27 September, 2020
Why a benchmark?
To compare, but against what, and for what purpose? Benchmarking has been a significant evolution of the asset management industry during the 80s. It has been instrumental in objectivising the dialogue between asset owners and asset managers. It is a natural extension of the efficient market hypothesis (although if it were truly efficient, we would only need index management). It reflects a view that investment is first selecting the right buckets of assets, then selecting the right assets within each bucket.
The main benefit of benchmarking is to shorten the time required to evaluate the quality of portfolio management. Without a benchmark, one has to wait for a full market cycle (i.e. 5 to 10 years) to determine whether the performance of a portfolio meets expectations. During bear markets, it is incredibly hard to make an opinion.
It makes a lot of sense when considering "pure" asset classes (i.e. US equities, Eurozone bonds etc.) because the selection of a benchmark is relatively straightforward. Whether you select the S&P 500 or the MSCI USA, you compare approximately the same thing.
For fixed proportion mixes, it is more or less the same: you compare a 40% US equity/60% US bonds with a composite benchmark 40% S&P 500/ 60% S&P US aggregate bond.
But for more fluctuating and more diversified portfolios, what is the right benchmark? Can you compare a fluctuating mix of listed equities, options, bonds, private equity and property with any meaningful index mix? A straightforward solution would be to compare against a combination of equity and bond indices having the same level of risk as the portfolio. However, it must overcome two hurdles.
First, one needs to be able to estimate the risks (volatility, Value-at-Risk, etc.) of both the portfolio and the benchmark. Then, although it answers the natural question "am I correctly rewarded for the risk I take?” The results will be hard to interpret when it comes to analysing the sources of over or underperformance.
For the first hurdle, there are several solutions, including using the past volatility of the portfolio, provided the mix didn't change too much over time, and there is a limited amount of illiquid assets. The second hurdle is harder to overcome. A convenient solution is to compare peers' portfolios. By peers, we mean investors that have the same broad asset classes, and ideally the same kind of investment goal. It is then possible to consider and interpret differences in weights and performances. The last required trick is to compare against peers’ aggregate portfolio entailing the same risk level so that such performance comparisons make sense.
Peers benchmarking is useful in the following ways:
Absolute return goal
Many investors have an absolute return goal, something like money markets +3%, or inflation +2%. Should that goal be the benchmark? Probably not. Although it is essential to keep that goal in mind in day-to-day management, it is very likely that every single year the performance will deviate strongly from that target. To be able to judge the quality of management on quarterly or yearly horizons, one would need to compare with a mix that has a good probability of meeting the investment goal.
Again, drawing peers' comparisons is often a good solution, as most peers will have the same kind of investment goal. It is necessary, though, to adjust for the difference in investment goals. A portfolio targeting money markets +5% will probably have a lot more risky assets than a portfolio targeting money markets +1%. Since it might be challenging to have peers disclosing their investment goal, the easiest solution is to adjust for possible risks involved.
Risk adjustment can be made naively by mixing peers’ portfolios with positive or negative cash positions. There are also smarter ways, sometimes called "portfolio morphing", by which the balance between risky and less risky assets is modified to reach a certain level of risk while preserving market visions that are implicit in the original portfolio.