On measuring comparative advantage: Further comments

Abstract

Forstner and Murray's (BFM) recent remarks suggest confusion on their part about (i) the motivation for constructing an index of comparative advantage and (ii) the relation between the NCA concept and the definition of comparative advantage upon which my indices are based. As is often the case, much of the confusion can be traced to semantics. Nonetheless, this confusion leads BFM to erroneously question whether the NCA concept is useful for inferring comparative advantages and also to conduct an improper analysis of the bias in a "production intensity" index. The following remarks are intended to point out and resolve these areas of confusion. I believe that most of the recent debate over the NCA concept can be traced to differences in the interpretation of the word "index". Anyone wishing to conduct an empirical study of comparative advantage immediately confronts two questions: How is comparative advantage to be defined and what variable is to be chosen to represent it. Assuming that the answer to these questions leads to the use of trade data, one quickly discovers that a direct comparison of trade volumes is likely to be misleading since both countries and goods differ in their importance in world markets. This leads one to choose a variable that can be used to scale the variable representing comparative advantage. The number that results when the variable chosen to represent comparative advantage is divided by the "scale variable" is what I have called an "index of comparative advantage". In contrast, I think BFM take the word "index" to be synonymous with the word "proxy" and therefore interpret the phrase "index of comparative advantage" to mean the variable representing comparative advantage. The central issue addressed in Bowen [1983; 1985] was the choice of a theoretically consistent scale variable and not the choice of a variable to represent comparative advantage. In particular, both the "net trade" and "production" indices were derived in a model of net trade and thus net trade was necessarily the variable chosen to represent comparative advantage. Within that model, the scale variable Sik = Yi(Qwk/YJ was derived under the assumption of identical and homothetic preference (IHP). Seeking an interpretation of this variable that was consistent with the interpretation given to previous scale variables, the NCA concept was invoked to demonstrate that Sik could be interpreted as the production that would exist in a