Even when researchers split people aged 50+ between top and bottom earners (ie people whose income is either above or below the median), the trend remains broadly consistent. People spend progressively less as they get older. However, high earners experience a much faster and steeper decline in spending after age 60 than bottom earners. Bottom earners have more stable spending patterns throughout their lifetime, and their total expenditure starts to decline only after age 70.
Fig. 31: Income and consumption expenditure at different ages. Source: ILC (2015) Understanding retirement journeys: expectations vs. reality.
Fig. 32: Income and consumption expenditure at different ages – top and bottom earners. Source: ILC (2015) Understanding retirement journeys: expectations vs. reality.
Fig. 33: Consumption by consumer segment. Source: ILC (2015) Understanding retirement journeys: expectations vs. reality
The authors concluded that older people spend consistently less than their younger counterparts regardless of their income. This trend is persistent, and it’s not a periodic effect.
‘Consumption in retirement starts relatively high and ends low. This pattern is common to both high and low-income groups, is robust to the inclusion of factors other than age and is not simply the result of the time period in which the data was collected.’
Amazingly, this trend is broadly consistent even when you look at people with different lifestyles. The authors note that even the ‘Extravagant Couples’ – those who spend nearly 40% of their total expenditure on recreational goods and services – spend more than their income in the first decade or so of retirement, as do those who are ‘Just Getting By’, The ‘Prudent Families’ and ‘Frugal Foodies’ consistently spend below their income over the duration of their retirement.
All of the groups save in later life. From age 75 onwards, even the ‘Just Getting By’ group, which is the lowest income group, starts to save. This is because they spend less on non-essential items towards the end of life.
These findings are consistent with findings19 in the US, where researchers found that retirement spending tends to decrease by at least 1% a year in real terms throughout retirement!
Additional research by JP Morgan Chase (2015)20 shows that, even for high-net worth (HNW) households (US), spending in retirement tends to decline as they get older. Care costs jump in later life but not enough to upend decline in other discretionary expenses.
Of course, many people will need care in later life, but this isn’t typical. Consider the following data from Age UK (2016)21.
Only 16% of people aged 85+ in the UK live in care homes.
The median period from admission to the care home to death is 462 days. (15 months)
Around 27% of people lived in care homes for more than three years.
Approximately 30% of people use some form of local authority-funded social care in the last year of life.
Fig. 34: Average spending pattern of various age groups (Chase Households with $1-$2million in assets)
It’s important to factor possible care costs into a retirement income plan, but the chances that someone actually ends up having to pay for care are relatively low. In any case, it’s entirely practical for people to rely on their property wealth to meet care costs in later life. And while it may not always be ideal, local authority-funded social care remains the last resort under current law.
So what?
The assumption that income withdrawal needs to keep up with inflation throughout the entire retirement period is not supported by empirical data on spending patterns in retirement. This has a major implication for how we calculate sustainable withdrawal from retirement portfolios.
An ideal sustainable withdrawal strategy should follow the typical spending pattern in retirement. This allows higher withdrawal at the early part of retirement and should progressively fall (at least in real terms) over their retirement period.
A major weakness of Bill Bengen’s SWR is that it assumes expenditure remains constant in real terms throughout retirement. This is inconsistent with research findings on how consumption actually changes in retirement.
Thankfully, a range of flexible withdrawal strategies has been developed to reflect likely changes in retirement spending.
ILC (2015) Understanding retirement journeys: expectations vs reality. International Longevity Centre
Blanchett, David, Estimating the True Cost of Retirement. Morningstar, 2013. http://corporate.morningstar.com/ib/documents/MethodologyDocuments/ResearchPapers/Blanchett_True-Cost-of-Retirement.pdf
JP Morgan Chase (2015) Spending In Retirement https://am.jpmorgan.com/us/institutional/library/retirement-spending
Age UK (2016) Later Life in the United Kingdom December 2016
CHAPTER 6
SWR 2.0: the power of flexible withdrawal strategies
‘Luck is not a strategy.’
