Distortions Arising From A Comparison With Synthetic Benchmark Indices
‘The Benchmark Effect’
The benchmark effect is a phenomenon that could be described as the benchmark mis-representing itself. It is a distortion bestowing a false halo of relative achievement to the portfolio manager, compared to the benchmark. The distortion is practically inescapable, and complicated, if impossible to quantify accurately, without detailed information on all changes to holdings within a portfolio. The effect can theoretically understate a benchmark’s performance, or overstate it. But more often than not, the effect will make portfolios look better than the benchmark, rather than worse.
The graph shown below illustrates how a benchmark clone (red line) could appear to outperform the benchmark, purely by virtue of this effect.
Before taking a look at what causes this, how strongly it manifests, and who is most affected by it, beneficially or otherwise, one should consider the three main investment styles found in institutional investment management.
Absolute return managers actively manage portfolios in the pursuit of specific objectives. These may vary from client to client, but will usually centre around the optimisation of several factors such as return, risk and capital preservation, or income versus capital gain, to name but a few. Typically, that causes conflicting priorities and thus requires a great deal of judgement.
Indices (benchmarks) are taken merely as simplistic descriptions of available investment possibilities. Blame for failing to achieve investment objectives can’t be placed on poor benchmark performance. This approach requires highly developed skills, experience, and not least, a particular set of character traits.
The sole aim of index-cloning (often referred to by the oxymoron ‘passive management’) is to replicate a benchmark as accurately as possible.
Index cloning has little to do with investing and ignores all consideration of investment return, or risk. It is an exercise in applied arithmetic and requires no investment skills whatsoever.
Given a computer, access to the composition of the benchmark, and a bit of time, any reasonably bright grammar school pupil will perform index cloning equally well as a Nobel Prize laureate.
Fees paid for this type of ‘investing’ are the triumph of salesmanship over common sense.
Semi-active managers do take investment decisions, but only in terms of weightings that are ‚over’ or ‘under’ those of a chosen benchmark index. If done skilfully, this investment style will fairly steadily accrue excess performance over the benchmark with time, even if at a slow pace.
A rapidly growing category of managers appears ‘semi-active’ while in fact being ‘pseudo-active’. Pseudo-active management uses minor weighting differentials that are effectively random in origin, to give itself the appearance of investment management, thereby deflecting attention from a blatant lack of investment competence. Pseudo-active management is like taking the scenic route around a circle large enough to allow the illusion of journeying.
Competent absolute return managers will generate significant value-added, measured in multiples of the management fee charged. Incompetent managers will soon and quietly migrate toward one of the other groups.
Due to management fees, a true index clone is guaranteed to generate a steadily widening, negative performance gap relative to the benchmark replicated. True index clones do not gain, or lose from the benchmark effect.
Pseudo-active managers can be hard to distinguish from semi-active ones under the shield of the benchmark effect. This group enjoys, and even seeks to take credit for, the windfall of seeming excess performance provided by the effect.
Exhibit-2 illustrates how a simplified asset allocation may look when anticipating a bear market in equities. Objective-driven managers will take drastic steps to shield portfolio values (switching to capital preservation mode, putting safety above possible gains, with low equity exposure, and high cash holdings). Index cloners are expected to not do anything and won’t. Semi-active investors will make some minor weighting adjustments, being more concerned with decision risk than with client’s value risk. Pseudo-actives will roll a dice and may act, or may not. In any event they will direct 95% of resources into drafting competently sounding reports, explaining why the market decline, was supposedly unforeseeable, and that ‘in the long term’, rewarding returns are supposedly inevitable, given sufficient patience.
Understanding The Benchmark Effect
Obviously, portfolios that invest in a multitude of assets will use multi-asset benchmarks for comparison. But asset classes don’t perform alike. This causes initial weightings within a benchmark index to deviate from the original, and intended asset mix. In an absolute return portfolio this is what is aimed for, but within a benchmark, or a clone of sorts, this needs to be erased regularly by re-balancing the constituents. Benchmarks that are subject to re-balancing are called ‘synthetic’. This applies to virtually all multi-asset benchmark indices and is a necessity.
