11.4.5 Differences between models
Research has continued to focus on differences in various cost estimates between models (Weyant, 2000; Weyant, 2001; Lasky, 2003; Weyant, 2003; Barker et al., 2006a; Fischer and Morgenstern, 2006). Weyant (2001) argues that the five major determinants of costs are: projections for base case GHG emissions; climate policy (flexibility, for example); substitution possibilities for producers and consumers; the rate and process of technological change; and the characterization of mitigation benefits. Turning to the base case, he notes the importance of assumptions about population and economic activity, resource availability and prices, and technology availability and costs. The key policy feature is flexibility, in other words whether trading over companies, nations, gases, and time is allowed. Substitution possibilities are governed by assumptions about the malleability of capital, economic foresight, and technology detail. Technology modelling includes assumptions about whether technological change is endogenous or exogenous, and whether technology costs drop with increasing use of technologies. Finally, mitigation benefits may be included in varying degrees in different models.
The factors accounting for differences between cost estimates can be divided into three groups: features inherent in the economies being studied (for example, high substitution possibilities at low cost), assumptions about policy (such as the use of international trading in emission permits, or the recycling of auction revenues), and simplifying assumptions chosen by the model builders to represent the economy (how many sector or regions are included in the model). The first two sets of factors can be controlled by specifying the countries and time-scales of the mitigation action, and the exact details of the policies, as in the EMF-16 studies. However, differences in modellers’ approaches and assumptions persist in the treatment of substitution and technology. The various factors can be disentangled by a meta-analysis of published finding (this may include an analysis of analyses). This technique was first used in this context by Repetto and Austin (1997) in a mitigation-cost analysis of GDP costs for the US economy. Fischer and Morgenstern (2006) conduct a similar meta-analysis dealing with the carbon prices (taken to be the marginal abatement costs) of achieving Kyoto targets reported by the EMF-16 studies and discussed in the TAR (Weyant and Hill, 1999).
The crucial finding of these meta-analyses is that most of the differences between models are accounted for by the modellers’ assumptions. For example, the strongest factor leading to lower carbon prices is the assumption of high substitutability between internationally-traded products. Other factors leading to lower prices include the greater disaggregation of product and regional markets. This suggests that any particular set of results about costs may well be the outcome of the particular assumptions and characterization of the problem chosen by the model builder, and these results may not be replicated by others choosing different assumptions.
Like earlier studies, the comparison of model results in Barker et al., (2006a) emphasizes that the uncertainty in costs estimates comes from both policy and modelling approaches as well as the baseline adopted. Uncertainty about policy is associated with the design of the abatement policies and measures (flexibility over countries, greenhouse gases and time) and with the use of carbon taxes or auctioned CO2 permits to provide the opportunity for beneficial reforms of the tax system or incentives for low-carbon innovation. In addition, targeted reductions in fossil-fuel use resulting from climate policies can yield benefits in terms of non-climate policy e.g. reductions in local air pollution. Uncertainty about the modelling approaches is associated with the extent to which substitution is allowed in terms of backstop technology, whether the economy responds efficiently (in terms of the use of CGE models), and whether technological change is assumed to respond to carbon prices, the topic of the next section. Uncertainty about the baseline is associated with assumptions adopted for rates of technological change and economic growth, and future prices of fossil fuels.