US inflation dynamics on long-range data

Abstract

In this article, we evaluate inflation persistence in the United States using long-range monthly and annual data. The importance of inflation persistence is crucial to policy authorities and market participants, since the level of inflation persistence provides an indication on the susceptibility of the economy to exogenous shocks. Departing from classic econometric approaches found in the relevant literature, we evaluate inflation persistence through the nonparametric Hurst exponent within both a global and a rolling window framework. Moreover, we expand our analysis to detect the potential existence of chaos in the data generating process, in order to enhance the robustness of our conclusions. Overall, we find that inflation persistence is high from 1775 to 2013 for the annual data-set and from February 1876 to May 2014 in monthly frequency, respectively. Especially from the monthly data-set, the rolling window approach allows us to derive that inflation persistence has reached to historically high levels in the post–Bretton Woods period and remained there ever since.

Keywords:

• inflation
• persistence
• Hurst exponent
• detrended fluctuation analysis
• Lyapunov exponent

JEL Classification:

• C14
• E31
• E60

Acknowledgements

We would like to thank two anonymous referees for their helpful comments. Of course any remaining errors remain our own. Vasilios Plaknadaras would like to thank Georgios Sarantitis for his insightful comments in the implementation of the DFA method.

Notes

¹ Due to space restriction, the interested reader is referenced to Mulligan and Koppl (2011) and the papers cited therein.

² For more information on the derivation of the test, the interested reader is referred to BenSaida and Litimi (2013).

³ Not surprisingly, unit root test results according to the ADF test (Dickey and Fuller, 1981), the Phillips–Perron test (Phillips and Perron, 1988) and the KPSS test (Kwiatkowski et al., 1992) suggested that both the monthly and the annual CPI series are I(1) in levels, but are stationary when converted to inflation rates. The details of these results are available upon request from the authors.

⁴ Numerical calculations are performed with the MATLAB code provided by Weron (2011).

⁵ All calculations were performed with the ‘nonlinear toolkit’ software of Ashley and Patterson (2000).

Additional information

Funding

Vasilios Plakandaras, Dr Papadimitriou Theophilos and Dr Periklis Gogas were partly financed for this research by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program ‘Education and Lifelong Learning’ of the National Strategic Reference Framework (NSRF) – Research Funding Program: THALES [MIS 380292]. Investing in knowledge society through the European Social Fund.