Causal linkages between US and Eurodollar interest rates: further evidence

Abstract

This study examines causal linkages between US and Eurodollar interest rates during 1983–2002. Recursive cointegration analysis shows that a stable cointegration relationship between the two interest rates emerges only since the early 1990s, when the Fed used federal funds rate targeting and eliminated the reserve requirement on Eurocurrency deposits. The study further reveals that bidirectional causality exists between the two rates over the period of 1993 to 2002, while unidirectional causality from Eurodollar rate to the US rate is found to exist over the period of 1983 to 1991. These findings consistently support increased interest rate linkages especially since the early 1990s.

Acknowledgements

The authors would like to thank an anonymous referee for several helpful comments, which substantially improved the article. The first author acknowledges financial support in the form of the College of Business summer research grant at Prairie View A&M University.

Notes

¹ A similar controversy exists in some studies involving nonUS markets. In particular, Kaen and Hachy (1983) documented unidirectional causality from domestic to external markets in UK. Lo et al . (1995) showed evidence of a strong causality running from Euroyen to Japanese interest rate with a weak feedback from the latter to the former.

² Though it might be more desirable, intraday interest rate data have not been used in this line of research using cash market rates, probably due to their unavailability to most researchers. Also note that the use of daily data would ensure this study to be comparable with most of previous studies, where much controversy exists.

³ Results of several conventional unit root tests are available on request.

⁴ The estimation with the appropriate heteroscedasticity and autocorrelation-consistent covariance matrix was also applied to address any possible conditional heteroscedasticity and autocorrelation problems. The appropriate χ²-statistic was used (instead of the F-statistic) to test the above null hypothesis and the conclusion remained qualitatively the same.