Research Review: Quality minus Junk – Asness et al 2013
posted on December 3rd, 2015 by Bellmont Research Team
While Robert Novy-Marx is widely acknowledged as the first academic to prove the existence of the ‘Quality’ factor (that has subsequently been accepted far more widely by the finance community), Cliff Asness of fund management giant AQR is undoubtedly one of the leading industry proponents of such an approach.
In this paper, Asness and his colleagues Frazzini and Pedersen attempt to both define ‘Quality’ in regards to stocks, and also determine the extent to which quality is priced. That is – are quality stocks more expensive than other stocks? If so, by how much? And importantly, can investors improve their returns by focusing on quality stocks?
Examining the US market from 1956 to 2012, as well as a broad sample of international stocks from 24 developed markets from 1986 to 2012, Asness’s first challenge was how to define quality. Rather than championing any one particular measure (which he felt would make the research results questionable, due to the potential for ‘data mining’), he chose instead to construct an aggregate quality measure based on multiple parameters, each one representing “..characteristics that investors should be willing to pay a higher price for, everything else equal.” Utilising the ‘Gordon Growth Model’ (aka Dividend Discount Model – one of the keystones of modern finance) as a guide, the resulting definition of quality could be simplistically described as “profitable, growing, safe and high (dividend) payout”. By summing the results of multiple objective measures of each of these favourable characteristics that had either been previously examined in academic research (e.g. – Piotroski’s ‘O’ score as a measure of safety), or widely utilised by practitioners (e.g. – ROE as a profitability measure), he assigned each stock a single aggregate ‘quality’ score.
The next step in the study was to identify the ‘persistence’ of these definitions of quality. That is, do stocks that exhibit these high quality characteristics today, continue to exhibit them in the future? The results overwhelmingly show that they do. That is, stocks that were profitable, safe, growing and well managed in the past, on average continue to display these positive traits 5 and 10 years into the future.
But being able to objectively identify quality companies need not lead to an improvement in investors’ returns, if these characteristics are appropriately priced by the market. That is, in an efficiently operating market, companies that exhibit these high quality characteristics should be readily identified by investors, and on average command higher prices, with these higher prices ensuring that subsequent investment returns are in line with the market as a whole (despite the fact that the underlying business may have performed better than average).
And indeed, whilst the study results confirm this broad thesis – that quality companies do in fact trade at higher prices – the degree of premium that they command is far less than might be imagined. In fact, quality explained only 12% of the variation of prices in the US sample, and only 6% in the global sample. That is, between 88% and 94% of a firm’s price is determined by factors other than the quality, growth and stability of a company’s earnings and its willingness to share them with shareholders – surely by far the most important determinants of investment desirability for investors!
Side Note – Segmenting this analysis further, Asness touched on some interesting observations in relation to the ‘size effect’. Since being identified by Fama and French in 1992 as one of the 3 factors that can explain a portfolio’s return relative to the market, the results of small cap stocks have on average failed to live up to expectations, leading many researchers to question the robustness of size as a factor. However, what Asness identifies in this study is that on average, small companies are much lower quality than large companies (which makes intuitive sense). And once you control for this variation of quality, small firms are substantially cheaper than their large company counterparts, and for a given quality level, returns from investing in them are markedly higher.
Having settled on an objective definition for ‘quality’ companies, shown that these characteristics are persistent over time, and that whilst priced by the market they only explain a very small variation in the prices of companies; Asness then moves to the returns available from investing in quality companies.
In both their long sample (US stocks from 1956 to 2012) and their broad sample (24 developed nations from 1986 to 2012), returns increased “..almost monotonically in quality such that high-quality stocks outperform low-quality stocks.” Improvements in risk-adjusted returns are even more marked, due to the inherently low volatility nature of quality stocks.
A zero cost “Quality minus Junk” portfolio was also formed, by buying (long) the top 1/3 of constituent stocks, and selling (short) the bottom 1/3. The results from this ‘Quality minus Junk’ portfolio backed up the long-only results, with significant excess returns in both the long and broad samples.
In order to gain a more granular understanding of the various quality measures, each quality characteristic was also tested separately to determine its individual contribution to the performance of the model, as well as examining the inter-relationships between the different characteristics.
Impressively, all four of the individual quality characteristics (measured as zero cost portfolios, using the same methodology as the “Quality minus Junk” portfolio described above) generated positive excess returns, both in the broad and long samples. Predictably, the excess returns for each of the individual factors was lower than that of the combined ‘Quality minus Junk’ measure, justifying the use of a combination measure. The most significant excess returns for the individual factors in both samples was generated by the ‘Payout’ component, suggesting that this aspect of quality is least appreciated and priced by the market (Note – it will be interesting to see in our own testing whether this remains the case in Australia, given Australian investors’ predilection for high dividends). Interestingly, the smallest net benefit was actually provided by the ‘growth’ factor, suggesting that these measures may be more accurately priced by the market on average.
The average correlations between the factors was also positive, at 0.4 for the long sample, and 0.38 for the broad sample, suggesting that “..while the quality components measure different firm characteristics that investors should be willing to pay for, firms that are high quality in one respect tend to also be high quality in other respects.”
The authors find that “..high quality firms do exhibit higher prices on average. However, the explanatory power of quality on prices is low… As a result, high quality firms exhibit high risk-adjusted returns.”
This conclusion adds to the results of Novy-Marx in making a very strong case for the inclusion of quality considerations in any systematic investment portfolio. The key differentiating points between the two relate to the combination of value and a simple quality measure for Novy-Marx, compared with a far more complex and comprehensive assessment of quality, but the exclusion of price from Asness et al. The results are the same however – investment returns in both an absolute and risk-adjusted sense can be substantially improved by careful consideration of the quality of a business, as well as its price.