Performance And Skill: Discerning The Signal From The Noise
By: Janet Rabovsky & Cameron Kerr
In some ways, it might be better if we lived in a world where outcomes were solely predicated on, and governed by, the amount of hard work, determination, and skill applied to a given task. Unfortunately, most of us are all too familiar with the timeworn, yet indisputable, notion that chance events often wreak havoc on the ‘best-laid plans of mice and men.’
Managing money – and, likewise, attempting to select external agents to whom to delegate this difficult job – is no different. The best investors may well have a special aptitude for assessing risk and reward, for researching trends that will move markets, for selecting securities that will outperform, and for constructing optimized portfolios. In short, they are skilled at managing money. Being imbued with skill then fosters an expectation that the results of these investors would surpass those of their less skilful peers. However, there is no special guarantee that riches will automatically accrue to ‘people of understanding.’ Unexpected circumstances – noise, for lack of a better term – can affect the results of the great and not-so-great alike.
Making a wrong judgment as a result of noise may damage a person’s or plan’s wealth. For example, if it is assumed that over the long run an active manager with skill will outperform an active manager with less talent, firing the skilful manager as a result of short-term noise that led to a temporary performance shortfall may be a poor decision from a financial perspective. That is, failure to appreciate the impact of noise not only can cause us to make wrong judgments, it can cause us to make wrong decisions as well.
In Chart 1, you can see that simply choosing a manager based on past performance does not provide guidance as to future performance. In fact, in our example, the underperforming managers are more likely to outperform over future periods. The ‘average’ remains the best performer for most of the period under review, suggesting a rotation of managers who have outperformed.
Illustrating The Impact Of Noise
The concept that noise plays a role in investment returns is implicit in the ‘past performance is no guarantee of future results’ boilerplate disclaimers attached to the bottom of tables of managers’ results. However, a hypothetical (yet instructive) example may help illustrate the impact of noise on the ability to make firm conclusions about the presence (or absence) of investment-related skill.
Many people in the investment industry would agree that a gross information ratio – that is, the relative return per unit of risk, or outperformance divided by tracking error – of 0.5 is very good. Suppose that a particular skilful manager could produce an outperformance of two per cent per year with a tracking error of four per cent per year. These assumptions yield a favourable information ratio of 0.5. Yet, despite being ‘endowed’ with skill, there is still a very good chance (19 per cent) that this manager would underperform the benchmark over the next three years.
One might ask, what length of track record would be required in this example to reduce these odds further, say to below five per cent? The answer is a longer period than one might suppose. A handy rule of thumb is that if the information ratio multiplied by the square root of the number of years for which that ratio applies is greater than two, there is about a 95 per cent chance for outperformance. Again, taking the hypothetical investment manager presented above with an information ratio of 0.5, one would need 16 years before the odds of registering an underperformance versus the benchmark had been reduced to just onein- 20. For this conclusion to be meaningful, over this entire period, the investment personnel would have to stay the same. And the environment – both within the firm and external market conditions within which the investment process was being applied – would have to remain unchanged.
The real world, however, is inherently filled with ‘known unknowns,’ not to mention ‘unknown unknowns.’ Given the role of ‘played by chance’ events, even a ‘good’ manager can produce bad performance from time to time, and a ‘bad’ manager can produce good performance over certain periods. But, as our example shows, it takes quite a long track record to be confident of one’s ability to sort the wheat from the chaff. Moreover, the two requirements noted above – continuity in both the personnel and market environment – are very seldom, if ever, achieved within the world of active management. This means that even long and deep datasets of past performance are unlikely to yield meaningful information regarding the existence or otherwise of a truly skilful manager.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -