It would be helpful to advisers and their clients if funds and model portfolios had some sort of common risk-grading system but, as Graham Bentley points out, we are currently a long way away from that being the case
What do you call your model portfolios? Do you use subjective human behaviours such as ‘Cautious', ‘Balanced', ‘Aggressive' … ‘Psychopathic'? Or do you prefer a zoomorphic approach, perhaps using degrees of cuddliness - ‘Bambi', ‘Baloo', ‘Banzai' or ‘Balrog'?
The regulator is concerned that portfolio names may not reflect varying exposures to equities, clients thus finding themselves in portfolios that may hold greater risks of capital loss than they were led to expect. Yet this represents a fundamental misinterpretation of the issues.
‘Asset allocators' such as Distribution Technology, Morningstar and Willis Towers Watson have respective recipes whose ingredients are represented by benchmarks - for example, the FTSE All-Share index for UK equities.
While the latter is considered the UK standard, however, there are other providers such as MSCI that include the UK in their range of global benchmarks. Allocators do not select their indices on a whim - the commercial cost of using those proprietary benchmarks in their support material is a tight constraint on selection. Here, then, is the first potential divergence.
Graham Bentley: Model portfolios v funds of funds
The use of a benchmark requires a licence from its provider. The more you diversify your index-providers, the greater the cost, so you may select only one. A number of funds, data providers and investment consultants have changed their index provider to reduce costs, as pressure on fund fees and transaction costs increases.
The selection of an index provider according to licence cost has a direct impact on the expected return for an asset class because their holdings may be very different from those of competitors. The expected risk/return on global equities might, for example, be measured according to the MSCI World index. This holds 1,634 large and mid-cap companies from 23 countries, but no emerging markets, and no China.
Alternatively, the FTSE All-World competes with 3,186 companies from 46 countries but includes no smaller-companies exposure. The MSCI licence holder might of course prefer to promote the MSCI AC World Investable Market index, featuring 8,909 large, mid and small-cap companies from 47 countries.
Index choice breeds variation in expected outcomes. While these variants might be small, the portfolio optimisation process is infamous for producing ‘tipping points', where a small change in data leads to the predominance - or complete absence - of an asset class in a portfolio.
As an example, distinct from its competitors, Old Mutual Wealth's asset allocations (as at 19 June 2018) bravely featured a complete absence of fixed interest or property beyond Risk Level 1, with cash being the dominant asset class right through to and including Risk Level 8 for its Collective Investment Account.
This susceptibility to extremes persuades allocators to introduce a ‘pragmatic overlay' or even an upper-and-lower constraint on the appearance of an asset class - the allocation being determined by what ‘looks' acceptable.
In most cases the process is finessed via an investment committee agreeing an estimated annualised return and volatility for each benchmark - and of course that is a matter of opinion, another level of divergence. Estimates of correlation between assets are then added to the mixture - a third level of divergence.
To further complicate matters, there is an almost infinite number of combinations of assets and weights that can produce the same estimated portfolio risk. This is refined by ‘optimisation', creating a string of portfolios, each of which has the highest expected return for a given volatility.
But if your Risk Level 5 has a range of volatility between say 8.41% and 10.5%, there could be thousands of portfolios along that frontier - and there could be significant differences in portfolio composition within that range. You may choose a volatility in the centre of the range but drifting to either end of the volatility range would not disqualify that asset allocation.
Thereafter, advisers' fund selection can have a distinct influence. Targeted Absolute Return is a case in point, where most allocators do not feature that asset class (since there is no definitive benchmark for it), yet many advisers incorporate it as a ‘satellite' investment, believing it adds flavour to their recipe.
Of course, it may completely upend the risk/return expectation, since long-short equity will behave very differently from multi-strategy. A tilt towards smaller companies or factor-based strategies, say, may similarly produce anomalous behaviour.
We then reach the point where the portfolio goes live and investors are subjected to real-world events. Markets have not behaved as we would expect over the last decade, as QE, ultra-low interest rates and inflation took effect - and equities in particular over recent years have shown historically low volatility. There is a temptation to call this the ‘new normal' and adjust long-term volatility expectations downwards as a result. Recent increased volatility is a reminder markets tend to revert to the mean.
Finally, while it probably makes sense not to describe a portfolio in terms diametrically opposed to the expected behaviour of its content, there are many combinations of assets that might be reasonably described as ‘lower risk' - after all, 98% equity is lower than 100%.
‘Balanced' implies equally weighted exposure to a variety of assets - however, those assets may prove to be highly correlated, so the risk-reducing benefits of diversification are absent. A 100% fixed income portfolio will almost certainly be described as ‘Cautious' but may prove to be a weak defence if interest and inflation rates rise dramatically. A structured product combining a long equity position with a put option could reasonably be defined as lower-risk, versus a long-only equity portfolio.
What about style?
What level of equity exposure delineates the border between ‘Defensive' and ‘Cautious'? Or ‘Balanced' and ‘Aggressive'? Why should the level of broad equity exposure determine risk? And what about style - a focus on smaller companies, for example, or a heavily value-oriented portfolio?
Many asset allocation models have too broad a definition of asset class - global equities, fixed interest and so on, defined by expected returns and volatility of the chosen benchmark - while the funds the adviser uses to populate these asset class boxes may refer to an entirely different comparator.
Models have ‘got away with it' versus their multi-asset fund counterparts, given they do not have sector constraints to adhere to. Yet even fund managers can deliberately place their funds in the Unclassified or Specialist sectors in order to remove the need to adhere to a standard at all.
If the provider chooses to name their funds or models to reflect risk, the regulator should perhaps consider non-evocative naming conventions that reflect a volatility target, since most investors go through some form of risk profiling in order to arrive at a suitable portfolio.
The new Summary Risk Indicator or ‘SRI' that applies under the PRIIPs regulation is an improvement over the less satisfactory Synthetic Risk Reward Indicator or ‘SRRI' currently used within UCITs fund KIIDS - not least because it accounts for credit risk as well as market risk. Perhaps this might be a good place to start in determining a common risk-grading system for funds and model portfolios alike.
Graham Bentley is managing director of gbi2