As a side effect, it produces distortions compared to a ‘natural’ portfolio. These are caused by:
- The frequency of re-balancing, or rather, the discrepancy between the re-balancing frequency within a benchmark and that within a portfolio
- Performance of benchmark constituents relative to one another.
In re-balancing, profits of good holdings are realised and losses of poorly performing holding are being ‘averaged down’. Since market prices move in trends, one could say that ‘good money’ is ‘thrown after bad’, while profits are being ‘curtailed’. That is the exact opposite of what generates performance.
Exhibit-3 shows the interior of a fictional portfolio with three holdings over a few time increments (N to N+3), together with a synthetic benchmark index. Decimals may be subject to rounding.
- Holding 1 rises strongly
- Holding 2 drops noticeably
- Holding 3 remains static
The differences in performance of the holdings to one another result in a pronounced change of internal weightings, ‘passive morphing’, without any transaction whatsoever. Over the three time increments, Holding-1 balloons from 33% of portfolio to 43%, Holding-2 shrinks from 33% to 28%, and Holding-3 from 33% to 30%. The benchmark re-balances every increment. It does so using the sum of ⅓ each, of all asset’s price change, to calculate it’s own monthly rate of change. The column to the far right calculates the cumulating performance difference between the portfolio and the benchmark. That is the benchmark effect, illustrated with purely fictional data.
To illustrate potential magnitude in the real world, an example benchmark with two assets, bonds, and equities was constructed, based on actual financial market data: Swiss Federal Bonds, and the Swiss Equity Market, with a mix of 50% bonds, and 50% equities. The benchmark re-balances monthly. All time series were set to a base value of 100 at inception (31.12.1984) and run through to 31.12.2018, spanning 34 years of actual market history.
Performance of the two asset classes, and the benchmark are shown in Chart-1, and Table-1.
To compare against this ad-hoc benchmark, and in order to gauge the benchmark effect’s magnitude, three distinct ‘cloning portfolios’ were calculated, investing in the same assets, in identical proportions, but differing in their frequency of re-balancing ‘morphed’ weightings, re-establishing the benchmark composition.
- Clone-A (monthly)
- Clone-B (quarterly)
- Clone-C (annually)
Since Clone-A re-balances with the frequency of the benchmark, its performance is 100% identical to it. In that sense, Clone-A represents a genuine index cloning portfolio, free of management fees. Clones ‘B’ and ‘C’, while labelled ‘clones’, should be understood as emissaries of pseudo-active management. Due to their slower re-balancing frequency, they differ from the benchmark only temporarily, effectively by chance (passive morphing between re-balancing).
Very Long Term Impact (34 Years)
The data of this simulation would suggest that the benchmark effect is sufficient to distort performance comparison measurably, even across very long periods of time, where one might expect the effect to produce a ‘net zero’. Particularly noteworthy in Table 2 below are the data for yearly ‘high’ and ‘low’ values of the distortion. Analyses for sample periods of shorter length are shown further down the page.
Effect Working To Investment Manager’s Advantage
Effect Working To Investment Manager’s Dis-Advantage
The impact of the benchmark effect may vary in its extent, depending on length of time measured, the kind of market cycle, and benchmark composition. Either way, with readings as significant as shown here, the effect is clearly pronounced enough to be taken into account when evaluating any portfolio with a high degree of congruence to it’s benchmark.
I often argue that any bright college kid would be capable of matching the performance of an index. To accept that statement as fact seems a challenge to the uninitiated to accept. Yet, it is too mild a comparison. A lazy college kid (re-balancing less often), is virtually guaranteed to exceed performance of a synthetic benchmark, all thanks to this distortion.
In a nutshell, practically any portfolio performance delivered consistently within a relatively narrow margin of a synthetic benchmark must be taken as evidence for ‘pseudo-active’ management. A genuine benchmark clone would fall further and further behind the index, due to management charges.