Search “What is the UK safe withdrawal rate?” and the results are disappointing. Google’s AI response offers 4%, which as we’ll see is just plain wrong. Meanwhile it’s hard to get a straight answer from the humans amid all the financial content marketing.
Indeed there’s plenty of sketchy chat about the 4% rule. It’s a much-misunderstood figure – predicated on US numbers that aren’t reflective of the data from most other developed world countries, including the UK.
This matters, especially considering that some US financial experts believe the 4% rule may be too high even for Americans.
Investigating the UK safe withdrawal rate (SWR) provides useful counter-evidence, a sober corrective, and a uniquely British perspective on not running out of money in retirement.
Not safe! I need to mention that the so-called ‘safe withdrawal rate’ is a complete misnomer. Applying the SWR rules naively does not guarantee safely completing your retirement with money still in the bank. A SWR number is just a rule-of-thumb. It may not be enough to keep your portfolio off the retirement rocks. For that reason, some prefer the term ‘sustainable withdrawal rate’. It’s the same metric, minus the misleading advertising.
Okay, before we get to the UK’s SWR number, let’s recap what the safe withdrawal rate actually is and does.
What is the safe withdrawal rate?
Figuring out a safe withdrawal rate is useful for:
- Retirees who want to know how much they can withdraw from their pension pot each year, while minimising the chances of exhausting it over some given timeframe. (Say 30 years).
- Anyone wondering: “How much should I put in my pension?” A realistic SWR helps you calculate how big your retirement savings should be before you tell The Man where to stick it.
The SWR itself represents the maximum percentage of your portfolio you can withdraw as income in the first year of retirement.
For example, a 4% SWR suggests a £500,000 portfolio can sustainably support a £20,000 annual income.
After year one, you discard the SWR and simply multiply your established income by annual inflation to calculate the next year’s income. Hence you effectively live off the same real-terms income every year. So £20,000 in this example.
The crucial thing is your initial withdrawal rate should be set low enough (based on historical precedent) that you can draw a stable real-terms income for the rest of your life, barring catastrophe.
How is the safe withdrawal rate calculated?
The devil is in this detail! The safe withdrawal rate for the UK or any other country is derived from backtests of asset class real returns.
Torturing the data reveals:
- The highest withdrawal rate that could have been sustained…
- …for a particular retirement length, by a particular portfolio…
- …during the worst sequence of returns faced by retirees…
- …that’s captured by your chosen historical record
Got that?
For example, the famous 4% rule was originally formulated by financial planner William Bengen.
Bengen discovered a US retiree should choose 4% as the maximum safe withdrawal rate (known as MSWR or SAFEMAX).
The small print
Bengen’s 4% number was – and is – conditioned on:
- A 30-year retirement. Longer retirements equal lower SWRs.
- A 50/50 US equity/bond portfolio. Different asset allocations and assets produce different results. As do different datasets.
- US inflation. It’s been fairly benign in comparison to the UK experience.
- Not incorporating costs and taxes. We pay those in the real world.
- The portfolio surviving the worst-case returns in recorded financial history. Clairvoyants can withdraw more if they know they’re living through better times. They should withdraw less if they predict fortune will deal their plans an unprecedented blow
- Annual rebalancing. Change the rebalancing rules and you change the number.
- Withdrawing the same inflation-adjusted income every year for the length of the retirement. No more, no less.
From that cluster bomb of caveats we can deduce that:
- Change any of the conditions and you change the SWR.
- The SWR doesn’t account for unparalleled future scenarios. (Especially if your only sample is from the most successful stock market on Earth.)
- SWRs are just heuristics. They need to be modified to suit real-world circumstances.
- Costs and taxes reduce your SWR.
- You may be able to improve your SWR number by using different asset allocations from those usually incorporated into standard SWR tests.
- US historical returns are exceptionally good. They fail to capture the worst-case scenarios embedded in other country’s datasets.
- There’s no reason to suppose that the US will continue to enjoy such favourable conditions.
- Retirees in other countries are ill-advised in adopting the US SWR simply because they can invest in US assets.
I’m not trying to put you off using an SWR here. Far from it.
The safe withdrawal rate lies at the heart of my own retirement planning.
But the 4% rule dominates this conversation like a big orange cheeto, so I want to lay out why that number can’t be taken at face value.
As far as I’m concerned, the starting point for British residents should be the UK’s safe withdrawal rate, not America’s.
The rest of this article will hopefully show you why.
Safe withdrawal rate UK
Traditionally, SWR studies calibrate on 30-year retirements, sustained by annually rebalanced equity/bond portfolios.
So we’ll start there and aim to improve our withdrawal rate, by applying some reasonable tweaks, later in this series.
Here’s the chart of UK SWRs from 1870 up to the last 30-year retirement cohort, the class of 1995:
Data from JST Macrohistory1, FTSE Russell, A Millennium of Macroeconomic Data for the UK and ONS. March 2025.
Each datapoint shows the maximum SWR that a retiree could employ to enjoy 30-years of inflation-adjusted fixed income withdrawals, for a retirement that began in any year from 1870.2
For example, the graph shows us that someone retiring in 1870 could initially withdraw 7.5% from their 60/40 portfolio without emptying it before the 30-year retirement’s end on New Year’s Eve 1899.
The best year was 1975 for most portfolios.3 The class of ’75 could have larged it up with a 15.9% SWR on a 100% stock portfolio. (But note: that’s not as amazing as it sounds because UK equities crashed 73% between 1972-74. So by 1975 the retirees had 100% of a lot less portfolio than they had in 1972.)
The worst times for Brits to begin their golden years were 1910 and 1937 (depending on your asset allocation).
Safety first
1910 and 1937 are the SAFEMAX years. Their numbers give us the highest SWR that would have sustained the portfolio for the given length of retirement, across all historical scenarios.
The best SAFEMAX UK safe withdrawal rate is 3.1% for 30-year retirements.
In other words, our version of the 4% rule in the UK is the 3.1% rule.
Everything’s bigger in America!
And you only drum up the 3.1% SWR using a 100% equities portfolio, too.4 The more bonds you add, the worse things get.
The 3.1% rule
I really think the 3.1% rule could catch on, you know. But before I get carried away with the trademarking, let’s find out how much it squeezes your income relative to the 4% rule.
- £500,000 x 3.1% = £15,500 sustainable real income.
- A 4% SWR provides £20,000.
What size portfolio do you need to support £20,000 with a 3.1% SWR?
