“Money flows into most funds after good performance, and goes out when bad performance follows.” (John Bogle)
We have all seen the wording discretely appended to mutual fund marketing stating that ‘past performance is no guide to future results’. Despite the ubiquity of this message we struggle to heed its warning. Instead, we operate in the powerful grip of outcome bias; where we assume that good results must be the consequence of skill and will persist, and that poor results are a prelude to ongoing disappointment. This leads to the damaging behaviour of performance chasing, where we sell our holdings in laggard fund managers and reinvest in recent winners.
Whilst such an activity may offer near term relief, it often comes with a long-term cost. In unpredictable financial markets relying on historic performance as a primary measure for quality is fraught with problems. There is no better example of how misleading past performance can be than the tale of a pharmacist turned fund manager.
Jayesh Manek was born in Uganda in 1956 and, after moving to the UK, studied to become a pharmacist at Brighton College before opening his own pharmacy in west London. He had great success in this field and eventually sold the chain of stores he had built, Dallas Chemists, to Alliance Unichem in 1999. His real passion, however, was for investing. A fact demonstrated by his success in a fantasy fund management competition run by The Sunday Times, which offered a £100,000 first prize.
In 1994, Manek beat thousands of entrants to win the stock picking contest after turning a paper £10m into a staggering £502m. Remarkably, he managed to repeat this feat in 1995, growing another £10m into an impressive, though modest by previous standards, £58m.
Manek had a predilection for small cap growth companies, and was ideally placed to benefit from the emerging bubble in technology stocks. It has also been suggested that he astutely exploited loopholes in the competition by making multiple entries[i], materially increasing his chances of success. But, in the face of stellar outcomes, we can often treat such details as something of an irrelevance. The results were leading us to believe that a major new investing talent had been unearthed, operating above a pharmacy in Ruislip.
Emboldened by his unprecedented success and the publicity surrounding it, Manek took the inevitable next step of launching his own fund; offering investors the opportunity to benefit from the acumen that had seem him twice crowned a champion stock picker. The story of pharmacist to fund manager not only captured the attention of the investing public, he also received £10m from revered investor Sir John Templeton. Manek had the validation, the captivating narrative and a track record of success – what could go wrong? Plenty, as it turned out.
Manek’s foray into fund management didn’t immediately turn sour. In the early years of his fund he persisted with his penchant for speculative, high growth stocks, which had brought him great such success in the competition. As Manek himself noted “at its peak, the fund was 75%-80% in technology”[ii]. From launch in 1997 to the height of the technology bubble, the fund outperformed the UK stock market by 152%.
The strong initial performance only served to bolster his reputation, and his fund grew to in excess of £300m. As the fervour for technology companies imploded however, so did Manek’s fund. In the three years from its peak at the end of March 2000, it fell in value by 75%. The poor results did not stop there and its travails continued in the years that followed, before its belated closure in 2017. Over its twenty year life Manek’s fund proved to be a disastrous investment, falling in value by 56%, whilst the broad UK stock market delivered 236%. This yawning performance disparity meant that Manek was landed with the unfortunate sobriquet: ‘Britain’s worst fund manager’[iii].
The experience of Manek is a stark reminder of the dangers of making inferences about both skill and future returns from historic investment performance. Manek not only had the unique achievement of consecutive victories in a fund management competition, he delivered dramatic outperformance with real money in his eponymous fund for two and a half years. Yet this sustained period of material stock picking success proved a prelude only to years of disappointment. Given how misleading past performance can be, do we really rely on it when selecting active managers? The evidence would suggest so.
Chasing Past Glories
In a 2008 study[iv] Amit Goyal and Sunil Wahal observed the investment decision making of 3,400 intuitional investors between 1994 and 2003, which included 8,755 active manager selection decisions across assets totalling £627bn. The researchers were seeking to identify patterns in the hiring and firing of fund managers.
The results were clear. The institutional investors typically appointed new managers following a period of strong excess returns, only to witness the outperformance evaporate in the following years. This behaviour not only applied to new manager purchases, but also firing decisions. Managers were frequently relinquished after a period of poor results often to witness an upturn in fortunes.
