Where the risk is hidden in CDC
Our response to the DWP consultation on CDC focuses on a level-playing field for risk disclosure. We ask for fair and not misleading communication of the risks, preventing any use of 'money illusion' to hide the risks that DB pension plan sponsors and financial advisers managing DC drawdown cannot hide.
You can read our two-page response to the current DWP consultation here. Our response only addresses the question of communication but that is one of a number of important aspects of product integrity it needs to be reassured about. In this article we take a little more space to set our views about authenticity of the CDC sales pitch in a broader market and historical context, and to show how it compares with our own approach to customised drawdown.
Return-seeking risk assets like equities bring with them significant risk of long windows of negative or very low real returns. (In our response to the DWP we quantify this, based on our model, in terms of one and two standard deviations of real returns for a globally-diversified equity portfolio over horizons of 10 and 20 years.) This is a market reality perfectly well understood by pension professionals including the consulting actuaries that have been attracted to the business opportunity presented by CDC fund structure, as a hybrid of DB (a market they are losing) and DC pensions (a market they are barely in). A viable hybrid offers competitive opportunities selling against both DB sponsors eager to close their schemes and financial advisers recommending or managing drawdown from personal DC pensions. (A reminder of the acronyms here. DB stands for Defined Benefit: member gets a defined outcome with negligible risk of shortfall. DC stands for Defined Contribution: member gets whatever pension the invested contributions will generate, so very risky outcomes. CDC stands for Collective Defined Contribution: members get an averaged pension across them all, so the outcome uncertainty is shared with the other members of the pool.)
What the consulting actuaries also know is that the simple act of collectivisation, based on the principle of unitising individual members' interest in the fund, does not make the inherent real-return risk go away. Pooling may have certain advantages, including dealing with longevity risk and lower costs, but not making capital-market risks disappear.
However the problem of retirement funding is framed, the key question is always 'can we use the risk premium, or positive real return trend, observed historically from risky assets to reduce the cost of funding a liability that arises a long time in the future?' Even if the answer is 'yes', it requires some mechanism for dealing with the brief or even very long intervals in which it doesn't work, so that confidence can still be sustained. That in turn requires some form of (no criticism implied) confidence trick. The obvious way is to conceal the path itself, so that everyone gets the benefits of lower funding costs without being tempted to give up along the way. DB pension schemes did this for a long time until international accounting conventions and UK regulations changed, forcing disclosure of what was happening along the way and preventing any illusion. With-profits policies sold to individuals used to report a notional, smooth path for the emergence of the expected long-term growth. They used reserves as a buffer to absorb the uncertainty about the actual path but having run down the buffers unwisely actuaries' smoothing errors were eventually exposed. Early 'institutionalised' DC pension schemes, before the era of SIPPs or personal pension accounts, relied on contract terms to keep members on board: they might experience panic but they couldn't panic out. All of these past forms of confidence trick lacked technical integrity, sales-pitch authenticity or even a bit of both.
Our approach, as managers of drawdown plans, includes a confidence trick, albeit a fully explained and conceptually consistent one. Since the incentive to low-cost funding relies on 'mean reversion' in real equity returns, in the sense that a positive trend is likely to be a systematic feature of capitalist markets, why not make the future expected real returns always precisely conditional on the current level of the portfolio? If markets fall, future real returns will be higher, making outcomes (the eventual real pension) relatively stable. Stable outcomes inspire confidence and composure even when the going gets rough. There still needs to be a risk structure, as a distribution of possible outcomes, even when assuming mean reversion will persist. That risk structure (and its relationship with time) is also observable from the data histories for deflated stock market indices and for exchange rates when adjusted for relative inflation. At the heart of the required solution is a constantly updated 'return-generating engine' applicable across as many different markets as can reliably be modelled.
Applying this engine to planning a rate of draw from a personal pension (or indeed from non-pension capital), we can solve daily either for the resources required or risk required to meet a distribution of possible real draw amounts, year by year, that obeys client-given constraints, such as not running out on a worst-case basis till age 100; or not tolerating a breach of a minimum real spending level; or avoiding declines in real spending greater than some tolerable degree, whether arising from steady erosion of spending power or sharp falls. The resulting draw rate will be smoothed, as long as the constraints require a smooth path for spending. But it will not be equivalent to a draw rate having only a 50% chance of being achieved, as that implies the chance of cuts is at odds with the client constraints. So the risk buffer is created by drawing at a lower real rate that has a higher probability of being sustained within the agreed constraints at every stage, creating an option in the form of higher spending later.
