Questions about FTA Models
Attached is a simple price elasticities model of world trade that
functions reasonably well in GAMS so long as the model's fta-parameter
is set equal to 0. If fta=1, the model attempts to solve for world
trade and prices in the presence of a free trade area formed by a
subset of k countries. The fta variant of the model is causing me
problems, I presume because it is nonlinear and discontinuous.
I have tried toying with a dnlp version of the model, but the attached
'square' mcp version seems to yield the same (unsatisfactory) results.
Within the logic of the model, if net regional trade in a commodity is
negative, then the regional to world price ratio should be greater
than unity. However, the model seems to follow this logic only on the
first iteration. After the first iteration, the model finds (declares)
a soln that does not observe the expected relationship between net
regional trade and relative regional to world prices.
Is there some way to dependably enforce this condition on the model
soln, within the framework of the present model? If not, could you
kindly outline how I might try to setup the model more appropriately
in Gams to obtain a proper soln.
Sorry, but the problems you are encountering are because of problems
in the model formulation. You are attempting to solve a
complementarity problem through a roughly implemnted enumeration
method which is unlikely to work reliably.
- Here is a fairly quick writeup which
introduces the model formulation.
- Here is an implementation of the
simple model which is formulated as a square system of equations. To
run this model you will need in addition a copy of the data file in gdx or xls
- Here is an implementation of the FTA model
in complementarity format. If you wish to return results in Excel
format, you may also wish to take a copy of results.xls
- Here is an revised implementation of the FTA model
which corrects some reporting problems and provides a loose constraint
preventing cross-hauling on intra-FTA trade flows.