– James Francis Cameron
Since Bengen’s original research in 1994, numerous studies have expanded the SWR framework to incorporate additional factors. I call this SWR 2.0.
In this section, I want to explore a number of rule-based withdrawal strategies. They aim to identify simple adjustments to withdrawal that mean money lasts a lifetime, but without excessive income volatility. Crucially, these strategies mean withdrawal can match the typical consumption path in retirement more closely.
There are several withdrawal rules – too many to discuss and model. I’ll focus on six different flexible withdrawal strategies. We can broadly classify them under two categories:
inflation adjustment
advance withdrawal rules
Inflation adjustment is a rule-based approach to how to adjust withdrawals for inflation each year. Advanced withdrawal strategies make further adjustments to the withdrawal over and above inflation adjustments.
Fig. 35: Rule-based sustainable withdrawal strategies
In practice, you can combine the advanced withdrawals with any of the inflation-adjustment rules. This gives a total of 16 strategies. Four inflation-adjustments (including Bengen’s CIA) on their own and 12 more combinations of advanced and inflation-adjustment rules.
For simplicity, I only discuss seven strategies in this book:
Bengen’s CIA
three other inflation adjustment rules on their own
the three advanced strategies combined with Bengen’s CIA
The method
I’ve modelled them all using and comparing historical data between 1900 and 2016, inclusive.
I compare the three inflation-adjustment strategies with the baseline strategy, which is Bengen’s CIA.
Each retirement block lasts 30 years.
The number of years of available data provided 87 separate scenarios but overlapping 30-year blocks of retirement.
I work with a 50/50 UK Equity/UK Bond portfolio, with annual rebalancing.
I give the gross safe withdrawal rate (ie no fees) and net safe withdrawal rate (ie net of 1% charges).
I’ve also selected some scenarios from the dataset and use them to illustrate the impact of inflation on the withdrawal during the retirement period.
Cliff Edge Scenario: this is the 10th percentile historical scenario. Think of this as one of the very worst 30-year scenarios since 1900. The hypothetical retiree experienced very poor sequence of return and high inflation. Nine out of 10 historical scenarios were better than this.
Comfy Scenario: this is the 50th percentile or median historical scenario, with relatively modest inflation and good sequence of return. Half of historical scenarios will be better than this.
Cloud Nine: this is the 90th percentile historical scenario. This is one of the very best historical scenarios for retirees since 1900. Our hypothetical retiree enjoyed very good sequence of return and subdued inflation. Only one in 10 historical scenarios are better.
Inflation-adjustment withdrawal strategies
These strategies use different ways to adjust the withdrawals for inflation.
Fi
xed withdrawal – no inflation adjustment
A key feature of Bengen’s SWR is that withdrawals are adjusted (up or down) for inflation every year. But there’s abundant research to show that spending in retirement isn’t static in real terms. In fact, spending tends to fall in real terms. Around 90% of annuities purchased are fixed annuities, without any index links or inflation adjustments.
So, what’s the sustainable withdrawal rate if we don’t adjust income for inflation?
The fixed or level withdrawal strategy defines withdrawal as a percentage of the initial portfolio. Subsequent amounts aren’t adjusted for inflation. For instance, a withdrawal rate of 5% from a £100,000 portfolio will produce an income of £5,000 in the first year. The retiree takes an income of £5,000 each year. This happens regardless of the portfolio value and the income isn’t adjusted for inflation. This strategy mimics level annuities.
Fig. 36 opposite compares fixed and inflation-adjusted withdrawals. It shows historical maximum sustainable withdrawal rates over a rolling 30-year period. A gross withdrawal rate of 4.5% (before fees) is the worst historical scenario for retirees opting for fixed withdrawals. This compares favourably with 3.1% for inflation-adjusted withdrawals.