- £20,000 / 3.1% = £645,161
That’s a sickener. Your starting portfolio needs to be 29% larger with the 3.1% rule versus a 4% SWR.
But hang in there! It will get better, but first, it’s gotta get worse.
Here’s the full safe withdrawal rate UK table including longer retirement periods:
Safe withdrawal rate (%) UK equity/bond portfolios by retirement length
| Years / Equities | 30 | 35 | 40 | 45 | 50 |
| 40% | 2.6 | 2.3 | 2 | 1.8 | 1.7 |
| 50% | 2.8 | 2.5 | 2.2 | 2 | 1.9 |
| 60% | 2.9 | 2.6 | 2.4 | 2.2 | 2.1 |
| 80% | 3 | 2.8 | 2.6 | 2.4 | 2.3 |
| 100% | 3.1 | 2.9 | 2.7 | 2.5 | 2.4 |
These SWRs delivered a 100% success rate. SWR research typically includes 0% to 20% equity portfolios. But I haven’t because I suspect they matter to few Monevator readers. For similar reasons I’ve excluded shorter retirements. It won’t be hard for me to dial ’em up if anyone wants the info.
Doubtless many Monevator readers will be hoping to stretch out their mortal coil a little longer than 30 years. And the trade-off is clear: Mo years mo money.
Sorry my Gen Z friends.
You’ll also notice the SWR uptick gained from reducing bond exposure (i.e. holding more equities) is quite large.
The 100% equities SWR is fully 35% larger than the 40% equities number for a 40-year retirement.
That’s very different from the optimal US withdrawal rate, which includes a substantial bond allocation.
Wade Pfau calculated that any US equity allocation between 20% and 44% lies within 0.1% of the best SWR for a 40-year period – while I previously found that an 80% global equities allocation was best for 30-year or longer retirements with the Timeline dataset, with 70% being a hair’s breadth behind.
We’ll look at a broader range of asset allocations in more depth later in the series.
Charging ahead
Finally, we best not forget portfolio charges. The rule of thumb is:
- Calculate your costs as a percentage of your portfolio’s value
- Reduce your SWR by half that percentage
For example, the running costs of our No Cat Food retirement portfolio are around 0.3%. That includes ETF OCFs, platform charges, and dealing fees.
So using that as a guide, we’d have to knock 0.15% from our chosen SWR.
Why Brits shouldn’t use the US 4% safe withdrawal rate
You can buy an S&P 500 ETF and US Treasury bonds. So why can’t we Brits just declare for Team America and supersize our SWR?
Because it doesn’t work. Here’s the SAFEMAX table for portfolios formed on US equities (unhedged, GBP returns) and US Treasuries (GBP hedged).
Safe withdrawal rate (%) UK: US equity/US Treasury portfolios
| Years / Equities | 30 | 35 | 40 | 45 | 50 |
| 40% | 2.8 | 2.7 | 2.5 | 2.5 | 2.4 |
| 50% | 2.9 | 2.8 | 2.7 | 2.6 | 2.6 |
| 60% | 3 | 2.9 | 2.8 | 2.7 | 2.7 |
| 80% | 3.1 | 3 | 2.9 | 2.9 | 2.8 |
| 100% | 3.2 | 3.1 | 3 | 3 | 2.9 |
SAFEMAX year is 1969 except for 40/60 portfolios – 35-50yr retirements, SAFEMAX 1965. Additional US data from Aswath Damodaran. March 2025.
Well, these are better withdrawal rates. Sometimes impressively so for longer retirements.
But the thing that’s jumping out at me is the distinct lack of 4 per cents. Or anything like them.
In fact, for the baseline 30-year retirement portfolio, buying American only upgrades us from the 3.1% rule to 3.2%. Disappointing.
The problem is that the UK’s inflation record is considerably worse than the US’s.5 And a safe withdrawal rate is founded on real returns. So while it’s true the S&P 500 has left our equities trailing, British price pressures still knocked the shine off American exceptionalism.
This means you can bedeck your portfolio in the Stars and Stripes, but there’s a risk it’ll be hampered like a Dodge Viper stuck in single-lane traffic on a Devonshire country road.
The bond-gnawing brutality of UK inflation also explains why the best US-orientated GBP portfolio is still 100% equities in contrast to more balanced Stateside recommendations.
Counterintuitively, Blighty’s SWR actually goes up if you bench US Treasuries and bring UK gilts back on.
Yes, there are differences in the maturities counted in the datasets. But a significant factor is also likely to be that British bond investors demand a fatter yield to compensate for bigger UK price shocks. Whereas US bondholders can happily settle for a skinnier inflation premium less suited to British conditions.
Ultimately though, a US equity/bond portfolio suffers from the same fundamental problem that a Union-Jacked one does: inadequate inflation protection.
Albion’s way
This isn’t a problem that’s likely to go away. US CPI increased 13.3% during the recent bout of high inflation from 2021-24. Ours rose by a chunkier 20%.
We have a more open economy than the US. One that’s more vulnerable to importing inflation from abroad.
As we’ll see later in the series, the appropriate response to this is to hold a better mix of assets.
Portfolio ruin: the danger of overcooking your SWR
Am I being nerdy and pernickity?
Well it wouldn’t be the first time – but just so you know this next chart shows what happens if you apply the 4% rule to a 30-year retirement starting in 1910:
These retirees went broke inside 17 years. The portfolio loses money in real terms for ten of the first 11 years.
Meanwhile inflation goes through the roof: up 158% from 1915 to 1920.
Every year our retirees scale up their withdrawal to cover inflation, swigging ever larger rations from their dwindling reserves.
Healthy returns arrive during the remaining five years of the portfolio’s shortened life but by then it’s too late. The portfolio has already entered a death spiral as the retirees withdraw 180% more in 1921 than they did in 1910.
There are simply too few assets left for growth to cover outgoings. The money’s gone before Christmas 1926.
Granted, in reality very few people would watch their resources evaporate like this without taking action. Budget cuts would ensue or you’d go back to work, or both.
But the point is that taking too much income from the start risks living later years in bleak austerity.
How often does the 4% rule fail in the UK?
