Climate Change 2001:
Working Group III: Mitigation
Other reports in this collection

7.6 Modelling and Cost Assessment

7.6.1 Introduction

The costs of climate policy are assessed by various analytical approaches, each with its own strengths and weaknesses. This section considers first the modelling options currently used to assess the costs of climate policy, and then the key assumptions that influence the range of cost estimates. The focus is on the general conceptual elements of cost assessment and on an evaluation of how model structures and input assumptions affect the range of cost estimates.

7.6.2 Classification of Economic Models

The models presented here are described and discussed in more detail in Chapter 8, in which a review of the main literature on these models is presented. However, it is useful to present an overview of the main modelling techniques applied in this kind of analysis here.

Input–Output Models
Input–output (IO) models describe the complex interrelationships among economic sectors using sets of simultaneous linear equations. The coefficients of equations are fixed, which means that factor substitution, technological change, and behavioural aspects related to climate change mitigation policies cannot be assessed. IO models take aggregate demand as given and provide considerable sectoral detail on how the demand is met. They are used when the sectoral consequences of mitigation or adaptation actions are of particular interest (Fankhauser and McCoy, 1995). The high level of sectoral disaggregation, however, requires strong restrictions that limit the validity of the model to short runs (5–15 years).

Macroeconomic (Keynesian or Effective Demand) Models
Macroeconomic models describe investments and consumption patterns in various sectors, and emphasize short-run dynamics associated with GHG emission reduction policies. Final demand remains the principal determinant of the size of the economy. The equilibrating mechanisms work through quantity adjustments, rather than price. Temporary disequilibria that result in underutilization of production capacity, unemployment, and current account imbalances are possible. Many macroeconomic models are available. They implicitly reflect past behaviour in that the driving equations are estimated using econometric techniques on time-series data. As a consequence, macroeconomic models are well suited to consider the economic effects of GHG emission reduction policies in the short- to medium-horizon.

Computable General Equilibrium Models
CGE models construct the behaviour of economic agents based on microeconomic principles. The models typically simulate markets for factors of production (e.g., labour, capital, energy), products, and foreign exchange, with equations that specify supply and demand behaviour. The models are solved for a set of wages, prices, and exchange rates to bring all of the markets into equilibrium. CGE models examine the economy in different states of equilibrium and so are not able to provide insight into the adjustment process. The parameters in CGE models are partly calibrated (i.e., they are selected to fit one year of data) and only partly statistically or econometrically determined (i.e., estimated from several years of data). Hence it is difficult to defend the validity of some of the parameter values.

Dynamic Energy Optimization Models
Dynamic energy optimization models, a class of energy sector models, can also be termed partial equilibrium models. These technology-oriented models minimize the total costs of the energy system, including all end-use sectors, over a 40–50 year horizon and thus compute a partial equilibrium for the energy markets. The costs include investment and operation costs of all sectors based on a detailed representation of factor costs and assumptions about GHG emission taxes. Early versions of these models assessed how energy demands can be met at least cost. Recent versions allow demand to respond to prices. Another development has established a link between aggregate macroeconomic demand and energy demand. Optimization models are useful to assess the dynamic aspects of GHG emissions reduction potential and costs. The rich technology information in the models is helpful to assess capital stock turnover and technology learning, which is endogenous in some models.

Integrated Energy-System Simulation Models
Integrated energy-system simulation models are bottom-up models that include a detailed representation of energy demand and supply technologies, which include end-use, conversion, and production technologies. Demand and technology development are driven by exogenous scenario assumptions often linked to technology vintage models and econometric forecasts. The demand sectors are generally disaggregated for industrial subsectors and processes, residential and service categories, transport modes, etc. This allows development trends to be projected through technology development scenarios. The simulation models are best suited for short- to medium-term studies in which the detailed technology information helps explain a major part of energy needs.

Partial Forecasting Models
A wide variety of relatively simple techniques are used to forecast energy supply and demand, either for single time periods or with time development and varying degrees of dynamics and feedback. The main content is data on the technical characteristics of the energy system and related financial or direct cost.

Limits of Economic Models Taxonomy
The macroeconomic and CGE approaches can be further classified as “top-down” methodologies, while the technology-rich dynamic optimization/partial equilibrium, simulation, and partial forecasting approaches can be considered “bottom-up” approaches. It is also noted that the dynamic optimization/partial equilibrium, simulation, and partial forecasting approaches are sometimes collectively referred to as the family of engineering–economic models.

While useful, this taxonomy has its limits. First, differences in parameter values among the models within a given category may be more significant than the differences in model structure across categories. Second, many differences emerge between the theory underlying a particular model group and the actual models. Third, most models are hybrid constructions linked to provide greater detail on the structure of the economy and the energy sector (Hourcade et al., 1998). A hybrid approach sheds light on both the economic and technological aspects of reducing energy-related CO2 emissions, but it does have its drawbacks. Consistent results require that a hybrid approach remove all the inconsistencies across the linked models. This process is often cumbersome and time consuming.

Other reports in this collection