Forecasting without Persistence

30 Pages Posted: 11 Jun 2012

See all articles by Christophe Boucher

Christophe Boucher

Université Paris I Panthéon-Sorbonne - Centre d'Economie de la Sorbonne (CES)

Bertrand B. Maillet

EMLyon Business School (Paris Campus)

Date Written: June 11, 2012

Abstract

The forecasting literature has identified three important and broad issues: the predictive content of explanatory variable is most of the times unstable over time, in-sample and out-of-sample results are often discordant and precise statistical inference with highly persistent predictors is challenging. In this paper, we simultaneously address these three issues, proposing to directly treat the persistence of forecasting variables before forecasting. We thus cut-out the low frequency components and show, in simulations and on real financial data, that this pre-treatment improves the predictive power of the predictor.

Keywords: forecasting, persistent regressors, detrending, expected returns

JEL Classification: C14, C53, G17

Suggested Citation

Boucher, Christophe and Maillet, Bertrand B., Forecasting without Persistence (June 11, 2012). Available at SSRN: https://ssrn.com/abstract=2081599 or http://dx.doi.org/10.2139/ssrn.2081599

Christophe Boucher (Contact Author)

Université Paris I Panthéon-Sorbonne - Centre d'Economie de la Sorbonne (CES) ( email )

106-112 Boulevard de l'hopital
106-112 Boulevard de l'Hôpital
Paris Cedex 13, 75647
France

HOME PAGE: http://ces.univ-paris1.fr/membre/boucher/christophe.html

Bertrand B. Maillet

EMLyon Business School (Paris Campus) ( email )

23 Avenue Guy de Collongue
Ecully, 69132
France