The Armington Assumption and the Size of Optimal Tariffs

There has been commentary on the seeming success of the world trading system responding to the large shock of the 2008 financial crisis without an outbreak of retaliatory market closing. The threat of large retaliatory tariffs and fears of a 1930s style downturn in trade have been associated with numerical trade modelling which project post retaliation optimal tariffs in excesses of 100%. In the relevant numerical modelling it is common to use the Armington assumption of product heterogeneity by country. Here we argue and show by numerical calculation that the widespread use of this assumption gives a large upward bias to optimal tariffs, both first step and post retaliation, relative to alternative homogenous good models used in trade theory.