In this article I will attempt to provide a brief overview of ten concepts that I believe are helpful for understanding investment performance. The latter is often presented as a binary issue – performance is either positive or negative. However I hope that, in reading what follows, it will quickly become apparent that appraising performance requires a much more nuanced approach than is sometimes adopted in the investment industry.
- Performance – the starting point for any analysis of investment performance is usually the total return of the portfolio. Total return measures the overall economic return on the portfolio – both income (interest and dividends) received and the capital gain of the portfolio are included. It is normally calculated on a time-weighted basis (see below). Total returns can be positive or negative. However, without any context such an absolute measure of performance can be misleading.
- Relative Performance – provides some of the necessary context for performance by specifying a benchmark against which the portfolio’s total return can be compared. For instance the performance of a UK equity portfolio would normally be compared to an index such as the FTSE All Share Total Return. For multi-asset portfolios, the benchmark will usually be a bespoke construction with different proportions given to different indexes according to the charity’s Strategic Asset Allocation (SAA). Each portfolio segment can then be compared to its individual benchmark (particularly useful if each asset has a different manager) while the total performance of the portfolio is compared to the total benchmark performance in order to see the overall picture.
- Indices – trustees need to be aware of the method of calculation for each index. Most indices, such as the FTSE All-Share, are arithmetically weighted according to each constituent’s market-capitalisation (number of shares multiplied by the share price). The larger the constituent as measured by market-cap the larger its potential impact on the index. Trustees should know that alternatives exist i.e. equal-weighted indices, although there is no clear consensus on whether or not they represent an improvement. Trustees should also note that some indices, for example both the Dow Jones and Nikkei 225, are rather bizarrely calculated on a share price-weighted basis.
- Costs – It worth understanding that the index’s total return assumes no transaction costs even though the constituents of the index may change periodically. In addition the returns normally assume that dividends are received on the ex-dividend date when in reality dividend payments are often received a couple of months later. Finally an index does not pay any investment management or custody fees. For all these reasons even an index-tracking fund is unlikely to deliver the same level of performance as an index. Thus some portfolio managers like to emphasise that their returns are ‘net’ as compared to the ‘gross’ returns represented by the benchmark.
- Time-weighted vs money-weighted – the industry standard is for time-weighted returns since the portfolio manager does not (normally) control the capital inflows and outflows into the portfolio. However for the charity itself, money-weighted returns can also be a useful measure particularly if the charity is active in making asset allocation changes. For example suppose a charity (of its own accord) allocates £10m to an investment manager who then delivers a 10% total return in the following year. Assuming no income is withdrawn (the total portfolio value is thus £11m), now suppose the charity adds a further £29m to the same manager (taking the total now invested to £40m). If the manager delivers a total return of -5% in the next year, then on time-weighted basis the total return is 4.5% or 2.2% per annum. It can be argued that this is the correct figure on which to judge the investment manager since he/she was not responsible for the decision to invest more money. However, on a money-weighted basis, the total return over the two years is -4.1% or -2.0% p.a. due to the charity’s decision to add extra capital at what (with the benefit of hindsight) was the ‘wrong time’.
- Performance Attribution – assuming we stick with the time-weighted return methodology, which is the industry norm, then the relative returns can be decomposed into constituent parts through attribution analysis. For a multi-asset portfolio this means breaking down the relative performance three ways: the asset allocation effect, the stock selection effect and the interaction effect. Consider a portfolio which holds 90% equities and 10% bonds and whose benchmark consists of 80% equities and 20% bonds. During the year the portfolio delivered a 11.2% return – the equities portfolio had a return of 12% and the bond portfolio had a return of 4%. In the same year the benchmark return was 9.2% – the equities index returned 10% while the bond index returned 6%. Note the equities index outperformed the bond index so being overweight equities relative to bonds added value from an asset allocation perspective. The asset allocation effect is calculated by adding the benefit of being underweight bonds (0.1-0.2) x (6%-9.2%) and the benefit of being overweight equites (0.9-0.8) x (10%-9.2%) which together total 0.40ppts. With regard to stock selection, this was clearly positive in equities (12%-10%) x0.8 but negative in bonds (4%-6%) x 0.2 with an overall positive impact of 1.20ppts. The interaction effect captures value-added which is not solely attributable to the asset allocation and stock selection decisions. It is positive when outperformance is overweighted and negative when underperformance is underweighted. In the example above this was positive (0.9-0.8) x (12%-10%) + (0.1-0.2) x (4%-6%) = 0.4ppts. So the total outperformance of 2% (11.2% v 9.2%) breaks down into 0.4ppts from asset allocation, 1.2ppts from stock selection and 0.4ppts from the interaction effect. Note it has become common practice to include the interaction effect within stock selection. Note also that for a single-asset portfolio, such as equities, performance attribution will normally be decomposed into the returns from sector allocation and stock selection although the principles involved are the same.
