SV mixture models with application to S&P 500 index returns

• 成果类型: Article; Proceedings Paper
• 署名作者: Durham, Garland B.
• 署名单位: University of Colorado System; University of Colorado Boulder
• 刊物名称: JOURNAL OF FINANCIAL ECONOMICS
• ISSN/ISSBN: 0304-405X
• DOI: 10.1016/j.jfineco.2006.06.005
• 发表日期: 2007
• 页码: 822-856
• 关键词: stochastic volatility; stock returns; forecasting

摘要

Understanding both the dynamics of volatility and the shape of the distribution of returns conditional on the volatility state is important for many financial applications. A simple single-factor stochastic volatility model appears to be sufficient to capture most of the dynamics. It is the shape of the conditional distribution that is the problem. This paper examines the idea of modeling this distribution as a discrete mixture of normals. The flexibility of this class of distributions provides a transparent look into the tails of the returns distribution. Model diagnostics suggest that the model, SV-mix, does a good job of capturing the salient features of the data. In a direct comparison against several affine-jump models, SV-mix is strongly preferred by Akaike and Schwarz information criteria. (c) 2007 Elsevier B.V. All rights reserved.