I am probably a little bit uncompromising,’’ says Noël Amenc, the founding CEO of Scientific Beta, the provider of factor indices and strategies. To those who know him that is an understatement.
In November 2018, his organisation released several papers advocating the use of “academically-validated factors” when building factor strategies.
The papers criticised the use of “non-standard factors” by established providers of factor strategies. In broad terms, Scientific Beta’s argument is that providers do not apply academic research correctly and therefore offer sub-optimal and, above all, non-robust results.
Amenc had a reputation as a steadfast academic and the controversy could only make it stronger. However, he refuses to see the disagreement as a dispute between academia and industry. “That is a false debate. It is not about being academic or practitioners. It is about being consistent and rigorous. Many practitioners can be just as rigorous as pure academics,” he says.
Amenc launched Scientific Beta in 2013, to provide investors and asset managers with factor indices and strategies. It is linked to EDHEC-Risk Institute, which Amenc led from its foundation in 2001 to 2015, and EDHEC Business School, where he is professor of finance and associate dean for business development.
Some argue that Scientific Beta has taken a confrontational approach to the matter because it is challenging established providers.
While that may be true, Scientific Beta maintains strong links with academia, and Amenc has made no secret of his purpose of bridging the gap between academia and industry.
The Scientific Beta papers have highlighted the divergence between providers that adhere to academic practice and those that allow themselves greater freedom. The papers have shown that several organisations, including FTSE Russell, MSCI and Research Affiliates, have reviewed their definitions of factors.
Amenc takes issue with this trend. He argues that factor definitions are modified by those who seek to optimise the back-tested results of strategies and indices.
“We have probably displeased these professionals by showing that the same research teams could, from one year to the next, qualify or disqualify a factor that is supposed to be long-term,” he says.
“One is forced to observe that rigour is lacking with some professionals.”
The controversy derives from the perception that common factor ‘proxies’ are deemed antiquated. The value factor is the best example. Academics still regard a stock’s book-to-price as the best proxy for the value factor, but many argue it no longer reflects intrinsic value. This is because it does not consider intangibles such as brand value or R&D, which have become more relevant.
This is what has led some factor providers to employ more sophisticated proxies. Research Affiliates’ chairman Rob Arnott argued for higher sophistication of the value proxy in a recent interview with IPE (May 2019 issue).
But this expansiveness in factor definitions is problematic, says Amenc. “It is a well-known cause of the out-of-sample weakness of results that are often optimised in-sample. The best way to maximise the risk of model mining, and more generally of manipulating back tests, is to allow the use of multi-criteria factors. Any serious researcher knows that as the number of possible combinations of criteria increases, the lack of robustness of the results displayed in-sample increases too,” he says.
If Amenc’s criticism is valid, it shows that there is confusion over the purpose of factor investing.
“A factor is not an investment theme that is liable to outperform the market over a given period if events are favourable to this theme occurring,” he says. “It is the expression of a sustainable anomaly, which means that for a given level of market risk we observe a greater return than that delivered by the market.
“The existence of this anomaly is not, however, a sufficient condition for one to be able to speak of a priced or rewarded factor. It is necessary to conduct empirical tests to show that, compared to the factors that already exist, this factor provides an additional and non-redundant explanation for the difference in return levels between different portfolios. It is only on the basis of these rigorous tests that we can speak of the existence of a new factor.
“A good factor proxy is one that meets acceptance criteria that are known to and widely accepted by researchers in the area,” he says.
Amenc does not rule out updates to its definition when it comes to the value factor. “It would be very naïve to think that an accounting metric could possibly capture the true value of a firm.”
For Amenc, a value stock is not inherently undervalued, it is simply less expensive. The two concepts are different. An accepted explanation for the relative inexpensiveness is simply that they are risky owing to the magnitude of their irreversible material investments, which can depreciate in the event of a cyclical change. “Presenting the book-to-market as a valuation proxy used by academics is misleading, because this ratio is merely a measure of the irreversibility risk and not a valuation ratio,” he says.
“If progress needs to be made in the area of the value proxy, then it is surely to take account of the irreversibility of intangible investments, whether involving technology, brands or, more globally, the firm’s expertise.”
This is tricky, but Amenc notes that it is producing extensive academic research. “In time, this research effort will probably be able to question the traditional definitions not only of the value factor but also probably other factors that incorporate similar accounting dimensions.”
Amenc regrets that the factor investing industry has become a battlefield over returns rather than risk management. “My hope was that people would emphasise risk management rather than return improvement, because return measurement is unfortunately too sensitive to the choice of sample, ” Amenc says.
“Providers do not seem to care about controlling the exposure to unwanted risks, such as those relating to sectors or geographies, and especially to market beta. As a result, the risk management frameworks of factor strategies are archaic. “Instead of competing on the definition of factors, factor managers should compete on risk management strategies.”