- Time/‘Mean Reversion’ – charities normally receive reports detailing the performance of their portfolio on a quarterly basis (some charities may even want monthly performance numbers). This is, of course, absolutely fine but trustees should be aware that the shorter the time period the less the scope for the manager to demonstrate investment skill and the greater the propensity for other (random) factors to dominate the outcome. A related concept is that of ‘mean reversion’ which was first observed by Sir Francis Galton in the nineteenth century. The basic idea, as applied to finance, is that returns can be very unstable in the short run but much more stable in the long run. This is because extremely positive or negative results tend to be followed by results that are closer to the average. Charity trustees need to comprehend this phenomenon when they are selecting managers – do not simply go for the manager with the best (short-term) performance numbers but rather spend more time on appraising the manager’s investment process and their team/organisation.
- Risk – investment risk is the chance that the return achieved will be different from that expected. Risk includes not only ‘downside risk’, the possibility of losing some or all of the original investment, but also ‘upside risk’ (returns exceed expectations). Note also that risk is normally considered on an expected (ex-ante) basis, since it is the future that is uncertain, rather than on an actual (ex-post) basis. Several different measures of investment risk have been proposed. At its simplest, and as alluded to above, it is the risk that the investor will suffer a loss to the initial investment (capital). Another measure, Volatility, considers in statistical terms how the value of the portfolio may vary over time. Note that this is based on analysing past price movements and extrapolating from this data set going forward. A related concept is that of Tracking Error which estimates the portfolio’s relative volatility compared to the benchmark index. Another related concept is that of Value-at-Risk (VAR) which seeks to specify how much a portfolio might lose (with a given probability), given normal market conditions, in a set time period such as a day. It is popular in the hedge fund industry but much less so with mainstream portfolio managers.
- Risk-adjusted returns – there are a number of equations that seek to estimate risk-adjusted returns. The Sharpe Ratio is most applicable were absolute returns are the focus. It is defined as the total return of the portfolio less the risk-free interest rate, divided by the volatility of the portfolio. It thus seeks to measure the return generated by each unit of absolute volatility (risk). This compares to the Information ratio which is most applicable where relative performance is the focus for analysis. It is defined as the return of the portfolio less the return of the benchmark index, divided by the tracking error. The Information Ratio thus seeks to measure the amount of relative return achieved per unit of tracking error (relative risk).
- Capital Asset Pricing Model (CAPM) – finally at the risk of complicating matters further, it is worth familiarising yourself with a few concepts associated with the CAPM which was developed in the 1960s by several economists building on Harry Markowitz’s Modern Portfolio Theory (MPT). Thus Beta is defined as the covariance of the portfolio’s returns with its benchmark returns, divided by the variance of the benchmark’s returns. A beta of 1.5 means that on average for every 1% change in the value of the benchmark, the portfolio’s value changes by 1.5%. Note that portfolios with betas of less than 1.0 are frequently described as defensive as they ‘should’ fall by less than the benchmark index in a bear market. Note once again that this is an expected result, which may not occur in future, based on analysing past data. Also derived from the CAPM, Alpha is a measure of how much the portfolio’s actual returns exceed expected returns, given its Beta. It is sometimes used as a short-hand way of quantifying whether the investment manager is demonstrating skill by adding value over and above what the Beta of the portfolio would imply for performance. Lastly note that the CAPM model itself has been heavily critiqued which, in turn, has led to the development of alternative so called Multi-Factor models (CAPM is a single factor model). With a Multi-Factor model, several betas are calculated to quantify the portfolio’s sensitivity to a variety of macroeconomic factors such as interest rates, commodity prices, company size, currency, etc.
Simon Jaffe, Director OLIM Investment Managers