RELATIVE PPP IN THE MEDIUM RUN

This paper tests the relative purchasing power parity (RPPP) hypothesi s on month-by-month, post-1972 data, and still obtains regression coef ficients that are close to unity. Two methodological aspects have cont ributed to this encouraging result. Firstly, although all RPPP tests a re vis-a-vis the USA, we nevertheless exploit our a priori knowledge a bout the implications of these USD-based equations on PPP relations be tween cross-rates. So, in a sense, we use all information implicit in all cross-rates too. Secondly, we selected an instrumental variable th at is specifically designed to cope with lead-and-lag effects in non-t raded vs traded-goods inflation. Our estimates indicate that most lead -and-lag effects seem to occur within a six-month window.