This next table shows you the percentage of retirement periods (%) that could not sustain a 4% SWR for different UK equity/bond portfolio allocations:
| Years / Equities | 30 | 35 | 40 | 45 | 50 |
| 40% | 36 | 45 | 53 | 67 | 75 |
| 50% | 29 | 37 | 47 | 54 | 63 |
| 60% | 24 | 31 | 41 | 45 | 54 |
| 80% | 17 | 21 | 26 | 30 | 36 |
| 100% | 12 | 17 | 22 | 24 | 27 |
Those are unacceptable failure rates in my view. And the sheer size of the numbers indicates that the UK’s problem isn’t limited to a few benighted retirements that were ravaged by two World Wars.
Superficially, the UK’s SAFEMAX years of 1910 and 1937 seem easy to dismiss. I can almost hear the devil on my shoulder saying: “Don’t worry about it. That’s ancient history. Nobody would be stupid enough to start a world war today.”
But I’m much less confident about that than I was.
I don’t know about you, but I’m feeling pretty twitchy about the prospect of British peacekeepers in Ukraine – operating without adequate American support. Not to mention the prospect of a rising China and the closing of Thucydides Trap.
But we don’t need to debate the probability of a war between World Powers. Just cast your mind back to the US asset-based portfolios we looked at earlier. There the SAFEMAX years were 1969 and 1965.
These pasty SWRs were induced by 1970s economic malaise, not by existential conflict.
UK OK
Alright, that’s enough doom and gloom.
This post has been about making public the kind of data that Americans take for granted.
And because we’re starved of information on the UK safe withdrawal rate, many Brits are told it’s fine to borrow the US one.
In truth though, the 4% rule does not travel well.
But while ours may not be as good as theirs, there’s plenty we can do to beef up a realistic, UK-centric SWR. We’ll explore how to do that in the next part of this series.
Take it steady,
The Accumulator
- Òscar Jordà, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, and Alan M. Taylor. 2019. The Rate of Return on Everything, 1870–2015. Quarterly Journal of Economics, 134(3), 1225-1298. [↩]
- We need a full 30 years of data to calculate the maximum safe withdrawal rate, hence no numbers yet for post-1995 retirements. The SWR number for each retirement cohort can only be known in retrospect. [↩]
- The 40/60 portfolio could only muster a relatively measly 10.9% SWR in 1982. [↩]
- For some reason, you’re allowed to assume you could have known the ideal retirement asset allocation in advance, which you couldn’t. This condition is called ‘Perfect foresight’. Not a gift of mine unfortunately. [↩]
- It’s debatable as to how much of the gap is an artefact of different headline inflation methodologies. For instance RPI vs CPI. [↩]
Thanks, if that’s the right word, for that rather chastening review of the SWR. As someone somewhat older than the average FIRE-fan would you mind dialling up the numbers for the 20 and 25 year retirements for the first, “Safe withdrawal rate (%) UK equity/bond portfolios by retirement length” table as you kindly offered?
Nice piece. Like the bond/equity time period table. I do however think that adding mortality percentages to it also adds perspective. For example, for the 100% equity portfolio and a 50 year old, a 25% chance of failure over 40 years may sound too high, but it’s likely greater than the probability of still being alive. We can all sweat over these percentages but the chances of still being on the right side of the daisies at certain ages helps to give some perspective.
A good book that covers a lot of this, along with variable strategies from a UK perspective is “Beyond the 4% Rule” by Abraham Okusanya.
Good article, thank you.
As a general comment, I notice this site is predominantly focused on retirement strategies/drawdown etc. Is there a reason for this, or simply based upon the site’s demographics? I say this as a (relatively) younger investor (mid 30s), and so struggle to see these as personally relevant (note, not a criticism at all – to the contrary, this site is incredibly useful) . Thanks again.
Hopefully some the global efficient frontier work at portfolio charts can suggest some alternative portfolios that may fair better, albeit only since 1970, when the charts above have all been excess 4% anyway.
https://portfoliocharts.com/charts/global-withdrawal-rates/
Also heavily reliant on the magic of gold, excluding it post 1970 is still only 3.7% (4.7% UK only).
Another great and thought provoking article. I believe that SWRs are often limited by poor returns in the early years of retirement where (thankfully) most will still be able to take steps to address (such as returning to work for a period as you point out). Also the SWR often seems odd to be applied as a fixed inflation plus figure for ever more as often income needs decline with age. Would be great to see some more information about the potential mitigations and alternate strategies in the upcoming articles!
@Ian — Glad you liked the article. I wrote about some general principles if you hit turbulence early into retirement back in March 2022, pretty presciently if I do brag so myself 😉
https://monevator.com/is-your-early-retirement-under-threat-from-an-unlucky-sequence-of-returns/
It’s not a data/maths heavy piece, more a general guide to evasive action if you hear the warning sirens.
Is there any importance as to the frequency and ratio from the portfolio the SWR x% is withdrawn? and does that change depending on what the market has done or is it a blindly remove x% each year from whatever flipping a coin decides? (stock/other).
I’m yet to find an idiot guide for me, saying how that x% withdrawal is made up.
Just my tuppence worth-walked the walk-so far!
23 years rtd -wife and I now aged 78
An initial 30/70 portfolio-at retirement-global equity index trackers-2 only plus a global bond index tracker ie 3 funds in total-these are cheap Vanguard funds on a cheap platform-Interactive Investor
Currently 35/59/6 where 6= 2 years living expenses in cash
Sustained an initial 3.3- 3.8% SWR for many years
Currently using 2.9%- now getting older and slower but IHT changes may cause a rise in my SWR
All very personal of course plus some helpful factors have occurred -index linked wife’s pension and the States pension have risen in value-There have been good stockmarket years lately etc etc
Portfolio much larger than at retirement – needs to be -inflation etc
xxd09
Will the next article in this series be more of a crowdpleaser?
Found the “Safe withdrawal rate (%) UK: US equity/US Treasury portfolios” table particularly interesting. Surprised (and disappointed!) investing in US assets doesn’t give more of an uplift in SWRs. I get the point about inflation being higher in UK than US, but isn’t there a compensating decline in the value of sterling vs. the dollar?
This is esoteric stuff, my takeaway is that the S in SWR should really stand for ‘Suggested’ rather than safe.
It is also a handy justification for putting in another ‘one more year’ before I reduce my hours.
