In parts one, two and three of this series on retirement withdrawal strategies, we looked at Bill Bengen's inflation-linked withdrawal strategies and the fifteen challenges that a U.K. retiree using his approach has to contend with:
Challenge One: U.S. data – 4% “rule” doesn’t necessarily hold for the U.K.
Challenge Two: Asset allocation of 50% stocks and 50% bonds.
Challenge Three: Limited historical data.
Challenge Four: Starting stock market valuations are not taken into account.
Challenge Five: Zero fees.
Challenge Six: 30-year retirement horizon.
Challenge Seven: Not evaluating whether the money is likely to outlast the retiree.
Challenge Eight: Not adjusting spending throughout retirement as remaining life expectancy changes.
Challenge Nine: Spending is assumed to increase with inflation each year.
Challenge Ten: Does not allow for spending flexibility
Challenge Eleven: Other income sources are not taken into account.
Challenge Twelve: Dying with too much money.
Challenge Thirteen: Not adjusting spending depending on how "lucky" your retirement outcome is.
Challenge Fourteen: Assumes perfect investor behaviour.
Challenge Fifteen: Being able to cope with drawdowns.
longevity
inflation
investment returns
making retirement planning a particularly challenging problem.
We believe there are typically three financial "worries" for a retiree.
Running out of money.
Having to take reductions in income during retirement.
Dying with too much money (challenge twelve in part two).
As a reminder, with Bengen's approach (withdrawals are adjusted for inflation each year over a thirty-year retirement), we do not adjust our withdrawals depending on how well the portfolio has been performing or how high inflation is. This leads to a wide range of potential outcomes. Using our baseline example (see challenge one in part one) with a portfolio of 50% stocks and 50% bonds, the historical worst case had the money running out after around 18 years, while the best case had the finishing balance at more than six times the starting balance (in real terms after 30 years). The reality of Bengen's approach is that there is a very good chance that you will die with more money than you started with.
What if we were prepared to be more flexible with our spending? Would this help a retiree avoid running out of money or dying with too much money? How much flexibility would we have to accept?
Guyton-Klinger guardrails
Variable withdrawal rate guardrails allow us to adjust our spending if markets are favourable, assuming we are prepared to cut spending during falling markets. I covered numerous approaches in my book 'Planning for Retirement: Your Guide to Financial Freedom', but for this article, I will focus on the popular Guyton-Klinger approach. In 2004, Jonathan Guyton released a paper titled Decision Rules and Portfolio Management for Retirees: Is the 'Safe' Initial Withdrawal Rate Too Safe?
Guyton established the following guidelines for determining the maximum safe initial withdrawal rate:
Never requiring a reduction in withdrawals from any previous year.
Allowing for systematic increases to offset inflation.
Maintaining the portfolio for at least 40 years.
Guyton looked at just one period, from 1973 to 2003, and used a more diversified asset allocation than Bengen's original work. Two portfolios were used, one with 65% stocks and the other with 80%. The 65% stock portfolio consisted of the following assets:
10% cash
25% fixed income
44% U.S. equities
15% international equities
6% real estate
Guyton determined that this approach allowed initial withdrawal rates of between 5.8% to 6.2%, far more than the (approximately) 4% proposed by Bengen (and that's not forgetting that Bengen's research was primarily over only a 30-year retirement period rather than the 40 years for Guyton's research).
There were three reasons for this improvement in initial withdrawal rates:
Better portfolio diversification.
Forgoing inflation-adjusted increase in withdrawals following a year in which the portfolio's total return is negative.
The maximum inflationary adjustment is capped at 6% (see challenge one in part one), which looks at U.K. inflation during WW1 and the 1970s, often much higher than 6%)
This paper differs from Guyton's original work in the following ways:
Three stock asset allocations are tested: 50%. 65% and 80%.
The 6% inflation cap is removed.
Two new rules are introduced, the capital preservation and prosperity rules, which act as "guardrails". The capital preservation rule is triggered when a current year's withdrawal rate has risen more than 20% above the initial withdrawal rate. If this is triggered, the withdrawal is reduced by 10%. This rule applies until the retiree is within 15 years of their predicted life expectancy. The prosperity rule is triggered when a current year's withdrawal rate has fallen more than 20% below the initial withdrawal rate. If this is triggered, the withdrawal is increased by 10%.
The paper concludes that 5.2%–5.6% initial withdrawal rates are sustainable at the 99% confidence level for a portfolio containing at least 65% stocks.
Has it 'fixed' our three worries?
We will compare the Guyton Klinger approach to the Bengen example above to see how this looks in practice.
