The cycle of institutional fund management can normally be described by the diagram below. The ‘objective’ is what the fund is trying to achieve – for example ‘meet the liabilities at lowest cost’. Through asset-liability modelling this objective can be expressed as a ‘strategic asset allocation’ – also known as a strategic benchmark. This strategic benchmark can reflect the particular circumstances of the fund: such as the nature, term and profile of the liabilities, the current and desired level of funding, the current and desired future level of contribution as well as the attitude to risk of both the sponsor and those with fiduciary responsibility.
‘Implementation’ refers to the actual fund management structure and appointment of managers that is employed to deliver the benchmark – or, more usually, deliver the benchmark and some quantum of added value, commonly known as an investment performance target.
Finally, but most importantly, ‘review’ evaluates the previous activities and provides management information to those who have to review previous decisions and make new ones.
Evaluation, historically, has a standard performance measurement on the ‘skill’ of the active investment manager(s). This skill was assessed predominantly in terms of a two-factor attribution model: active asset allocation and stock selection. This was entirely appropriate when all funds were actively managed against a peer group (WM Universe) average and the management structure was typically one or more – depending on fund size – multi-asset ‘balanced’ managers. The underlying rationale was to identify any weak link or, more usually the ‘weakest link’ in terms of poorest performing manager and dispense with its services. This despite WM research which shows that giving more money to the poorest performer and releasing the best performer often makes more financial sense! In changing manager there is a considerable danger of mistiming relative performance cycles. The almost inevitable result is that the new manager disappoints, the released manager recovers and the fund incurs transition costs.
Fund benchmarks and fund management structures are now much more fund-specific. The simplicity of peer benchmarks has been replaced in very many cases by scheme-specific benchmarks. There is also a greater variety of fund management structures. Some are truly unique, but the majority combine various elements of more simple structures in a hybrid arrangement.
Evaluation in this new world is more complicated. With complex funds, the first task is to identify the decisions that affect the fund’s performance. This can easily be overlooked. It is preferable to understand who is accountable for each identified decision. Only then should each decision be evaluated, separately and collectively.
Fund management structures can be tested. Evaluation focuses on the probabilities of success, usually defined in terms of meeting – or exceeding – the investment performance target, and the probability of failure, such as breach of any downside risk limit. But other less complex analysis can, and should, be made. It still amazes me how many complex funds have no total fund benchmark. Even more have no total fund target or risk tolerance indicator.
A number of multi-managed funds have a more insidious mismatch. It makes little sense if all a fund’s portfolios achieve their benchmark and/or their investment performance target return but the total fund does not. In this case, an investment decision has been made, possibly implicitly. Portfolio benchmark rebalancing is inconsistent with the total fund.
Returning to our two-factor attribution model, evaluation of a portfolio will still include an analysis of any active asset allocation, as well as stock selection within each asset.
Evaluation should reflect the owner’s perspective and encompass the extent to which asset manager discretion is exercised, and its impact – for example, investing in non-government index-linked when the benchmark comprises only government. Some sympathy and understanding, however, can be offered in terms of the manager’s particular investment process and style. For example, geographic market exposures may not represent relevant attribution for a bottom-up stock-picker.
With multi-managed structures, it is recommended that the total asset, market or segment is analysed to verify the benefits of the multi-manager arrangement. The net IR – defined as the relative return after fees, divided by the relative risk– should be a higher positive for the multi-managed arrangement than for alternatives, such as indexation or a single active manager.
Evaluation of active managers now encompasses both relative returns and relative risk; the latter also known as active risk, investment risk, tracking error and sometimes – confusingly – as benchmark risk. Despite differing terminology the risk that is being measured is the uncertainty of the fund or portfolio returns relative to the benchmark. If the fund achieves a return consistently at a specific level above or below – or at – the benchmark each month there is little uncertainty, relative to the benchmark, and therefore little relative risk. If the fund achieves very volatile performance unlike the benchmark there is considerable uncertainty and therefore high risk.
Risk is not, in itself, a bad thing. It can be assessed relative to the portfolio investment performance target and investment discretion available to the manager; as well as any prescribed risk indication such as annual downside limit. What matters is that the risk is measured and understood enough to build confidence that it will be satisfactorily rewarded. If it is not, then there is no reason to take the risk. If it is rewarded it may be worth spending more of the available risk budget in that area.
Over the years, from both a fund and asset manager perspective we have seen tactical asset allocation risks decline as a proportion of total risk. Ironically for those who are strong advocates of scheme-specific benchmarks, the experience is that there is typically even less active asset allocation in such funds than in the traditional peer benchmarked funds. Our analysis has shown there is almost no evidence of significant, added-value from active asset allocation on a consistent basis.
Recently, focus has been directed on the amount of voluntary trading by the asset manager within portfolios or, more pertinently, the trading and other costs incurred by such activity. Transparency of these, and all other costs has become a central tenet for fund owners.
Eric Lambert is head of the performance consultancy for WM Company in Edinburgh