Risk Premia Strategies - Equity Markets
The Art of Selectively Harvesting Attractive Risk Premia
The value risk premium:
  • Value strategies are some of the best known stock selection strategies with origins dating back the work of Ben Graham and David Dodd. The value premium essentially consists of buying undervalued stocks and selling expensive ones, using some form of fundamental valuation metric, with the most commonly cited being low versus high price to book value, as described by Fama and French in their paper.
    Given the irrefutable logic of the argument that buying low and selling high ought to be profitable, the value premium is probably the least disputed of all equity risk premia. However, harnessing this is not as easy one might think. Long periods of negative performance
    are common and losses associated with the strategy can be particularly severe, and worst of all usually occurring during periods of economic stress. For example, since Fama and French published their paper over 20 years ago, the factor has averaged a return of 3.1% per year in the US but has shown a negative return over the last five years (-0.3%) and lost 20% during the recent financial crisis.
  • Numerous academic papers have investigated the value premium and provided alternative explanations.
    • Supporters of the efficient market hypothesis (Fama & French being among them), argue that the positive excess return earned from value companies is compensation for the higher risk (e.g. distress risk, ability to adapt in difficult economic environments, etc.), hence investors are paid a premium for this additional risk that most often materializes during periods of economic stress.
    • Detractors of the efficient market hypothesis point to behavioral biases such as over-reacting to past performance and extrapolating historical growth too far into the future as potential sources of the value premium. We suggest that both explanations are not mutually exclusive anc can co-exist, in that there is a premium for buying problem stocks (and therefore having to suffer near-term losses whilst that problem is resolved) and also that there is a premium for avoiding richly valued glamour stocks.
    • Numerous academic papers have investigated the value premium and provided alternative explanations.
  • Our value risk premium definition
We define value using the equal-weighted quintile score of a set of five traditional value factors, which have all been associated with positive returns in academic literature. All factors are taken relative to the median value of the sector, plus where the metric is not applicable (i.e. free cash flow to price for financial companies), we exclude it from the calculation:
  1. Book to Price
  2. Earnings to Price
  3. One-Year Forward Earnings to Price
  4. EBITDA to Enterprise Value
  5. Free Cash Flow to Price
After the average score is calculated, we rank all companies in the FTSE World universe and define the top quintile as the long portfolio and the bottom quintile as the short portfolio.

    • A reward for suffering losses
The goal of value strategies is obviously to buy undervalued stocks, although the profile of those undervalued stocks can vary significantly. A value portfolio might invest in companies that have suffered significant price declines and are having difficulties. These difficulties can be company-specific or related to the macro situation at the time. In both cases, the success of the value strategy will be determined by the ability of these companies (and the investor) to sustain losses and eventually fix their problems. Market sentiment is also very important as it affects the degree to which the market is discounting the ability of these companies to bounce back.

Therefore, it is not surprising that we typically see the best performance from value strategies during periods of improving sentiment and economic recovery. For example, from March to November 2009 value strategies gained north of 40% (on a long-short basis). On the other hand, during a downswing the returns of the strategy suffer significant drawdown, with value strategies usually losing more than 20% during 2007/2008. Potential mechanisms for timing entry and exit points for value strategies would warrant a book in themselves but ideas might include valuation dispersion, absolute valuation, levels of equity volatility and an overall view on the macro cycle.

    • A reward for being dull
Value portfolios might also include companies that are simply out-of-favour, as seemingly they do not offer the same excitement and growth opportunities which investors think they can get elsewhere. Unlike more cyclical value opportunities, these companies are not necessarily riskier but are quite often very well run businesses that have been left undervalued courtesy of their ‚non-glamour‛ status or are just classified as dull.

For example, during the dot-com bubble, value strategies (again on a long-short basis) suffered a -20% drawdown as the strategy avoided most TMT stocks (a loss that put many value-managers out of work at the time), but then they rallied 70% the first 12 months after the bubble collapsed. Understanding the exposures of a value portfolio for different types of company is clearly important when assessing the risks (and potential rewards) of the strategy.