What is of particular note in this study is the type of investor on which it trained its focus. These were not inexperienced, individual investors, but large institutions benefitting from sophisticated and seasoned insights from trustees and investment consultants. Investment practitioners who should have been well aware of the problems inherent in performance chasing, yet still they succumbed.
The findings of this research were corroborated in a broader context in a study conducted by Itzhak Ben David, Jiacui Liu, Andrea Rossi and Yang Song[v]. They analysed flows into US mutual funds between 1997 and 2011 and sought to observe the determinants of flows across a range of risk factors. The two primary drivers identified were Morningstar ratings and recent returns. Whilst the Morningstar rating evolved through the evaluation period, it is a quantitative metric based on peer group relative outcomes (risk and return). In other words, it is a measure based largely on past performance. The authors argue that investors seek: “easy to follow signals”.
It is not only that performance chasing is reliant on data that provides no guide to future returns, but that strong performance in one period can be a forebear of poor future returns. Research by Brad Cornell, Jason Hsu and David Nangian found[vi] that managers with recent weak performance earned higher benchmark relative returns in the future than those that had outperformed. The tendency of mutual fund returns to experience mean reversion shows that our propensity to sell stragglers and buy recent winners is not just pointless; it is often the exact opposite of what we should be doing.
Whilst performance chasing is endemic in mutual fund selection, it is not necessarily deliberate. Although some of us will explicitly target historic returns, many will consider a broad range of factors and undertake a rigorous due diligence process. The problem is as soon as we observe past performance it shapes our view on every other aspect of our research. The judgements we make about factors entirely unrelated to performance are indelibly tainted. We chase performance even when we are not consciously attempting to.
This damaging behaviour is driven by outcome bias. Its influence over our decision making is not just that it gives results pre-eminence over all else, but that it changes the way we perceive the other relevant issues.
The Origins of Outcome Bias
The classic origin study for outcome bias was carried out by Jonathan Baron and John Hershey[vii]. They asked participants to rate the quality of a hypothetical medical decision:
A 55 year old man with a heart condition was offered a bypass operation that would improve his standard of living and increase his life expectancy; however, 8% of patients who undertook such an operation died. The decision was taken by the physician to carry out the operation.
Participants were presented with this scenario twice (amongst an array of different questions) with a single difference – in scenario one the patient lived and in scenario two they died. Despite the result being irrelevant to the quality of the decision, and it being the same participant making the choice, the scenario that ended with success was rated as being a significantly superior judgement.
Outcome bias has also been shown to be hugely influential in areas where we might expect our views to show more stability – our ethics and values. Francesca Gino, Don Moore and Max Bazerman carried out a study[viii] where they presented participants with a range of scenarios featuring ethically dubious behaviour. In one case this involved a company knowingly selling a toy containing harmful chemicals that was hazardous to children. The ethical condemnation of this activity by participants was far greater when the outcome was negative – six children died – than in the positive version, where no children were injured.
Even when it should be inconsequential, results have a material influence on the views that we hold and the judgements that we make across many domains, and investing is no exception. Indeed, there are features of financial markets which make us particularly vulnerable to its more damaging consequences.
Outcome Bias and Randomness – A Potent Cocktail
Although it is easier to observe the negative elements of outcome bias, in many aspects of life it can be viewed as an effective adaption. A shortcut to making quick judgements based on limited information. I can with some confidence assume that a carpenter has skill by observing their finished item. I don’t need to watch her work with a chisel for hours.
The challenge for investors is that outcome bias is only useful when we can be confident that a good process will consistently result in positive results. If outcomes are unstable or subject to randomness then they become hugely misleading. Unfortunately, we don’t adjust sufficiently for this when using outcomes to inform our judgments (we are poorly calibrated). We can even be swayed by outcomes when an activity is purely and categorically the result of pure chance.
In a 1975 study[ix], Ellen Langer and Jane Roth asked a group of Yale University students to predict a sequence of 30 coin tosses. Although the overall outcome for each participant was fixed to be right 50% of the time, the study was designed so that half of the group enjoyed early success with their predictions and the other half struggled at the start. Emboldened by indications of their otherworldly foresight, those who were correct with their initial guesses were more likely to consider themselves to be better than average at predicting the toss of a coin. Startlingly, 40% of the students also believed that they could improve with practice (which reflects well on the ethos of Yale students, though not necessarily their grasp of probabilities).
