Scientific Beta

This paper presents two contributions on the subject of factor measurement. In the first contribution, entitled "Exposed to Nonsense? Spurious Factors in Popular Investment Tools," the authors question the choice of factor menu and factor proxies that are supposed to represent, and above all to standardise, the measurement of a portfolio's factor exposure. Their conclusion is indisputable: the factor definitions used by popular tools offered to investors are not supported by serious academic research. The second contribution, entitled "Mismeasurement of Factor Exposures in Score-Based Analytics Tools," questions the way in which the measurement of factor proxies is implemented. 

This paper presents two contributions on the subject of factor measurement. 

In the first contribution, entitled "Exposed to Nonsense? Spurious Factors in Popular Investment Tools," the authors question the choice of factor menu and factor proxies that are supposed to represent, and above all to standardise, the measurement of a portfolio's factor exposure. Their conclusion is indisputable: the factor definitions used by popular tools offered to investors are not supported by serious academic research. In very concrete terms, four major conclusions can be drawn from their contribution:

• Analytic tools do not employ academically-grounded factors and their factor-finding process maximises the risk of ending up with false factors;
• Thus, most factors used in commercially-available analytic tools are likely false;
• Non-standard factors lead to mismeasurement of exposures and may capture exposure to redundant factors;
• Analytic tools for investors distort the key idea of factor investing because they lack transparency and expose investors to providers' conflicts of interest.

The second contribution, entitled "Mismeasurement of Factor Exposures in Score-Based Analytics Tools," questions the way in which the measurement of factor proxies is implemented. Here again, the observations should tell investors something:

• The use of factor scores instead of factor betas for the measurement of portfolio factor exposures is a cause for concern because factor investing literature uses beta-based models for factor premia tests and for portfolio style analysis;
• The major drawback factor scores suffer from is "double counting" of exposures, which is due to their lack of regard for the correlation structure of factors. This makes factor scores a very poor proxy for factor betas;
• Another common practice is to combine single factor scores into a composite factor score. This leads to additional problems because combining two scores that have a skewed joint distribution disproportionately favours one factor over another and using these composite definitions to measure score exposures leads to misidentification of factor tilts.