Predicting the pattern of international trade in the neoclassical model: a synthesis (Q834726)

The analytical backbone of the paper is the notion of predictability (prediction set): given a well-specified set \(\Omega\) of conceivable events, the prediction set is a subset of \(\Omega\). Different theories generate different prediction sets. If prediction sets for two theories are equal, then the theories are not discernible. As a starting point of trade analysis it is observed that the exchange patterns can be (at least partially) determined based on a country's data without reference to information about world/other countries -- this explains why Ricardo was right, when he predicted bilateral trade pattern on the grounds of just one country data. Then the formulations of prediction sets are presented for single-economy/multi-country trade and with number of goods from two through continuum, and two production factors with CRS technology represented by production functions. In case of single country formulations the predictions are based on autarky prices of factors/commodities. These prices are not usually observable, and it is proposed to avoid this problem by applying integrated equilibrium formulations, where trade decisions are made on the basis of observable production factors prices. The basic message of the paper is that 1) when the autarky prices are observable, then they wholly determine the predicted pattern of trade; 2) in integrated equilibrium models specialization in trade reflects the efficiency of production (cost minimization). In all cases the sets of conceivable trades and the predicted sets are represented either by hyperplanes or by -- usually positive -- half spaces containing the origin (or their intersections). Many useful diagrams accompany the analysis.