  • Jegadeesh and Titman (1993) demonstrated the positive performance of previous winners versus previous losers. Since then, momentum has been one of the most commonly applied factors in stock, country, sector and style selection as well as in strategies across other asset classes. In its basic form, momentum strategies simply buy stocks with the best historical returns and short stocks with the worst returns, thus trying to capture trends in the market. Typically, they look at the last three to 12 month returns, skipping the last month to avoid short term reversal effects.

As with the value effect, there is a continuing debate about possible explanations of the momentum effect. With efficient market hypothesis supporters failing to convincingly tie momentum profits to some sort of compensation for additional risk, behavioural finance explanations have been more prominent. The most popular ones draw from behavioural biases such as overconfidence and loss aversion and link momentum profits to under- or over-reactions to company-specific news.

Momentum investing came under pressure after the great momentum crash of 2009, when a traditional momentum strategy (12-month lagged by one-month) lost more than 60% in the space of a few months. Whilst this clearly provided some ammunition in favour of a risk-based explanation, momentum stock selection during that period was mainly driven by the macro environment rather than capturing any reaction to company-specific news. As we have discussed before, during periods when macro conditions influence stock performance, momentum becomes a bet on the underlying macro story. As a result, if the macro direction changes abruptly (as it has a tendency to do), momentum will incur heavy losses.

Since 2009 researchers have been proposing risk/macro adjusted momentum metrics to mitigate the impact of macro and better isolate momentum effects. Whilst we recognise that it is always easier to go back and fix a problem after the event, not learning from past mistakes is unwise. QMS Advisors utilizes residual momentum approaches to investing, as it mitigates some of the macro effects and offers a far more appealing risk/return profile than adjusted momentum returns.
  • Instead of looking at the overall return of a stock, residual momentum focuses on the stock-specific part, i.e. the part that cannot be explained by the beta of the stock and market performance, hence diminishing overall market and macro effects. In order to calculate residual momentum, we first estimate the below equation using 36-month rolling returns and then calculate the residual monthly returns (ϵit). Residual momentum is defined as the cumulative residual return over the last 12 months, skipping the most recent month:
rit = αi + βitrmt + ϵit

where rit is the total return of stock i, rmt is the total return of the stock's local market index and αi , βi the parameters to be estimated from the regression.

We have also found that avoiding shorting stocks that have fallen too much can further improve the profile of momentum strategies. We utilize a "drawdown filter" using the drawdown of the stock’s price (i.e. the percentage from its prior one-year peak) and exclude stocks that have fallen by more than 70%. After excluding these stocks, we rank companies in quintiles based on their residual momentum and define the long and short portfolios using the top and bottom quintiles.

Our backtested momentum factor has averaged 9% per year since 1995. Nevertheless, the strategy has still suffered significant losses during the period following the dot-com bubble (2000/2001) and post March 2009. During the latter, our momentum strategy has seen its worst performance, with an 18% drawdown (which would have been more than 60% without controlling the macro risk). When comparing the performance of momentum to that of value during those periods, we see that they followed contrasting patterns with momentum doing well before 2000 and 2009 and falling afterwards whilst value was struggling before 2000 and 2009 but strongly recovered thereafter. The negative correlation of value and momentum during such volatile periods is the main reason why this combination has been so commonly applied in the quant world and why many include momentum in a risk premia portfolio.

  • A high dividend yield strategy involves buying high yielding stocks and shorting low yielding ones, hence it can be seen as a carry trade strategy for equities. It can be also be considered a value strategy as it selects stocks by comparing the price of a stock versus a fundamental driver, i.e. dividend per share, though it is often shown separately as the purpose of the strategy is somewhat different. Here, we look to buy stocks that offer us a good dividend yield rather than underpriced companies. Unlike its value counterparts that look to provide positive returns mainly through price appreciation, yield strategies outperform due to the reinvestment of the high dividend received. The focus is therefore more on the income rather than the fundamental value of the stock.
Historically, dividends have represented the greatest component of returns, and dividend yield is the more consistent component of returns through time.

