Many are the times where use of general equilibrium methods is being met by harsh criticism. Mostly it happens when the associated model results concerning some policy reform do not provide politicians or voters with what they want or had expected. Sometimes it happens through more fundamentally-based reservations against the modeling strategy per se. Critics taking either point of departure support each other in envisaging general equilibrium models as fictional and idealized Utopian worlds. Hence, they are not to be taken seriously (and they are often claimed to be used merely to promote some free-market agenda).
As I see it, the critics often overlook (deliberately or not) that general equilibrium models are perfectly able of portraying strongly malfunctioning economies. Of course, they are fictional (as they are models), but they are definitely not Utopian. A recent example caught my attention today. In a World Bank publication, Economic Premise, Lúcio Vinhas de Souza writes on “An Initial Estimation of the Economic Effects of the Creation of the EurAsEC Customs Union on Its Members.” (pdf file). In the note, he attempts to provide an estimate of the GDP effects on the recent formation of a customs union within the Eurasian Economic Community (consisting of former Soviet states Belarus, Kazakhstan, Kyrgyzstan, the Russian Federation, and Tajikistan). He uses the GTAP model (which is available in the public domain), which is a computable general equilibrium model. The results are a nice example of how second-best analysis can provide answers, which I doubt any back-of-the-envelope/ Keynesian-cross analysis would adequately provide.
The result is that the harmonization of tariffs within this group of countries will reduce GDP for all in a range of experiments. The reason is that those countries that reduce tariffs (in particular Kazakhstan) are already trading “too much” under the existing conditions. The harmonization therefore exaggerates existing distortions. I am not a trade economist, so I cannot critically comment on the results as such, but I can guarantee that the model does not portray a Utopian world I would like to live in.