How does CDC deal with the same risk? It cannot use capital buffers (reserves being considered unfair to different generations of members) and it needs to pay out at a rate having a 50% chance, not a higher chance, (as the payoffs to the option created by drawing at a lower level would accrue to someone else). So it needs another form of buffer. Following in a long tradition of UK financial products, it has adopted a less conspicuous buffer: money illusion. Converting the real-return model of capitalist financial markets into a nominal return model by adding inflation tells us that the two elements historically contributed broadly similar amounts to the total return. So treating nearly half the expected nominal return as a buffer goes a long way, perhaps even further than necessary, to replace the capital buffer equal in size to the real-return risk or range of possibilities at a high level of confidence, or to replace the buffer or upward only-option created when an individual chooses to draw at a rate that has a much higher probability of being sustained than just 50%.
Money illusion worked in the past because most people did not make any distinction between nominal and real returns or nominal and real outcomes. Even real wages could be eroded without workers reacting as if actual pay had been cut. House prices in real terms could fall by as much as 25% (in one episode by 37%) without owners panicking as long as the nominal value never fell. But a long period of high inflation, followed after a relatively brief interval of normality before starting a long period of frozen nominal incomes, has gone a long way to banishing money illusion across most age groups. So it is less easy to use it in an exploitative way, where one party knows better than the other what is really going on. But it is still (apparently) capable of being exploited, as long as it is not too obvious. And this is what CDC, if allowed to, does with its inflation buffer.
In our definition of the typical constraints an individual imposes, sustainable smoothed spending (even if it has a time profile added such as more early in retirement and less later) treats a reduction caused by asset prices failing to keep pace with inflation as the same as a cut caused by falling asset prices. We see this as no less realistic than employees reacting as badly to a pay cut caused by poor trading as one caused by the combination of unchanged pay at a time of high inflation. Its importance stems from consequences, such as a squeeze on discretionary spending that at low levels of pay might even bite into non-discretionary spending. Consequences are blind to the source.
The CDC version of the confidence trick when relying on risk assets to reduce funding costs to work only works if in fact scheme members do make a differentiation between the sources, even if the consequences are the same. It also relies on 'success' being measured as avoiding nominal cuts from one year to another, together perhaps with the period it takes to restore the cut. This is not a negligible risk but not as likely to arise as periods in which nominal pensions are maintained but left to erode in real terms.
The CDC form of money illusion is quite similar to that relied on by so-called 'third-way annuities' with guarantees. An example is the recently closed MetLife retirement product that included an upward-only guarantee of a minimum nominal pension. The guarantee would ratchet up when returns earned were high enough and would be frozen otherwise. Even without seeing the underlying modelling (we asked them), it was obvious that the real equity return risk would show up as a high probability of long windows of erosion of a spending level that was not (in real terms) sustainable. The MetLife product, like almost all earlier versions of third-way annuities that relied on a subtler form of confusion between real and nominal incomes, was withdrawn last year. It may be that the illusion did not fool advisers, who they relied on to distribute the product. But equally what killed it (and other versions before it) could have been the fact that the embedded option (a guaranteed nominal minimum with an upward-only ratchet) was priced off the market and so, even if the risk was being concealed, the high cost of hedging it was not.
Third-way annuities were able to exploit money illusion because the product was positioned as an alternative to a level annuity. But this is too easy a target to beat when the objective is properly described in terms of real retirement spending. CDC promoters have tried to use the same argument but it no longer works now that Pension Freedoms have removed the bias or default to a level annuity for DC plan members. Communication to members must measure up against the possible outcomes that exploit the freedoms: drawdown.
If we could be persuaded that a client did in fact define a good outcome in the way the CDC promoters have framed it, so that differences in the origin of real spending variance mattered even if the consequences were the same, and if we thought the real income meeting particular constraints was likely to be higher than we could generate with drawdown subject to the same constraints and the same definition of a good outcome, then we would want to incorporate into our holistic retirement planning any CDC product designed to deal with decumulation. (This would apply in exactly the same way as we assess the merits of retaining a DB pension as part of a holistic spending plan, unbiased by fees because we charge flat, not asset-based, management fees.) But we need a level playing field. The risks to the sustainability of real, not nominal, incomes at all time horizons must be spelt out clearly so an individual can reveal directly their true definition of welfare (i.e. utility, or what constitutes a good outcome).
Fair and not misleading communication of the sustainability of real pension income is vital to the authenticity off the claims made by sponsors of CDC, be it an employer (like Royal Mail), a promoter of a collective scheme set up to attract decumulators in direct competition with drawdown or indeed the consultants and pension professionals lining up to offer services in this area.
Fair and not misleading communication is also vital to ensuring firms like ours are able to compete in a regulatory environment that is not biased to one solution or another. The last thing we want is to find the FCA or the FOS believes, on incomplete information, that collective schemes are superior because they have removed a risk that remains in DC drawdown. 'Moved', yes; 'removed', absolutely not.