Factors in asset pricing models exhibit cyclical behavior as they offer a premium in the long run while going through bull and bear phases in the short run. In our research, we investigated the possible explanation to these cyclical dynamics, focusing mainly on value, quality, momentum and low-risk factors and an equally-weighted multi-factor portfolio.
From a rational asset-pricing perspective, factor premiums are seen as risk premiums, reflecting rewards for certain macroeconomic risks. This would imply that factor performance is related to the business cycle. But many studies have failed to establish a robust empirical link between the two.1 Indeed, results from our study also illustrated the tenuous nature of this relationship.
Business cycle indicators can’t fully explain the cyclicality in factor returns
In general, what we found was that the multi-factor portfolio had similar returns in contrasting macroeconomic conditions, i.e. expansions versus recessions; inflationary versus non-inflationary; and high ISM purchasing managers’ index (PMI) versus low ISM PMI.
The outcome varied slightly from a single-factor perspective. For example, we observed what we would characterize as strong returns from the momentum factor during inflationary periods, while the low-risk factor struggled, consistent with its known ‘bond-like’ properties.2 Meanwhile, the value and quality factors appeared largely immune to both inflationary and non-inflationary conditions.
We also took into account the Baker and Wurgler investor sentiment index, a popular sentiment indicator in academic literature.3 We established that the returns for the value, quality and low-risk factors were strong when investor sentiment was positive, and weak when investor sentiment was negative. Only the momentum factor appeared to be resilient to the various sentiment states.
Perhaps the difficulty in establishing a relationship between macroeconomic risks and factor premiums lies in the flawed notion that factor premiums are risk premiums. If the source of factor premiums is indeed behavioral, then this would explain why the Baker and Wurgler sentiment index appears to be more effective at distinguishing between high and low factor returns. However, even this indicator is only able to pick up a small portion of the much larger time variation in factor returns.
In our view, the cyclicality in factor returns is driven by sentiment which can best be inferred directly from factor returns. In other words, our premise is that factors essentially follow their own behavioral cycle, and although other macroeconomic and behavioral indicators may pick up some of these dynamics, the full picture can only be uncovered by studying factors themselves.
Mapping out the quant cycle
In our study, we determined the quant cycle by qualitatively identifying peaks and troughs that correspond to bull and bear markets in factor returns. Following this approach, we identified a cycle consisting of a normal stage, which prevails about two-thirds of the time, punctuated with occasional large value drawdowns, which tend to be followed by subsequent reversals.
We found that the value drawdowns were caused either by growth rallies or value crashes that occurred roughly once every ten years, typically lasting about two years. These were usually followed by sharp reversals, which were characterized either by a crash of the growth stocks that outperformed strongly in the previous stage, or a strong recovery rally in the stocks that underperformed in the preceding phase. Our historical definition of the quant cycle is depicted in Figure 1.
Figure 1 | The quant cycle, 1963 to 2020
Source: Robeco Quantitative Research
Breaking down the different stages
During the normal stage, we observed that all factors delivered solid positive average returns, typically above their long-term average premiums. Luckily for multi-factor investors, this stage prevailed about two-thirds of the time. However, the relative peace and quiet of the normal period was disturbed by events that unfolded during the remaining one-third.
We witnessed that the growth rallies were characterized by large negative returns for value and large positive returns for momentum. This clearly illustrates how the two factors typically diversify so well with each other during these extreme phases. Low-risk typically took a hit and quality usually performed well in this phase, although not in every instance for both factors. Altogether, the multi-factor approach delivered a roughly flat return on average during growth rallies.
Our sample period only contained one value crash, namely the 2007-2009 global financial crisis. We saw that the factor performance was remarkably similar to the one observed in growth rallies, with negative returns for value and low-risk and positive returns for momentum. Based on this single observation, quality seemingly fared better during value crashes. This resulted in a small positive return for the multi-factor portfolio.
We also picked up two types of reversals: bear and bull. Bear reversals were distinguished by large positive returns for value due to a growth crash. They also tended to be highly favorable for quality, momentum and low-risk. Given that all the factors tended to be effective during these periods, we noticed that the multi-factor mix also delivered strong returns.
By contrast, bull reversals were characterized by large negative returns for momentum due to a rally in stocks with poor momentum. There were also large negative returns for quality and mixed results for value and low-risk. From a multi-factor perspective, we saw that bull reversals presented much tougher challenges than bear reversals.
Using the quant cycle as a framework -07
Thus, we conclude that investors should focus their efforts on better understanding the quant cycle as implied by the factors, rather than adhering to traditional frameworks. We believe this can helpful in contextualizing the cyclical dynamics of factors. It can provide investors with the basis to formulate a multi-year outlook, by providing insight on how the quant cycle could potentially unfold based on the prevalent market environment. The model can also give them a frame of reference to examine the robustness of new alpha factors across the various stages of the cycle.Read the full paper
1See: Ilmanen, A., Israel, R., Lee, R., Moskowitz, T. J., and Thapar, A., February 2021. “How do factor premia vary over time? A century of evidence.”, Journal of Investment Management (forthcoming).
2See: Blitz, D., September 2020, “The risk-free asset implied by the market: medium-term bonds instead of short-term bills.”, Journal of Portfolio Management 46 (8): 120-132.
3See: Baker, M. and Wurgler, J., August 2006, “Investor sentiment and the cross-section of stock returns.”, Journal of Finance 61 (4): 1645-1680.
本文由荷宝海外投资基金管理(上海)有限公司(“荷宝上海”)编制, 本文内容仅供参考, 并不构成荷宝上海对任何人的购买或出售任何产品的建议、专业意见、要约、招揽或邀请。本文不应被视为对购买或出售任何投资产品的推荐或采用任何投资策略的建议。本文中的任何内容不得被视为有关法律、税务或投资方面的咨询, 也不表示任何投资或策略适合您的个人情况, 或以其他方式构成对您个人的推荐。 本文中所包含的信息和/或分析系根据荷宝上海所认为的可信渠道而获得的信息准备而成。荷宝上海不就其准确性、正确性、实用性或完整性作出任何陈述, 也不对因使用本文中的信息和/或分析而造成的损失承担任何责任。荷宝上海或其他任何关联机构及其董事、高级管理人员、员工均不对任何人因其依据本文所含信息而造成的任何直接或间接的损失或损害或任何其他后果承担责任或义务。 本文包含一些有关于未来业务、目标、管理纪律或其他方面的前瞻性陈述与预测, 这些陈述含有假设、风险和不确定性, 且是建立在截止到本文编写之日已有的信息之上。基于此, 我们不能保证这些前瞻性情况都会发生, 实际情况可能会与本文中的陈述具有一定的差别。我们不能保证本文中的统计信息在任何特定条件下都是准确、适当和完整的, 亦不能保证这些统计信息以及据以得出这些信息的假设能够反映荷宝上海可能遇到的市场条件或未来表现。本文中的信息是基于当前的市场情况, 这很有可能因随后的市场事件或其他原因而发生变化, 本文内容可能因此未反映最新情况,荷宝上海不负责更新本文, 或对本文中不准确或遗漏之信息进行纠正。