Buckets, MPT and the 5 year rule……

MPT theorists have a clear and general disdain for the bucket theory of investment allocation.  To tell you the truth, I think it all depends on how you structure, optimise and manage risk and return that matters.  

If you want to swap the communication of complexity with an understandable version of how a portfolio would work, then what is wrong with this, providing of course you do not dumb down the portfolio structure itself?  

I do of course have issues with some bucket approaches and in particular rules of thumb that place insufficient low risk assets in the dedicated portfolio, but not all bucket strategies are the same I am sure.  Labels can be misleading.

The idea behind the bucket approach is that it can take years for market and economic risks to work themselves out of the system and so you do not want to be exposed to these risks.   But, higher risk equity investments can take 5, 8 10 or more years at times to fully recover their previous nominal highs, let alone provide a total real return or beat a comparable low risk alternative.   If you are having to draw down on equities during these periods you risk either depleting at a faster rate than anticipated and/or selling assets at distressed prices.   What a dedicated low risk allocation structure is designed to do is to provide stability and support to a withdrawal strategy during such turbulent times as well as important forward planning opportunities for complex asset liability situations that are not provided for within MPT structures: this greater level of flexibility is an unsung benefit of portfolios driven along liability frameworks. 

The key is of course making sure that drawing down low risk assets does not leave you ridiculously exposed to equity based investments.   In my opinion, following a 5 year low risk allocation alone, can result in far too high an equity allocation and too low a low risk allocation.  A five year bear market would leave your low risk allocation depleted, leaving your portfolio wholly exposed to equities. 

If you are looking at structuring a dedicated low risk portfolio within an overall portfolio there are some rules that I think you should follow:

Number 1: you need to use an optimisation engine that integrates all components of the portfolio, relating liability and yield profiles, risk and performance preferences and central investment strategy for both low risk and higher risk asset allocations to the portfolio outcome.   Portfolio components should not be allocated separately and then stuck together.

Number 2: you need to be clear about the operational function and the period of the significant stock market and economic risk you are looking to cover.  You also need to determine the minimum low risk allocation that your portfolio should not breach in the event of a risk event and the risk events that could put your structure at risk.  For example if you are looking to cover a 5 year risk period and you would never wish to let your portfolio go below 3 years of cover, you really only have a two year window.   Your market risk time frame needs to be the minimum required + the time frame of significant risk.   The structure should not in of itself have a reasonable chance of being a risk to financial security.

Number 3: the low risk portfolio should not actually be run down during fair to upward trending markets.   Excess equities returns, less rebalancing needs, should be crystalized and transferred to the low risk portfolio, for future or immediate consumption.  This provides a portfolio discipline focussed on return management.  During such time frames a liability driven portfolio will operate pretty much as a normal portfolio. 

Number 4: your period of significant market and economic risk is to cover market and economic uncertainty and should be able to cover a reasonably wide band of market valuations.  Nevertheless, as we saw in the late 1990s, markets can move to extreme levels; simple modelling of economic and market cycles should uncover the risks to withdrawals posed by exceptionally elevated markets and advanced structurally impaired economies.   I consider it a valuable and rationale action for portfolio managers during periods of extreme valuations to enhance a portfolio’s ability to deliver short term financial security.   This is also in keeping with maintaining a healthy balance between consumption, saving, investment and production decisions within an economy.   This action is also similar to that of rebalancing highly valued equity allocations to lower risk assets in your static benchmarked portfolio, just that we are framing it within risks to consumption terms and tracking its impact in liability terms.  

Number 5: assumptions used to determine allocations to markets, securities and assets should be different from those used to model the ability of liabilities to meet financial needs over time.  Note the platform is a risk management platform that should be designed to minimise risks to withdrawals of both income and capital.  This means the use of conservative assumptions for inflation, stock market returns and for the longer and higher risk end of the lower risk portfolio (longer dated, corporates, high yield, preference shares) assumptions that cover adequate credit risk, liquidity risk and duration risks.   In all truth, asset allocation is a relative valuation exercise and should not be dependent on specific return expectations, while liability management is an absolute valuation exercise and at risk from inappropriate return expectations.

