By LARRY SWEDROE
The basic hypothesis of behavioural finance is that, due to behavioural biases, markets make persistent mistakes in pricing securities. An example of a persistent mistake is that the market under-reacts to news — both good and bad news are slowly incorporated into prices, leading to the momentum effect. Another well-documented phenomenon is recency bias. It is the tendency to overweight recent events or trends and ignore long-term evidence. That leads investors to buy after periods of strong performance — when valuations are higher and expected returns are now lower — and sell after periods of poor performance — when prices are lower and expected returns are now higher. Buying high and selling low is not exactly a prescription for successful investing. Yet, it is the way many individuals invest.
An interesting question is: Does recency bias provide us with information as to the cross-section of returns? The hypothesis is that because investors typically use past return distributions to form expectations of future performance, their perception of the past is distorted, leading to systematic biases in return forecasts — biases which, in turn, can lead to structural mispricing in the stock market.
New research
Nusret Cakici and Adam Zaremba sought the answer to that question in their May 2021 study, Chronological Return Ordering and the Cross-Section of International Stock Returns.
The authors began by noting: “Market participants affected by the recency bias will overestimate the probability of the most recent returns and underestimate the probability of distant values. If the recent returns were comparably high relative to the distant ones, investors would overshoot the future returns — leading to eventual overvaluations. The emerging mispricing is then corrected by arbitrage forces, driving the prices down again. On the other hand, if recent payoffs were relatively low, the behaviourally provoked supply and demand may drive stocks below their intrinsic value. Again, when arbitrageurs step in, the underpricing is corrected and the prices proceed to rebound.”
Methodology
To capture this pattern, Cakici and Zaremba calculated a chronological return ordering (CRO) variable and analyzed its importance for the cross-section of returns. The original framework for the CRO methodology was developed by Hannes Mohrschladt, author of the study The Ordering of Historical Returns and the Cross-section of Subsequent Returns, published in the April 2021 issue of the Journal of Banking & Finance.
The CRO measure relies on the correlation between the historical stock returns and the number of days that have passed since the return realisation. The interpretation of CRO is intuitive and straightforward. Low CRO numbers indicate relatively low distant returns and high recent payoffs. On the other hand, high CRO values indicate that the distant returns were comparably high and the recent returns were low. Hence, low CRO values should imply low future returns; conversely, high CRO scores should signal high future returns. Their baseline CRO used a monthly formation window. However, in tests of robustness, they also analyzed estimation periods from three to 36 months—using both daily and monthly data frequency.
Cakici and Zaremba’s data sample included 23 developed markets and 26 emerging markets, covered the period 1990-2020 and accounted for survivorship bias (both live and dead stocks were in the data). They also dropped the 10 percent of securities with the lowest price at the beginning of each month.
Findings
The following is a summary of Cakici and Zaremba’s findings:
Globally, the value-weighted (equal-weighted) decile of stocks with the highest CRO outperformed the decile of stocks with the lowest CRO by 0.91 percentage point (0.63 percentage point) per month.
The phenomenon exists in the U.S. and across developed markets. The top CRO quintile of stocks historically outperformed the bottom quintile in 84 percent of the tested markets; furthermore, the difference was significant at the 5 percent confidence level in 59 percent of the countries. However, the picture of emerging markets differs dramatically, as there was no CRO anomaly—neither average returns nor alphas in the long-short portfolios produced any significant profits.
The abnormal returns in developed markets were not explained by the common asset pricing factors (including beta, illiquidity, size, value, reversal, momentum, idiosyncratic volatility and skewness) and were robust to many considerations—CRO did not exhibit a visible relationship with any of the other control variables.
The CRO anomaly does not depend on any particular industry.
Unlike many other anomalies that are concentrated in micro-cap stocks, the CRO effect is powerful and significant in both small and big firms. This is important in terms of implementability. The results were statistically significant at the 1 percent confidence level.
The long-short CRO portfolios continued to deliver significant alphas even when the portfolios were reconstructed only once in six months.
Two variables play a critical role in the CRO effect: individualism and investor protection. The third of countries with the highest individualism (linked in the literature with overconfidence and self-attribution biases) score produced monthly CRO long-short strategy returns higher by 0.26-0.40 percentage point (depending on the measurement approach) than the most collectivistic countries. And the CRO spread portfolios in the top third of markets with the highest anti-self-dealing index outperformed their low-investor-protection counterparts by 0.27- 0.45 percentage point per month. This observation accords with the supposition that stronger investor protection attracts retail investors, who are typically considered as less sophisticated and more prone to behavioral biases than institutional players.
CRO profits are more substantial during periods of elevated market volatility—positive and significant coefficients were observed for both historical and implied volatility measures. This observation complies with the reasoning that high volatility tends to impede arbitrage activity.
Market crashes augment the global CRO-driven mispricing. The anomaly was particularly pronounced following extreme volatility and significant down markets — outside these difficult periods, the profits on global CRO portfolios hardly differed from zero.
Of interest is that the lack of CRO evidence in emerging markets perhaps can be explained by the finding that countries with the highest individualism scores showed the highest CRO. Developed countries tend to be more individualistic. And individualism tends to correlate with self-attribution bias and overconfidence. The result is that individualistic investors focus on short-term performance, and they also tend to be overconfident and sensitive to extreme returns. Thus, the differences in cultural dimensions may explain the differences in the CRO anomaly between developed and emerging markets.
Their findings led Cakici and Zaremba to conclude: “The chronological return anomaly is particularly pronounced in challenging market conditions: during high volatility states, following market crashes, severe bear markets, and during tight funding environments. This is consistent with the behavioural view on market anomalies; asserting that they should be the strongest during periods when arbitrage activities are limited.”
The authors also observed: “The chronological ordering anomaly is particularly pronounced in individualistic cultures and markets that are characterised by high investor protection.” Developed markets tend to have these characteristics, providing a behavioural explanation for the anomaly.
They added: “The pattern is stable through time and not specific to any particular sub-period.” They also noted: “Following substantial payoffs in years 1990-2002, no clear pattern emerges in the years 2003-2008. This is followed by another, more extended, bull period for the CRO anomaly, and the curve flattens again in the most recent years.”
Conclusion
The takeaway for investors is to not make the mistake of recency bias, as the research demonstrates that companies with comparably low recent returns and high distant ones significantly outperform their counterparts (companies with relatively high recent returns and low distant ones). Investors should be particularly careful to avoid getting euphoric because when something is highly popular, it runs the risk of being bid up in price and thus is prone to disappointment.
The bottom line is that adhering to your plan by rebalancing your portfolio will help you avoid the mistakes caused by recency bias and overconfidence.