Issue: June/August 2008
Getting to grips with investment risk
Most people have a good feel for colloquial definitions of risk and an understanding that, intuitively, one should expect some extra reward for exposing oneself to an uncertain outcome. The ever-expanding number of tools and models made available to investors and their advisers, accompanied by a sea of jargon, makes it increasingly difficult for even experienced investors to understand the benefits and limitations of these measures. And using the wrong measure can lead to sub-optimal decision-making.
The importance of risk is not merely limited to the probability of outcomes being very different to expectations, but can be extended to the concept of risk-adjusted returns where the merits of different investments or the skill of the underlying manager can be evaluated.
"It is increasingly difficult, even for experienced investors, to understand the benefits and limitations of measurements of risk"
Absolute risk and return measures
Lets deal first with the concept of the total riskiness of a portfolio, link this to total returns and then to tools that measure risk-adjusted returns in an absolute sense. These absolute measures (as opposed to relative measures) are arguably the most appropriate to any investor in that it is what "hits the pocket". Total risk is often referred to as volatility and is usually measured by standard deviation. Standard deviation is often not well understood and is actually very simple to get to grips with.
If an investment has an expected return of 15% and an annual standard deviation of 25% then, two thirds of the time, the funds return is expected to fall between -10% and +40% (15% + or- 25%) in any given year. This measure becomes very useful in determining the likelihood of underperforming or outperforming things such as inflation, interest rates or the value of the initial investment.
Risk-adjusted return measures (such as the Sharpe ratio) use standard deviation as a measure of volatility to compare different investments on a risk-adjusted basis. This aims to determine whether the investor was compensated for taking additional risk compared to the alternative of investing in the "risk-free" asset, namely cash. Simply, it measures the performance of the portfolio relative to cash and adjusts this using the volatility of the underlying investment.
Relative risk and return measures
Risk and return numbers are measured relative to a benchmark, as well as risk-adjusted returns derived from these, , measure something quite different. Active risk is typically referred to as a tracking error, which simply determines how likely it is that the portfolio's performance may be different to its benchmark. This has nothing to do with the total risk (volatility) of the portfolio, but indicates how different the composition of the portfolio typically is to that of the underlying benchmark.
Active return, or alpha, simply measures the difference between the portfolio return and the benchmark return. Why should the benchmark return be relevant? The usefulness really lies in attempting to determine and attach some level of skill to the portfolio manager.
The question is whether the investor derived any value from having his portfolio actively managed after fees, and, if so, how significant was the value added through skill? Put simply, did the portfolio manager generate excess returns by building a portfolio that is different to the benchmark? Logically, if the manager is skilful, the bigger the positions taken relative to the benchmark, as measured by active risk or tracking error, the higher the expected excess performance over the benchmark. By measuring actual performance relative to the benchmark (alpha) and comparing this to how aggressively his view was implemented (tracking error), one can develop a sense of how much any excess performance had to do with skill. This is what an information ratio measures - the performance added relative to the size of the positions taken.
There are however some major pitfalls!
Where a manager runs a portfolio strictly around the portfolio's benchmark, active positions are regarded as the difference between the benchmark weight in a stock and the weight of that stock in the portfolio. For example, if a manager views a share as expensive and only holds 2% of that share in the portfolio, versus a benchmark weight of 5%, this would indicate a negative active position of 3%. Nevertheless, the portfolio still holds 2% in a share that is considered overvalued and is expected to underperform! If the share does subsequently underperform, the portfolio will do better than the benchmark, generating positive alpha. If this approach to investing is adopted, it is reasonable to use active and return measures to test for skill.
What if, however, another portfolio manager sees the world differently? This manager may be more interested in generating the highest total return over time for the investor with little reference to the benchmark.
Remember, this approach need not result in a riskier portfolio. Here, a stock that has a weighting of 2% in the portfolio will be regarded as positive 2% position regardless of the benchmark weight. If the manager doesn't like a share, it gets left out. In this case, a skilful manager may well generate superior returns over time for investors. However, as the stock weights may differ substantially from the benchmark (as positions are not being measured with reference to the benchmark, but relative to not holding the share at all), the manager will be seen to have taken substantial relative positions and may run a high tracking error. Hence, measures such as information ratios may in this case look relatively poor and give a very distorted indication of skill and value added.
What then are our conclusions?
Keep it simple. If you want to know how risky a portfolio is, simply check how volatile it is - this is usually represented by standard deviation in the surveys. Be careful though. Most general equity portfolios exhibit very similar levels of volatility and often this changes somewhat over time and one cannot always expect the risk "ranking" of portfolios to persist. So, if volatility is similar across portfolios, don't stress too much about evaluating portfolios on small differences in risk for a given portfolio type. In these cases, simply compare returns over time.
"Using the wrong risk measurement tool can lead to sub-optimal decision making..."
Where the volatility of portfolios differs significantly within a sector, do incorporate risk for the basis of comparison. This helps us determine whether we are comparing apples with apples. You can't compare the performance of two portfolios when one has a volatility of 15% and the other 5%. Where portfolios have similar volatility, you can compare returns. Where absolute return type portfolios have different risk profiles, it is valuable to use risk-adjusted returns such as Sharpe ratios to validly compare them.