions and how and why their models differ from each other. But let's also not forget that they have a place in this business, albeit perhaps a bit less than IAM builders might have you believe.
Schneider and Schneider and Lane also have nice critiques of IAMs.
Comments
Post a Comment