The coin toss example is consistent with Langer’s broader research on the ‘illusion of control’, which is the idea that we consistently believe that random events are under our own or someone else’s influence. If we are prejudiced by the outcomes of a coin toss (where the randomness is certain), our desire to see order in the vacillations of capricious financial markets is a given.
Whilst the result of a coin toss is pure chance, active fund management combines some fortune and some skill. Although exactly how much luck or skill is involved is itself highly uncertain, our performance chasing behaviour suggests that we greatly overstate (either directly or indirectly) the role of skill. On an individual basis the influence of outcome bias on our investment decisions is profound, but we also need to consider how our collective susceptibility to the bias serves to compound the problem.
The Collective Impact of Outcome Bias
When we discuss behavioural biases it is typically about their influence on us as individuals, however, it is important not to neglect their impact on group behaviour. What does it mean if we are all subject to outcome bias? If the majority of investors are in the grip of outcome bias and engage in performance chasing behaviour as a result, then taking action that is contrary to that becomes impossible for most professional investors. Imagine the reaction of your clients and colleagues to investing money with a poorly performing fund.
Investing in underperforming managers and selling strong performers might be a beneficial strategy over the long-term, but it comes with profound career risk. It is acceptable to buy an outperforming manager that subsequently struggles – this is what we are all prone to do. Investing in a poor performer that continues to deliver weak returns is likely to be seen as unpalatable. Wasn’t it obvious that they had no skill?
Attempting to mitigate outcome bias and prevent performance chasing behaviour means overriding our natural instincts and also having a willingness to fail unconventionally. Neither of these is simple, but that does not mean there is nothing we can do.
How Can We Prevent Performance Chasing?
Outcome bias cannot be switched off. Whilst awareness is a starting point, it is evident from our continued performance chasing behaviour that it alone is insufficient. We need to make clear and focused interventions to change our behaviour:
1) Stop Using Performance Screens: Mutual fund performance screens are ubiquitous across the investment industry, and I have never worked in a team where they have not been employed in some form. Wealth managers, investment consultants, ratings agencies all use some form of historic performance screen to rank funds. Whilst each will differ based on the metrics and time horizon used, they are all fundamentally doing the same thing – assessing funds based on past performance. They are a behavioural disaster.
Outcome bias and the performance chasing behaviour that follows is difficult enough to avoid even if you are not actively employing tools that encourage it. Investors will argue that it is difficult to reduce such large universes of funds without some kind of performance filter, and also that it is only a starting point. However, we know that observing performance in this way will have a dramatic impact on the decisions we make. Virtually any other way of cutting down a potential investment universe would be better.
2) Create decision rules: A simple step to avoid performance chasing behaviour is to create fixed decision rules that strictly prohibit it. For example, you might state: “We will not purchase a fund that has outperformed its benchmark by more than 10% over three years”. This sounds straightforward but is fiendishly difficult to apply given that it runs counter to our typical approach to selecting active managers.
It will prohibit buying of those anomalous managers that continue to generate strong returns even after a stellar run. On average, however, it should be an effective means of avoiding the cost of purchasing active managers with a high potential for severe mean reversion.
3) Go Passive: The best behavioural interventions are the simple ones. Anything that requires behavioural discipline or continued effort raises the prospect of failure. Given this, what is the best way to avoid performance chasing in active mutual funds? We can restrict ourselves to buying only passive market trackers. This changes the context of the decision and entirely nullifies the influence of the bias from our fund selection decisions at least.
4) Specify the activity in which you believe skill exists. When investing with an active manager, we are taking the view that the underlying manager has some form of skill. We tend, however, to be very vague about what we actually mean by this.
When seeking to identify whether a fund manager is skilful, we need to be specific about the activity in which someone may possess it. If we suggest that they are skilled at ‘beating the market’ then we are entirely beholden to whether performance is good or bad, irrespective of the reason. If we instead make precise claims about where they may hold an advantage – for example, in identifying companies where the potential earnings growth is greater than market expectations – we can at least begin to assess whether the outcomes are directly related to an explicit process.