I listened to a really interesting podcast about the 4% rule from its creator recently. I’m curious your take on it. He says the opposite of a lot of what you’re stating here and claims 4% is the LOWEST retirees should go. It’s meant to be the ultimate safe withdrawal rate. It’s Afford Anything episode #560. https://open.spotify.com/episode/2im86e3jL1nc68gltH8D0p?si=Qo_ZLKYkTVaVt9BI1MtsjQ
After reading this I started worrying about the 12% chance I might burn through my savings within 30 years but then looked at the Government ONS Life Expectancy Calculator which tells me I have a 90% chance of being worm food by then anyway. Think I’ll take my chances with 4% and eke out the last few years in front of the telly on the State Pension in the unlikely event it becomes necessary.
Round it up to 3.141592653589793… and make it the Pi rule. Much catchier!
@Azamino
You’re indeed right about it being “suggested” but it’s definitely not “esoteric”. As final salary schemes roll off these are exactly the issues the man in the Clapham omnibus will have to grapple with. It’s going to be disastrous for him and this is where the super/supra schemes will have to be created and come in and essentially manage it for a large percentage of retirees. The savvy readers of Monevator are a tiny minority.
Soul destroying,depressing article..
I’m planning on the die with zero approach.. after 80 I’m probably done with spending and travelling and most likely prefer a quieter life. Therefore my portfolio won’t need to keep pounding out the money the same for 30 years..
@Stroud – I sure will. Will probably be able to rustle up those shorter timelines over the weekend.
@Vic – I fully agree that it’s worth modifying your SWR by your life expectancy (and/or the joint-life expectancy of a couple, if there are two retirements on the line). There’s a practical guide on doing that here:
https://monevator.com/life-expectancy-for-couples/
@parguello – We used to write exclusively about accumulation. The deeper you go into the archives the more you’ll find on that. I’ve tried to rebalance that more recently because decumulation / retirement is a much trickier problem to solve and because there’s much less coverage of how to manage it. You’re not wrong either to suggest I’ve got a few more silver whiskers these days 🙂
On this article though, I found good SWR data incredibly useful when I was starting out because it helped me work out how much to save before I could FIRE. It was – and I think still is – hard to find credible alternatives to US-centric views. I’m glad you’re finding the site useful!
@Ian – Much of the ground is covered here: https://monevator.com/how-to-improve-your-sustainable-withdrawal-rate/
@Miner – The percentage of the portfolio taken is key and the rules you apply to withdrawals beyond year one are critical. The basic SWR rules advocate a very simple post-yr 1 withdrawal system.
The main pros are: Easy to understand and stable inflation-adjusted income (unless things go wrong.)
The main con is doesn’t respond to market conditions (so leaves a lot of money on the table when things go well, can run down your portfolio very fast if you don’t respond when the market is historically bad.)
There are alternative methods. We haven’t written an in-depth piece about them but we have reviewed a good book that does: https://monevator.com/review-living-off-your-money-by-michael-mcclung/
Re: frequency – do you mean does it matter whether you withdraw monthly versus annually? It would change the SWR but AFAIK nobody has proposed that as a meaningful variable.
Hopefully this answers your question although I’m not sure?
@CGT101 – Yes, I think so 🙂 It won’t get any worse than this unless I take a detour into the Japanese SWR.
I used unhedged US equities so they did benefit from the depreciation of the pound and still produced a poor result. I decided against running the same analysis for unhedged Treasuries because, ex-ante, that would be a bad strategy for everyday investors. Essentially investing your entire bond allocation in unhedged UST because of a backtest seems like a case of overfitting to me? You’ve got me thinking though, it would be a good idea to see what holding 10-20% in unhedged overseas bonds looks like. Certainly some Monevator readers hold at least some of their bonds as unhedged USTs.
@Azamino – “Suggested” makes much more sense.
@Niomi – I’d guess Bengen has a US audience in mind? I’ve listened to and read a lot of his stuff and I don’t remember him paying much attention to non-US data. He’s always been very bullish on his version of the number and doesn’t seem to worry much about the fact that the US may not perform so well in the future.
Wade Pfau wrote a great paper some time ago showing that only 4 developed countries out of 17 made the 4% rule work:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1699526
4% is an artefact of a particular historical path. For example, Pfau shows that Japan’s SWR was 0.5%.
Since then Bengen has introduced other modifications to beef up his version of the number. For example changing the glide path and investing in small cap stocks.
But much as I admire Bengen, I’ve not seen him pay heed to the international experience and he’s far more optimistic than other researchers in this field. One of the best I’ve read on the topic runs a website called Early Retirement Now. He does not agree with Bengen that the SWR for early retirees is anything like 4%. His site is well worth your time if you’re interested.
@Michael – one of the issues with the SWR is it focuses our attention exclusively on the worst-case scenario. The median SWR for a 60/40 portfolio (30yrs) is 5.3%.
@Dawn – I’m sorry! Remember I’m using a 4% SWR on the No Cat Food portfolio – by utilising techniques that go beyond the basic SWR rules. Accepting there’s a strong chance you’ll do/spend less later in life is also a very reasonable strategy. Not one that fills me with cheer but I’m seeing it play out in my parents’ generation.
Thanks for this, great article.
https://www.hl.co.uk/retirement/annuities/best-buy-rates
So a 55 year old can obtain a 4% RPI linked annuity.
That’s a lot better than the 2.4% your analysis has indicated for 45 years with 40% equities?
Acknowledging there are pros and cons to an annuity vs self drawdown, does this mean
– your analysis is too conservative given one’s ability to build a long term index linked bond ladder with positive real interest rates currently?
– or retirees should strongly consider at least a partial annuity to provide at least some kind of floor to their income as currently this looks more attractive? I do.
– annuity providers are underpricing annuities?…feels unlikely
I am surprised index linked annuity rates are so far out of alignment from your swr analysis given both are hedging uk inflation risk…in theory!
Interesting piece, thanks @TA!
I got to my rule of thumb for “SWR” as expected real return. Calculated from real yield on inflation linked gilts (ILGs) + equity risk premium (ERP). This has two offsetting limitations – a) ignores sequence of return risk and b) assumes you don’t want to be eat into your capital.
When real yields were low, this give similar answers to 50 years/50%.
Now that real yields on ILGs are positive that rule of thumb feels way too punchy.
AFAIK none of the SWR back-tests cope well with a portfolio of held to maturity ILGs (not to mention they’ve only been available since the 80s). The success rate will be hugely influenced by the real yield at entry and how actual inflation compares with what was implied (expectation priced into the market) at outset.
When I first started following them ILGs had a real yield of 4%. If you could earn 4% real guaranteed (as long as the UK government doesn’t default) on a ladder of held to maturity ILGs then taking 4% inflation adjusted out each seems pretty safe.
