After collecting data from all transfer specialists in the market, the FCA is ‘concerned and disappointed’ with the number of transfers from Defined Benefit to Defined Contribution pensions and is promising further action.
Why is the data either surprising or worrying? If anything, the surprise is the low proportion that has hitherto been advised to transfer.
Fowler Drew contributed to the data-gathering exercise the FCA conducted last year. We were at the time critical of the questions asked which we did not think would cast enough light on whether or not there is a problem with the volume of transfers already made or that other members might want to consider in the future if DB schemes continued to derisk and real interest rates remained very low.
Here is what the FCA said yesterday when publishing its findings after surveying the activity of 3,015 firms advising on transfers between April 2015 and September 2018:
2,426 firms had provided advice on transferring their DB pension.234,951 scheme members had received advice on transferring.Of those 162,047 members had been recommended to transfer out and 72,904 had been recommended not to transfer.The total value of DB pensions where transfer advice had been provided was £82.8bn with an average value of £352,303.1,454 firms had recommended 75% of more of their clients to transfer. One reason a firm may be recommending a large number of clients to transfer is if the firm has a robust initial guidance service triaging clients. 1,346 firms reported data on the total number of clients who had not proceed past the firm’s initial guidance. The total number of clients reported as not proceeding to advice was 59,086. When these triaged clients are factored in 55% of clients were recommended to transfer.
Fowler Drew’s own data was remarkably similar, which we found was alone telling, as our advice process is driven by interaction with an ‘engine’ and we are confident the model structure and the interaction with it are designed to be indifferent between staying or replacing. Including triage where clients did not go on to a full advice process, transfer was supported by that process for 60% of all cases versus the population finding of 55%. (Press reports may well focus on the higher proportion of ‘75% or more’ because that is the headline figure it is apparently so disappointed to discover.) Even our average transfer value was very similar (although the sums will be much smaller as a proportion of total wealth given the high net worth of our clients). The big difference is the volume and frequency per adviser – data not collected but we know it’s a production line for some and genuinely bespoke for us.
The difference made by triage in these findings is highly significant. It implies that having a process of guidance that is generic and stops short of being a personal recommendation (under the rules) was doing the job it ought to do of ‘correcting’ uninformed and largely emotional opinions about the improvements in personal welfare available by transferring. However, we doubt if many of these triage processes were entirely generic, given that they were taking place before the FCA spelt out the risk of a breach of the rules if any comments were made that were specific to the individual case and could be construed as a comment on the merit. The boundary is inevitably fuzzy because commercial criteria can so easily (and legitimately) serve as a proxy for welfare criteria (e.g. past experience of risk taking or total wealth positions could either qualify a transfer or qualify a client). The actions taken by the FCA to prevent this boundary being crossed must look perverse now, if not before, in the light of this data.
At Fowler Drew triage is a process that can involve interaction with a model. This ‘transfer calculator‘ is an adapted version of our standard drawdown model. As inputs it needs a notional resource, which could be the Cash Equivalent Transfer Value or an estimate of it, and start date (planned draw) and end date (how long the money must last). Risk is a variable that can be experimented with. The output is a probability distribution for sustainable annual draw in real terms. This is made comparable with a target to beat, which could of course be the present value of the projected DB pension income. Apart from making the drawdown model more specific to a transfer, different iterations help explain the key principles of any decision involving replacement of a risk free asset (or income) with a risky asset (or income). Since the trade-off involves long-run real outcomes (sustainable spending) and short-run nominal volatility, the model also outputs the latter, including peak to trough falls.
This is vital in the case of a transfer because it is lack of past exposure to volatility (and all the emotional or behavioural stresses and errors it entails) or past examples of poor persistence that most often invalidate a transfer. Mathematically this must be so at a time when the transfer values are bound in many cases to offer greater welfare, in the form of a very low chance of falling short of the risk free pension amount. (This prima facie welfare gain is what predictably arises when schemes have ‘derisked’, real discount rates are low or negative and equities are fairly normally valued. For the FCA to ignore this economic or market context and persist with the default position that transfer is not usually beneficial implies an assumption about the frequency of low experience or low risk tolerance in the population of DB scheme members. We wonder what it bases that on.)
Because we put this model online, where we cannot see the inputs and outputs, we are not able to count how many cases there were of prospects deciding not to proceed with advice after using the model to estimate the chances of any welfare improvement from investment returns alone. If we could include those, the proportion whose transfer was supported would be even lower than the 60% we reported to the FCA.
DB transfers have never been an important business line for us and we did not support the transfer calculator with a serious SEO marketing budget. (In fact now it has none, as we are unable to advise on any new transfers until our Professional Indemnity insurer has worked out what to charge for underwriting our past business.) But it did occur to us that this would have been a valuable resource for members of the British Steel Pension Scheme, faced with very difficult choices under time pressure and in an atmosphere of frenzied and often unprincipled advice activity. We had one conversation prompted by a steel worker finding the application online. He told us about his past experience of an equity-based fund in an ISA that he had decided to sell when the price behaved badly. He couldn’t be sure that the same wouldn’t happen again. We wouldn’t be allowed to say that he therefore should ignore the transfer opportunity and not waste his money with an adviser. But we could say he did not fit our criteria for taking on discretionary clients with drawdown plans. Either way, he would have discovered what he needed to know just by thinking about the outputs of the application: he could tolerate the chance and size of a worst-case shortfall in real spending and was highly motivated by the material spending upside but he wasn’t confident he could live with the volatility that went with taking investment risk.
A useful finding of the FCA exercise was that the feature that characterised the feeding frenzy around steelworkers, unregulated introducers, is not significant in the wider market. That’s a relief.
The volumes do not alone tell us much about the prevalence of transfers to date, or potentially in future, as a proportion of all DB scheme members. They appears very small but that could be because the scheme conditions for increasing utility, namely derisked asset exposures and full funding of liabilities, do not more generally apply. Now, of course, we can add greater capacity constraints because the FCA has effectively driven advisers out of the market.
Finally, the data calls into question the representativeness of the FCA’s selective, file-based findings of poor levels of ‘suitability’ in adviser’s recommendations. These have been widely quoted without acknowledging (partly due to the FCA’s own slowness to do so) the fact that these file searches were based on information about possible poor practice, not on random selection or complete data gathering. Our feeling was that, if this was prevalent and was a function of biased advice or failure to identify risk preferences (rather than just minor process shortcomings), it would show up in a much higher proportion of advised transfers than the market-based findings indicate.
This is not to say standards cannot and should not be raised. But, together with the wording of this latest release, it does imply the FCA has unreasonable and unfounded expectations of the likely scale of transfers. That in turn might be because it lacks a proper understanding of the source and scale of the utility gain from transfers generally, or sees only what it wants to see.