There is no such thing as a ‘skilled investor’ as much as there is a ‘skilled surgeon’ – I wouldn’t be entirely comfortable with even the most skilled urologist performing brain surgery on me. When discussing skill always ask – skilled at what, exactly? Then test it.
5) Understand how much randomness there is an activity. In his book ‘The Success Equation’[x] Michael Mauboussin suggests an intuitive check for resolving the quandary about gauging whether a particular activity is more subject to luck or skill – simply ask the question: is it possible to fail deliberately?
Take a lottery – it is impossible to purposely fail when playing; provided the ticket is correctly completed, the probability of success cannot be impacted by the combination of numbers selected, and the result is entirely arbitrary. Contrastingly, chess is a game dominated by skill, with limited influence from chance or randomness; it is easy to intentionally lose a game by recklessly sacrificing key pieces.
The more randomness there is in any given activity, the higher the threshold must for claiming that skill exists. Always apply the ‘can you fail deliberately’ test.
6) Extend your time horizons: Our susceptibility to outcome bias is greatly influenced by the time horizons involved. Whilst investment management is always a combination of luck and skill, the length of the assessment period alters how much we can glean from performance alone. If we assess investment performance over one day it can be considered to be pure luck, but as we extend the period skill can exert more of an influence. You can still be lucky over ten years, but the outcomes should be less influenced by chance.
We would not necessarily expect Warren Buffett’s stock picks to beat those of my six year old daughter tomorrow, but over a decade we would. The fact, however, that over a ten year period you couldn’t be 100% certain that the world’s most revered stock picker would beat the returns of a six year old tells you a great deal about the problems of performance chasing and outcome bias.
Unfortunately, if anything we are seemingly becoming more myopic in our investment judgements. Three years is considered long-term, and performance is increasingly assessed over ridiculously brief periods of time such as one month and one quarter. If you are assessing outcomes over such time horizons you should understand that they are likely devoid of meaning.
Performance chasing behaviour is, of course, not isolated to our selection of active fund managers. It is readily apparent in how we invest in everything from major asset classes to individual stocks. It is also not entirely driven by outcome bias. There are a multitude of factors at play; including our willingness to extrapolate recent history into the future and our susceptibility to a good story.
Active manager selection decisions are, however, the perfect breeding ground for outcome bias to run amok. The use of past performance to guide judgements seems to offer some sense to a highly uncertain environment – accepting that past performance is largely random may lead to some difficult introspection. Its importance is also embedded in the system. It is what everyone cares about – boards, clients, trustees, risk teams, CIOs – it is measured and therefore it matters, and usually a great deal to your career.
This is not to say that outcomes do no matter. Of course, all investors are seeking better long-term results for their clients. The problem is not that we care about performance, it is that we have become reliant on historically noisy data to make unsubstantiated or plain incorrect inferences about the future. If we want to invest in active managers, we need to think far more about decision quality and process, and far less about yesterday’s performance.
[iv] Goyal, A., & Wahal, S. (2008). The selection and termination of investment management firms by plan sponsors. The Journal of Finance, 63(4), 1805-1847.
[v] Ben-David, I., Li, J., Rossi, A., & Song, Y. (2019). What Do Mutual Fund Investors Really Care About?. Fisher College of Business Working Paper, (2019-03), 005.
[vi] Cornell, B., Hsu, J., & Nanigian, D. (2017). Does Past Performance Matter in Investment Manager Selection?. The Journal of Portfolio Management, 43(4), 33-43.
[vii] Baron, J., & Hershey, J. C. (1988). Outcome bias in decision evaluation. Journal of personality and social psychology, 54(4), 569.
[viii] Gino, F., Moore, D. A., & Bazerman, M. H. (2009). No harm, no foul: The outcome bias in ethical judgments. Harvard Business School NOM Working Paper, (08-080).
[ix] Langer, E. J., & Roth, J. (1975). Heads I win, tails it’s chance: The illusion of control as a function of the sequence of outcomes in a purely chance task. Journal of personality and social psychology, 32(6), 951.
[x] Mauboussin, M. J. (2012). The success equation: Untangling skill and luck in business, sports, and investing. Harvard Business Press.