Within the risk management modelling structure, your portfolios should already be hit with the risks that you would want your portfolio to be able to cope with.  In other words, such a structure should not just aim to provide support during market volatility, but an almost fail safe platform for delivering income and capital security.   Where many of these approaches fall down is in the use of inappropriate risk management assumptions and the lack of a robust optimisation engine.  The probability of increasing withdrawals at a real rate should be higher than the probability of reducing withdrawals.  Unfortunately this approach needs to assume that markets are not efficient and that future price movements are too a large extent influenced by prior price movements: you should not be coming up with a 9%+ return expectation for equities at the top of the 1990s bull market, as for example Ibbotson Associates did.

Number 6: All liability profiles are different and not everyone is going to want an average withdrawal over a number of years.  An optimised, dedicated low risk allocation that would take over in a risk event is much better at handling significant portfolio outflows and differing liability profiles.   MVO structures that structure in risk and return at a point in time are not adapt at managing point in time liability interactions.  We need to operate in risk and return space and liability space at the same time and a portfolio needs liability tracks on which to plan and structure itself.  

Number 7: investors will have different preferred short and long term risk return trade offs: some will prefer greater short term security and certainty in return for lower potential equity or other higher risk asset class returns.   A liability driven framework allows you to frame their decisions in terms of the ability of a portfolio to meet their liabilities over time, and the impact of significant risks to that ability.  It is just so much more flexible and intuitive.  In other words, in addition to the point in time standard deviation risks, we have the absolute valuation or market risks and their impact on the ability of assets to meet financial needs over time.  We can therefore better frame the risk and return profile across the twoi necessary dimensions: risk and return space and liability space.  MPT theorists deign to consider the interaction of the two, save for where human capital theory interacts with efficient MVO allocation.

Finally, I was at a recent conference where one of the speakers criticised the drawdown of assets from a dedicated portfolio component.  My point was this: in a falling market where the low risk allocation rises in price relative to the equity allocation, an MPT structure would recommend a sale of low risk assets to transfer into equities with a withdrawal being made from the rebalanced low risk/equity portfolio.   Does it not make sense to go straight from the low risk portfolio to consumption. 

The point where a rebalancing portfolio will more likely to result in a greater return is where the duration of the downturn in equities is short and sharp: in this case a large rebalancing will dwarf the withdrawal demand and result in a higher rebalanced return.  The problem here is that the low risk portfolio in liability terms is declining, in other words you are reducing the structural integrity of the portfolio in liability space.  There is a conflict.  

Indeed, there are a whole host of counterbalances within the two competing structures: a liability driven portfolio would be raising higher risk assets, at the margin, at market peaks, but more likely to consume those returns than reinvest them on the way down.  It would have a limit on how much of the low risk portfolio it could reallocate away from liabilities in sustained and deep bear market, which may well turn out to be a benefit.  

Personally, I see no reason why liability frameworks should not be imposed on any particular portfolio construction methodology.   This would surely have benefits.

Important background information:  I have been involved in developing dynamic integrated ALM structures focussed on the two dimensions (risk & return, and liability space) since the late 1980s and developed my first integrated models in 1990/1991.  More advanced integrated models were developed during 1997 to 1998.  Much of my early work was embodied in the UK Institute of Financial Planning’s investment course, which I wrote for the Bristol Business School, which was at the time part of Bristol Polytechnic and now I believe the University of New South Wales.  Personally, I think MPT and its disdain of liability driven frameworks forged by an unrealistic belief in efficient markets has set back innovation in this important area of the market place.  Likewise simple industry solutions that rely on narrow rules of thumb and over optimistic return assumptions, and high costs, has also impaired the development of liability driven frameworks.

Leave a Reply