Having said that the 30 year experience for any of the equity portfolios from that time from the chart indicates more like 7%. Which funnily enough is not far off my real yield on ILGs + ERP…
A lot of the risk can be taken away by removing the inflation adjustment each year. Then if the market does well you get a boost and if not then you have to tighten your belt just like everybody else.
@Seeking Fire – The SWR framework doesn’t account for linkers as @Prospector says. To that extent, the two metrics are measuring different things. SWR views risk through the historical lens of a volatile portfolio. While an annuity is priced using the net present value of bonds with a known payout / mortality tables / whatever other special sauce the insurance provider brings to the table.
It’s also worth mentioning that there’s no pooled risk with an SWR. That is, you choose the SWR to hedge against the risk of being a member of one of the worst retirement cohorts in history. As an individual investor that risk falls entirely on your shoulders.
If we were part of a collective pension fund then the risk could be spread across multiple cohorts. For example, the SWR for the UK 60/40 (3o yrs) for the class of 1969 was 4%. But could be raised if the risk was spread among adjacent cohorts:
1967: 5.4%
1968: 4.8%
1969: 4%
1970: 4.7%
1971: 5.3%
Personally-speaking, I’m very tempted to take a 4% offer from an annuity provider, though I’m still too young! Just.
@Prospector – really interesting. I think I’ve seen a paper from someone simulating linker returns going further back. Oh, I’ve found it. It only goes back to 1987 which doesn’t improve matters for the UK market but does for many others. It may be useful to you to provide your numbers with an out-of-sample test? Here’s the link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2894869
@Ducknald – Absolutely. What kills the 191o retirees is those inflation adjusted withdrawals. They start off taking £40,000 from a £1 million portfolio but they’re taking £112,000 in the 11th year. Meanwhile the portfolio is shrinking. They could at least stave off the inevitable by forgoing some of that. Would mean loss of purchasing power and ultimately reduction in quality of life eventually.
I’m not sure that these calculations are very useful, they depend on a set of assumptions which are unrealistic.
When bad stuff happens you don’t plough on , you modify behaviour…. The future is unknowable but some bad stuff will very likely happen and you will react.
The audience for this article will panic and worry far more than the average, they will be just fine.
I would pick up the point about holding unhedged US equities and hedged US government bonds, I would be unhedged on both , that’s the diversifying effect of a different economy AND currency.
The future is unknowable, the idea of replaying a fairly small number of historical events is just going to be wrong, being aware of the past is enlightening but a flexible approach with backstops and multiple streams of income will provide a reasonable defence against the unknowns…
@SeekingFire. Annuity rates are not as far off the historical SWR analysis as headline numbers suggest. Remember that the SWR analysis is based off discounting using the RPI inflation index. While the linker real yields that drive annuity rates are also RPI-linked, after 2030, RPI will be redefined as CPIH. As a result, a 30-year linker can be considered a 5-year RPI linker and a 25-year CPIH linker starting in 5 years. So, the dominant driver of annuity rates is now CPIH, not RPI. With RPI typically around 1% higher than CPIH (very back of envelope), then you need to adjust for the change of inflation metric. If you could recalculate a historical SWR analysis with CPIH, you’d find the SWR would be higher.
This will bring the SWR and annuity somewhat back into line. Albeit, it’s true annuity rates are still somewhat above historical SWRs. This also explains some of the difference between UK and US SWRs. US CPI is less volatile than UK RPI, partially due to the country difference, but also due to the poor construction of RPI that makes it more volatile.
I would agree with you though that the current annuity rate for your age is an implicit floor to any SWR calculation. The SWR is a simplistic model that isn’t even coherent mathematically. The annuity rate is tradeable. As a result, I think the current SWR could be reframed as:
Current SWR = max(annuity rate for age, historical backtest SWR).
Pursuing Seeking Fire’s point: withdraw 25% of your SIPP tax-free and spend the other 75% on index-linked annuities. That way you get the equivalent of your 3% SWR as income – but guaranteed income, not cross-your-fingers income.
You can invest the 25% capital for growth as insurance against any bad effects of tying up so much capital in an annuity. That capital would presumably be invested principally in equities.
Blessed be the 25% tax-free lump sum. Note too that you’ve removed your SIPP money from exposure to whatever attempt will finally be made to exact death duties on pensions.
If your TFLS portfolio were by chance to yield the SWR being discussed, i.e. 3% real over 30 years, it would have grown to 60% of the SIPP capital you started with in inflation-linked terms. The balance of advantages here, in terms of worry avoided, might be attractive to many people.
Interesting analysis from a UK perspective, albeit annoyingly pessimistic. A more relevant article would have been looking at the impact of dynamic withdrawal strategies rather than a static inflation plus strategy. I suspect the SWR would look a lot more positive.
@Hariseldon — I’m with you (natural yield basically) but acknowledging (a) we’re heretics (b) much larger pots are required to begin with, given you’re not spending capital (c) probably a lower average total return from a natural yield portfolio, since income is definitely not free alpha and probably the opposite, but still I believe with a very strong possibility of a central acceptable outcome, as suggested by the long history of equity income trusts for instance and (d) lots of portfolio money left on the table at the end, unless you go crazy in your 80s, if that matters to you because you don’t care about inheritances etc.
I will cheekily say that the more this SWR discussion has developed over the years, the less I’ve felt somehow like an antediluvian simpleton for thinking natural yield based on reasonable assumptions about the characteristics of different asset classes might be an acceptable alternative to spending down a pot of cash to zero by some particular date based on historical backtests.
But I don’t want to derail TA’s article here. This is not weird stuff he’s writing, it’s the mainstream approach these days and such it’s important for everyone to understand.
And I again stress my (a) to (d) above, and perhaps even add (e) natural yield is probably best for lovers of the game of investing.
So on that note I’ll not comment further here, but perhaps do a natural yield portfolio suggestion for Moguls later this month. 🙂
Any idea why your results are so much worse than those Pfau gives in his international SWR paper for the UK?
You’re going all the way to 50 years in calculating the SWR, but skipping the PWR (Perpetual Withdrawal Rate, goal of amount at end >= start amount)? I feel like aiming for a 30 year PWR is the safe choice, even though it’s probably close to the natural equity yield…
@Hariseldon – I personally found the SWR a very useful metric to plan against. I guess others do too given the amount of further research and commentary it’s prompted.
