Integrated Assessment Models: PNNL GCAM

The IPCC is in the process of developing the next generation of emission scenarios. The scenarios are used to generate anthropogenic radiative forcings that drive Global Circulation Models. The so-called Integrated Assessment Models (IAMs) are being developed by government agencies and research laboratories in the U.S., the Netherlands, Japan and Austria. It is worth trying to understand and simplify these models because they are critical to conclusions the IPCC is drawing about climate change.

In the U.S., the Pacific Northwest National Laboratory (one of US DOE's ten national laboratories) has developed the Global Change Assessment Model (PNNL GCAM) which will be used to generate scenarios for the next release of IPCC documentation.

The PNL GCAM model is essentially the same partial equilibrium model you saw in introductory economics textbooks. Basically what these models (and their big brothers, general equilibrium models) do is compute long-run equilibrium prices. Partial equilibrium models look at one market (energy) while general equilibrium models look at all prices. Actual prices are assumed to be deviations from rationally determined long-run prices.

In partial equilibrium models, population growth, technology and gross domestic product are given exogenously. All the model does is calculate energy prices. The existing capital stock is assumed fixed.

It's hard to understand how a partial equilibrium model would be of much use in studying global climate change. Wouldn't there be impacts on food prices from oil prices and fertilizer prices? Wouldn't food prices have an impact on population and production? Wouldn't energy prices have an impact on technology and the existing capital stock e.g., replacing gasoline with electric vehicles?

These models were originally used to help the DOE study energy demand. They were recruited by the IPCC for the study of climate change.

PNNL also has a computable general equilibrium (CGE) model or Second Generation Model (SGM) that is described here. I'll talk about that model in a future post. CGE models are not necessarily an improvement.