Long-Range Dependence in Daily Interest Rate

Abstract

We employ a number of parametric and non-parametric techniques to establish the existence of long-range dependence in daily interbank o er rates for four countries. We test for long memory using classical R=S analysis, variance-time plots and Lo's (1991) modi ed R=S statistic. In addition we estimate the fractional di erencing parameter using Whittle's (1951) maximum likelihood estimator and we shu e the data to destroy long and short memory in turn, and we repeat our non-parametric tests. From our non-parametric tests we And strong evidence of the presence of long memory in all four series independently of the chosen statistic. We nd evidence that supports the assertion of Willinger et al (1999) that Lo's statistic is biased towards non-rejection of the null hypothesis of no long-range dependence. The parametric estimation concurs with these results. Our results suggest that conventional tests for capital market integration and other similar hypotheses involving nominal ...