Some factors are now more equal than others

Some investors have strong hands during such downturns, enabling them to benefit from the upswings, while others may try to dynamically allocate to factors. Our research shows that over a five-year horizon, value is currently the most attractive factor.

The considerable variation in the rolling five-year returns of value, quality, low risk and size for the US equity market can be seen in the chart below, using data going back to 1968. 

The long-term factor premiums, excluding costs and fees, can be seen in the table below. 

Size: The long-term outperformance of small cap stocks relative to their large cap counterparts (the ‘size effect’) is positive, at 2.49%. However, the market risk-adjusted return (the alpha) of 0.91% is not statistically different from zero, as its t-statistic is below the threshold of two. 

Value: The raw outperformance of high book-to-market stocks relative to their low counterparts is strong at 4.03%. while its alpha is 2.75% per year. The value premium is also strong for the post-1963 period, despite the insignificant alpha of 2.14% since 1990 in the US. 

Low risk: This factor’s premium has been one of the strongest in history, with a long-run average return of 5.99%. This is close to its alpha by design, as the factor becomes neutral to market risk by leveraging up the low-risk portfolio and leveraging down the high-risk portfolio. 

An important question is whether these historical premiums will persist in the future. We believe they will, based on three convincing explanations. The first is that they are the reward for bearing systematic risk; the bad news is that factor investing then is risky, but the good news is that the premiums are persistent. 

The second is institutional: the way the finance industry is organized, including its regulation, may give rise to incentives that are inconsistent with traditional finance. For example, capital requirements in solvency regulation are not dependent on the risk of the equity portfolio, which may lead to a low-risk premium. 

The third relates to behavior, in that investors use signals, heuristics or preferences that are inconsistent with rational behavior. Based on the assumption that fear and greed are hardwired in investor behavior, this leads to persistent factor premiums. 

However, if all these explanations no longer hold going forward, we are open to reconsidering the existence of these premiums.

The second step is to find indicators of an asset class being cheap or expensive. Valuation indicators, for example, have been found to have predictive power for returns. Research shows that when factor premiums are cheap, returns tend to be high in the five years that follow – a pattern that is particularly visible for value and size. 

The chart below shows the time variation of the valuation signal for the four equity factors over time, where a measure above one indicates that the factor is cheap. Over the past two years, the value factor has been extremely cheap, peaking at even higher levels than previous highs, which were reached just before the dot-com bubble burst. 

Based on the current readings of these valuation spreads, we expect higher returns for value, low risk, and quality, and average returns for size.

Research shows that following expansionary signals, the size and quality factors tend to perform poorly, while the value factor is relatively strong. Low risk is the strongest factor during recoveries.

There is a popular belief that value factor returns correlate negatively with bond returns. While this may have been the case over the past couple of years, research show that the relationship between value and interest rates over the long run is virtually zero. The size factor has a significant negative relationship, and the low-risk factor a positive one.

The final ingredient is embedded climate risks. This looks at the carbon exposure of the generic size, value, quality and low-risk factors alongside that of the market index. Value and quality are currently the factors most exposed to climate risk, while size is close to the index, and low risk is substantially less exposed.

This leads to a negative climate tilt for the generic value and quality factors, and a positive tilt for low risk. Such tilts for generic value and quality can though be eliminated through strategy enhancements, which we implement for many of our clients.