Drawing down from capital: the ‘managed outcomes’ the investment industry should, but cannot, delive
Recognising the need
Retirees want to know how much they can draw from their capital to meet their lifetime spending, confident they will not spend too little or run out of money. Trustees need to know how much investment income to target so as to balance equitably the needs of income and capital beneficiaries. Divorcees want to know how much spending a clean break settlement will safely sustain. Philanthropic families want to know how much they can safely give away without risk to their own lifetime well-being, which means their own needs have to be planned first.
What these real-life needs have in common is ‘drawdown’. An investment portfolio is being regularly drawn on (whether the cash stream is limited to income or includes capital) to support present welfare without prejudicing desired future welfare.
The principles are clear, as a set of cash flows. But to financial planners and investment managers, they represent some of the most complex tasks imaginable.
They involve massive inconsistencies in personal definitions of welfare, which need to be confronted and informed so tradeoffs can be made
They require a ‘total return’ approach to investment returns that ignores the traditional but artificial separation of income and capital
They introduce goal-specific constraints on portfolio management that cut right across the standardisation of investment solutions, including risk and return tradeoffs, that the industry business model relies on
Their complexity calls for ‘quantitative’ techniques but up to half a century after the formalisation of portfolio management by algorithms these still insult the egos of the vast majority of investment managers and are beyond the pockets of under-capitalised IFAs, even if their egos allowed them
These are also the investment tasks with the greatest likelihood of failing to achieve their required outcomes whenever market or inflation risks have been badly estimated. Worse, they tend to be the tasks for which failure carries the most serious consequences.
In terms of retirement income, we believe many older people are spending less than they need to, younger retirees spending more than they can sustain and accumulators are under funding their expected spending. We see trusts that are over-distributing to income beneficiaries and divorcees who are overspending. In a prolonged period of poor investment returns or revived inflation, many of these payments will be slashed and (ironically) some by more than they need be. This is no way to run a drawdown plan.
Making drawdown safe Credit, then, to Pauline Skypala, editor of FTfm, the Financial Times’ weekly fund management supplement, for spotting the significance of the criticisms of the ‘primitive’ state of personal pension advice and management made by three eminent British academics. They first aired their critique in April, as a contribution to the World Bank’s ‘Pension Competition and Asset Allocation Policies for Mandatory Defined Contribution Funds’ Research Project. It sounds mundane but it strikes at the heart of a serious weakness affecting customers of the financial services industry right across the wealth spectrum and in most countries. FTfm published on May 19 a short article by the three, David Blake (Pensions Institute, Cass Business School), Andrew Cairns (Maxwell Institute for Mathematical Sciences, and Actuarial Mathematics and Statistics, Heriot-Watt University and Kevin Dowd (Centre for Risk & Insurance Studies, Nottingham University Business School).
Though the article and the paper are both specific to pensions, this is only one case of drawing down from accumulated capital. The authors likened the current state of defined contribution pension planning to the early stages of airplane design, when poorly understood technology led to crashes. They contrasted it with the present state of planning plane journeys, when safe outcomes are largely assured by the underlying engineering and the focus is on choices the travellers themselves make that best suit their preferences.
We agreed with the essentials of their critique, in particular that the entire process is too rarely driven from desired outcomes back to what is required to deliver them. Outcomes-driven planning is to personal pensions what ‘liability-driven investment’ is to final-salary, occupational pensions. It involves the same harsh realities, or inconvenient truths. We agreed that planning lifetime journeys requires a ‘holistic’ approach to household risks and not a narrow investment one, which is why investment and financial planning should not be separated. They identified the right areas in which techniques needed to alter so as to ensure complete integration of the setting up, accumulation and decumulation stages (likened to take-off, cruising and landing). They also understand what sort of mathematical techniques would get the job done better.
Our criticism of the paper is that it assumes that the techniques that best deal with complex relationships between different sets of uncertainties (personal circumstances, capital markets, inflation, longevity) are too ‘difficult’ to be adopted by the industry and that ‘deterministic’ short-cuts will suffice.
In the pensions context, this capitulation would mainly impact on retirees (at all level of accumulated capital) who are deprived of well-engineered drawdown options instead of annuitisation at an early age. The authors recognise the importance, with long lives in retirement, of inflation protection. But they gloss over the very high cost in spending outcomes and hence the probability of achieving much higher spending outcomes with continued, but tightly-managed, risk taking.
Here is their summary.
“Like an aircraft journey, a DC [Defined Contribution or money-purchase] pension plan must be designed from back to front from desired output â€“ a desired life-long consumption stream in retirement â€“ to required inputs, particularly the contribution amount and investment strategy.
Other factors that need to be taken into account include the target retirement date, the value of the plan member’s human capital (especially when human capital can be adjusted via flexibility in the member’s supply of labour) and housing wealth. When human capital and housing wealth are taken into account, the optimal initial investment in equities is higher than when it is not (in order to counterbalance these bond-like assets). The higher equity weighting will result in pension wealth being more volatile and if it falls short of the level needed to provide an adequate retirement income, the plan member must be prepared to contribute more or retire later. The optimal weight in equities declines over time as pension wealth increases.