I don’t think people are so easily panicked. One of the best things about SWR discussions is they help surface the levers we have to pull in retirement and the factors worth thinking about when decumulating. I do agree that it’s only a model and therefore isn’t an accurate description of how retirees actually spend down their portfolios. As ever, the real world is more complex than any model.
Re: Unhedged US bonds – I guess you have something else in the mix too. Some income stream or GBP-priced asset or cash reserve that doesn’t expose you to the vagaries of the currency market? Or you will when the time comes? It’s not mainstream advice to only hold unhedged foreign assets and for good reason. Hence I haven’t modelled a two-asset portfolio with two unhedged holdings. I can allocate a chunk to unhedged bonds when I model more diversified portfolios later in the series.
@Al Cam – Pfau uses the DMS dataset which is paywalled. Also, his paper adds bills into the mix and introduces a perfect foresight assumption:
“For each country and in each retirement year, the paper optimizes across the three domestic financial assets, finding the fixed asset allocation that provides the highest sustainable withdrawal rate over the next 30 years.
Perfect Foresight Assumption: Much of the analysis provides a best-case scenario for increasing the SAFEMAX by assuming in each year for each country that the new retiree has perfect foresight to choose the fixed asset allocation that maximizes the withdrawal rate for the subsequent 30 years. Obviously the assumption is not realistic and artificially inflates the
SAFEMAX, but even so, the traditional 4 percent withdrawal rule will still perform surprisingly poorly. This assumption avoids accusations that a poor-performing asset allocation was chosen to discredit the 4 percent rule.”
@Pikolo – Fair enough, I agree PWR is safer still but the trade-off is requiring a bigger portfolio. I feel like the SWR is safe enough, especially once you add in some of the additional measures discussed in the thread. Still you’re talking The Investor’s language with natural yield.
I’m going to bang the same drum as I did the last time this SWR issue came up – although I’m sure I’ll be shot down again. There are plenty of “dividend heroes” that will pay over 4%. One can just spend the income with literally ZERO risk of running out of cash. I’ve heard every reason under the sun as to why one shouldn’t do this – but none of them make this a worse option than taking a low 3% SWR because you might run out of money. Hell, even the FTSE 100 natural yield is more than that. If one thinks THAT is more risky than whatever SWR one chooses, then one needs a gun and some beans. And a tin opener!
The annuity comparison as discussed above fails to mention the upside of the SWR. The table at presented only provides the failure rate, the “value at risk” on the upside is also substantial and it’s obviously traded away when buying an annuity. The different pay off profiles should be compared in a relative value analysis.
Thanks TA, I’m delighted you’ve taken the time to go into this depth to consider SWRs for UK investors. Get where the comments are coming from re: scenarios/complexity/pessimism etc but everyone’s preferences and circumstances will vary so wildly. I’ll use the insights as more a helpful list of considerations to calculate my own SWR based on my own circumstances/plans etc. Look forward to the rest of the series. Keep up the great work
@Matt (#29)
I know a lot of people who subscribe to the spend the dividend leave the capital philosophy. My Gran may have been one of the one of the first people to share any retirement saving tips explaining the virtues of blue chips paying stable dividends in her modest savings portfolio.
I’d say there are some risks:
There may be a bit of money illusion with this strategy, as I’m not sure that dividends have kept up with inflation. The Barclays Equity Gilt study includes a dividend index from which it’s possible to see that UK dividends haven’t kept pace with inflation for large periods of time (most recently they still haven’t recovered from 2012 levels thanks to the 2021/2022 bought of inflation)
And there is the risk a company goes under (think Carillion)
Another great article and excellent comments as always. The key reason to stick with Monevator! I second the comment that GBP depreciation may have slightly helped and not sure it appears in the outcome (or if using US does it convert spend at spot GBP each year ?)
Lastly, in my modelling, taxes become quite significant. I may (not) be in Finimus/FirevLondon levels, however tax rate in retirement is likely to be quite significant. Scaling the pretax number to get a post-tax does make things a lot more challenging. For those on the higher end of spend, this is probably going to be the biggest impact vs the naive 4% (4% pre-tax is 2.8% posttax)
(I also model GBP vs EUR vs USD vs AUD draws from portfolio similar to FirevLondon and hope that the unhedged portfolios help in mitigating somewhat localised inflation and expropriation issues)
@TA (#28):
OOI, if I have read Pfau’s Figure 2 correctly* his perfect foresight assumption for the UK requires somewhere between 80% and 90% equities. Your calcs show 100% equities gives the peak. Pfau’s paper (Table 1) shows that UK bonds and equities are the most correlated. Could such a small percent of UK bills alone explain the near 20% difference between the UK SWRs: 3.1% (you) and 3.8% (Pfau)? Maybe, maybe not – it is hard to tell without the actual data.
Also, Pfau’s worst case year (1900) performs a lot less badly than yours (1910). His lasts 26 years (and 17 years at 5%) – see his Table 3 – and yours just 17 years at 4%. That seems to be a pretty significant difference.
I think there may be other factors at play than just the composition of the dataset and perfect foresight. Most likely the values of the dataset. Are you able to calculate the stats Pfau gives in his Tables 1 and 2 for your dataset? I ask as these stats might just be helpful.
The [so called] SWR is an attribute of the dataset and nothing else. Change the dataset and you will most likely get a different SWR – even though they claim to represent the same historical period.
* I happen to find it a bit tricky
P.S. totally agree that frictions (inc. taxes) make all SWR results worse too. As you highlight at the outset of the post, SWR is not safe, nor IMO is it particularly realistic/helpful. FWIW, I remain of the view that a floor and upside approach (which is also not without flaws) is better – as several other folks have already mentioned above.
@Vic – great point about the upside. I think the SWR should just be a starting point. What are the reasons it’s so low? What are the steps you can reasonably take to improve it? You’ve already mentioned one… adding in life expectancy. Dynamic withdrawal rates have been mentioned in the thread. Partial annuitisation, the State Pension, the tendency towards lower spending in very old age. By the time, these elements are layered on, the SWR looks much healthier.
I understand the pessimism about the bald SWR but things look much brighter once you add in all the things you can do to improve it.
@AndyfromAus – GBP depreciation is included in the unhedged US equities returns but doesn’t figure in the hedged US bond numbers.
@Al Cam – Yes, I agree there must be differences in the values in the DMS dataset. Most probably on the bond side but I can get those figures for you so we’ll see shortly. Quite possibly there are differences in the inflation dataset too as Pfau’s paper was published before the BOE released the Millennium of Macroeconomic data. My read of figure 2 is that there’s little difference in the GBR SafeMax from 70%-90% equities.
