The Fisher effect: a Kalman filter approach to detecting structural change

Abstract

This article uses quarterly data on short-run nominal interest rates and inflation rates over the last four or three decades collected from Australia, Japan, Malaysia and Singapore to test whether the Fisher relation has empirical support. Since meaningful Fisher effect tests critically depend on the integration and cointegration properties of the variables, we present some empirical evidence on these issues and we also apply the Kalman filter to estimate the time-varying parameters. The results show that the data are generally rejecting a full Fisher effect. This implies that nominal interest rates do not respond point-for-point to changes in the expected inflation rates. The possible reasons for the inability to detect a full Fisher effect are also discussed.

Acknowledgements

We would like to thank the editor of this journal and anonymous referees for their useful comments that resulted in improving the article. The usual disclaimer applies.

Notes

¹ See also Yuhn (1996), Thornton (1996), Choudhry (1966, 1997), Hamori (1997), Carneiro et al . (2002) and Million (2003).

² It should be mentioned that Harvey (1993) calls these kinds of models as structural models.

³ This methodology follows a similar algorithm applied by Hatemi-J(2002) to test for time-varying cointegration.

⁴ For an overview of the empirical tests for the presence of a Fisher relation, see Haliassos and Tobin(1990).