Abstract
The latest generation of multi-sectoral trade models has abandoned the conventional strategy that regards the similarly named exports of different countries as homogeneous products. Armington's assumption, that the products of different countries compete as imperfect substitutes, has become the theoretical rationale for an alternative, heterogeneous modeling procedure. This paper describes some problems with the assumption and develops a more rigorous framework for the heterogeneous strategy. The processes that lead to heterogeneity in trade are modeled explicitly: comparative advantage theory is applied at a sub-sectoral level to explain the detailed structure of international specialization. Implications of the new framework for empirical estimation are also explored.