Fidelity’s new retirement solutions: why semi-customisation of large asset pools does not work
Fidelity are launching in the IFA community a new solution combining the accumulation of funds for retirement spending and regular drawdown from a fund after retirement. The offering recognises some key principles of retirement planning:how long we live means there is uncertainty about how long the fund will live if we deplete it at the ‘wrong’ rate; retirement dates are nowadays fuzzy rather than precise planning horizons; the assets funding retirement spending plans are not necessarily all held in a vehicle called ‘pension plan’.
The innovation is a series of date-stamped funds for accumulation. Fine, but the flaw in the thinking is in the single pool for post-retirement drawdown. It is the coincidence of volatile asset paths and a stable target drawdown in sterling terms that defines the key planning risk in self-funded retirement income. Fidelity have it the wrong way round. Liability mismatches have a far greater impact in the decumulation stage than in the accumulation stage. The long life of a drawdown plan (particularly when outside a pension wrapper) means the switch to bonds in the later stages of accumulation, and the heavy non-indexed bond exposure in the decumulation fund prove that mass customisation of asset allocation does not get the job done in a large pooling structure. Individual investors reading this item will recognise links to money illusion and my items on bonds.
My purpose Fidelity have a sensible objective in seeking to combine the economics of mass pooling of capital with customers’ need for solutions tailored to their time horizons, targets and risk aversion. Much of the thinking behind their solution, billed to IFAs as ‘a couple of blockbusters from Fidelity’, is superficially appealing. However, if a solution does not start from the right first principles it will probably not advance the customer’s choice or, eventually, their outcomes.
Rather than review the offering in detail, which is not the purpose of this site, I want to use it as an excuse to reprise the nature of the planning challenge and as an example (one of many) of the errors in much conventional thinking about asset allocation and goal matching through time.
Finally, I link this to some solutions that I believe can combine the economics of pooling (on a far greater scale than even Fidelity can provide) with the customisation of the dynamic mix of portfolio building blocks that solve the problem of individual retirement planning. The latter was featured in my ‘product of the future’ submission to the recent FT/AMRO competition.
‘A pioneering solution to investors’ lifetime retirement needs’? Fidelity’s claim is based on the idea that investment risks at any stage in a person’s life (whether still working and accumulating, retired and drawing down or somewhere between the two) can be roughly matched to needs defined by age and stage.
The solution is to create a series of separate date-stamped accumulation funds one of which the investor would select to match as closely as possible their expected retirement date. The literature shows six such funds, five years apart, from 2015 to 2040. These funds are differentiated by their asset allocation. They may not start with a different allocation but the allocation will evidently shift as the time remaining shortens, so there should be an apparent logic in the asset allocation differences at any point in time on the spectrum from longer to shorter-dated accumulation funds.
The marketing literature suggest the actual difference may only be apparent from 14 years out, so the differences between this ‘dynamic’ approach and crude, linear, lifestyle fund approaches may be more apparent than real.
At the point of retirement, or when drawdown starts, the investor only holds accumulation units in a single ‘Retirement Income Fund’. This is supposedly divesified but is expected to hold 70% in bonds and the balance split between property and equities with a small allocation to a fully-collateralised (ungeared) commodity index futures strategy.
Asset allocation Any attempt at customisation therefore relies on asset allocation. The specific application of asset allocation is based on traditional thinking about the main asset class differences, notably equities and bonds, rather than ‘new’ thinking about multi-asset class diversification favoured, for instance, by a small minority of pioneers in endowment management (led by Harvard and Yale in the US) who also have to grapple with the problem of setting a stable but sustainable rate of drawdown from a capital fund, given particular objectives for residual capital at distant horizons.
Mathematically, there is no difference if the residual capital target is zero at age 100 or the same level (in real terms) as today in, say, 25 years time. The only difference is the solved-for drawdown rate. However, there are differences in the implied asset allocation. The multi-asset class approach is technically valid whatever the target residual capital because of the assumption that the errors in the expected return path from the asset strategy are not serially correlated (more alpha than beta and not mean reverting) and are independent of the time horizon. These ex ante assumptions may not be realised (there may be loads of hidden beta, low assumed correlations may turn out to be much higher and tail events may be under-estimated). But there is at least an intention in the investment strategy to match the nature of the problem being solved.
In a conventional approach to conventional asset classes, as the main portfolio building blocks, there is no implied internal logic that the same fund can serve different objectives for drawdown rate and residual capital or different time horizons. Path and outcome risks are significantly different because of the mean-reverting characteristics of the equity investments and the duration effects of the conventional bonds.
I do not believe the 4% allocation to commodities will make a difference. The main component of the expected return to a passive commodity futures strategy is the risk free rate (earned on the 90% or so of the fund applied to the collateral). The property exposure is via real-estate related securities and it is silly to assume that these will not be inconveniently highly-correlated with other equities.
Residual capital and the drawdown rate Fidelity do not make particular claims for the sustainability of a planned drawdown amount as a function of the planned low volatility of the single retirement stage fund and I am not suggesting they are being deliberately misleading.
In fact, there is one pronounced oddity in the literature which is the unfamiliar emphasis on residual capital. This makes sense if part of the household retirement spending is met from assets held outside a pension wrapper and therefore not subject to inefficiencies in the rate of extraction or tax treatment of residual capital passing as inheritable wealth.
Unfamiliar it may be but it is not unrealistic in all circumstances (think about high-income families and the impact of a lifetime pension fund limit). However, it also provides Fidelity with a convenient excuse to ‘gloss over’ the real problem: setting an optimal rate of draw where the investor’s utility treats underspending as being undesirable, not just overspending.
