A comparison of CAPM variants : which market proxy and information set best predicts future asset returns?

Davies, Matthew D. (2003). A comparison of CAPM variants : which market proxy and information set best predicts future asset returns? Honours Thesis, School of Business, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE17291.pdf Full text application/pdf 5.29MB 7
Author Davies, Matthew D.
Thesis Title A comparison of CAPM variants : which market proxy and information set best predicts future asset returns?
School, Centre or Institute School of Business
Institution The University of Queensland
Publication date 2003
Thesis type Honours Thesis
Total pages 105
Language eng
Subjects 15 Commerce, Management, Tourism and Services
Formatted abstract This thesis determines the optimal specification of the single factor CAPM for pricing Australian industry and size-sorted portfolios. Previous studies have left open the issue of comparative performance of various CAPM specifications in forecasting the systematic returns on assets. This thesis addresses the issue by testing an unconditional rolling regression model against a conditional generalised autoregressive conditional heteroskedasticity (GARCH) model of variance and covariance, using both a domestic and international market proxy. Measuring the forecast errors before and after deregulation also tests the pricing impact of the deregulation of the Australian capital market. For all portfolios, the domestic proxy is superior to the international benchmark. However, a significant finding is that the pricing performance of the domestic CAPM is largely the same regardless of whether conditioning information is incorporated into the portfolio beta estimates. Conversely, the international market proxy does not have explanatory ability without the conditional GARCH specification. Differences in the pricing error measures pre- and post-deregulation are associated with a shift in volatility rather than change in the power of the CAPM variants. These results suggest that the parsimony of the unconditional domestic model would be preferred unless a more effective conditional model is identified.

 
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 624 Abstract Views, 7 File Downloads  -  Detailed Statistics
Created: Mon, 22 Nov 2010, 13:04:28 EST by Muhammad Noman Ali on behalf of The University of Queensland Library