A SELECTIVITY CORRECTED TIME-VARYING BETA RISK ESTIMATOR Robert D. Brooks*, Jonathan Dark*, Robert W. Faff** and Tim R.L. Fry*** * Department of Econometrics and Business Statistics, Monash University ** Department of Accounting and Finance, Monash University *** School of Economics and Finance, RMIT Business ABSTRACT This paper explores two issues in beta risk estimation, namely, time variation of risk and the impact of thin trading (data censoring). Within a multivariate GARCH setting, the paper first conducts an analysis of the importance of assumptions made about the correlation structure. Collectively, the results of Monte Carlo analysis and an empirical application to a sample of individual Australia stock returns, demonstrate that it is preferable to allow for time variation in the correlation structure. The paper then develops a selectivity corrected time-varying beta risk estimator. The results of a Monte Carlo experiment demonstrate that the new estimator performs well in handling censored data. Finally, we confirm that when the new model is applied to our Australian dataset it successfully captures the impact of censoring and thin trading on time-varying beta risk. Keywords: Time-varying betas; GARCH; Censoring; Thin trading; Sample selectivity model JEL Classification: G12, C24 Acknowledgements: The authors acknowledge the financial support of Australian Research Council Discovery Grant DP0345680 – A Complex Systems Approach to Modelling Time-varying Risk in the Presence of Market Frictions. The authors wish to thank Diana Maldonado and Michael Gangemi for their work as research assistants on this project. The authors also wish to thank the participants at the 2005 Quantitative Methods in Finance conference and seminars at Deakin and ANU for their helpful comments on earlier versions of this paper. Correspondence to: Robert Faff, Department of Accounting and Finance, Monash University, VIC 3800, Australia, Phone: 61-3-99052387, Fax: 61-3-99052339, Email: robert.faff@buseco.monash.edu.au