The factors behind the return

There seems to be something like a distant relationship between low volatility and the quality investment approach, but nothing more: this was the verdict of the first two parts of this series of articles, where we subjected the two investment strategies to a descriptive (Part 1) and qualitative (Part 2) comparison. However, the value portfolio also examined seems to have almost nothing in common with the two approaches mentioned above. In this article, we will use statistical indicators to analyze whether the findings so far also stand up to quantitative evaluation. For this purpose, the sensitivities of the three factor portfolios to the Fama-French factors known in the financial literature are determined.[1]

Explanatory power times five

The three portfolios examined all rely on factors that empirically demonstrate better risk/return characteristics than the market index over the long term. The selection of these factors is no coincidence. Theory and practice have been dealing with the question of which systematic factors explain the excess return of investments for decades. From a multitude of answers, different convictions regarding the best investment strategy have emerged. Eugene Fama and Kenneth French's "five-factor model", an updated version of their famous "three-factor model" from 1993, provides a sound comparison of the return composition of the three investment approaches shown here.[2]  While Fama and French in the 3-factor model still assume that the return on a share can essentially be explained by the "market risk", the "size of the company" (size) and the "valuation" (value), the two renowned scientists have expanded their universe of yield-driving factors in the 5-factor model by the - not undisputed [3] - dimensions "profitability" and "investment activity". While the market risk factor reflects the risk premium of a market portfolio diversified across the entire equity market compared to a risk-free investment, the size factor takes into account the fact that shares with low market capitalization (so-called small caps) have historically achieved a higher return than large-capitalized securities (so-called blue chips). In contrast, the value factor is based on the observation that shares with a low price/book ratio (P/B) yield a higher average return than shares with a high P/B ratio. The two newly added factors target different aspects of "quality". While the profitability factor is based on an increased return on equities with good operating profitability, the investment factor assumes that high investment activity, measured as a strongly growing balance sheet total, leads to declining profitability and increases risk.[4] The "five-factor model" is of interest to us because it measures how strongly an investment strategy is exposed to the major return drivers identified by Fama and French. Accordingly, the five-factor model is well suited to locate the differences and similarities between the investment returns of the three portfolios we examined.

Volatility is a part of it

Since we examine a value, a quality and a low volatility portfolio, it makes sense to include a volatility factor in the statistical analysis in addition to the five-factor model, which already includes value and quality. The volatility factor is based on the empirical effect known as the "low volatility puzzle", according to which securities with low volatility often perform better in the long term than those with high volatility. In line with Fama-French's methodology, we construct the volatility factor as a long-short portfolio, with equities with historically low volatility receiving a positive weight (long) and equities with high volatility a negative weight (short).

When explaining the returns of the three portfolios using the Fama French factors and the additional volatility factor, clear patterns can be seen (see table and chart). As expected, the portfolios have, in addition to the market factor, the highest exposure to the factor that should be included in it according to the construction criteria. Thus, the value portfolio is mainly driven by the value factor, the quality portfolio by profitability and the low-volatility portfolio by the volatility factor.

The analysis of the coefficients and their statistical significance helps us to better understand the similarities and differences of the three portfolios examined. The low-volatility and quality portfolios show that both are influenced by risk and profitability factors. Although the returns of the two portfolios are positively driven by lower risk and high profitability equities, there are clear differences in the level of coefficients. The low-volatility portfolio is strongly driven by the risk factor, while the quality portfolio has a statistically significant but much smaller exposure than the volatility factor. In addition, there are other differentiating features. The investment and value factors show that low volatility and quality are not equivalent substitutes for each other. While the low-volatility portfolio is mainly made up of companies with low investment activity that make a positive contribution to returns, no comparable statement can be made for the quality portfolio. The corresponding regression coefficient here is even slightly negative, but not statistically significant. With the value factor, the difference between the two investment paradigms becomes even clearer: while the low volatility portfolio has no significant exposure to the value factor, the quality approach has a strong negative impact on the value factor. This negative sensitivity proves to be critical for the quality portfolio, as securities with a high valuation have a negative impact on the return. This is a fact that must be taken into account in the construction of the portfolio.

Finally, the analysis of the regression coefficients of the value portfolio provides an important insight into stock selection. As is to be expected from the construction, the value portfolio shows a strongly positive sensitivity to the value factor. Among the securities selected by the value portfolio, however, there is also a large number of securities with high volatility, which is reflected in a negative coefficient relative to the volatility factor and is an important reason for the comparatively weak performance of the value portfolio. When constructing a value portfolio, the risk component of the individual securities must therefore be explicitly taken into account. This is important in order to avoid a so-called "value trap". This is the only way to prevent a value portfolio from investing in equities that look cheap in valuation terms but are also very risky or even on the verge of bankruptcy. In short, we conclude that when constructing a value portfolio the risk should be taken into account and when structuring a quality portfolio the valuation should be taken into account.

The coefficients of the three portfolios compared to the other two factors market and size do not provide any differentiating features. As expected, the market factor proves to be the most important and comparably strong driver of returns for all portfolios. The fact that the size factor has a negative sign for all three portfolios is due to their composition, as only equities of medium-sized and large companies were used to construct the portfolios.

Volatility and quality: two independent factors

According to our analysis, the returns on the low-volatility and quality portfolios are thus both influenced by risk and profitability, but to varying degrees. The same picture emerges with regard to the sensitivities to the value and investment factor: here, too, there are considerable differences. Paired with the findings from the first two articles, this demonstrates that there are commonalities between quality investing and the low-volatility approach, while the value approach focuses differently. However, if one looks at it in a more differentiated way, it becomes clear that there are fundamental differences between all the above-mentioned investment paradigms. Low volatility and quality have certain points of contact, but they are definitely independent investment approaches.

The question remains as to how the three portfolios will behave during different market cycles. We will examine this in more detail in the coming 4th part of the analysis.

[1] The sensitivities correspond to the coefficients of a multiple linear regression of the portfolio returns to the Fama-French factors. A statistical test (t-test) can be used to check the significance of the results.

[2] Fama and French 2015

[3] Blitz et al. 2018

[4] See Novy-Marx 2013 and Fama and French 2006.