Not logged in.

Contribution Details

Type Master's Thesis
Scope Discipline-based scholarship
Title Variance-based Risk Overlays
Organization Unit
Authors
  • Guillaume Bourquenoud
Supervisors
  • Alexandre Ziegler
  • Thorsten Hens
Language
  • English
Institution University of Zurich
Faculty Faculty of Business, Economics and Informatics
Number of Pages 84
Date July 2019
Zusammenfassung This thesis analyzes the impact of applying variance-based risk overlays to the U.S. and ex-U.S. aggregate equity markets. For both equity markets, three categories of variance indicators were constructed: (1) the realized variance measures, (2) the model-free implied variance measures, and (3) the variance risk premiums. All three categories were computed in a symmetric and asymmetric way, creating in total six variance indicators to time equity markets. These six variance indicators were selected to develop variance-based risk overlays due to prior evidence of their ability to predict return or variance. Specifically, the academic literature has recognized that realized and implied variance measures have predictive power for future variance, while the symmetric and asymmetric variance risk premiums have been reliable predictors of future returns. Once the variance indicators were computed, their in- and out-of-sample predictive powers for fu-ture return and variance were assessed. Through this analysis, we confirm that realized and implied variance measures have strong predictive ability for future 1-month U.S. and ex-U.S. equity market variance, with out-of-sample R2s of up to 45%. Meanwhile, the predictive ability of variance risk premiums for future return is only found in the U.S. equity market and peaks for 3-month future excess return, with significant out-of-sample R2 of 4.5%. To evaluate the financial impact of variance-based risk overlays, we back-test different risk overlays employing return and/or variance forecasts obtained directly or indirectly from the variance indicators. Considering realistic implementation conditions, we demonstrate that variance-based risk overlays have historically improved the risk adjusted performance. Indeed, assuming transaction costs of 25 bpbpss, we found that unleveraged variance-based risk overlays: • Applied to the SPX from 1998 to 2017 would have produced annualized Sharpe ratios ranging from 0.31 to 0.39, compared to 0.23 for a passive investment. • Applied to the EFA from 2008 to 2017 would have produced annualized Sharpe ratios ranging from 0.22 to 0.39, compared to 0.09 for a passive investment.
Export BibTeX