Forecasting hedge fund volatility: a Markov regime-switching approach

Abstract

The article addresses forecasting volatility of hedge fund (HF) returns by using a non-linear Markov-Switching GARCH (MS-GARCH) framework. The in- and out-of-sample, multi-step ahead volatility forecasting performance of GARCH(1,1) and MS-GARCH(1,1) models is compared when applied to 12 global HF indices over the period of January 1990 to October 2010. The results identify different regimes with periods of high and low volatility for most HF indices. In-sample estimation results reveal a superior performance of the MS-GARCH model. The findings show that regime switching is related to structural changes in the market factor for most strategies. Out-of-sample forecasting shows that the MS-GARCH formulation provides more accurate volatility forecasts for most forecast horizons and for most HF strategies. Inclusion of MS dynamics in the GARCH specification highly improves the volatility forecasts for those strategies that are particularly sensitive to general macroeconomic conditions, such as Distressed Restructuring and Merger Arbitrage.

Keywords:

• hedge fund (HF) risk
• Markov switching (MS) model
• volatility forecasting

JEL Classification:

• G11
• G17
• G23

Acknowledgements

The authors thank the Editor, the anonymous referee, Luc Bauwens, Helmuth Chávez, Christine Choriat, Georges Gallais-Hamonno, Germán López, Yuliya Lovcha, Bertrand Maillet, Carlos Méndez, Anna Naszódi, Alejandro Pérez-Laborda, Hélène Raymond and participants of the following conferences: Forecasting Financial Markets 2009 Conference, Luxembourg, May 2009; Monetary Economics, Banking and Finance Conference, Orleans, June 2009; European Economics and Finance Society Conference, Warsaw, June 2009; 29th Annual International Symposium on Forecasting, Hong Kong, June 2009; XVII Spanish Finance Association Meeting, Madrid, November 2009. Additional thanks go to seminar participants at Universidad de Navarra and Universidad Francisco Marroquín for helpful comments and discussions on previous versions of this article. The first author acknowledges the research financing of the PIUNA project from Universidad de Navarra. The second author acknowledges financial support: S2009/ESP1685, ECO2009-14457-C04-03 and ECO2010-17625 of the Spanish Ministry of Science and Innovation.

Notes

The tables and figures not reported in this article are available from the authors electronically in separate appendices.

See the results in separate appendix available from the authors.

The results are available from the authors as a separate appendix.

Parameter estimates of the common factor model and the evolution of the common latent factors over the period 1990–2010 are available from the authors as a separate appendix.

Available from the authors as a separate appendix.

The expressions for the MS-AR and MS-GARCH covariance stationarity conditions and the computation of the expectations in the last two equations of the MS-GARCH formulation are available from the authors as a separate appendix.

See some details of the computation of the likelihood function are available from the authors as a separate appendix.

Figures on the evolution of the filtered probability of the first regime over the period 1990 to 2010, i.e. over t=1, …, T with the corresponding HF index excess return series are available from the authors in separate appendix.

The estimation results are available from the authors as a separate appendix.