|Working Group III: Mitigation|
|Other reports in this collection|
9.4 Why Studies Differ
This section consolidates the explanations for the different findings in both the macro studies reviewed in Chapter 8 and the sectoral studies in this chapter. It extends and complements the methodological discussion in the SAR (Hourcade et al., 1996, pp. 282-92), particularly in the role of assumptions leading to differing results.
In assessing the economy-wide effects of mitigation, considerable use has been
made of top-down models (macroeconomic, general equilibrium, and energy-engineering),
while specific sectoral studies use both top-down and engineering-economic bottom-up
models. Critical differences in the results come from the type of model used,
and its basic assumptions. Repetto and Austin (1997), in a meta-analysis of
model results on the costs of mitigation for the USA, show that 80% of predicted
impacts come from choice of assumptions. They find that four assumptions are
critical in leading to lower costs of mitigation. These are that:
They conclude that under reasonable assumptions, the predicted economic impacts
from the models for the USA in stabilizing CO2 emissions at 1990
levels through to 2020 would be neutral or even favourable.
Most early studies are focused on the costs, rather than on the benefits of mitigation11. More recently, top-down modellers have studied the impact of using the revenues collected from carbon taxes (or from auctions of carbon permits) to correct economic distortions in some sectors of the economy (typically to reduce taxes on labour, taxes on incomes and profits, or taxes on investment).9.4.1 The Influence of Methods
The adoption of top-down or bottom-up methods makes a significant difference to the results of mitigation studies (see 8.2.1 and 8.2.2 for discussion and results). In top-down studies the behaviours of the economy, the energy system, and their constituent sectors are analyzed using aggregate data. In bottom-up studies, specific actions and technologies are modelled at the level of the energy-using, GHG-emitting equipment, such as power-generating stations or vehicle engines, and policy outcomes are added up to find overall results. The top-down approach leads easily to a consideration of the effects of mitigation on different broad sectors of the economy (not just the energy and capital goods sectors), so that the literature on these effects tends to be dominated by this approach.
Table 9.10 compares the methodologies. They have a fundamentally
different treatment of capital equipment and markets. Top-down studies have
tended to suggest that mitigation policies have economic costs because markets
are assumed to operate efficiently and any policy that impairs this efficiency
will be costly. Bottom-up studies tend to suggest that mitigation can yield
financial and economic benefits, depending on the adoption of best-available
technologies and the development of new technologies. Some hybrid models include
both approaches (see Laroui and van Leeuwen, 1995, for an example).
There are two main types of macroeconomic models used for medium- and long-term
economic projections12: resource allocation models (i.e. CGE) and time-series
econometric models. Their main differences being the assumptions made about
the real measured economy, aggregation, dynamics, equilibrium, empirical basis,
and time horizons, among others.
The main characteristic of CGE models is that they have an explicit specification
of the behaviour of all relevant economic agents in the economy. In the mitigation
applications they have usually adopted assumptions of optimizing rationality,
free market pricing, constant returns to scale, many firms and suppliers of
factors, and perfect competition in order to provide a market-clearing equilibrium
in all markets. Econometric models have relied more on time-series data methods
to estimate their parameters rather than consensus estimates drawn from the
literature. Results from these models are explained not only by their assumptions
but also by the quality and coverage of their data. It is usually argued that
CGE models are more suitable for describing long-run steady-state behaviour,
while econometric models are more suitable for forecasting the short-run. However,
the models have increasingly incorporated long-run theory and formal econometric
methods, and several now include a mix of characteristics, from both resource
allocation and econometric models; see Jorgenson and Wilcoxen (1993), McKibbin
and Wilcoxen (1993, 1995), Barker and Gardiner (1996), Barker (1998b) and McKibbin
et al. (1999).
Other reports in this collection