Three essays on modelling and testing the conditional risk premium

Hoang, Trinh Anh Khoa (2016). Three essays on modelling and testing the conditional risk premium PhD Thesis, UQ Business School, The University of Queensland. doi:10.14264/uql.2016.66

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s4159653_phd_submission.pdf Thesis (open access) application/pdf 2.23MB 0
Author Hoang, Trinh Anh Khoa
Thesis Title Three essays on modelling and testing the conditional risk premium
School, Centre or Institute UQ Business School
Institution The University of Queensland
DOI 10.14264/uql.2016.66
Publication date 2016-01-31
Thesis type PhD Thesis
Supervisor Robert Faff
Mamiza Haq
Total pages 118
Language eng
Subjects 1502 Banking, Finance and Investment
Formatted abstract
This thesis, a collection of three essays, focuses primarily on modelling conditional expectations and investigates the central research question: is the ex-ante market risk premium always positive?

Many asset pricing studies have focused on testing linear restrictions imposed by asset pricing models and largely ignore another important restriction: the positivity of the market risk premium. Aiming to enhance our understanding in this area, the first essay applies a novel market-based measure of conditional expected return, namely, the implied cost of capital, and examines whether or not the ex-ante risk premium is always positive. Employing economically meaningful information variables and a multiple inequality constraints framework, I find evidence that this positive condition is violated not only in the US, but also in other major international markets including Japan, Italy, and Germany. In stark contrast, when the realised return is used to proxy for the expected return, there is insufficient evidence against the hypothesis due to a high degree of noise embedded in the realised return proxy. Accordingly, I argue for the use of the implied cost of capital in modelling the time varying expected return.

In the second essay, I further examine the positive risk premium hypothesis. Specifically, I introduce a new two-stage method, involving Principal Component Analysis and Boosted Regression Tree to model conditional expected return. With these techniques, I address potential pitfalls associated with existing methods of capturing the true identity of investors’ information sets, and how investors use the information in forming expectations. Consistent with the first essay’s finding, the positivity condition of the risk premium is violated in the US. Collectively, the evidence suggests the rejection of the Conditional Capital Asset Pricing Model.

The implied cost of capital is appealing in its own right, yet its validity as a proxy for conditional expected return is open to debate. Returning to this measure, the third study investigates when and to what extent this estimate deviates from true expectations. In a simulation study allowing for time varying discount rates, I find that due to the constant term structure assumption embedded in the research method, the variation of the implied cost of capital is significantly lower than that of the true expected return. This feature reveals that in a standard regression, the economic significance interpretation of the coefficient is no longer appropriate. Additionally, I show analytically that when analyst forecasts are biased and/or unable to capture the full information of cash flow expectations incorporated in the market price, the derived implied cost of capital is contaminated. Moreover, there exists a real issue regarding spurious regressions involving the implied cost of capital.
Keyword Positive risk premium
Implied cost of capital
Boosted regression tree

Document type: Thesis
Collections: UQ Theses (RHD) - Official
UQ Theses (RHD) - Open Access
Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Wed, 20 Jan 2016, 00:03:36 EST by Trinh Anh Hoang on behalf of Learning and Research Services (UQ Library)