Covered interest parity (CIP) is tested in conventional and Fisherian
frameworks using three-month treasury bill rates denominated in eight
currencies against the dollar. Results of cointegration and coefficien
t restriction tests are slightly more favourable for conventional CIP,
while model selection tests are more strongly in favour of this versi
on of the theory. One of the conclusions derived from the empirical ev
idence is that arbitrage, rather than the integration of financial and
commodity markets, is the force maintaining this international parity
condition.