A study of the outward transmission of U.S. interest rate shocks.

Abstract

The primary object of this research is to analyze effects of a shock in U.S. interest rates on the interest rates for countries with whom the U.S. has close economic ties. Simulations are conducted to track the distinct effect of changes in U.S. rates. The data for this study are composed of daily observations for three-month maturity yields on five Eurocurrencies, which include the U.S. dollar, German mark, British pound, Japanese yen and Canadian dollar. The data source is the Bank of International Settlements (BIS). The sample period extends from January 3, 1980 to October 31, 1994.

Cointegration analysis is used to assess the stability of long-run interest rate relationship, while Granger causality analysis is used to assess the short-run relationship. Bivariate reaction functions are estimated, and based on these functions, simulations of the responses of foreign rates to a shock emanating from the United States are conducted. The study suggests methods by which the results may be introduced into the teaching of economic principles and money and banking courses.

The cointegration tests reveal no systematic interest rate relationships in the long run between most pairs of Eurocurrency rates. This finding may be attributed to the structural breaks, the nonstationarity of either expected exchange rate movement or the risk premium, or shifts in expected inflation between the various countries in this study. The Granger causality tests reveal contemporaneous effects of a U.S. rate shock on the other rates, and the presence of reverse causality from foreign Eurocurrency rates back to the U.S. Eurocurrency rates.

Stability tests show that, except for the Canadian Eurocurrency rates, there are comparatively brief periods within the 14 year sample period in which each relationship can be considered stable. Various explanations are offered for this phenomenon.

Reaction functions linking the foreign and domestic Eurocurrency rates were estimated for the full sample period and for the stable sample period. The simulation results were based on the functions estimated over the stable period.

The study concludes with a review of several issues regarding the teaching of financial integration in economic principles and money and banking courses. The simulations are the key element of that process.