The Cointegration of International Interest rates

Cointegration is a new way of looking at the relationship between economic variables. If rNL and rDE are interest rates on the same instrument in the Netherlands and Germany respectively we would expect that they would be the same i.e. (1) rNL - rDE = 0 At any instant in time there may be a differential between the two interest rates i.e. (2) rNL - rDE = d If the action of the market through arbitrage or other processes tends to reduce d towards zero, whenever a non-zero d arises, the process will tend to move towards the relationship (1), even though (1) may never hold exactly. A relationship of this form is known as a cointegrating (or equilibrium) relationship. Such relationships may take many forms, may involve more than two variables and are often found in economics. The present analysis concerns the forms of cointegrating relationships that hold between international interest rates and the conclusions that can be drawn as regards the international convergence of interest rates. In a risk-free world with fixed exchange rates and perfect capital mobility interest rates will be the same in all countries. Any variations that occur will tend to vanish after a period of time. This equality of interest rates will define an equilibrium or cointegrating relationship between them. In a less than perfect world there can be other cointegrating equilibrium relationships. In Bretton Woods days, for example, the interest rate differential will have been determined mainly by capital controls and tended to some constant value. In such cases modern statistical analysis would have found equilibrium or cointegrating relationships even though interest rates had not converged. If we consider a period of time made up of three parts - the first in which such cointegrating relationships held and the second a time of transition to a third in which interest rates had converged. In this instance there is no unique equilibrium relation between the interest rates over the extended period and statistical analysis should fail to find a cointegrating relationship. Thus, the interpretation of findings of cointegration between interest rates does not imply convergence to zero differentials. The existence of convergence from one equilibrium with non-zero differentials to one with revised differentials implies that cointegration tests should fail. If we are living in a world of increasing financial liberalisation and harmonisation of world economies, then it is likely that such equilibrium relationships are changing and that we will not be able to find cointegration even though there may be a tendency to converge. The literature which has been reviewed in this regard has not addressed the points raised in the above discussion. Hence, it is not surprising that the results of the empirical evidence reviewed have presented divergent and sometimes ambiguous results. The relevance of international interest rate linkages to macroeconomic policy implementation suggests that the cointegration of long and short interest rates across boundaries warrants further attention.