Long-Memory Modeling and Forecasting: Evidence from the U.S. Historical Series of Inflation
Document Type
Article
Publication Date
10-27-2020
Publication Title
Studies in Nonlinear Dynamics and Econometrics
First page number:
1
Last page number:
5
Abstract
We report the results of applying several long-memory models to the historical monthly U.S. inflation rate series and analyze their out-of-sample forecasting performance over different horizons. We find that the time-varying approach to estimating inflation persistence outperforms the models that assume a constant long-memory process. In addition, we examine the link between inflation persistence and exchange rate regimes. Our results support the hypothesis that floating exchange rates associate with increased inflation persistence. This finding, however, is less pronounced during the era of the Great Moderation and the Federal Reserve System's commitment to inflation targeting.
Keywords
Long memory; Time-varying persistence; US inflation; Wavelet analysis
Disciplines
Econometrics | Economics | Social and Behavioral Sciences
Language
English
Repository Citation
Boubaker, H.,
Canarella, G.,
Gupta, R.,
Miller, S. M.
(2020).
Long-Memory Modeling and Forecasting: Evidence from the U.S. Historical Series of Inflation.
Studies in Nonlinear Dynamics and Econometrics
1-5.
http://dx.doi.org/10.1515/snde-2018-0116