For the following tests, we will use a portfolio of 65% developed market stocks and 35% global bonds, historical U.K. inflation and a 40-year retirement horizon. We will use the same clients in the previous articles, David and Samantha.
David Smith is 60 and recently retired from ABC Chemicals. His wife Samantha is 55 and also retired. Together, they have a retirement portfolio of £1,000,000 and are looking for a sustainable income of £40,000 per annum (please visit part one for the assumptions used).
We will use a starting withdrawal of 5.2% (which is at the lower end of that specified by the Guyton-Klinger (GK)) methodology, but at £52,000 per annum, far more than David and Samantha are seeking).
We remove the capital preservation rules 15 years from the plan's end as GK specified. Timeline does not allow the prosperity rule to extend to the end of the plan, so we have set it at 35 years (five years from the end of the plan).
If we look at our three worries:
1. Running out of money.
2. Having to take reductions in retirement income.
3. Dying with too much money.
The GK approach seems to have solved worry 1(running out of money), with the red line (investment balance) not going to zero, even for the worst-case 40-year historical scenario (when David will be 100 and Samantha 95). It also has a good stab at preventing retirees from dying with too much money (worry 3/challenge twelve (dying with too much money)), with the best-case (green line) reducing from circa £6m to around £3m at the thirty-year point (albeit we are taking out a higher starting withdrawal rate than with the Bengen approach).
How has it fared with worry 2 - having to take reductions in retirement income? If we view expenditure in nominal terms, we can see that in the worst case, income drops very little from the initial £4,333 per month starting figure.
However, looking at real income, we can see where potential problems lie. Our 1973 retiree would've had to take a near 50% cut (£4,333 vs. £2,200 per month) in real income within eight years of retiring!
Furthermore, this "allowed" spending remains at this level for almost four years. Many retirees may find this unacceptable, especially those without other sources of secure income (e.g. state pension).
If we reduce our starting withdrawals to 4% (as David and Samantha desire), our spending in real terms will still be cut by around 45%, from £3,333 to £1,800 per month for our 1973 retiree.
This big issue with the GK approach is rarely addressed, with most literature focusing on the upsides compared to Bengen's approach. Karsten Jenske wrote a series of articles on his EarlyRetirementNow blog in 2017, concluding:
"If you want to use GK yourself, make sure you’re aware of the downside (literally!), i.e., be prepared to curb consumption by 50% if things don’t work out."
More recently, an article titled "Why Guyton-Klinger Guardrails Are Too Risky For Most Retirees (And How Risk-Based Guardrails Can Help)" was published on Kitces. It again highlighted the issue of potential cuts in real income, a 45% cut during the Great Depression, which was close to the 50% our example U.K. retiree experienced in 1973.
Has it addressed the challenges?
We've addressed the three worries: how does the GK approach stack up against the fifteen challenges we covered for Bengen's strategy(see parts one, two and three)? We've already looked at Challenge 12 (dying with too much money). What about the rest?
Challenge One: U.S. data – 4% “rule” doesn’t necessarily hold for the U.K.
As demonstrated above, the GK approach does an excellent job of preventing a retiree from running out of money by cutting (real) income. That's not to say that you can't "break" it. For example, increasing the starting withdrawal from 5.2% to 10% does exhaust the money in the worst historical case within a decade, with the capital preservation rule not being able to cut spending quickly enough!
Therefore, the issue with the GK is not so much about initial withdrawal rates and safe withdrawal rate levels as it is about potential downward adjustments in real income.
Challenge Two: Asset allocation of 50% stocks and 50% bonds
Like challenge one, changing asset allocation tends to most impact cuts in real income rather than starting withdrawals with the GK approach. For example, using the U.S. total market for our portfolio leads (rather than 65% developed market stocks and 35% global bonds) leads to dramatic cuts in real income in our 1973 example, from a starting amount of £4,333 to £1,700 per month!
Challenge Three: Limited historical data
Guyton & Klinger had the benefit of conducting their research around a decade after Bengen, so they benefited from a slightly larger dataset (1928-2004). They also used Monte Carlo modelling, which allows many more simulations (sometimes over 10,000) versus Bengen's approach of rolling thirty-year samples.
The issue we have with our analysis is that we are using Timeline historical data (rather than Monte Carlo), so we are constrained by around 100 years of data.
Challenge Four – Starting stock market valuations are not taken into account.
GK did not address this.
Challenge Five: Zero fees
This was not explicitly addressed in the GK work, but we can easily evaluate the impact of adding fees to the GK model. Recall that our initial starting withdrawal rate of 5.2% led to a near 50% cut in real income to £2,200 per month for the 1973 retiree.
Adding a total cost of 1% brings withdrawals down to £1,900 per month at the same point.
