Doctoral thesis

Smart Beta & Factors Investing

Low Volatility Factor - Analysis of the Anomaly In Different Interest Rate Environment. We analyzed portfolios sorted by realized volatility on the US equity universe ove ...

Author(s) :

Bacem Rezgui, PhD

Chief Executive Officer at Reevalo, France

Abstract :

Low Volatility Factor - Analysis of the Anomaly In Different Interest Rate Environment. We analyzed portfolios sorted by realized volatility on the US equity universe over a long historical period (from 1966 to 2016) and confirm the low volatility anomaly; that low-risk portfolios show a higher risk-adjusted return than a high-risk portfolio. We found that, during months of long term interest rate decrease, the low volatility portfolio shows high non-explained return (four-factor alpha around 14.1% annualized) compared to -3.3% (annualized) for the high volatile portfolio. Controversially, during months of rising interest rate, the high-volatility portfolio’s alpha is positive +12.6% (annualized), and low-volatility alpha is negative around -8% (annualized). We show that applying sector-neutral adjustments diminishes alpha. We provide evidence that equity portfolio sorted by realized risk load to interest rate risk premium that we price around 6% annualized. We propose a modified version of the CAPM model that conditions on interest rates (rising and falling), explaining better returns than CAPM, Fama-French-Carhart, and Fama-MacBeth models. The result (historical and conditional analysis) are robust to the choice of risk metric and weighting scheme.

Keywords: Low volatility, CAPM, Fama-French-Carhart model, Interest rates.

Smart Beta - A Long Horizon Investigation. In this paper, over a long historical period (more than 50 years), we investigate alternative weighting schemes or "smart beta" strategies (to traditional capital-weighted (CW)). On the US equity market, risk-adjusted return (Sharpe Ratio) is better than the capital-weighted portfolio, even after applying conservative transaction costs of 0.5%. We provide evidence that the CW portfolio suffers from poor diversification, whatever the risk measure considered. We identify important style bias of smart beta (compared to CW) to value and small-cap. Adding volatility factor to the Fama-French-Carhart, we fully explain the performance of optimized portfolios: global-minimum-variance (GMV), equal-risk-contribution (ERC), and max-diversification (MD). For non-optimized portfolios: Equally-weighted (EW), inverse-volatility (IVO), and inverse-variance (IVA), alpha still positive and significant but not statistically different from CW portfolios. We decompose portfolios’ performances into weighting and active rebalancing effect. We confirm the literature finding that constant rebalancing adds value. We show that optimized portfolios do not benefit from this constant rebalancing. Finally, we identify that small-cap exposure is mainly due to the weighting scheme’s choice. In contrast, value exposure is a direct consequence of rebalancing.

Keywords: Smart Beta, Weighting, Rebalancing, Factors, Bonus, Fama-French, CAPM, EW, GMV, MD, ERC, CW, IVO, IVA.

 

Date : 24/11/2021
Thesis Committee :

Supervisor: Lionel Martellini, EDHEC Business School 

External reviewer: Robert Kosowski (Imperial College London)

Other committee member: Emmanuel Jurczenko, EDHEC Business School

See Also

EDHEC PhD IN FINANCE FORUM 2022'S DETAILED AGENDA IS NOW AVAILABLE
News
- 27-04-2022
The fifth EDHEC PhD IN FINANCE FORUM will be held online on 7 June 2022. Discover the...
ALUMNI STORY – VIJAY VAIDYANATHAN (PhD IN FINANCE 2012): INTERNATIONAL HORIZONS
News
- 08-03-2022
Vijay Vaidyanathan, a Californian alumnus of EDHEC Business School’s PhD in Finance,...
Professor Gianfrate, contributing scientist of the IPCC report on climate change
News
- 02-03-2022
The Intergovernmental Panel on Climate Change (IPCC) released its latest Climate Report...
What were EDHEC-risk’s top 10 most read articles in 2021?
News
- 23-12-2021
A year in research: focus on retirement investing (goal-based investing,...