Measuring and evaluating investment performance

MEASURING AND EVALUATING PORTFOLIO PERFORMANCE

Measuring and evaluating a portfolio’s performance tells us something about how a portfolio performed relative to its investment goal or objectives. This is an essential step of any investment journey. If a portfolio is not meeting its objectives, it runs the risk of falling short of meeting desired long-term investment goals. 

However, interpreting performance figures is one of the most misunderstood areas of the investment management function – and these misinterpretations wreak significantly more damage on members’ outcomes than poor investment performance. So, trustees must be absolutely clear about what performance numbers can tell them about the efficacy of a portfolio, before they effect a change that may not necessarily be in members’ best interests. 

What follows is a compilation of some important points and insights to raise the financial consciousness on how to better evaluate the meaning of investment performance and its role in the broader framework for driving better outcomes for members.

Performance driven investment

Perpetuating the cycle of disappointing investment outcomes
Today, performance measurement and monitoring is predominantly carried out in reference to whether a portfolio has delivered superior results against a pre-defined benchmark or industry peer portfolios. In many cases, the benchmark is a proxy for an alternative investment strategy (usually a capital market-related index or a composite of multiple indices) that is designed in a much simpler and more cost-effective way. If the portfolio outperforms the benchmark, it is delivering an efficient outcome. Performance surveys publish comparative industry-wide portfolio performances and allow investors to monitor the quality and consistency of a portfolio pertinent to the environment in which it operates. If a portfolio ranks high relative to industry peers, it is deemed to be competitive within the context of similar offerings in the market.

Both benchmark and peer cognisant metrics offer retrospective comparative performance insights over a specified period, and is taken at a particular point in time. Both metrics have also become the easiest on which to base investment choices, simply because it is easy to communicate and obtain. However, both metrics have limited actionable insights in identifying how portfolios should be selected, whether securing better value for money or securing members’ long-term investment goals.

In order to help clients manage expectations, make meaningful assessments and settle on the right investment decisions, we need to rethink what is being measured and why.  

Past performance

Basing investment decisions on past performance causes accidents
Basing investment decisions on past performance is a common mistake many clients make. The problem with evaluating past performance is that a snapshot of history is never an accurate indication of the future, in any context, including investment performance.

Everyone wants the best performing portfolio, so it is a natural tendency for clients to anxiously compare their portfolio’s performance relative to its mandated benchmark as well as to what other portfolios in the market are doing. Clients who are not with the best performing portfolio experience regret – an experience that could lead them to impulsively question whether their current portfolio is doing what their fund requires.

However, pinpointing which portfolio will be the best and when it will deliver clients’ the best is easier said than done. This is because all investment types, strategies and asset managers behave differently and move through different periods of outperformance or underperformance depending on prevailing market and economic conditions.

The reality is that no one ever really knows when various market and economic cycles will play out. Knowing which portfolio will be the best performing and when would be the right time to invest in that portfolio is near impossible. Despite the numerous warnings on the dangers of basing investment decisions on past performance, clients are still easily tempted to switch to the next best thing, disinvesting prematurely when they see a portfolio underperforming for a period and locking in losses.

Asset allocation

The more predictable driver of long-term investment performance
Global research¹ has shown that the long-term variability of returns are more stable, more predictable and, hence, more controllable through the asset allocation decision than they are through active management skills (i.e. stock selection and market timing). Simply put, what this means is that being a top performer is more of function of the excess risk taken than it is of asset manager skill.

This does not mean that the highest returns go to the portfolio that gets its asset allocation timing right. Rather, it means that the differences between long-term fixed asset allocations are what account for whether a portfolio’s absolute returns are on top.

We know that growth assets such as equities and property outperform defensive assets such as bonds and cash over time, regardless of skill. Although growth assets are more likely to produce higher returns over time relative to defensive assets, they do also exhibit a relatively higher dispersion of returns. The return dispersions on growth assets are unpredictable over short-term periods and the magnitude of the dispersions represents the excess risk assumed for their higher return potential. That is, the level of uncertainty that a portfolio’s returns will be different to what they were expected to be.

Although the wide range of returns experienced by growth assets can be unsettling at first sight, these assets have delivered the most impressive performance relative to bonds and cash over the 50-year period. In fact, not only did growth assets deliver better monthly returns relative to the other asset classes over this period, they shared the highest percentage as best performing asset classes, even though they experienced the most negative months. Therefore, the differences in the mix between growth and defensive assets across a spectrum of portfolios can, and do, translate into significant differentials in performance. For example, a portfolio that has a long-term strategic (fixed) exposure of 65% in equities is expected to outperform a portfolio with a strategic exposure of only 63% in a bull market for equities.

When clients are faced with retrospective performance comparisons (benchmark or industry related), the most important measure that matters here is knowing whether the best performing portfolio is on top as a result of accentuated positions which amplified small successes relative to their portfolio. What this then implies is that clients need to evaluate performance comparisons on genuine like-for-like basis, and assess the differences in mandated strategic asset allocations between the different portfolios.