Even after accounting for a 1% fee, the 50/50 UK portfolio still supports a fixed withdrawal of 4% of the initial portfolio over a 30-year period in the historical worst-case scenarios.
Most retirees appear to be comfortable with this approach, because they tend to spend progressively less as they get older.
However, with a fixed approach there’s the danger that the real value of their income will fall faster than what they need. To understand the impact of inflation, we can compare the real income under this fixed withdrawal (FW) with Bengen’s constant inflation-adjusted withdrawal (CIAW) strategy.
For this comparison, consider two hypothetical retirees, James and John. They each have a portfolio of £100,000. The asset allocation is the same 50/50 UK portfolio, with annual rebalancing and total portfolio charges of 1%. Our first retiree James adopts Bengen’s CIAW method. He takes £2,600pa from his portfolio, which is then adjusted for inflation each year. Our second retiree John adopts the FW approach. He takes £3950pa from his portfolio but it’s not adjusted for inflation in future years.
Fig. 36: Initial sustainable withdrawal rate: inflation-adjusted vs. fixed spending (no fee deducted)
Fig. 37: Sustainable withdrawal rate: inflation-adjusted vs. fixed spending (1%pa fee deducted)
Fig. 38: Real income under Bengen’s constant inflation-adjusted withdrawal vs. fixed withdrawal (1%pa fee deducted)
Fig. 38 compares income with Bengen’s CIAW and the FW method in the three scenarios.
With Bengen’s CIAW approach, the real income of £2,600pa is maintained under all the historical scenarios. So, we only see one glide path.
Under the FW approach, income starts at a much higher level of £4,000pa. But this declines rapidly under both the Cliff Edge (10th percentile) and Comfy (50th percentile) scenarios. These represent a decline of 73.62% and 40.98% respectively in real income by the 15th year of retirement. This is way more than the typical spending decline of 35% from age 65 to 80.
Over 90% of annuity purchases are level annuities rather than index-linked annuities. So, it appears many retirees are quite happy to tolerate inflation risk. This may be because the State Pension tends to keep up with or exceed inflation and it compensates for the decline in real income elsewhere.
So, if a client adopts the fixed percentage withdrawal, they can spend more in the early part of retirement, but the impact of inflation on their income may be too much for them.
Guyton inflation-adjustment withdrawal
The prospect of an income that falls progressively in real terms may not appear all that attractive to many retirees. And if investment returns turn out to be favourable (as you’d hope they would in the vast majority of circumstances), it would be nice to enjoy the upside.
One of the earliest proponents of rule-based spending is US-based financial planner Jonathan Guyton. His main point is that retirees don’t want to play a game of chicken with their retirement portfolios and are often prepared to make compromises.
Guyton developed a set of rules and an approach to make sure withdrawals last a lifetime. These rules can be applied together, but they’re also effective individually.
In this section, I want to model one of Guyton’s individual rules. The rules state that: withdrawals are adjusted for inflation annually except in the years after a negative portfolio return.
This rule is a half-way house between Bengen’s inflation-adjusted withdrawal and the fixed withdrawal.
Fig. 39 compares the historical withdrawal rates under the Guyton inflation adjustment (GIA), the fixed withdrawal and Bergen’s CIAW.
Fig. 39: Initial sustainable withdrawal rate: Inflation-adjusted vs. Guyton inflation adjustment vs. Fixed spending (no fee deducted)
A GIA strategy produces a sustainable withdrawal rate of 3.7%, compared to 3.1% for inflation-adjusted withdrawal and 4.5% for fixed withdrawal. This is before fees of course.
Even after we apply 1% for charges, SWR under the GIA strategy is still a tidy 3.2%, compared to 2.6% under Bengen’s CIAW.
Again, in Fig. 40 we’ll compare the spending glide path in real terms under both strategies using our three scenarios – Cliff Edge (10th percentile), Comfy (50th percentile) and Cloud Nine (90th percentile).
