Calculations in AnalyseGE versus similar ones in TAB files

The question as to what exactly is being calculated by AnalyseGE (given that the Coefficient values used are pre-simulation ones) is a rather technical issue. We recommend that you do not think too much about until you have used the program a few times. Your natural instincts as to the meaning of what is being calculated will be more than adequate for the purposes of analysing simulation results.

The following example aims to explain the small differences you might find if you evaluate the same expression using AnalyseGE or via an equation in a TAB file.

Consider the following equation from ORANIG.

Equation E_x3 # Source-specific commodity demands #

(all,c,COM)(all,s,SRC)

x3(c,s)-a3(c,s) = x3_s(c) - SIGMA3(c)*[ p3(c,s)+a3(c,s) - p3_s(c) ];

If you wish to analyse simulation results in the context of this equation, you will probably be interested in the relative sizes of the expansion term x3_s(c) and the substitution term (which is the rest of the RHS). You might add the following statements to your TAB file:

Variable (Change) (All,c,COM)(All,s,SCR) sub3term(c,s) ;

Variable (Change) (All,c,COM)(All,s,SCR) lhsx3(c,s) ;

Variable (Change) (All,c,COM)(All,s,SCR) exp3term(c,s) ;

Equation (Linear) E_sub3term (All,c,COM)(All,s,SRC)

sub3term(c,s) = - SIGMA3(c)*[ p3(c,s)+a3(c,s) - p3_s(c) ];

Equation (Linear) E_sub3term (All,c,COM)(All,s,SRC)

exp3term(c,s) = x3_s(c) ;

Equation (Linear) E_lhsx3 (All,c,COM)(All,s,SRC)

lhsx3(c,s) = x3(c,s) - a3(c,s) ;

If so, you would find that the simulation results for the "variables" sub3term, exp3term are slightly different from what AnalyseGE reports if you ask it to evaluate the relevant expressions. [Of course the two sets of numbers would be identical if you carried out a Johansen simulation.] However, the differences between these two sets of numbers would be relatively small in most cases. Certainly the numbers produced by AnalyseGE should be more than sufficiently accurate for the purposes of simulation analysis.

Secondly, in the example above, you would find that the results for the variables sub3term and x3term add exactly to the results for lhsx3 (even after a multi-step simulation). But If you evaluated in AnalyseGE the corresponding 2 expressions (substitution and expansion terms) you would find that their sum was slightly different (perhaps by a few percent) from what the results AnalyseGE gives for evaluating the expression on the LHS of the equation. [The fact that they do not add up exactly to the LHS is because the equation above is a linearized one and so should not be satisfied exactly after a multi-step calculation.]

Although the LHS and RHS are not equal for many equations when you decompose them in AnalyseGE, they should always be equal for various types of equations (adding-up equations) which occur frequently in GE models. Examples include market clearing equations and equations adding up values or quantities.

URL of this topic: www.copsmodels.com/webhelp/analysege/hc_ancalcs.htm

Link to full GEMPACK Manual

Link to GEMPACK homepage