Drawing conclusions solely on performance rankings without understanding the reasons behind return differentials can lead to sub-optimal investment decisions. Understand that if it was a slightly higher equity exposure that led to the best performance when markets were rising, so too would it hold that that same equity overweight would also potentially lead the portfolio to underperform when markets are falling. Often, the winners in retrospective performance comparisons are not a reflection of the best portfolio, but merely the bravest rewarded for the magnitude of risk taken.  

The key point to understand is that one portfolio is not necessarily better than another – it would simply be a better reflection of the risk appetite of the trustees (i.e. a board of trustees might prefer to hold a higher equity exposure at all times, but then they take the chance that returns may be very different to what they expected).

Benchmarks

Different benchmarks mean different risk referencing
Another factor explaining return differentials between portfolios is the fact that different portfolios employ different underlying benchmarks, particularly equity benchmarks.

There is no perfect index, but a key attribute of an appropriate index is that the index should have adequate risk diversification, representing good economic equity exposure in terms of the South African equity market. After all, market indices are often used to benchmark active performance, so it is critical that they are a true reflection of markets. Benchmarks therefore need to be transparent, investable, replicable and readily available.

Historically, significant structural shortcomings have seen concentration risk creep into South African equity indices. Concentration risks exists when a few shares form a significant part of an index (by market capitalisation), thereby dominating the index from a risk and return perspective. The Johannesburg Stock Exchange has experienced concentration in its indices in the past and have launched new indices over the years in an attempt to address it. The SWIX, or Share Weighted index, was an improvement on the ALSI, or All Share index, and the Capped SWIX was a subsequent improvement on the SWIX. Again, neither index can be deemed perfect and it is in the best interest of all investors to continually research and identify good benchmarks that are a true reflection of markets. Good benchmarks increase the proficiency of performance evaluation, highlighting the contributions of skilful managers. Poor benchmarks obscure manager skills. Good benchmarks enhance the capability to manage investment risk. Poor benchmarks promote inefficient manager allocations and ineffective risk management. They also increase the likelihood of unpleasant surprises, which can lead to counterproductive actions and unnecessary expense on the part of the investor.

The tables below show the sources of risk across each of the three JSE indices.

Super sector breakdowns as at 31 August 2020

Top 10 holdings in the various listed indices as at 31 August 2020

Given that the various JSE indices display varying levels of concentration and diversification, it naturally follows that the performance differentials between these benchmarks are not always trivial over time. For example, the SWIX performed significantly better than Capped SWIX over all periods from YTD and beyond as at the end of August 2020.

What this tells us is that even if we are presented with two portfolios of identical strategic equity allocations, performance differentials would exist between them if each portfolio had a different underlying benchmark or performance target mandated to them.

For example, of all three indices, the ALSI has the highest resources sector exposure. When the market collapsed in 2008, resources were instrumental in leading the collapse. It should be no surprise then that growth portfolios (with a heightened asset allocation to equities) referencing the ALSI would underperform similar growth portfolios not benchmarked to the ALSI.  However, in the three-year period preceding the crash, exactly the opposite outcome was reflected. More recently, the underlying ebb and flow of super sector-driven performance still influences the return trends of portfolios with different benchmarks.

Another example looks at portfolios that employ the Capped SWIX index. This is an index in which all constituents with a weight greater than 10% in the index will be capped at a fixed level of 10%. The Capped SWIX addresses the diversification and concentration concerns inherent in the SWIX (which addressed the resource super sector concentration prevalent in the ALSI, but not individual stock concentration). If we compare a Capped SWIX-referenced portfolio to an ALSI or SWIX-referenced portfolio, where a share like Naspers could represent well over 20% of the index, we would expect the Capped SWIX-referenced portfolio to underperform the ALSI and SWIX-referenced portfolios during a time when Naspers rallied.

In both examples, the issue of comparably higher or lower absolute returns is a function of risk referencing to different benchmarks. This will necessarily lead to different performance outcomes – up to a point – depending on the index concentration biases at play.

Managing concentration risk reduces the probability of a single market event affecting a portfolio’s value. The management of a portfolio’s concentration risk, therefore, should form an integral part of the portfolio management process. This could take the form of applying risk allocation limits on single stock exposures on both an absolute and relative basis. For example, monitoring single stock risk exposure within certain parameters, say between 10% and 15%, of total portfolio risk. It must be pointed out that this level of risk mitigation is more applicable to equity-only portfolios where this concentration risk is exacerbated relative to multi-asset portfolios.

So, if we are comparing performance figures between various portfolios, it’s not enough to only ensure that they are genuinely comparable from a strategic asset allocation perspective, but we also need to assess each portfolios performance in the context of their mandated benchmarks. Because of the varying magnitude of risk biases inherent in each index, there will be times when the performance differentials between like-for-like portfolios that employ different underlying benchmark indices could be significant.

