Mercer's David Scobie explains how referring to past performance has its uses when selecting investment products, but can be a comfort blanket with many holes

By David Scobie*

If you have attended a fund manager presentation or browsed related marketing material, you will be well aware of the prominent role historical returns can play in promoting a manager’s capabilities.

Despite the ever-present health warning that “past performance is not an indicator of future performance,” there are plenty of theories in behavioural finance literature to suggest that investors allow past performance to influence their decisions.

This is entirely understandable – it is one of the few pieces of readily available information upon which investors can base their decision-making. Therefore, it is important to consider whether there is any meaningful information in such data that can assist investors in consistently selecting the winners of tomorrow.

Crunching the numbers

Mercer recently examined the subsequent performance of a wide range of strategies based on their past performance relative to their peers.

Performance over three- and five-year periods were considered over a 20-year period for survey universes including Global, US and Emerging Market Equities.  

The study found that past performance is a very weak indicator of future performance, both in terms of the probability of future outperformance and the scale of the performance delivered. 

Past outperformers and past underperformers both had close to a random (50%) chance of outperforming in the future.  (Preliminary analysis on the impact of survivorship bias suggested that it’s influence on the findings was relatively minor.)  More detail on the study can be found here.  

This analysis should not be interpreted as suggesting that there is no persistence in performance for equity strategies at all.

The range of outcomes within each past performance group is broad, and invariably there will be some managers within each universe that do have the requisite capabilities to deliver returns that can persist over a long timeframe.

However, what this analysis does suggest is that, on average, allowing three- or five-year relative performance to influence manager selection is unlikely to improve outcomes for investors.

It is important for investors to recognise that performance data, particularly over the shorter-term, tends to contain significant elements of market noise, luck and stylistic head- and tailwinds.

Ultimately, performance attributed to these factors will be more susceptible to reversion in the future.  Investors may underestimate the role that these factors play in a track record and mistakenly interpret positive or negative past performance as a gauge of a manager’s skill.

The reality is that skilled managers will experience periods of sustained underperformance due perhaps to an unfavourable style or bad fortune, while unskilled managers will have periods of strong performance for the opposite reasons.  

Over the very long term, skilled managers are able to deliver attractive relative returns because their investment acumen will persist, while luck, noise and investment styles tend to balance out over cycles.

To illustrate this point, consider the eight top-performing Global Equity managers in the Mercer database over the past 15 years.

These managers have, on average, outperformed the median strategy by 2.7% p.a. - an excellent outcome for investors over such a long time horizon.

Yet strikingly, despite their stellar long-term track records, these strategies have, on average, underperformed in nearly six of the 15 calendar years captured.

An investor in these strategies would have been experiencing below-average returns 40% of the time, and those influenced by one-, three- or even five-year performance may have been inclined to give these strategies a wide berth at numerous points during those 15 years.

Options for investors

Being aware of one’s behavioural biases, in particular a natural tendency for performance chasing, is undoubtedly a good first step. 

When undertaking fund manager selection, it is tempting to recommend constructing manager short-lists without any reference to track records, meaning that such data is less likely to dominate their perception of a manager’s skill.  

However, this is an extreme approach. 

When dissected with some sophistication, and in conjunction with qualitative analysis, past performance can yield meaningful insights as to whether returns have reflected true skill at, or not at, work.  (Qualitative analysis in this context refers to factors as the strength of the manager’s investment team, philosophy and process, and the ability to effectively implement stock ideas in the market.)    

For example, Mercer considers questions such as the following:  Were the returns of a manager with a natural bias (versus the benchmark) to small cap stocks largely influenced by the fortunes of that market segment?  If the manager has a growth- or value-type investment style, does their performance resemble an index which reflects that inherent approach?  Has the investment team significantly changed, meaning that historical returns were delivered by capability no longer in place?  Was past performance achieved with the investment firm having far lower (or higher) funds under management, meaning future outperformance could be far more (or less) challenging? 

These questions are not easy for the average investor to deliberate upon, but moving from “naïve” to “informed” analysis of past performance is critical to deriving useful insights.  The nature of this conundrum is summarised in the table below.

Assessing Future Outperformance of Fund Managers

The thinking described in this article underpins the growing area of Factor Analysis, where investors focus on detecting what particular return and risk factors may be at work on an ongoing basis and which drive investment performance.  If these factors can be identified, they could potentially be obtained at cheaper cost via systematic or more passive-like strategies. 

“Not everything that can be counted counts, and not everything that counts can be counted.” - Albert Einstein

In summary, investors’ behavioural biases mean that returns generated in the past are often used as an uncomplicated means to indicate what future returns may be. 

Unfortunately, this is too simplistic an approach, and can be hazardous. 

The historical performance of a fund manager has the potential to be a useful source of information, but if not overlaid with an understanding as to exactly how and by whom that performance was generated, reliance on such data may be detrimental to arriving at optimal portfolio outcomes.

*David Scobie is Head of Consulting at Mercer Investments, based in Auckland.

This article does not contain investment advice relating to your particular circumstances. No investment decision should be made based on this information without first obtaining appropriate professional advice and considering your circumstances.