I have always questioned the traditional approach of benchmarking portfolios, using Sector Averages or industry-wide indices such as the APCIMS. As you can imagine, we see this fairly often, hence why I probably spend far too much of my waking hours thinking about this sort of thing.
We recently got asked to review a firm’s model portfolio and this issue came up again, so I thought I’ll share my views. The firm told us that it wasn’t always clear what was being measured and clients weren’t really familiar with the indices they were using (should they be?) So, as part of the work, we were asked to consider building composite benchmarks, based on the strategic allocation of the portfolios using sector averages. I argued that my preferred approach was to build composite benchmarks using market indices.
I guess it depends on what you are measuring but I think the key question you want to ask with active portfolios is this: are we adding any value through tactical asset allocation and funds picking.
To my mind, if you use a common industry benchmarks such as the APCIMS, it is very difficult to gauge exactly where your return is coming from. Is it down to the fact that the strategic allocation of the portfolio markedly different from the benchmark you are using? Or Is it down to your tactical overlay and fund picking? This is what hotheads in the investment management world would call ‘performance attribution’ but you are more likely to find those words in the IMC or CFA study texts than on the investment committees of IFA firms.
So anyway, composite benchmarks based on the strategic allocation of the portfolios makes a lot of sense to me, because it is easier to tell if your calls to outweigh or underweight certain asset classes are paying off or if you are just shooting blanks.
Coming back to the issue of using indices rather than sector averages. I have to be honest, I don’t know who came up with that idea of comparing fund managers with the ‘average’ in their sectors or why.
First of all, most active managers destroy so much value after cost and the poor ones are always going to drive down the sector average. And of course, we are not interested in comparing with the ‘average’ of the poor-performing majority.
Then there is the issue of variation within each sector. Take the UK All Companies for instance, you could have anything from Standard Life UK Equity Unconstrained Fund to Premier ConBrio Sanford Deland UK Buffettology Fund! (I’m not making it up BTW, that’s a real fund!) My point is that the mandates can be markedly wide that comparing these funds is like comparing apples and candy (apples and pears are too similar, they are both fruits and I think they both grow on trees.) Sector Averages hide the real truth.
My final point is simply that an investor can’t buy sector averages. They can buy an index, minus the cost of course. The goal is to see if a manager is adding value, over and above what the market delivers.
Of course, there are no rules as to how advisers should benchmark portfolios or funds. Perhaps there should be. But it seems I’m not the only one who doesn’t get the absurdity of sector averages. Certain Nobel Laureate couldn’t get it either. William Sharpe’s famous 1991 article “The Arithmetic of Active Management” should be required reading for anyone involved in running portfolios. In fact, no one should be allowed anywhere near your firm’s portfolio without reading it. Here’s what the last paragraph of that article says;
“An important corollary is the importance of appropriate performance measurement. “Peer group” comparisons are dangerous. Because the capitalization-weighted average performance of active managers will be inferior to that of a passive alternative, the former constitutes a poor measure for decision-making purposes. And because most peer-group averages are not capitalization-weighted, they are subject to additional biases. Moreover, investing equal amounts with many managers is not a practical alternative. Nor, a fortiori, is investing with the “median” manager (whose identity is not even known in advance).
The best way to measure a manager’s performance is to compare his or her return with that of a comparable passive alternative. The latter — often termed a “benchmark” or “normal portfolio” — should be a feasible alternative identified in advance of the period over which performance is measured. Only when this type of measurement is in place can an active manager (or one who hires active managers) know whether he or she is in the minority of those who have beaten viable passive alternatives”
That was 1991! I just don’t get why not much has changed in 25 years but I do take comfort in the fact that I’m probably just as naïve as this Nobel Laureate.