Dynamic alpha risk budgeting
Dynamic management of core/satellite allocations might seem an ideal strategy for maximising return while minimising costs. But this approach can be difficult to execute, argues Daniel Wallick
There can be many variations on a core/satellite approach, but in its most straightforward form, this type of allocation is represented by a core composed of beta products, such as index funds or ETFs, and a satellite composed of alpha-producing active managers that seek to outperform the core and/or mitigate systemic risk at the portfolio level. Institutions will set a strategic core/satellite allocation based on a given risk/return trade-off, which takes the form of a tracking-error target. The portfolio is monitored over time but, generally, the allocations to passive or active investments are fairly static.
Some new risk budgeting techniques suggest that a more dynamic management of this core/satellite allocation is appropriate in order to maintain a given risk budget. Practically all active managers, with the exception of market-neutral strategies, provide exposure both to the systematic risk of the markets and idiosyncratic risk of their process. The returns derived from either source will fluctuate through time. To better control their portfolio's overall exposure to market risk, institutions might seek to reduce their allocation to satellite investments as their active managers' beta exposures increase, and reallocate those assets to the lower-cost core portfolio. In turn, as active managers' alpha increases, the institution would shift assets back to the satellite.
On its surface, this approach seems an appealing way of maximising return while minimising costs, and an effective method of offsetting potentially converging correlations. However, based on existing evidence, this approach may be difficult to implement. The decision to reallocate funds among managers within a portfolio is fundamentally dependent on three key factors - the nature of a manager's returns, future expectations for each manager, and the liquidity of each investment.
The most straightforward of these factors is liquidity. Any reallocation among different managers presumes the ability to transfer funds in and out. This will typically not be an issue among most long-only funds, with the exception of those funds that impose a transaction fee to enter or leave the fund. Private investments, on the other hand, typically are structured with liquidity clauses like gates and side pockets that limit an investor's ability to move in and out of the investments. As a result, the dynamic reallocation among managers is effectively limited to public funds, thereby shrinking the universe of available investments for the satellite portion of the portfolio.
It is important to note that even traditional assets are not always liquid when periods of crisis cause discontinuous markets, as we have seen in recent history in the fixed income markets, or during the bursting of the technology bubble of 2000. These events are rare, but should still be a consideration if such an approach is to be implemented.
With all funds, it can be valuable to deconstruct where returns are generated, especially when thinking about staying with or changing a manager. Returns are a function of broad systematic risks (often defined as beta) and excess returns (those returns not defined by the broad market). More recently, many practitioners have been further deconstructing excess return into two parts, alternative beta and alpha. It is fairly straightforward to note that betas are available via index funds at a fraction of the fees active managers charge. So, to the extent active managers are charging active fees for beta that can be obtained much more cheaply in an index fund, the decision would seem clear.
Difficulties can arise with this concept, however, when one looks to deconstruct the return sources of active managers. Certainly, what one wants to pay for with an active manager is alpha, not beta. However, the source of alpha that managers generate can often involve the use of beta. Therein lies the complication. Long-only active manager alpha comes from one of two places - security selection or moving beta exposures (often thought of as tactical asset allocation). Excess return, on the other hand, is a different matter than alpha. It's the difference between the benchmark return and the manager's return. Excess return can come from either the sources of alpha mentioned above, or the systematic beta overweight to market factors that differ from the benchmark.
As a result, to the extent a manager has static market overweights, those betas can be more effectively replicated via an index fund. But to the extent that a manager is shifting betas with some regularity, in effect implementing some form of tactical asset allocation, this is a contribution to returns that is not easily replicated to an index. Consider the manager represented in the chart below, who has drastically shifting beta exposures. Looking at this past performance does not help an institution to predict where the next beta exposure might be, in order to replicate it. An institution will find it difficult to time these beta shifts effectively, because it is, by its nature, a forward-looking exercise based on backward-looking data - and if a long-term analysis reveals that a manager is unexpectedly providing beta-like performance over a full market cycle, then a change of manager, rather than a temporary adjustment to their portfolio allocation, is likely warranted.
A slightly different variation of the same concept would be in the consideration of alternative betas. Again, assuming that private investments' liquidity restrictions limit their inclusion in this process, we focus again on long-only funds. To the extent managers are using static alternative beta exposures, such as the carry trade or volatility trade, one can replicate these risk factors independent of the manager to the extent that such exposures are desired by the investor. On the other hand, if the manager is shifting alternative beta exposures over time, we do not view it as viable to attempt to time a manager's movements in and out of these trades.
In our view, implementing a dynamic allocation based on past performance is difficult to execute effectively. Core/satellite is meant to be a strategic approach, rather than a reactive one. Indeed, it is a process based on future expectations where alpha generation is likely to ebb and flow over the short term. As a result, changing allocations to managers should not be done lightly, nor be based solely on short-term results. If an allocation to active is made, an institution must be willing to live with both beta and alpha-driven returns from an active manager, and must have the courage and long-term perspective to stick to a strategy in bad times as well as good. What price an institution pays for that active management will depend on where the manager falls on a spectrum of risk, but holding an active manager for outperformance requires faith in that manager's approach. If a manager is not providing the investment process advertised there is cause to consider changing managers on a permanent basis.
Daniel Wallick is a principal in Vanguard's investment strategy group where he is responsible for developing portfolio strategies