Full implementation of the optimal accumulation-stage investment strategy, stochastic lifestyling, is difficult and some simpler solutions are considered: deterministic lifestyling, threshold and portfolio insurance.
At some point, the member will retire and draw down his accumulated pension wealth. At least some of this needs to be in the form of a life annuity in order to hedge against longevity risk. The decision about when to annuitise can be thought of as an option and will depend on the survival credit, the degree of risk aversion, annuity income risk, and the fund size. The decision about how much to annuitise depends on such factors as the size of the state pension, the equity premium, the member’s health status and the desire to make a bequest.
As in the accumulation stage, full implementation of the optimal decumulation-stage investment strategy is difficult and some investment-linked retirement-income programmes that try to deal with some of these factors are considered. Stripped down to its bare essentials, a simplified DC plan would involve a deterministic lifestyling investment strategy during the accumulation stage and an index-linked life annuity purchased with the accumulated fund on the retirement date.”
Our reply in the FT With ten years of experience modelling an integrated solution for take-off, cruising and landing, we thought we could explain just what proven investment technology can already do for all three stages. FTfm published a response two weeks later written by myself and Chris Drew (MD, Lambda Investment Technology and consultant to No Monkey Business).
The features we listed in our article are those that characterise the ‘Lambda’ quantitative model No Monkey Business uses to plan and manage ‘Defined Outcome’ portfolios. These are dynamic portfolios, responsive to changing market conditions, organised to deliver client-specified outcomes (expressed in real, or constant purchasing-power terms) at defined points in time. Their function is to maximise their owners’ welfare, as defined in the planning stage. Welfare may have several complex attributes (such as avoidance of regret, smoothness of the draw, sustainability of the draw or different preferences between current and future spending) that are specific to each individual and probably to this rather than other goals. The most common use of our Defined Outcome portfolios is for a range of tasks involving drawdown and the model is also used to define assets surplus to lifetime welfare.
The planning of these portfolios is equivalent to the take-off stage. This is also described rather dismissively by Blake, Cairns and Dowd as ‘the marketing stage’ but it should of course be the foundation of the entire process of funding future cash streams.
These are the features we listed.
One integrated model is required for generating individual asset returns, optimising portfolio asset allocation, projecting plan outcomes, managing the assets and determining if and when to annuitise
The technical form of model is ‘stochastic’, generating thousands of possible scenarios by combining some ‘knowable’ but weak relationships with the dominant force in all forecasting which is randomness. Stochastic models can provide complete information where a deterministic approach provides only partial information, such as one scenario or estimate
Inflation is the least predictable of all the different risk exposures involved in the journey so using nominal asset returns and a separate inflation variable is inferior to modelling real returns directly from deflated time series
The driver of all sustainable investment planning is ‘total return’: the return assuming that the independently-derived income proportion is retained or reinvested and that any return component ‘consumed’ represents a drawdown from the stock of capital
Reversion to the mean in real total returns from equities is fundamental to good model construction (even though it will not help short-term forecasts)
Diversification is not enough: uncertainty in real outcomes from equity markets (including time-varying correlations between them) has to be managed by ‘dilution’ using index linked gilts (as the only risk-free asset for long-term real outcomes)
Conventional non-indexed bonds are not risk free when planning in real terms and are no more a rational building block in accumulation than level annuities are a rational choice for decumulation
The asset allocation that keeps the portfolio ‘on plan’ is dynamic and cannot be managed using fixed rules like linear ‘lifestyle’ changes in exposure
If the investment within a market is to be actively managed, the outcome ranges must allow for the costs and uncertain pay-offs of active management which are independent of the asset allocation strategy uncertainties
Plan outcomes should assume drawdown in the landing stage but treat annuitisation as a dynamic, parallel assessment (when to leave the casino). For drawdown capital not trapped in a pension scheme, the value that clients place on leaving an estate (the ‘bequest utility’) is a key element in this assessment
The customer benefits depend on them sustaining the plan, so progress reports must provide messages about resources (when accumulating, today’s capital plus required contributions; when drawing down, the latest fund value) that are more stable than the volatile market values of the portfolio. Hence the importance of mean reversion and making plan outcomes conditional on actual portfolio values as opposed to the conventional approach of assuming ‘normal’ returns and applying them to any portfolio value, however high or low.
Chris Drew and I believe not just that it is possible to integrate these decision processes in a systematic, mathematical way but also that it is impossible to take these kind of decisions consistently any other way.
It is striking, therefore, that this kind of technology adoption is very rare in portfolio management for private clients, let alone in financial advice which is where we might reasonably expect professionals to want to establishment their superiority (and command of the available fee income) relative to fund managers.
Why technology matters
When drawdown is unmanaged or poorly managed, we can relate the differences that could have been made by technology both to the source and to the consequences for the owners of portfolios.
