The poor performance of the global share benchmark (MSCI Index) over recent years has highlighted the need for managers to focus on absolute risk rather than tracking error (TE) — a measure of relative risk — to assess a fund’s absolute performance.
What is tracking error?
Put simply, tracking error is a measure of the extent that a portfolio’s performance varies from the benchmark’s performance — the standard deviation of the difference between portfolio and benchmark. While there are different methodologies to calculate TE, the above definition is appropriate.
So while TE has a purpose in understanding a portfolio’s positions away from benchmark, it does very little to explain the absolute volatility in returns. TE statistics have to be examined in the context of the relative performance, as well as absolute performance and risk statistics. The reality is that this rarely happens and a common misconception in investment circles is that high tracking error equates to high volatility in returns, when in actual fact it can mean low absolute volatility in returns.
The industry is increasingly using TE as the preferred measure of risk. With this in mind, and the general investor demand to reduce risk, the temptation for managers to reduce their TE has become too great. Table 1 demonstrates this fact.
Over the past five years, two trends have occurred in the TE of international share managers. Firstly, the median TE (the bold line) has declined significantly from 6.1 per cent to 2.1 per cent. Even more significant is that the range of manager TEs has halved.
While TE is generally accepted as a more sophisticated measure of volatility, it can be dangerous if not taken in the right context. The TE of a portfolio that consistently underperforms its benchmark can be very low or even zero and this is not an ideal outcome. Similarly, a portfolio can have a high tracking error and always outperform the benchmark (but with varying magnitude).
More importantly, a focus on TE implies that benchmark-like volatility is good and that, by default, a portfolio that looks like the benchmark is also good. The benchmark itself has some fundamental flaws and structural biases that make it an inappropriate basis for constructing portfolios.
If, as most commentators seem to agree, markets have become more volatile, it follows that the benchmark is going to be more volatile too. So what use is a low TE if that means tracking a benchmark with volatile returns? Table 2 shows the standard deviation (absolute volatility) of the benchmark and, for good measure, the median manager. We saw in Table 1 that, over the past five years, the TE of international share portfolios has been rapidly decreasing. Despite that fact, the chart below shows us that the absolute volatility of returns increased over most of this period. So, has a lower TE reduced volatility for the client? The answer is simply no.
Standard deviation
What is most important from the investor’s perspective is the absolute volatility of returns — in other words, the standard deviation of returns.
A portfolio with low standard deviation of returns would deliver a more desirable experience for the investor. This can occur with or without a high TE (that is, a manager can have a high or low TE and still produce a low standard deviation in returns) and there are plenty of examples of both. Standard deviation is more relevant if reducing absolute volatility is the aim. Neither measure is necessarily an indicator of either absolute or relative performance.
Assessing a manager’s portfolio from a risk perspective is important in any analysis. The challenge is to ignore the herd and actually use a risk measure that better reflects what the investor is seeking.
Peter Walsh is an investment specialist with Zurich Financial Services Australia .