The literature does offer a specific suggestion. “Our research, using long-term asset class returns and sophisticated modelling techniques suggests that a 4% withdrawal rate, certainly in the early years of retirement, may provide an optimum mix of income and longevity”.
I am reminded of the early years of endowment plans, when return expectations (and therefore premiums or contributions) were based on a gilt yield and capital growth ended up being far more substantial than expected. The reason ‘low-cost endowments’ were developed, particularly for house purchase, was that the utility assigned to ‘surplus’ capital can be less than the utility of setting a realistic and sustainable contribution rate.
Because we manage money for wealthy individuals, most of our clients expect to have sizeable residual capital. However, good planning requires that to be identified in advance and separated form the retirement spending goal and then invested accordingly. We would argue that a client needing to withdraw only 4% per year should have a radically different asset allocation from the Retirement Income Fund.
Inflation Though the 26% of the fund not invested in bonds and (for counter-intuitive reasons) in the commodity index strategy may be expected to compensate investors for inflation, whatever the future rate, there is more uncertainty about the bond exposure itself.
Most of the bond return (whether running yield or redemption yield) represents the market’s estimate of future inflation. Implicit in the suggestion of a withdrawal rate of 4% is that the purchasing power of the expected residual capital will decline, as the compensation for inflation is being consumed rather than added to capital. It will decline faster or slower as a function of the accuracy of the market’s inflation estimates. Historically, the errors in estimating inflation have been cumulatively very significant for the real value of bond investment outcomes, as many asset allocation articles on this site have pointed out.
Inflation risk is a key element of retirement planning that needs to be explicit. It can only be made explicit by making the drawdown target an actual amount of money and expressing it in constant purchasing power terms. It does not need to be fixed (it could taper with age, for instance) but it must be estimated. If the target is fixed, it is the drawdown rate, as a percentage of current values, that fluctuates.
The flaw in most UK retirement planning is that the income target is planned in money terms as a level rate – and indeed the most popular form of vesting is a level annuity. Whether annuitised or matched to a level annuity, the plan adequacy is largely at the mercy of unexpected inflation. Inflation will erode the real pension, as if the targets were planned as real but tapering with age. The problem is that the taper rate is unrelated to needs or desires and is completely unknown (by definition, since it it depends on unexpected inflation). In this key respect, Fidelity has nothing innovative or progressive to offer.
Consequences for Fidelity investors Most of the logical errors relate to bond holdings.Investors in the accumulation funds will not have their allocations matched to the likely duration of the retirement capital and (unless consuming much of it quickly or annuitising it quickly) will have sub-optimal bond holdings.
Investors in the retirement fund will be consuming capital in real terms in an unplanned manner and with unarticulated prospects for residual capital.
The sales pitch implies low volatility for the retirement fund so experienced volatility may well cause concern that the product is not what they thought it was
Using the fund for pension-wrapped assets is likely to be suboptimal relative to an annuity (unless an investor distinguishes between losing out to the Treasury and to longer-lived annuitants)
There is no scope to customise the risk aversion of the investor to underlying asset allocation differences (although clever advisers may cheat by selecting a deliberately incorrect date stamp for the accumulation fund)
The form consequences take is generally sub-optimal choices: under or over-spending relative to needs or personal trade-offs between consumption and residual capital. But in extreme, the fund structure does not prevent more adverse consequences arising from excessive drawdown where unexpected inflation is high.
Alternative approaches to mass customisation We start from the assumption that most people cannot justify the cost of face to face advice on any regular basis and need the cost savings inherent in large pools of capital. These ideas were summarised most recently in a post about a meeting with the FSA.
We see the solution as being characterised by:
technology-delivered planning solutions (not requiring experienced planners or investment professionals to deliver the planning process)
customised and dynamic asset allocation (also requiring good technology to manage and monitor) and, finally,
the right asset allocation building blocks at the right price – this being where the economies of pooling come in
If the building blocks are entirely passively managed, the best people to deliver the solution will not be a predominantly active management asset gatherer like Fidelity but a giant tracker manager like Barclays, State Street or Legal & General.
What is also apparent is that the added-value elements are logically separable: the technology-rich elements of the offering, interfaces and accounting, can be separate from each other as well as from the underlying investment management; the branding driving the marketing can be separate from the technology building and certainly from the investment.
The model that could implement this approach, though within limitations that Fidelity sought to break down between pension and non-pension assets, is Personal Accounts: the proposed government-sponsored pension savings scheme.
A key investment distinction in our thinking is that risk is controlled and customised not just by diversification but by dilution: by combining exposure to an optimal set of diversified risky assets with a dose of risk free assets. The definition of the risk free asset is given by the nature of the target for the goal and its time horizon. If it is a spending amount in real terms, distant horizons for thsi target can only be met with certainty by holding index linked gilts. These are the ‘power money’ in modern asset allocation and financial planning for individuals.
If the Government were serious about encouraging more effective personal provision, it would see easy access to date-stamped inflation-proofed securities as a key enabler. Funds do not provide the same security, as the duration is not customised. Mismatched duration counts more when the path of index-linked returns is more volatile than conventional bonds, as inevitably arises from the lower coupons (perhaps one reason Fidelity rejected going the more innovative ‘real terms’ route).
For many years, the Debt Management Office and National Savings have had as their mandate to minimise the cost of government borowing, consistent only with a mix of maturities and instruments. It is a long time since there was a social goal, one recognising that the opposite side of public sector borrowing is public sector saving, for which it does have some clear objectives. We would like to see this being debated.