How Can You Tell When a Factor Stops Working?

There is persuasive evidence that a select group of risk factors in equity investment provide investors with a return premium – that is they offer a long-term performance advantage relative to the broad market, on a risk-adjusted basis.  The most established and robust are value, momentum and size.  All three of these meet the crucial criteria of being present over an extended time horizon, pervasive across markets and supported by a behavioural / economic foundation.

Based on this, I consider it prudent to have exposure to these factors in a portfolio. However, this acceptance raises an immediate question – if I believe in the efficacy of these factors now, is this decision irrevocable? If I take a view about there being certain structural risk premia in equity markets, is it possible to change my mind? What could the rationale be for such a shift of view?

It is important to make a distinction here – I am not referring to the cyclicality of returns from a factor, I fully acknowledge that the performance of all equity risk factors will wax and wane, often over long cycles.  Rather, my concern is a situation where the long-term expected risk premium for a particular factor evaporates because of some change in the drivers of the anomaly.  I do not worry that the size factor will fail to deliver for sustained periods (for example), but that the premium no longer exists.

Whilst such an occurrence may be considered to hold a low probability, it surely cannot be impossible. Therefore a risk exists, yet one that is difficult to identify and quantify for a variety of reasons:

–  The pillars of all credible equity risk premia are economic and behavioural rationale. Structural excess returns cannot simply be present in the data; there must be a reason for their existence – I have written previously about behavioural drivers for value and momentum.  The problem is that these are only suppositions, in most cases it is impossible to precisely and unequivocally identify the causes of certain risk premia. It therefore follows that we cannot confidently observe when these alter.

–  Long term horizons (large sample sizes) are a prerequisite to build confidence in the robustness of equity risk premia. If we require multiple decades of data (at the very minimum) to support our behavioural / economic hypothesis, we need similar periods of time to ‘disprove’ them.  By the time there is sufficient evidence to support the contention that a risk factor no longer effective, you are likely to be enjoying retirement.

– The problem is exacerbated by the fact that even if a currently sound factor stopped working it would still behave, at times, as if it did.  Let’s assume that from this day on there was no value premium in equity markets – even in such a scenario there would continue to be extended periods when value outperformed.  Expensive stocks can (and do) generate prolonged periods of excess returns, despite there being no evidence of a structural premium.

–  In the event that you are willing to declare the demise of a previously robust equity risk premium, you are likely to be wrong, and at a particularly inopportune time – the sounding of the death knell for value investing will almost inevitably be prelude to resurgent performance from the factor.  It is easy to mistake the cyclical for the structural, and abandon a style of investing when it is depressed and the valuation dispersion wide.

– It is simply not in the interests of the majority of groups working within the burgeoning factor investment sub-industry to state that an prominent risk premium has lost its efficacy – informing clients that the factor they have invested billions in is unlikely to work in the future seems to be an unwise and unlikely business decision.  There are undoubtedly investors who would seek to exploit the change in factor dynamics, but these would be the exception rather than the rule.  It is not an industry renowned for parsimony.

As with all investment judgement, the consideration of factor viability it is about uncertainty and subjective probability. On the weight of evidence, it is likely that established equity risk premia will continue to deliver over the long term, however, there is a low probability that they will not, and it is important to incorporate this possibility into your decision making.  It is also prudent to assess how rich or cheap a particular factor is relative to its own history, which may provide some protection in the event that the presumed structural risk premium is no longer present. Neither of these options directly addresses the central question posed in this article, simply because there are no obvious answers.