In a recent opinion piece published on Pensions & Investments' website titled “Factor indexes: Choose with care,” John Chisholm, chief investment officer of Acadian Asset Management, sought to make a distinction between the “simplistic” “factor factories” of index providers and an “alpha modeling” approach to factor investing that would be representative of the “true” quant approach that only active managers can offer.
One can understand that the success of new passive investment offerings worries active managers whose performances often fail to justify their high fees. One can also find that it is good business practice when faced with investors' increasing interest in factor investing, which actually gives access to betas beyond the broad market premium, to attempt to communicate and to position oneself in such a way as to preserve the search for alpha.
It is nonetheless regrettable that this marketing strategy is based on approximations or assertions that do not have any real academic justification.
The first of these assertions is that the sole source of smart beta's outperformance, and more specifically of the premiums associated with the factors, relates to anomalies in price formation on the market, even though a wide-ranging section of the academic literature has actually shown that the greater return associated with the consensual long-term rewarded factors — which are value, momentum, size, profitability, investment and even low volatility — can to a large extent be attributed to rational pricing. Indeed, the bulk of the rewards offered by these factors corresponds to risks that are marginally higher than those to which traditional cap-weighted indexes are exposed. It is because long-term investors can bear these risks, which only become a reality in situations that rarely occur, that they can collect the premiums associated with them.
By portraying factor premiums as purely anomalous, Mr. Chisholm prepares his readers and prospects to hear an argument that revolves around alpha, which is in essence the search for pricing anomalies. By affirming that the value added by factor investing should be based on “alpha modeling to identify the stock characteristics or combinations of characteristics associated with high risk-adjusted returns,“ Mr. Chisholm is actually only renaming the old practice of stock picking.
Unfortunately, this facelift cannot cure the fragility of the out-of-sample performance of this traditional approach, since the portfolio is ultimately strongly influenced by the manager's returns forecasts. Even though academic research has shown that it is vain to try to optimize or construct a portfolio on the basis of estimated returns, whose dispersion is a source of lack of out-of-sample robustness, Mr. Chisholm makes it one of the key elements in his factor solution construction process.
If there is merit in smart beta and factor investing, then it is firstly in not trying to find alpha or precisely predicting the future value of stocks or their returns. It simply involves measuring and managing risks in a systematic way by relying on risk parameters, which, since they are converging estimators, can be estimated robustly, as is the case for volatility or covariances. Moreover, the whole object of 60 years of research in empirical finance and asset pricing theory is to try to improve these estimations. And it is indeed this search for the right estimation of variations in returns that led to the discovery of systematic risk factors other than market beta.
Another assertion that potentially presents grave dangers for investors is the idea that the added value of factor construction relies on its complexity compared to the purportedly excessive simplicity of smart beta indexes. Here too, Mr. Chisholm forgets a fundamental portfolio construction principle, namely respecting rules of parsimony. These rules actually limit the risks of data-mining that increase the sample dependency of performances. In this area, one can only recall that what led to the success and credibility of the academic research that founded factor investing was that it relied on the discovery of factors that were defined in a very simple way and supported by economic reasoning that was also easy to grasp.
If factor providers want to add genuine value by relying on a set of more sophisticated quantitative methods, they should do so not in order to pick stocks, but to improve the risk diversification of the portfolio of stocks selected. Such diversification is the real guarantee of improving the risk-adjusted performance of factor investing. This is the choice that ERI Scientific Beta has made by providing a totally transparent index platform that allows a choice of factor tilts and weighting schemes to be combined in order to ensure sound diversification of both factor risks and specific risks.
NOEL AMENC
professor of finance at EDHEC Business School
CEO of ERI Scientific Beta
Nice, France