RegFin models
In a CGE simulation the researcher analyses how a policy or other shock affects the economy
Static or dynamic approach
Static CGE models don't reveal the path of the economy from the benchmark equilibrium to the new equilibrium when a shock enters. If the researcher is interested in the path, he should use a dynamic model. Including dynamics is a quite demanding and laborious process. The dynamic version of the RegFin (RegFinDyn) model is of a recursive type. The model is solved for a sequence of several years, say 2009-2020. The researcher can then follow the adjustment of the economy to a policy or other shock on a year-to-year basis.
The following figure represents a typical static simulation framework.
Social Accounting Matrix
The starting point of the analysis is the data. The Social Accounting Matrix (SAM) is constructed for one so-called benchmark year. This means that the CGE models are lighter in their data requirements, so their maintenance costs are lower compared with econometric macro models. The SAM describes the money flows between sectors (primary, industry, services) whether they represent intermediate or final demand. The cost structure of the sectors is also present. The SAM also shows the factor income flows of the consumers and the structure of their final demand. The public sector tax revenues and subsidies from/to different sectors and consumers are included in the SAM. Finally, there are money flows of domestic and foreign exports and imports. The SAM thus describes the overall structure of the economy and shows the relationships and linkages between the decision makers.
Parameter values
The analysis then continues, and the values for the parameters are specified. These are divided into two parts: primary and secondary. The primary parameters are typically substitution elasticities, and their values are taken from estimates of the econometric research or typically used values from the literature. The secondary parameters are different efficiency and distribution parameters. Their values are dependent on the values of the primary parameters and are calibrated to a level that reproduces the benchmark data as a base case solution of the CGE model. The model is well behaved when it passes this replication check. One should always keep in mind that the results of any numeric economic model are usually quite sensitive with respect to the parameter values chosen. Sensitivity tests are therefore important.
Mathematical structure
Mathematically, a CGE model is a system of linear and non-linear definition, equilibrium and behaviour equations. The model is built by using the GAMS/MPSGE or GEMPACK software, which are high-level programming languages. The Walrasian assumption is that prices adjust and equilibrate the economy after a policy or other shock.
Comparative statistics
The next step is to parameterise the changes in economic and other policies. After the changes have been fed into the model, the numeric algorithm will find new equilibrium prices and quantities for the production factors (capital, labour and land) and all goods and services. At this point the researcher has two solutions for the model: the benchmark equilibrium and the new equilibrium that prevails after the changes in economic policy have been made. It is then possible to do comparative-static calculations to find out how much the changes in the policy have affected the key variables, such as GDP, employment, income, consumption, investments etc.
Sensitivity tests and policy recommendations
A good CGE analysis includes sensitivity tests using different values for the elasticities. It is quite probable that the results of the simulations are sensitive with respect to the elasticity values. It is, however, a usual outcome that the sign of a result does not change. The researcher can now calculate the confidence intervals for his simulation results. Sensitivity tests should also be done with respect to the size of the shock.
The last phase in the CGE analysis is to give economic policy recommendations. Through model simulations the researcher can provide the decision makers with solid calculations and recommendations that will improve their ability to make better decisions.