Global models of world trade are often used as input into assessing the possible effects of liberalising trade. As simplifications of the economic processes – and the counterfactuals that they are designed to explain – they include many assumptions on which their results depend.
Some of the crucial assumptions in global models of world trade include how – and which – gains from trade are captured. One of the main features of global models is the Armington assumption which differentiates products according to their geographic origin and gives rise to certain types of gains from trade. Although some trade is differentiated, trade in homogenous products exists and is responsible for other types of gains from trade.
The purpose of this paper is to provide an intuitive explanation for a modification of the standard Armington specification of trade models, to represent gains from trade in differentiated and homogenous products.
A toy model illustrates that in an A-H-O model:
• terms of trade effects from trade liberalisation are smaller than in a standard Armington model, and
• an A-H-O database combined with standard GTAP code is a good approximation for full-blown A-H-O model.
Illustrative experiments with a 25-country GTAP aggregation indicate that the A-H-O specification can quadruple the estimated effects of reducing tariffs globally in manufacturing.