While fees of 2.5% drop income by a further £200 to £1,700 per month.
Challenge Six: 30-year retirement horizon
GK was based on a rigid 40-year retirement horizon, so this doesn't offer any additional flexibility over Bergen's research and may not reflect an individual retiree's requirements.
Challenge Seven: Not evaluating whether the money is likely to outlast the retiree
As with Bengen's work, the GK methodology focuses on portfolio longevity (40 years) and doesn't evaluate the likely chances of the retiree outlasting the retirement portfolio.
Challenge Eight: Not adjusting spending throughout retirement as remaining life expectancy changes
The GK logic removes the capital preservation rule within 15 years of retirement, but other than that, it doesn't offer the facility to alter 'allowed' spending based on remaining life expectancy.
Challenge Nine: Spending is assumed to increase with inflation each year
As shown above, with the potential cuts in real spending, the GK approach leaves the retiree at the mercy of poor markets and high inflation. It also doesn't allow the option of a higher starting withdrawal with a periodic reduction in spending throughout retirement (e.g. reduce spending by 1% in real terms every year).
Challenge Ten: Does not allow for spending flexibility
Like Bengen, spending tends to be led by rigid portfolio rules rather than allowing flexibility to align with 'real world' retirement spending.
Challenge Eleven: Other income sources are not taken into account
As with Bengen, other income sources are not considered in the GK research, which could lead to unnecessary cuts in early retirement spending.
For example, in our baseline case in early retirement, our 1973 retiree's income recovered to around £3,100 per month twenty years into retirement (after falling to around £2,200 eight years into retirement).
If we add two state pensions into our plan, David and Samantha's real income drops slightly less after eight years into their retirement (£2,700 vs. £2,200) as David's state pension has just begun payment (reducing the withdrawal rate on the portfolio and therefore reducing the chances of downwards adjustments in (real) spending).
But after twenty years, the income is now at £4,200 (vs £3,100 without their state pension). You could, therefore, argue that such severe cuts were not required in the early years, as the state pension would be available to support withdrawals for most of their planned retirement.
Challenge Twelve: Dying with too much money
Covered above
Challenge Thirteen: Not adjusting spending depending on how "lucky" your retirement outcome is
This is where GK shines. In contrast to our 1973 retiree, who was unfortunate enough to have to reduce their spending to £2,200 per month eight years into retirement, our 1921 retiree has increased their initial monthly spend from £4,444 to £8,000 at the same point!
Challenge Fourteen: Assumes perfect Investor Behaviour
Both Bengen and GK's rules are just tools and do not take into account potential (mis)behaviour. In Using risk-based guardrails for retirement planning, we address the value that an adviser can add when helping clients stick to the plan.
Challenge Fifteen: Being able to cope with drawdowns
Over the short term, the GK model doesn't differ much from Bengen, with some inevitable short-term drawdowns, as shown in our 1973 example below. This is no criticism of the GK model as there is a limit to what it can do in the short term (cutting expenditure not having much of a short-term impact when the markets are falling heavily).
However, over the longer term, and provided the initial starting rate is not too high, GK does a good job of keeping the nominal investment balance reasonably steady, even in the worst historical cases.
Positives
As with Bengen's work, we may seem overly critical of the GK logic. That is deliberate - we only get one chance at retirement, so it's worth fully understanding the downsides to ensure no nasty surprises once you enter retirement and decide how much of a 'worst case' you want to plan for.
As an example, Timeline offers the following scenario outcomes.
If we look at the lower bound of likely (30% percentile) outcomes, we can see that real spending has reduced from £4,333 to £3,800 per month eight years into our retirement, a much better outcome than the £2,200 per month for our 1973 retiree. A retiree may be comfortable with this being a reasonable 'not great, although not terrible' lower bound and base their planning around this.
Conclusion
In this article, we looked at GK and its positives and negatives compared to Bengen's approach. As with Bengen's work (perhaps more so), we believe that real-world challenges mean that the approach should perhaps remain an academic concept rather than something that a retiree should implement, particularly with risk-based guardrails becoming a more established alternative, and is an example of retirement planning ideas that sound good in theory but not in practice.
Want to find out more?
If you would like to know more about designing and implementing a robust withdrawal strategy, please get in touch.
About us
The team at Pyrford Financial Planning are highly qualified Independent Financial Advisers based in Weybridge, Surrey. We specialise in retirement planning and provide financial advice on pensions, investments, and inheritance tax.
Our office telephone number is 01932 645150.
Our office address is No 5, The Heights, Weybridge KT13 0NY.
Please note: This blog is for general information only and does not constitute advice. The information is aimed at retail clients only.