Benchmark risk sounds rather incongruous. A benchmark is by its very name something that is supposed to be without risk. Active managers attempt to add value over a benchmark, and judge the behaviour of the market by the behaviour of the benchmark; passive managers track the benchmark because that is the risk-free strategy, and so on. Surely a benchmark cannot itself be a source of risk?
Unfortunately for many pension funds and their managers, it can be, and is. Benchmark risk does exist. Worse, unlike systematic risk, which pension funds are paid to take, there is no theoretical reason why benchmark risk should be rewarded by the markets.
To understand what benchmark risk is, however, we first have to understand what a benchmark is. The answer to this is simple. A benchmark is a portfolio that is used to explain the return of a group of stocks. The group of stocks that is of interest, of course, is the opportunity set available to the investor. Now, an investor’s opportunity set is very easy to define: everything that the investor could possibly buy, and nothing that they cannot buy. The return on this group of stocks, then, is what a benchmark portfolio is trying to measure.
Now, at least one portfolio will exactly match the return of this opportunity set: a portfolio with exactly the same constituents in exactly the same proportions. By definition this will have exactly the same return, and exactly the same risk, as the opportunity set. It will also correlate exactly. Conventional wisdom, however , says that as long as we use a portfolio which has a very high correlation with the opportunity set, and a similar volatility, the resulting measured return will be the same too. All of the monthly differences in return caused by the differences in holdings and weights will cancel themselves out in a short period of time.
Unfortunately, this turns out not to be true. The chart shows the cumulative return of a range of US market indices over a five-year period to the middle of this year. The correlations between each pair of these indices are extremely high: 0.99 or 0.98. In addition, they have almost identical volatilities. However, the end result to the investor was as much as $300m more or less depending on the index chosen for every $1bn invested.
That is a huge difference in return. Worse, it is a huge difference in return caused by only one thing: the choice of the benchmark. More precisely, it is a huge difference in return caused by benchmark risk. These return differences are caused because the stocks that the various benchmark portfolios hold are different, and are held in different proportions. Each of the benchmark portfolios is simply a portfolio run by a particular set of rules. There is no reason for assuming that the errors caused by a set of rules in one period will be cancelled out by the same rules in the next. And if these differences will not vanish within a short time period, there will appear cumulative differences in return.
Another way to look at this, of course, is in pure risk-return terms. One very common definition of risk is variability in return. One assumes that variability in return will be compensated by greater return. Here, the returns are different: there is clearly a difference in variability of return. This is benchmark risk.
Where do these holdings differences come from? There are two basic sources. If the box in the chart represents the whole equity market, the index is represented by the left hand column. The free float, the stock actually available in the market, is represented by the top row. Now, an alternative name for the free float is the opportunity set: this really is everything that you could possibly buy, and nothing you cannot. As can be seen, there is a box in the bottom left which represents the unavailable capital of index companies. This capital should be excluded from the index, as it is not available to the investor. Generally, however, most indices include at least some unavailable shares. The other anomaly is the box in the top right. This is free float capital available for ownership that is excluded from the index: stocks that could be bought.
Benchmark risk, then, derives from the interaction of these two boxes. Only by using a benchmark that accurately measures the free float available in the marketplace sponsors and managers effectively eliminate benchmark risk. The question fiduciaries should ask themselves today is this: if they can eliminate it, can they justify not eliminating it?
Ian Toner is global head of marketing, global equity index system, at Schroder Salomon Smith Barney