With the GIA approach, income starts at a higher level than Bengen’s CIAW, at £3170pa. But it declines progressively under both the Cliff Edge (10th percentile) and Comfy (50th percentile) scenarios. These represent a decline of 38.62% and 16.09% respectively in real income by the 15th year in retirement. This is more in line with the typical spending decline of 35% from age 65 to 80.
By the 30th year in retirement, spending has declined by a total of 44.11% and 29.54% under the Cliff Edge and Comfy scenarios respectively.
The Guyton inflation-adjustment approach is much more in sync with a typical retirement spending pattern. It means you can start spending at a higher level than Bengen’s CIAW, but the decline in real terms is much more constrained than the fixed withdrawal strategy.
Fig. 40: Real income under Bengen’s constant inflation-adjusted withdrawal vs. Guyton inflation adjustment (1%pa fee deducted)
Cap and collar inflation adjustment
This is another halfway house between constant inflation adjustment (CIAW) and fixed withdrawal (FW) strategies. The cap and collar inflation-adjustment approach addresses the problem of extreme inflation and deflation.
Under Bengen’s rule, you adjust withdrawals for inflation, regardless. This includes historical periods of double-digit inflation, particularly between 1915-1920 and most of the 1970s.
There were also periods of deflation – for instance in 1921 when prices fell by over 20%. Retirees would have had to cut their withdrawals consecutively over a period of five years in the late 1920s. While the real spending power of each year’s withdrawal remains the same, it could be a challenge explaining this to the client. These extremes are a key reason why Bengen’s SWR for the UK is so low in the early 1900s.
A rule-based approach could be used to cap and collar each year’s inflation adjustments. This way, retirees can avoid extreme changes to their income withdrawal, at least in a nominal sense. A cap-and-collar approach also matches the typical retirement spending pattern. Income increases in real terms but at a slower pace than inflation.
Here I’ve applied a cap of 5% to inflation and a collar of 0%. (Advisers can model any combination of cap and collar through Timelineapp.co.)
Fig. 41 shows the Cap and Collar 5/0 approach gives very similar withdrawal rates to Bengen’s CIAW: a SWR of 3.6%, compared to 3.1% under Bengen’s CIAW and 3.7% under the GIAW, before charges.
After applying a 1% fee, the cap and collar 5/0 (CC) produced a net sustainable withdrawal rate of 3.1%, compared to 2.6% under Bengen’s CIAW.
Wit
h the CC, income starts at a higher level than Bengen’s CIAW – £3120pa – and it’s maintained under the Cloud Nine scenario.
Under the Cliff Edge (10th percentile) and Comfy (50th percentile) scenarios, real income declines progressively by 50.25% and 9.5% respectively. This decline under the Cliff Edge scenario is higher than the typical spending decline of 35% from age 65 to 80.
By the 30th year, real income has declined by about 60.9%.
My view is that the cap and collar strategy works well under most scenarios, with low or even moderate inflation. But, if inflation constantly exceeds the cap, then real spending is decimated.
For the cap and collar strategy, by the 15th year of retirement, the real income has declined by 38.62% and 16.09% respectively under the Cliff Edge and Comfy scenarios. This is in line with the spending decline of 35% of an average retiree, from age 65 to 80. By the 30th year in retirement, spending has declined by a total of 44.11% and 29.54% under the Cliff Edge and Comfy scenarios, respectively.
Fig. 41: Initial sustainable withdrawal rate under four inflation-adjustement rules compared
Fig. 42: Real income under Bengen’s constant inflation-adjusted withdrawal vs. cap and collar inflation adjustment (1%pa fee deducted)
Advanced withdrawal strategies
The advanced withdrawal strategies make further adjustments to the withdrawal over and above inflation adjustments. In practice, there are 12 possible strategies. You can combine the advanced withdrawals with any of the four inflation-adjustment rules discussed in the earlier section.
Beyond The 4% Rule Page 6