Whilst we favor including quality characteristics in a high yield strategy _to avoid investing in stocks with unrealistic or unsustainable dividends_ here we analyze the dividend yield strategy by simply considering the trailing dividend yield. We define the long portfolio as the top quintile of stocks with the highest dividend yield and the short portfolio as the bottom quintile (excluding stocks with a zero dividend yield).

Like the value and momentum strategies above, the yield strategy has also outperformed significantly since 1995, averaging 9.8% per year on a long/short basis. Although it is highly correlated (+0.74) with the value strategy, and sharing the same periods of underperformance, dividend yield has seen more severe drawdowns during these periods and, as such, has underperformed our value index. Without the quality controls that QMS Advisors normally include in a high yield portfolio, the profile of the yield strategy is a lot riskier and certainly a lot less appealing than that of value strategies.

  • Quality investing has received a lot of attention in recent years with many products now offering exposure to high quality or low volatility stocks. The main intuition behind quality investing is that counter to financial theory and efficient market hypothesis, investors are not compensated for the higher risk they take when investing in low quality stocks (often referred as the low volatility anomaly). This was first documented by Robert Haugen and James Heins, who, using data from 1926 to 1971, found that in both equity and bond markets the relationship between risk and return was negative. Their findings were not very well received by the academic community at the time, with their article published four years later, in 1975.
That said, with equities suffering two heavy drawdowns since 2000, investors have been reducing equity allocations and looking for ways to manage the risk of their equity portfolios.

Explanations of the low volatility anomaly are mainly focused around behaviour biases, like the lottery effect, representativeness and over-confidence. These biases lead investors to overpay for the more exciting/higher upside, lower quality stocks whilst undervaluing the less exciting/limited upside, high quality companies. Several have again highlighted benchmark and business restrictions to also be important and we have also found that analyst overoptimism and an under estimation of the dangers of leverage are key drivers of the anomaly. Most define quality by simply looking at historical price volatility of the stock or the sensitivity of the stock to market movements (i.e. beta). QMS Advisors utilizes a combination of the volatility of the stock and its leverage, and using the Merton model we find that it is the combination of high volatility and high leverage that compounds problems.

There is also the option to define quality using company fundamentals, as suggested by Joseph Piotroski, which we believe can add value on top of market implied quality measures. As the Piotroski score can range from 0 to 9 (depending on how many of the binary fundamental tests a company passes), QMS Advisors combines it with the Merton quintile score using a weighted average. The top quintile of stocks with the highest weighted scores within the FTSE World universe comprises our high quality portfolio and, similarly, the bottom quintile makes up the low quality portfolio. Unfortunately, neither the Merton nor the Piotroski models can be applied to the complicated capital structures of financial companies, which are therefore excluded from our quality portfolios.

  • Along with market and value factors, Fama and French (1992) also included a size factor in their famous three factor model. They found that small cap stocks provide excess returns versus large cap stocks. Earlier Banz (1981) had also reported similar conclusions. Most popular explanations of the size premium are in line with financial theory and suggest that small cap stocks earn higher returns because they carry higher risks (such as liquidity risk, bankruptcy risk, information uncertainty, etc.). Behavioural biases like the tendency to extrapolate past growth too far into the future, have also been suggested as contributors to the size premium. Whilst we are not big supporters of simply selecting companies based on their market cap, the size premium is often used as a risk factor and thus we wanted to include it in our portfolio of risk factors and investigate the benefits in diversification and performance.
To be consistent with the construction of the previous strategies, we have again used the FTSE World index to define our size benchmark. We use the free float market cap of each stock and define the long portfolio as the bottom quintile and the short portfolio as the top quintile. Obviously, by focusing on a large/mid cap universe, like FTSE World, to define the size portfolios we expect to find weaker effects than those reported in academic literature.

In our study small cap companies have, on average, outperformed large caps by 2.5% since 1995 with the positive return mainly being generated during two periods, 1997-2005 and the strong market recovery post March 2009. The higher return of small caps is proportional to their higher risk, as measured by volatility, and so risk-adjusted returns are similar.
SelectionFile type iconFile nameDescriptionSizeRevisionTimeUser