FWIW, Pfau published an update 2 years later that presented the UK SAFEMAX as 3.36% (50% stocks / 50% bills):
https://retirementresearcher.com/the-shocking-international-experience-of-the-4-rule/
3.05% here (50% stocks / 50% bonds)
http://advisorperspectives.com/newsletters14/pdfs/Does_International_Diversification_Improve_Safe_Withdrawal_Rates.pdf
There’s another here: 3.6% SAFEMAX for 100% equities (This is Okusanya as mentioned by Vic)
https://finalytiq.co.uk/lower-equity-allocation-retirement-reducing-risk-shooting-foot/
Okusanya doesn’t mention his source in the article but his Timeline app, around that time, used 10-yr UK gov bond returns from Global Financial Database.
UK equity sources are the same but UK gov bond figures are much more variable in my experience. Macrohistory database UK bonds are long in the safemax years.
The numbers do fluctuate by database but the key is the concrete steps you can take to improve your chances of spending down your portfolio. Like you, I’m a big fan of the floor and upside strategy.
Excuse the market timing point, but I guess now might not be the time to be considering unhedged US Treasuries exposure given US policymakers’ desire to weaken the dollar. (Or US Treasuries exposure generally given plans being aired for the US government to effectively default!)
A meta (and speculative) point that might be better suited to a different post…. Given the fiendish complexities involved in decumulation, it doesn’t feel all that sustainable for individuals to be expected to manage these risks. Maybe we’re in an interregnum where the transition of middle income private sector pensioners from DB to DC pensions hasn’t quite yet reached a critical mass and so this hasn’t quite become a political issue yet. Perhaps abandonment of the triple lock will be a trigger for significant change. It feels like we’re in the middle of an “if something cannot go on forever, it will stop” scenario. Are there plausible changes which could undercut the major underlying assumptions for this kind of SWR analysis (e.g. removal of pensions freedoms)? There have been some interesting allusions in the comments to risk pooling solutions, which should surely be on government’s agenda if they’re not.
It is interesting that Reddit (UK) personal finance sites, when the question of SWR comes up, usually settle on something between 6-8%..! Based on nothing more than ‘that is what the Stock Market has produced over the last 20-30 years’ or so. People that suggest 3-5% are generally dismissed.
Nice piece. I must admit I’ve found SWR helpful as a tool to try to address the “how much pot do I need?” question and when to trigger the RabitE piece.
I’d always resisted the 4% as too US centric though and conservatively viewed 3% as my number. I think this broadly validates that, though I expect the reality will be some self-imposed frugality if SRR bites hard in the early years.
The main problem I have with the dire outcomes modelled is that, in reality, people would not continue spending blindly as their portfolios depleted excessively – they might buy into the safe haven of annuities or adopt much more constrained lifestyles or even venture back into some employment if possible.
SWR is only ever a broad tool, or perhaps a “permission” to spend for those of us inclined to the conservative. Not something that I expect anyone lives to the letter on.
@TA (article and comment #35) and @Al Cam, #34
As well as the Pfau paper (which is, IMO, significantly marred by his use of the misleading ‘perfect foresight’ approach – AFAIK, this is the only paper of his that uses it but I think I understand why he did), the paper by Estrada (https://blog.iese.edu/jestrada/files/2018/03/MaxWR.pdf) provides a much more comprehensive survey of safe withdrawal rates across a range of countries. An examination of Exhibit A2 suggests that the MSWR (i.e., P1 value) ranges from 3.6% (100% equities), 3.5% (80%), 3.2% (60%), 2.9% (40%) for a 30-year retirement. Again, these are higher than those calculated in TAs post.
There are several reasons (some of which I can quantify).
1) Bond duration. The duration of bonds in the macrohistory dataset is very long (undated, i.e., ‘perpetuals’ before 1962, 20 year maturity until 1990, 15 year maturity until 2016 and then ‘all stocks’). I do not know what the bond duration of the DMS dataset used by Estrada is. However, I found (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4742456) that bond duration made a difference of about around 40 bp (depending on the equity allocation – obviously zero for 100% equities) on the MSWR and up to a few percentage points on the SWR for individual years (see Figure 1 and Figure 2 in the paper). FWIW, the ‘best’ choice was an intermediate bond fund of 5 to 10 years, under 10 years, or 5 to 15 years (which is a relatively well known outcome from US research). I also found that intermediate duration bond funds came out best for percentage of portfolio withdrawal strategies too (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4827947).
2) Construction of the equity index. The macrohistory returns are based on those in the Barclays equity gilt study from 1908 to 1963 (a new 30 company index to 1935 and the FT30 from 1935 to 1963) and the all-share index from 1964. The way the index is constructed (i.e., which stocks are included) and how dividends are reinvested (e.g., at settlement or at end of year) makes a significant difference to the returns. There is insufficient detail in the Barclay’s study to know precisely how this has been done, but I know that it differs from that in the DMS database (A Dimson et al. paper at https://www.annualreviews.org/content/journals/10.1146/annurev-financial-082123-105515 discusses how to construct indices).
3) Inflation. Leaving aside the difference between RPI and CPI/CPIH, prior to 1948, there are at least four inflation indices for the UK (ONS RPI, Measuring worth, Barclays equity gilt study, and Millennium of macroeconomic data (which is the one used in the macrohistory database). Application of the different databases leads to a difference in MSWR of about 20 bp.
4) Time resolution. I have only compared annual and monthly data for the US (since I don’t have monthly equity data for the UK), but the difference in MSWR amounts to about 20bp for equity allocations of 60% and 80% and about 75 bp for 100% (the latter entirely driven by 1929).
In other words, the overall uncertainty in determining the historical MSWR probably amounts to 50 bp and therefore quoting the historical 30-year UK MSWR as falling somewhere in the region of 3.0 to 3.5% is probably good enough.
Future SWRs are, of course, unknown and unknowable.
@CGT – Agree wholeheartedly on the need to look at better risk pooled options. As I understand it, other countries took this route (e.g. the Dutch) while we followed the US model.
@Kraggash – That is astounding and in many ways reflects the anecdotal state of UK financial advice on withdrawal before Bengen’s work became more widely known.
@BBBobbins – Very nicely put! SWR is a blunt tool in its most basic form.
