Modelling & training > Our suite of models > E3ME: An energy-environment-economy model of Europe > The model's estimated equations
Econometric specification
Economic models consist of two types of equations: Identity relationships and behavioural relationships. Identity relationships are given by accounting identities, such as GDP being equal to the sum of its component parts, and are usually non-controversial. Behavioural relationships, however, are unobservable and are given by model parameters. These parameters may be calibrated (CGE models, see below), estimated (econometric models) or imposed by the model user. Typically they vary between countries, between sectors and in the short and long run.
Most conventional macroeconomic models which are operational in government describe short and medium-term economic consequences of policies but with a limited treatment of longer-term effects; this limits their ability to analyse long-term policies. On the other hand, Computable General Equilibrium (CGE) models have been widely used to analyse long-term E3 (energy-environment-economy) policies. CGE models specify explicit demand and supply relationships and enforce market clearing, and are therefore seen as desirable characterisations of long-term outcomes in which markets are assumed to be in equilibrium; for this reason they have been developed particularly in the US for the analysis of environmental regulation.
However, CGE models are not generally estimated by time-series econometric methods and they have not typically been subjected to rigorous historical validation, either in terms of the values of the model’s parameters or, more broadly, the underlying assumptions with respect to economic behaviour. They also typically tend to impose the dynamics of the model solution, and so cannot be used for historical validation of the overall model. In addition the analysis of short and medium-term impacts of policy changes tends to arise from the assumptions inherent in the model (including perfect knowledge and foresight, and optimal rational behaviour and expectations). These assumptions, and the models in which they are incorporated, are being increasingly questioned as to whether they provide an adequate representation of complex real-world behaviour.
E3ME combines the features of an annual short and medium-term sectoral model estimated by formal econometric methods with the detail and some of the methods of the CGE models, providing analysis of the movement of the long-term outcomes for key E3 indicators in response to policy changes. It is essentially a dynamic simulation model of Europe estimated by econometric methods.
The econometric specification of E3ME gives the model a strong empirical grounding and means it is not reliant on the assumptions common to Computable General Equilibrium (CGE) models, such as perfect competition or rational expectations. The econometric techniques used to specify the functional form of the equations are the concepts of cointegration and error-correction, allowing short-term dynamic (or transition) outcomes, moving towards a long-term trend. The dynamic specification is important when considering short and medium-term analysis (e.g. up to 2020) and rebound effects, which are included as standard in the model's results.
In brief, the estimation process involves two stages. The first-stage is a levels relationship, whereby an attempt is made to identify the existence of a cointegrating relationship between the chosen variables, selected on the basis of economic theory and a priori reasoning. For example, for employment demand the list of variables contains real output, real wage costs, hours worked, energy prices and measures of technological progress. If a cointegrating relationship exists, then the second stage regression is known as the error-correction representation, and involves a dynamic, first-difference, regression of all the variables from the first stage, along with lags of the dependent variable, lagged differences of the explanatory variables, and the error-correction term (the lagged residual from the first stage regression).
The main endogenous variables in E3ME are determined from functions estimated on historical time-series data on European economic indicators, energy and material use. There are a relatively small number of variables for which econometric equations are estimated; around 33 in all. However these variables may be disaggregated in two dimensions, for example in the equations for energy demand there are 19 energy users in 33 countries.
The econometric specification of E3ME is described in more detail in the model manual.
E3ME: An energy-environment-economy model of Europe E3ME purpose and design
Econometric specification
Sectors Recent developments Publications Licensing E3ME E3ME manual
For more information contact:
Hector Pollitt
Associate Director, International Modelling