Controlling risk of losing money
The continuing falls in equity markets have brought home the reality to investors that equity investment has its risks. Across Europe, pension fund solvency and managing investment risk is an increasing concern for all trustees and sponsors responsible for funded pension arrangements. What does this mean for risk budgeting and the use of risk controls in portfolio management?
The last three years serve as a salutary reminder that the main investment risk is of being in an asset class that fails to keep pace with a pension fund’s liabilities. Yet much of our industry chooses to focus the risk budgeting process on a very different type of risk – that of failing to keep pace with a benchmark index. How many trustees spend more time and effort quizzing their fund managers over a 1% or 2% shortfall against the benchmark, than asking them about the absolute level of returns?
Risk budgeting relative to the benchmark has its place. Active management is all about taking and understanding risks to gain the corresponding returns. The danger is taking this element of the overall risk out of context, and forgetting to devote sufficient time and effort to more important, strategic investment questions. Benchmarks are simply a means to an end, a tool to serve a greater goal. By overemphasising its role, and giving it a golden position of ‘no risk’ in the risk-budgeting process, we give it a life of its own and an importance which it doesn’t deserve.
Error or opportunity?
Obsession with returns relative to the benchmark has spawned a whole industry and even added new terms to our vocabulary, such as ‘active risk’ and ‘information ratio’. Let’s take, for example, the concept of ‘active risk’. In itself, the alternative name of ‘tracking error’ suggests that moves away from the benchmark are a mistake! Active risk is a statistical measure of the degree to which the portfolio return might be different to the benchmark index. The closer the portfolio is to the index, the lower the active risk statistic. The tracking error name ties in very well with its original use of monitoring the inaccuracy of an index-tracking fund relative to its base index.
Does an index-tracking fund have zero risk? Certainly not. Simply that the fund manager is not adding any risk to that which is already implied by the benchmark index itself. So far so good. Yet what does it mean to have a portfolio that has an active risk of, say, 3%? Is an active manager with a 3% tracking error more risky than an index-tracking fund?
To some extent it depends on what has contributed to the active risk. If, for example, the active fund manager has invested a large part of the portfolio in volatile smaller companies, then it might well be true that this portfolio is more risky than an index-tracking fund. However, it is also possible to ‘create’ active risk by being underweight in stocks that are large components of the benchmark index.
For example, if you held a portfolio which mimicked a market index but excluded a single large stock (such as Vodafone in the UK), you could easily have created a portfolio that showed an active risk of 3%. It seems odd that our imaginary index fund that doesn’t hold Vodafone is more risky than the index-tracker that has, say, 5% of the portfolio in that company. Granted, our Vodafone-free portfolio is exposed to the risk that the investor misses out on good performance when Vodafone does well and outperforms other stocks in the market. But most investors are currently more concerned about the risk of actually losing money if you invest and the stock goes down. Here, concentrating your bets in a possible loser is a greater concern, a greater risk.
So where active risk derives from underweight positions, a portfolio with a higher active risk may well be less risky than an index-tracking fund. In fact even those statistical packages that are commonly used to calculate active risk would have shown that the Vodafone-free portfolio would have had lower volatility than an index-tracker.
Better models or clearer thinking?
It is tempting to think that this deficiency in the active risk statistic can be overcome by looking for more sophisticated measures of risk or software packages. But this is not a statistical problem, and the answer does not lie in more complicated mathematics. Rather what is required is a clearer idea of the purpose of the benchmark. Instead of asking the question: “Is the portfolio structured with a reasonable amount of risk given the benchmark and performance objectives?”, we should be asking “Do the benchmark and performance objectives incentivise the fund manager to structure the portfolio with a reasonable degree of risk?”
Pension funds invest in equities because they are expected to produce better returns than the asset that is the closest match to the liabilities – typically inflation-linked bonds. Ultimately, the risk in any equity investment is that the portfolio underperforms the liabilities. Yet active risk simply measures the extent to which the portfolio differs from the benchmark index. It tells us nothing about the risk of the portfolio underperforming the liabilities. A portfolio with high active risk might actually be less risky when compared with the liabilities than an index-tracking fund that has zero active risk.
What is of interest is measuring the risk of a particular equity portfolio underperforming the liabilities? Certain types of equities may carry a lower risk of underperforming the liabilities than others. Even if we only discover a very rough understanding of this sort of risk, it may be a more helpful approach to risk budgeting than the active risk statistic. It helps focus on the real risk which pension funds care about when margins are tight – the downside, not missing the upside.
Downsides not upsides?
Measuring and budgeting for this liability-related risk measurement is an ideal but may be too complicated for current investment processes. Where does that leave trustees who wish to control risk in their equity portfolio? One solution may be to measure and control the risk of losing money in absolute terms. This approach to risk budgeting might take the form of a series of limits on stock and industry positions.
The objective to the fund manager could be expressed in a number of ways. One example might be “To maximise the return on the portfolio subject to investing no more than 2% in any stock and 5% in any one industry”. Admittedly, under these terms of reference, it would be more difficult, but still possible, to tell whether the fund manager had done well. Yet, surely it is better to have a less precise performance result with a degree of risk that is intuitive and directly relevant to the needs of the fund, rather than a very precise performance result with a degree of risk that is potentially misleading and of no direct relevance to the fund?
In my experience, most fund managers are reluctant to engage with this sort of challenge to conventional thinking and risk management. That is unfortunate. By insisting on living only in a world framed relative to the benchmark, fund managers are leaving the trustees to take the most important investment decisions, namely that of choosing the benchmarks and the extent to which they will tolerate deviations from them.
Active risk is probably as good a tool as any to measure the extent to which the fund manager is exercising his judgement when he has been asked to beat a benchmark index. However, it would be misleading to think that the active risk statistic provides any guide about the degree of risk the portfolio is running in the total risk budget for the fund.
Kerrin Rosenberg is an investment consultant at Hewitt Bacon & Woodrow in London