Investors who in the past have planned some form of drawdown are likely to have unwittingly misspecified to agents their preferences (for capital market risk, inflation risk, inter-temporal spending preferences, bequest utility) if the principle of sustainability in real terms was not explicitly addressed
The consequences are not likely to be allayed by complaint about the sales or advice process as the complaints procedure relies heavily on the recorded preferences exhibited by clients, even if they are ‘trapped’ by artful questionnaires
Many older investors will have chosen irrational degrees of risk taking because the nature of the payoffs implies a time shift from spending now to spending later, such that spending outcomes are likely to be either lower or higher than they need (or ‘most value’)
A special case of this misspecification is investors’ aversion to annuities, if it is not challenged (particularly pension annuities, where the balancing factor of ‘bequest utility’ is very weak now that penal taxation applies)
Investors expecting long periods of further draw are likely to have made inadvertent time shifts in the opposite direction, drawing more than is consistent with sustainable outcomes. This implies either lower spending later or instability in their spending (something they may also want to avoid)
Younger investors in the accumulation stage are likely to have exhibited greater risk aversion if it was based on discussion of volatility than if it had been based on the range of possible real spending outcomes
Many forms of drawdown are relying on conventional bonds which are suboptimal as soon as real outcome uncertainty is the key matrix. Bonds are the asset whose real outcomes are least well described by short-period risk and return characteristics and correlations. They are the asset most commonly used (in ignorance or mischief) to over-distribute income (see a recent item on bonds for a detailed explanation)
The worst adverse consequences are likely to stem from a prolonged bear market in equities. Short-period standard deviations do not reflect the probability of cumulating falls (‘long strings’) and the slow speed at which ‘mean reversion’ may subsequently be achieved. The evidenced probability of long strings is much greater than investors realise and the Japanese example of delayed mean reversion is not as ay typical as they realise
Even when the asset allocation strategy does not rely on non-indexed bonds, we think that drawdown plans are vulnerable to high and unstable inflation rates because the rate of draw relies on an overestimation of the correlation between nominal equity returns and inflation
Any increase in portfolio volatility will lead to greater instability in advised rates of draw than is necessary, because ‘normalised’ return assumptions are so widespread. This applies equally whether advice is not al all technology-based or is supported by off-the-shelf resource-planning applications.
Our estimate is that investors requiring high certainty of withstanding these bad scenarios are typically overdrawing by a factor of 50-100%.
This sounds shocking but it is actually entirely consistent with previous large-scale errors which arose from banking on average returns, such as occupational pension scheme deficits, endowment shortfalls and catastrophic losses in splits and precipice bonds.
The fallacy of funding on a 50:50 basis, which is what banking on average returns means, is being roundly challenged in occupational pension schemes. A good recent example also comes from FTfm, in an article by pensions consultant John Ralfe: Public sector pensions: at what future cost. It should strike a chord with ordinary investors because they are also the taxpayers who are underwriting the risk of adverse outcomes for a lucky group of employees. These are mineworkers. They enjoy fully indexed pension underwritten by the taxpayer and they also enjoy 50% of any actuarial surplus as added benefit, which cannot be taken away. The trustees run the two very mature mineworkers’ funds with 70% in equities. Actuarial valuations that give rise to these upward-only bonuses are arrived at by discounting the pension promises at ‘the expected return on assets’ with that chosen asset strategy.
Ralfe tells us that if the promises were backed not by equities but by index linked gilts, as the risk free matching asset, the latest valuation would have thrown up a deficit of £0.9b. As it was, the actuaries reported a £1.9b surplus.
There are two fallacies in this case. First, the funding rate is based on a 50:50 investment outcome. Second, projected surpluses with a 50% chance of being made are being signed away. Taxpayers might reasonably be appalled to see how they are funding a really shaky drawdown plan but how many of them conduct their own plan in the same speculative way?
Conflicts Many of these industry shortcomings are also tainted by conflicting agendas.
Agents generally benefit from higher fee streams where the client’s supposed inter-temporal preferences and risk tolerance favour higher exposure to risky assets
They lose fees when annuities are preferred over drawdown from capital or surplus resources are given away
Their charges, and the charges of the typical investment vehicles that agents favour, have a far greater impact on real wealth outcomes than is implied by the percentage they represent of short-period nominal returns
In fairness to agents, however, consumers make monkeys of themselves when they resist paying a fee for high-quality techniques. There would be a greater uptake of modern investment technology if private clients were more discriminating. This was true when Chris Drew and I were consulting to retail financial service firms between 2000 and 2004 and has barely altered since. There are no product or service providers in the mass market offering industrial-scale solutions such as we have described.
Fee resistance persists even where the application of technology is combined with low-cost implementation, in spite of the fact that the combination can significantly increase plan outcomes and pay for itself many time over.
Paying for ‘clever’ implementation rather than integrated goal-based planning and management plays right into the hands of an exploitative industry, whether the actual model is stockbroking, wealth management in private banks or the typical investment service provided by IFAs.