Computable General Equilibrium Models in Transport Appraisal

At a glance

A computable general equilibrium (CGE) model is an economic analysis tool with an economy-wide focus that estimates changes in key economic indicators at the national level, for individual industries and often regions or small areas, as a result of external changes or policy changes. CGE models were developed for the purpose of estimating impacts of policy changes with widespread impacts across the economy and high flow-on impacts such as changes to taxes and tariffs, taking into account interactions between all markets (hence the term ‘general equilibrium’). In contrast, cost–benefit analysis (CBA) was developed to estimate changes in the community’s welfare as the result of initiatives that have their primary impact in a single, localised market. CBA considers only the market directly affected by the initiative and closely related markets (hence, the term ‘partial equilibrium’).

A general and a partial equilibrium analysis will give the same welfare change result if there are no ‘distortions’ (prices different from marginal social costs) in markets other than those considered in the partial analysis. Sources of distortions include taxes, subsidies, unpriced externalities, regulatory restrictions on land-use, and imperfect competition. CBA takes account of distortions in closely related markets (for example, competing and complementary transport modes and routes) and in other markets through ‘wider economic benefits’ and ‘higher value land-use benefits’. Beyond that, the welfare effects in more distantly related markets are thought to be small because the shifts in demand or cost curves would be small in those markets and the distortions through-out the Australian economy are, for the most part, not large. This view, however, is contested and evidence is needed to resolve the question.

In the meantime, the ATAP Guidelines takes the view that, for the vast majority of transport assessments, the partial equilibrium analysis of CBA is adequate because it gives a good approximation of the results that would be obtained from a general equilibrium assessment. The ATAP Guidelines does not discourage use of CGE models to estimate benefits of initiatives where the model has a suitable welfare measure and level of disaggregation, but recommends that the CGE analysis be accompanied with a standard CBA, the results compared, and detailed explanations given for major differences. CGE models forecast changes in a wide range of economic indicators (GDP, private consumption, investment, exports, employment and industry outputs) at the macro-economic level and at the level of individual industries. If a regional or spatial CGE model is used, these results will be available at the regional or small area level as well as at the national level.  Estimates of GDP changes from a CGE model and net benefits from a CBA are not interchangeable, nor additive. However, a CGE model can be set up to estimate changes in economic welfare consistent with a CBA. The more recent spatial CGE models, including urban CGE models, have welfare measures. They also feature high zonal disaggregation, transport costs and travel times, and land-use markets. They can contribute to CBAs by making land-use change forecasts.  CGE models can provide useful information about the distributional impacts of transport initiatives and estimate nationwide greenhouse gas emissions.

The ATAP Guidelines make no recommendation about circumstances where CGE modelling ought to be undertaken. CGE modelling is viewed as an optional extra. Proponents of transport initiatives should weigh up whether the additional information obtained is worth the cost of the modelling. CGE modelling would be considered only for very large transport initiatives with significant geographical and distributional impacts Transport appraisers are not expected to build or operate CGE models by themselves. Outsourcing to specialist modellers is the usual practice. To avoid the CGE model being treated as a ‘black box’, the report from the modellers should describe the model used and the changes made to the model to represent the project or policy change (called shocks) in detail, and interpret the results with qualifications. Input-output (I-O) analysis might be considered a less expensive alternative to CGE models to supplement CBA. Results of I-O analysis have major limitations and are likely to be biased upward. I-O analysis should not be used in project appraisal except possibly for small projects considered in isolation (not groups of projects) in areas of high unemployment where economic stimulus is a policy objective.