Freedom and choice have driven a big increase in drawdown sales and a growing sense that more science is needed to help people manage their savings as safely as possible.
With the exception of flexible drawdown (which, let’s face it, only applied to those with significant pension wealth) pre-April drawdown came with a safety mechanism: GAD limits.
The rules preventing savers from withdrawing income greater than a comparable annuity stopped people depleting their drawdown savings to zero. Granted, even with these limits in place, significant damage to a drawdown portfolio was possible given bad market conditions.
But the GAD limits are no more and savers and their advisers are now responsible for imposing their own sustainable income rules.
This has led to increasing discussions around issues such as sequencing and volatility risk or, as some have described these dangers, ‘pound-cost ravaging’.
So, where do we start? Most look to the US, where drawdown without GAD limits has always been possible. The science there is better developed, although there remains considerable room for further study to build on the excellent work of a few US financial planners.
The starting point for the US drawdown theory is the ‘safe withdrawal limit’, originally devised by Bill Bengen. Bengen examined the historical performance of a portfolio consisting of 50 per cent US equities and 50 per cent medium-dated treasury bills.
What Bengen found was that for any 33-year period starting between 1926 and 1976, you would not run out of money as long as you took no more than 4 per cent of the initial capital each year, increasing in line with inflation, from your drawdown portfolio.
This work led to the so-called safe withdrawal limit of 4 per cent or ‘Safemax’. In calculating Safemax, Bengen did not take into account charges, which would reduce the maximum withdrawal but (due to the maths) not necessarily by the same amount as the charge.
Further work by Wade Pfau in 2010 calculated the UK Safemax to be just 3.43 per cent, again before charges and other costs are deducted.
Safemax is a rather extreme concept, accepting a zero risk of running out of money over any historical 30-year period. It also fails to manage a drawdown portfolio dynamically to optimise the amount of income that can be taken.
If clients are prepared to accept a risk of running out of money – even a small one such as 2 per cent or 3 per cent – then withdrawal rates (based on historical analysis) can be improved.
Applying withdrawal rules to historical modelling can also significantly increase the initial level of withdrawal.
A 2006 paper by Jonathan T Guyton and William J Klinger outlines a number of rules that could be applied to a drawdown portfolio, such as banking outperformance as cash to be drawn as income in fallow investment years. It also sets rules that determine which asset income withdrawals are extracted from, what happens when overall portfolio returns are poor and what happens when they are good.
Guyton and Klinger found it was possible to draw 6.3 per cent a year over 40 years from a 65 per cent equity, 25 per cent fixed interest, 10 per cent cash portfolio by following these rules with a zero chance of failure. This, again, was based on US asset returns.
Accepting just a 2 per cent chance of failure, the initial percentage withdrawal increased to 7.3 per cent.
Like Bengen, Guyton and Klinger also assumed withdrawals would increase in line with price inflation – in their case up to a cap of 6 per cent and, in years where portfolio returns were poor, there was no inflation increase at all.
This meant total withdrawals over 40 years did not equal 40 times the initial withdrawal in real terms. In the zero failure scenario, total real income was 93 per cent of 40 times the initial withdrawal and in the 2 per cent chance of failure case 88 per cent.
This research provides an excellent platform on which the UK financial planning profession can build.
Although this analysis focuses on the past (and, as we are reminded daily, the past is no guide to the future), some of the research stretches back almost as far as reasonably possible – in Pfau’s case over 100 years.
Another possibility is to use stochastic models (Guyton and Klinger also used stochastic modelling to determine probability of failure). Stochastic models can help predict the probability of ruin using thousands of different potential future scenarios, including scenarios similar to the early to mid-1970s with disastrous asset price falls and runaway inflation.
Such modelling creates the basis for a discussion about the risk of running out of money over any given period relative to the amount being withdrawn, so clients can clearly see the trade-offs.
John Lawson is head of financial research at Aviva