After looking over Krugman's Trade and Wages Reconsidered the only thought that struck me was this:
How do we reconcile the data with static general equilibrium models when the data generating process is neither static nor in equilibrium at the time it is recorded?
I had expected a paper which would have stated the following:
In the following model, volume of trade is endogenous. If the elasticity of substitution between skilled and unskilled labor is y and the volume of trade exceeds a, then wage inequality will be b.
I imagine such a proposition is out there somewhere but I'm not a follower of trade theory enough to go out looking for it and perhaps I should.