Appropriateness

Good performance does not necessarily mean the right performance for your needs
Because pension funds have their own unique needs and risk profiles, an appropriate portfolio should be one designed to deliver the highest likelihood of meeting required investment objectives over the long term and in line with specific risk constraints. The validity of a portfolio must be considered in the context of a fund’s own long-term objectives and risk budget.

What is not immediately obvious from performance figures is the amount of risk taken over time to deliver the performance. Today, most portfolio comparisons are supplemented with a risk measure alongside the performance figures over different time periods. But, it is only when looking at how the risk profile of the fund has changed over time that you really get a better sense of whether a portfolio's design manages risk carefully and consistently, within a certain ‘risk budget’. In other words, assessing whether the portfolio has been positioned to take on varying degrees of excess risk in the pursuit of delivering superior performance.

Depending on portfolio management guidelines, portfolios can, and do, take big investment positions that can work out well and set them apart as a top performer for a period. But, they can also go wrong and it could then take a portfolio a long time to make up the lost ground, let alone deliver the returns fund’s need to achieve investment objectives. Taking more risk does not always equate to higher returns. Ultimately, funds need to choose the portfolio that matches its appetite for risk. The only way to do this is to get a deeper understanding of what that risk profile has looked like over time.

Performance differentials can also be a function of the differences in how risk management is defined in a portfolios’ mandates. In other words, is the portfolio’s risk constraint defined as capital loss, absolute return volatility or deviation to the benchmark (tracking error)? The answer to this could yield significantly different structural designs between portfolios that determine how much excess risk they can assume. 

So, when comparing headline performances, it is important to establish what whether a portfolio is one that demonstrates a consistent risk profile or one that demonstrates considerable variations in its risk positions through time. Funds that remain conscious of their long-term investment objectives as well as the degree of risk they are willing to take to reach those objectives will stand a better chance of considering and evaluating those portfolios right for them. 

Performance path

The importance of a meaningful timeframe
Performance figures are reflected over a particular point-in-time and taken over specified time periods (one-, three- and five-years). Often, performance is also reported in rolling windows of time over such periods. The problem with this is that these performance numbers are as much a function of the latest month’s performance data as it is the dropping off the first month’s performance data. That’s because the reporting timeline moves onto the next month, and then the next month and then the next month after that.The graph below is one such example which illustrates this point using a timeframe from October 2018 to October 2019, but there are many other periods one could use to arrive at a similar conclusion.

Performance can shift wildly because of events that may have happened one, three or five years ago.Unless funds can assess the full performance history of a portfolio, it is very difficult to evaluate its “bigger picture” performance path. For instance, a portfolio’s performance path could have been lumpy. It could have experienced a number of years of bad performance and only recently delivered a short burst of great performance – and this ‘wild ride’ may not be evident in the headline numbers or in specific snapshots in time.

The reality is that different portfolios require demand different time-frames for the performance potential to manifest. Once a fund has aligned itself with a portfolio that meets its needs and risk budget, it is important to understand the time-frame it requires to reflect its performance potential and allow enough time for it to deliver.

Significance

Be clear about what performance numbers say about the efficacy of a solution
A performance comparison can demonstrate the absolute performance differentials between similar portfolios by an analysis of their past returns. While examining differences in performance, consider whether these differences are significant in the context of:

• Switching costs
  Is the difference between the ‘best’ and the ‘mediocre’ portfolio of a magnitude that matters? Research has shown that switching portfolios in a long term pension fund structure can lead to costs that outweigh subsequent top-quartile outperformance over the long term. This means that recouping losses on the back of unnecessary portfolio switches can be highly problematic.
• Fees
  How does it compare to differences in fees? Are the headline performance comparisons represented consistently in terms of the fees accounted for (i.e. gross of fees versus net of fees)? Once fees are accounted for, has time been taken to analyse associated fee structures and determine whether they are appropriate and fair for what the portfolio seeks to deliver? Fees can have a significant impact on performance figures especially if portfolios with different level and fee structures are being compared.

In the long term, it is important to evaluate whether the differences today of real consequence to members’ future outcomes or a small distraction in the overall context of prudent trustee management.

Conclusion

The purpose of this document has been to raise clients’ financial consciousness on the pitfalls of focusing on up-to-the-minute performance and taking absolute performance comparisons to inform investment decisions. The very little difference in future performance on offer is more likely to be overshadowed by the certain costs incurred by doing so.

Headline performance comparisons should be used as a tool to ensure a more concerted effort is made to evaluate a portfolios strategy and whether it is still appropriate for the objectives and risk appetite of the fund. The insights gained from peeling back the layers of performance differentials between portfolios can be tremendously valuable in helping clients gauge what drives performance and how their portfolio is expected to perform. Once this is clear, clients will have a better understanding of what should really matter to them and whether their portfolio is aligned to their needs and expectations.

Reference

1 Determinants of Portfolio Performance; Brinson, Hood, Beebower, 1986 The Asset Allocation debate; Vanguard, 2007