Long-Run Exchange Rate Modeling: A Reply to Moosa

Abstract

N HIS comment on my paper, Imad Moosa makes some interesting points regarding recent tests of the purchasing power parity (PPP) hypothesis. More specifically, he raises some issues that may be interpreted as general points on the PPP literature and some that are specific comments on my paper. In this note I shall address only these points that apparently conflict with remarks made in my paper. It is correct to say, as Moosa argues in the second paragraph of his comment, that if prices in equation (17) of MacDonald (1995) are I(1) then a pure first difference version of PPP (with an error term added) would represent a violation of conventional PPP and support for efficient markets PPP. I agree, and spent some time making this point in the paper. Moosa argues that I should not have used the expression Cassellian PPP because "Cassel's view of PPP ... is simply an extension of the quantity theory of money applied to the case of an open economy." I am in total agreement that in his writings Cassel placed considerable emphasis on the monetary aspects of PPP (and indeed this view is noted on p. 441 of my paper). However, this is not inconsistent with my label of Cassellian PPP. First, it is clear from Cassel's writings (see, for example, Officer (1976)) that it is arbitrage in internationally traded goods that links prices across countries (why else would Cassel spend time discussing the effect impediments to international trade could have on PPP holding?). If there is no arbitrage why would exchange rates necessarily track their PPP value rather than some other value? It is also very clear from Cassel's writings that this arbitrage need not necessarily force PPP at all times (see Officer (1976) for a fulsome discussion of this point). My term "Cassellian PPP" was simply a convenient way of making the point that a proponent of traditional PPP would not necessarily expect exchange rates always to be at their PPP values.