Asymmetric Predictive Abilities of Nonlinear Models for Stock Returns: Evidence from Density Forecast Comparison

Yong Bao; Tae-Hwy Lee

doi:10.1016/S0731-9053(05)20021-X

Login

Welcome:
Guest

Search for:

Browse:

Product Information:

Resources:

Advances in Econometrics

ISSN: 0731-9053
Series editor(s): Thomas B. Fomby, R. Carter Hill, Ivan Jeliazkov, Juan Carlos Escanciano and Eric Hillebrand

Subject Area: Economics

Content: Series Volumes | Current Volume RSS

Options: To add Favourites and Table of Contents Alerts please take a Emerald profile

Document request:
Asymmetric Predictive Abilities of Nonlinear Models for Stock Returns: Evidence from Density Forecast Comparison

Document Information:
Title:	Asymmetric Predictive Abilities of Nonlinear Models for Stock Returns: Evidence from Density Forecast Comparison
Author(s):	Yong Bao, Tae-Hwy Lee
Volume:	20 Editor(s): Thomas B. Fomby, Dek Terrell ISBN: 978-0-76231-273-3 eISBN: 978-1-84950-388-4
Citation:	Yong Bao, Tae-Hwy Lee (2006), Asymmetric Predictive Abilities of Nonlinear Models for Stock Returns: Evidence from Density Forecast Comparison, in Thomas B. Fomby, Dek Terrell (ed.) Econometric Analysis of Financial and Economic Time Series (Advances in Econometrics, Volume 20), Emerald Group Publishing Limited, pp.41-62
DOI:	10.1016/S0731-9053(05)20021-X (Permanent URL)
Publisher:	Emerald Group Publishing Limited
Article type:	Chapter Item
Abstract:	We investigate predictive abilities of nonlinear models for stock returns when density forecasts are evaluated and compared instead of the conditional mean point forecasts. The aim of this paper is to show whether the in-sample evidence of strong nonlinearity in mean may be exploited for out-of-sample prediction and whether a nonlinear model may beat the martingale model in out-of-sample prediction. We use the Kullback–Leibler Information Criterion (KLIC) divergence measure to characterize the extent of misspecification of a forecast model. The reality check test of White (2000) using the KLIC as a loss function is conducted to compare the out-of-sample performance of competing conditional mean models. In this framework, the KLIC measures not only model specification error but also parameter estimation error, and thus we treat both types of errors as loss. The conditional mean models we use for the daily closing S&P 500 index returns include the martingale difference, ARMA, STAR, SETAR, artificial neural network, and polynomial models. Our empirical findings suggest the out-of-sample predictive abilities of nonlinear models for stock returns are asymmetric in the sense that the right tails of the return series are predictable via many of the nonlinear models, while we find no such evidence for the left tails or the entire distribution.