Investors are worried about the high valuations of stocks in general and low-volatility stocks in particular. And so are we! In relative terms, low-volatility stocks have become more expensive during the last two years, but it’s not the first time. It happened first in 2008 and again in 2011.
Concerns over low-volatility’s valuation are nothing new. Low-volatility stocks have been trading at higher valuation multiples than the global equity market for years. The MSCI World Minimum Volatility Index, for example, has been more expensive than the broader market since 2009 and as early as 2012 we addressed valuation concerns in our Robeco research paper ‘Enhancing a low-volatility strategy is particularly helpful when generic low volatility is expensive’. This report argues that the current post-2008 environment is similar to the post-1930s era, when generic low-volatility stocks were also ‘expensive’.
During such periods, which can stretch into decades, there are large performance differences between generic and enhanced low-volatility strategies. When low-volatility is expensive (about one third of the time), low-volatility stocks that score well in terms of valuation and momentum outperform generic low-volatility stocks by a hefty 6% per year.
Recently, the valuation topic received more attention when Rob Arnott, founder and chairman of Research Affiliates, published a paper with the ominous title ‘How Can “Smart Beta” Go Horribly Wrong?’. In it, he states that “The unsurprising reality is that many of the new factors deliver ‘alpha’ only because they’ve grown more expensive”. He claims that valuation has been a major driver of smart beta returns over the short and even the longer term and so factor returns, net of changes in valuation levels, are actually much lower than recent returns suggest.
Arnott concludes that if rising valuation levels account for most of a factor’s historical excess return such returns may not be sustainable in the future. In a response to Arnott’s claims, in his ‘My Factor Philippic’, Clifford Asness, co-founder of AQR Asset Management, states that he finds factor valuations moderately expensive, but not as expensive as Arnott. He asserts that multiple expansion is not the sole driver of factor returns.
To correct returns for multiple expansion, we use a method that is similar to that of Arnott but instead of using only the price-to-book ratio (P/B), we take four additional multiples: price-earnings (P/E), forward price-earnings, dividend yield (DY) and EBITDA to enterprise value (EV). This method is more robust as it does not rely solely on one specific value metric. At each moment in time we calculate the equal-weighting relative (%) difference between these value scores for a low-volatility portfolio and for the market portfolio.
The MSCI World Index is fixed at 100% throughout the period to control for market movements. Using this methodology, Figure 1 shows the relative valuation of the MSCI Minimum Volatility World Index and Robeco Global Conservative Equities over the period September 2006 - July 2016.
This ten-year period covers the life of our Conservative Equity strategy since it was launched. Back in 2006 the fund was about 15% cheaper than the market. When we presented our new strategy then, many potential clients thought it was just another value strategy, since low-volatility investing was not widely known in the market. In addition, based on their previous experience in 2001-2003, many investors wrongly assumed that value always offers protection. With hindsight MSCI made an excellent decision in launching their low-volatility index in April 2008. When the crisis hit that same year, it became clear to everyone that low-volatility stocks offered downside protection, while value stocks failed to do so on this particular occasion.
As a result of the good relative performance, the multiples of both the Conservative Equities fund and the Minimum Volatility Index quickly expanded by about 20% during 2008. After multiples contracted in the following years, we witnessed another increase in multiples during the sovereign debt crisis of 2011 (in Greece and the European periphery). This was also followed by a period when multiples contracted again. Since 2013 we have again observed a gradual increase in relative valuations. Anyone who believes in the merits of value investing, including us, should be worried about generic low-volatility stocks becoming more expensive.
But what the figure also shows is that Robeco Conservative Equities is currently priced at a 5% discount to the market, while the MSCI Minimum Volatility Index is priced at a 15% premium. A valuation gap of 20%.
The MSCI World Minimum Volatility Index and our global developed Conservative Equities strategy have shown time-varying multiple expansion during the past ten years. The net effect has contributed positively to past returns. Over the 2006-2016 period, the multiples of both low-volatility strategies have expanded by roughly 20% compared to the market, which can be translated into roughly 2% additional performance per year. To be more precise, 1.7% of the return of Conservative Equities can be attributed to this effect and 1.9% of the MSCI Minimum Volatility Index. Figure 2 shows the annualized excess returns of both low-volatility strategies, with and without adjusting for multiple expansion.
Even when adjusted for multiple expansion, our Conservative Equities strategy still shows a substantial excess return over the MSCI World. This is not bad for a strategy where the main objective is to reduce risk. Most of the excess return of the MSCI Minimum Volatility Index can be attributed to multiple expansion. Since risk-reduction is also a benefit to the clients, we should asses the performance of low-volatility strategies on a risk-adjusted basis.
In order to do this, in Figure 3 we show the excess return in combination with the realized volatility over the past ten years. Both strategies where effective in reducing risk. Conservative Equities and the Minimum Volatility Index reduced volatility by 27% and 28% respectively. The risk-adjusted outperformance was therefore higher than the simple excess return. Jensen’s alpha (CAPM alpha) was 4.71% for Conservative Equities and 3.32% for the Minimum Volatility Index.
When we adjust for multiple expansion, returns are lower as shown in Figure 4. If we also correct for market beta, the Minimum Volatility Index still shows a Jensen’s alpha of 1.33% per year, while Robeco’s global developed Conservative Equities strategy then delivers a Jensen’s alpha of 2.94% per year.
We find that a proportion of low-volatility strategy returns can be attributed to multiple expansion. However, this does not explain all the alpha. The Minimum Volatility Index and Conservative Equities both reduce risk and show better risk-adjusted returns than global equities before and after correction for the multiple expansion of the past ten years.
We do agree with Robert Arnott that valuation is an important factor to look at and to consider when evaluating the performance of strategies. We also agree with Clifford Asness that one should consider multiple valuation ratios and that more factors matter than value alone. For example, besides value we also include risk factors like distress and momentum in our active low-volatility strategy.
Ever since their launch ten years ago, our Conservative Equities strategies have taken multiple value and momentum factors into account when selecting the best low-volatility stocks. Currently, the valuation of Robeco’s global developed Conservative Equities portfolio is on average 20% lower than that of the Minimum Volatility Index based on a robust mix of five valuation multiples (P/E, Forward P/E, DY, EBITDA/EV and P/B). This offers our active low-volatility portfolio a substantial margin of safety compared to a generic low-volatility portfolio. For Robeco’s Emerging Markets Conservative strategy this valuation gap is 25%.
The valuations of generic low-volatility strategies are currently relatively high, but this has happened on several occasions since 2006. At this moment, Conservative Equities has a 20% margin of safety compared to generic low-volatility stocks.
Over the past ten years, the Robeco Conservative Equity approach has lived up to its promise to deliver equity-like returns with less risk. The ten-year real-life alpha is significant, also when corrected for multiple expansion. We believe that it is necessary to include valuation to achieve superior long-term returns. A previous long-term study on this topic, with data going back to the 1920s, supports an enhanced low-volatility approach which includes valuation and momentum. Especially when generic low-volatility stocks are relatively expensive, as they are today, it is important to mind the valuation gap.