On real interest rate persistence: the role of breaks

Abstract

The role of structural breaks in long spans of ex-post real interest rates for 10 industrialized countries is studied. First, the persistence of the real interest is assessed with newly proposed low-frequency tests of Müller and Watson (2008). Second, the test of Leybourne et al. (2007) for a change in persistence of a time series is applied to the real interest rate. The results show that real interest rates over the full sample period do not fit a covariance-stationary or unit-root model, nor a fractionally integrated, near-unit-root or local-level model. Instead, the persistence of real rates changes over time and there are periods when the real rate is covariance-stationary and other periods when it follows a unit-root process.

Keywords:

• real interest rates
• persistence of a time series
• breaks in persistence

JEL Classification:

• E43
• C22

Acknowledgements

Most of the research for this article was carried out while the author was visiting the Econometrics Institute at the Warsaw School of Economics in Warsaw, Poland. The author wishes to thank the Institute for its kind hospitality and seminar participants at the Institute for their helpful comments. The author also thanks the anonymous referees for their useful comments. The usual caveat applies.

Notes

¹ Neely and Rapach (2008) provided details on the role of the real interest rate in various theoretical models and I draw on their paper in this paragraph.

² There is a related literature that studies the relationship between nominal interest rates and inflation, based on the Fisher hypothesis (e.g. Haug et al., 2011). I do not pursue this line of research here.

³ See also Bai and Perron (2003, fn. 15, p. 17).

⁴ Lai considered only one endogenous break but studied eight industrialized and eight developing countries.

⁵ Fisher (1930, p. 43) argued that the one-for-one relationship of nominal interest rates and (expected) inflation is a long-run process so that long-term real interest rates over long spans would seem appropriate, though I do not directly test the Fisher hypothesis in this article.

⁶ I apply the demeaned version and set , following Elliott et al. Also, the lag order for the test is chosen with sequential t-tests and a 10% level of significance, following Ng and Perron (1995).

⁷ I use a 5% level of significance throughout the article.

⁸ The reason for the different behaviour of the French real interest rate may reflect the differences in monetary policies compared to those of the other countries in my sample. For example, Sims (1992) found that prices responded very strongly, statistically significantly and persistently to a nominal interest rate shock in France, in contrast to the responses for other countries in his sample, except for Japan. Morton and Wood (1992) documented idiosyncrasies of French monetary policies after World War II. Markets were tightly regulated, unlike in other major developed countries (except for Japan). France had extensive exchange rate controls, regulated interest rates and had ceilings on credit growth, among other measures, until liberalization and deregulation began in the 1980s. Similarly, French monetary policy stands out as different in the interwar period between World Wars I and II. Irwin (2010) pointed out that France accumulated gold at a much more rapid pace than the United States and other countries, and ‘…increased its share of world gold reserves from 7 percent to 27 percent between 1927 and 1932 and effectively sterilized most of this accumulation. This gold hoarding created an artificial shortage of reserves and put other countries under enormous deflationary pressure’ (Irwin, 2010, Abstract, title page).

⁹ The parameter , except for the two smallest samples when for Sweden over period 1922–1950 and for Denmark over the period 1971–2006, for which and 0.30, respectively.

¹⁰ See, for example, Mitchener and Weidenmier (2010).