The low volatility premium – An analysis of factor exposures of minimum variance strategies
Minimum variance strategies have gained significant traction especially since the global financial crisis. They aim at reducing or minimizing variance, i.e. the square of volatility as measured by standard deviation, or, in this case, price fluctuations of portfolio prices around their mean. As such, they are of much interest to risk-aware and risk-averse investors. While assets linked to investments that follow such strategies increased substantially in recent years, the theoretical foundation underlying minimum variance concepts has been around for decades, going back to Markowitz . In the years following his work, a vast number of empirical studies looked into characteristics of minimum variance portfolios with a focus on how to best implement such strategies in practice. In recent years, however, academia has shifted its focus to the explanation of the so-called low volatility factor.
The traditional Capital Asset Pricing Model (CAPM) explains asset returns in excess of the risk-free rate as compensation for systematic, i.e. non-diversifiable, risk. In a market equilibrium, investors would hold risky assets only if they received a premium that is positively related to the extent of systematic risk they take on. Hence, the higher an asset’s exposure to systematic risk, the higher the expected return. In violation of this explanation, however, empirical studies have found that stocks that exhibit comparatively low levels of risk tend to outperform riskier stocks. This phenomenon is often referred to as the ‘low volatility anomaly’.
Are minimum variance concepts that attempt to minimize variance, really a guarantee for outperformance relative to traditional market-capitalization-weighted benchmarks? We attempt to answer that question in this paper, by breaking down the performance of minimum variance strategies into the underlying systematic factors. We further isolate the performance contribution of the low-volatility premium, while simultaneously assessing contributions from other style factors identified in the related literature that arise as a consequence of the minimum variance optimization.
Why minimum variance strategies and not low volatility strategies? Minimum variance strategies typically aim at minimizing the variance of a portfolio. The construction of such a portfolio considers both stock price volatility as well as correlations among stocks. Hence, the strategy does not directly target the low volatility factor, unlike low volatility concepts that simply select stocks based on their price volatility. However, over the last decade or so, minimum variance strategies have become more popular, in part driven by the fact that their risk-adjusted returns have been higher than those of low volatility strategies in many cases. Therefore, we focus on minimum variance strategies in our analysis, rather than on the more heuristic low volatility strategies.
Our empirical analysis reveals that, relative to market-capitalization-weighted benchmarks, minimum variance strategies not only reduce volatility but also enable investors to harvest positive and statistically significant low-volatility premia. This observation holds true even after controlling for industry and country factors as well as for other style factors. However, the empirical results further show that selecting and weighting stocks in order to minimize portfolio risk leads to statistically significant exposures to other style factors versus the benchmark, sometimes with persistent negative performance contributions over time, thereby cannibalizing parts of the low-volatility premium. This observation is found to exist across geographies.
These findings are likely of great importance for investors while implementing minimum variance strategies. Investors need to be aware of interactions among factors in order to apply appropriate countermeasures. In this context, the empirical findings indicate that applying constraints to limit unintended style factor exposures relative to the benchmark may help reduce the negative performance contribution and its statistical significance. However, this typically comes at the cost of a lower reduction in portfolio risk and a reduction in the low-volatility premium.
 Markowitz, H. .
 See e.g. Behr et al.  and Chow et al.  for an analysis of the impact of constraints on minimum variance portfolios.
 The CAPM goes back to William F. Sharpe, John Lintner and Jan Mossin, who developed the model independent from one another in the 1960s. See e.g., Sharpe , Lintner  and Mossin .
 See e.g. Ramos and Hang .
 All of the performance attribution analysis is carried out using Axioma’s Portfolio Analytics system with attribution carried out using the Axioma AX-WW 2.1 World-Wide Equity Factor Risk Model.
Please fill the form below to attend :