Retirement planning and sustainable withdrawal rates – an update

Most of us will be familiar now with the starting point for most discussions of this kind – the Bengen 4% rule.

This research was a landmark in establishing a new kind of calculation, but Bengen's research only dealt with US historical returns, and our intention here is to outline how Bengen's approach might be applied in a UK retirement context today, using modern stochastic techniques.

First of all, the use of cash-flow plans

As readers will know, cash-flow models are prescribed for DB transfers, but in his recent report 'An-ex-regulator's guide to cash-flow planning', Rory Percival explains why cash-flow plans should be taking centre stage in a wider context of advice:

  • 'Cash-flow planning can be about any form of income and outgoings and does not have to be related to the client's overall position.' A cash-flow plan need not be a holistic view of a client's finances.
  • The second point is that the cash-flow plan can establish a client's capacity for Loss for most clients. In fact, writes the author 'it should be the main way of assessing Capacity for Loss for most clients'. Capacity for Loss is the lynchpin of compliance and meeting the funding requirement for future income is central to any plan. An adviser needs to make an expert, objective assessment of Capacity for Loss, or using its FCA definition, identify the level at which loss may make a 'material impact on the standard of living'.

Synaptic users can use growth assumptions from the Synaptic Risk profile, but the calculations remain deterministic.

Use of stochastic modelling to demonstrate the possible investment outcomes

  • Financial planning in Synaptic Modeller is built around a forward-looking methodology that takes the markets as they exist now, and studies the whole range of 'viable' outcomes. This is increasingly proven as a superior technique to basing assumptions on previous market conditions.
  • The great gift of the Moody's Analytics stochastic model is the ability to ascribe the extent of losses to a portfolio in a bad year (covered in subsequent article) and the ability to ascribe probability to possible outcomes.

To illustrate the dynamics of risk over the period of a client's retirement, we ran some scenarios in Synaptic Modeller, a tool that accesses the Moody's stochastic model. We looked at the probability of success for a couple who are approaching retirement (age 59), looking to retire at 65, and we are modelling to try and understand what investment strategy to apply to ISA holdings that have accumulated over the years (so no tax implications). As our clients have their bills covered with pension income and state income, this is purely discretionary spend. Our client and his wife however both have a balanced risk profile, so it will be difficult to recommend a solution that takes them over this threshold. The research however, points to the need to take risk on.

These illustrations are 'real' so include outcomes across the full range of viable variants for inflation and interest rates. (One advantage of the stochastic approach is the built-in stress testing).

The percentage withdrawal rate is an 'initial withdrawal rate', which is subject to an annual uplift of 2.7% (current CPI) to maintain purchasing power.

In our first graph, the green line represents the probability-based outcomes for £100k invested in the Balanced portfolio in our C.I.P., (projections are calculated using asset allocation). The red line represents the projections from our C.I.P.'s Adventurous strategy, and the blue line is Cautious.

We are aiming for a nominal £50k remainder in the pot at 95 years old as our definition of success.

Here are the results. Probability of success (£50k remaining) from our scenario:

Strategy / Withdrawal Rate 2% 4% 6%
Cautious 34% 0 0
Balanced 73% 32% 8%
Adventurous 76% 53% 33%
Risk Level 10% Best Outcomes 10% Worst Outcomes Range (90th to 10th Percentiles)
Cautious >£9.6k <-£43.8k £54k
Balanced >£129k <-£35k £165k
Adventurous >£432k <-£48k £480k

This is what the graph looks like, mapping investment outcomes from the simulation per £100k invested against probability for the three strategies…



It can be inferred from our research that a 'safe initial withdrawal rate' would be significantly less than the 4-6% I will recommend to my client. Looking at the £0 final fund value, I can see a balanced strategy would suggest a 65% chance of success for the strategy to last the distance. Remember that Bengen didn't allow for costs, where I have allowed 1%.

What really hits home is the impact of inflation. If I were run the scenario on a nominal basis the results are obviously very different. The temptation is of course to show the nominal projections to the client… 85-90% chance of success… but this would be half the story:


The ranges of possible outcomes is large (I have shown the range for 80% of the outcomes in the table). Clearly a plan will need to be carefully monitored, because adjustments can and should be made depending on the sequence of returns in the early years. If all is well, more money will be able to be extracted, but a poor sequence of returns initially will necessitate a constriction of the withdrawal rate and recovery time for the fund.

A major concern of course is the cognitive ability of the client (and the adviser!) in later years, so the due diligence and research needs to be very thorough, including understanding of how the reviews will be undertaken. The sensible approach here is to work with a Withdrawal Statement that sets out the expectations and rules to address different market conditions. This is also the best way to demonstrate the suitability of the plan, as the rules will be designed to prevent the depletion of the principal and breach of the client's Capacity for Loss.

There are many interesting considerations that come out of this research, for example how different asset allocations and investment styles might work; whether the use of natural yield to determine withdrawal rates might help, and my favourite, allowing the asset classes to drift.. which again provides some really interesting results and options for the adviser. Maybe a topic for next time.

All illustrations are from Synaptic Modeller. You can access the Moody’s stochastic engine via Synaptic Modeller for £50 per month (network discounts may apply). Please call us on 0800 783 4477 or email us at