@Alan – Thank you very much for that overview which is extremely helpful. BTW, I read three of your papers on gilts while researching this topic and thought they were fantastic. They really helped me understand the historical depth and composition of the UK gilt market and how that alters the numbers thrown up by the archives. I’ve also downloaded your figures for different gilt indices and would very much like to use them as part of this series if you have no objections.
@Al Cam – I have the numbers you suggested calculating and I think they do help show why UK SWRs calculated from the Macrohistory Database are at the low end relative to the DMS database used by Pfau / Estrada. I’ve calculated the figures for 1900-2008 to align with Pfau’s timeframe.
UK equities:
Geometric mean
DMS/Pfau: 5.09
Macrohistory/TA: 4.75
Standard deviation
DMS/Pfau: 20.06
Macrohistory/TA: 20.09
UK bonds:
Geometric mean
DMS/Pfau: 1.39
Macrohistory/TA: 1.o5
Standard deviation
DMS/Pfau: 13.75
Macrohistory/TA: 12.44
Inflation:
Geometric mean
DMS/Pfau: 3.95
Millennium of Macroeconomic Data/TA: 4.o7 (Slight variation on Macrohistory pre-1915. Macrohistory = 4.03)
Standard deviation
DMS/Pfau: 6.64
Millennium of Macroeconomic Data/TA: 5.98
I haven’t calculated bills as they didn’t feature in my post.
So Macrohistory’s UK bond returns are quite a bit worse, while equities and inflation are a little worse relative to the DMS database.
If you’re conservatively minded, use Macrohistory. If you’re feeling lucky, use DMS. Or split the difference and add half of Alan’s 0.5% variance estimate to the Monevator numbers. (Or perhaps use it to cancel out the obligation to add portfolio costs.)
@TA, @Alan S,
All good stuff. Thanks very much for providing your various additional inputs.
To my eyes, we are in danger of obsessing about a fundamentally flawed number*. However, if one takes the view that the SWR construct is in some way useful the focus IMO should not be on the SWR but rather the extent of failures (and not just avoiding [some of] them). That TA’s worst case failure at a 4% w/d is a whole 9 years worse than Pfau’s is truly worrying. That is, coming up 4 years short rather than 13 years short [in a 30 year scenario] is a whole different type of failure in my opinion.
*quoting it to more than one decimal place is false precision and declining to add any error bars might confer it some form of total undeserving legitimacy
@CGT:
Are you familiar with Henry Tappers blog?
Thanks for a great column, full of interesting insight and lots of numbers to play with and think about.
At the risk of being black-balled from the Monevator website, I’ve thought about it and am currently in the process of annuitising my SIPP. I know, I know. This is probably anathema to the majority of readers, but I do think that annuitisation can play a role, especially in the drawdown phase of life.
Annuity rates (15 year gilt yields in the UK are a good proxy) are now back at levels last seen in 2008 – 17 years ago.
Just as a single, random data point, I’m currently 58 and in reasonably good health (so no ill-health enhancements). I’ve found a provider that will currently give an annuity rate of 4.5% (single life, no guarantee, RPI escalation).
OK, annuities are pretty (very?) boring, and there is no upside potential, but 4.5% is a pretty good return for a risk-free (hopefully!! depending on the financial soundness of the financial institution), inflation-proofed income, and well worth considering as an option (or part-option).
@NickH – I think an RPI-linked annuity is a great choice. It’s a fundamental part of a floor and upside strategy as mentioned by @Al Cam. Do you intend to maintain a smaller portfolio alongside?
@Matt – Have you seen these posts?
https://earlyretirementnow.com/2019/02/13/yield-illusion-swr-series-part-29/
https://earlyretirementnow.com/2019/03/04/the-yield-illusion-follow-up-swr-series-part-30/
https://earlyretirementnow.com/2020/10/14/dividends-only-swr-series-part-40/
The point is, there is no 4%. There is no single figure. I’ve spent a career modelling financial assets and people always want certainty and a single number to hang their hat on and it doesn’t exist.
The future is unknown. Bond and equity returns are unknown and your longevity is unknown. We just have a range of probabilities to work with. As has been pointed out, that’s ok as you should be able to adjust along the way.
The data set that we’re working with, whilst being the complete list of asset returns since modern finance was invented is still incredibly small. If we take 30 year time periods over the last 120 years, that’s still only 91 observations. That incredibly small sample size that makes the standard deviation of the failure probabilities huge, so renders them, in statistical terms, practically meaningless. However, they do still represent the entire lived financial experience, so we can’t dismiss them.
4% isn’t an answer, it’s a guide. 30 40 50 years are an age in financial markets. Like an airplane traversing an ocean, there’s a lot of variables that can knock you off course, and like an aeroplane you need to be monitoring trajectory and adjusting course regularly.
If people are uncomfortable with uncertainty, then ILG real yields or annuities look pretty good at the moment. It’s very much a question of attitude to risk. Many solve the problem by just being extremely wealthy and others “solve” it by being completely unaware. I think most of us here fall in-between those camps, hence the interest in it.
Hello @Accumulator.
Yes, the index-linked annuity will be in addition to a small portfolio.
I regard the annuity as an income stream into my portfolio, and as a floor to my income, should the worst happen and my portfolio run out (of course, the absolute worst would be if the annuity provider disappeared as well, but let’s not think about that particular scenario if we want to sleep at night).
I did quite a lot of spreadsheet modelling, and an index-linked annuity provides protection against inflation shocks as well as market shocks. It also removes the sequence-of-return risk, which can be quite critical at the start of the decumulation phase.
As an interesting aside, annuity rates have moved more than 5% in the last two two weeks, which reflect underlying gilt yield movements. I’m not an expert in gilt yields, but I was surprised at the size of the movement (absent a Truss-like moment).
It strikes me that the key things are 1) the Trump Govt, and 2) the German fiscal expansion. Clearly the markets are pricing in future uncertainty and higher longer term interest/inflation rates.
An index-linked annuity would also offset the risk of any US/UK inflation mismatch. as alluded to in the original SWR article.
I agree with pretty much everything in Vic Mackey’s last post (having also spent a significant proportion of my working life modelling financial assets).
I would also add in the opportunity cost of time spent thinking about these things, monitoring and adjusting portfolios and keeping abreast of all the issues (as well as worrying about them).
At the end of the day, some of us want security and financial freedom in order to be able to pursue other activities (ie not finance related).
In my case, I wanted to free up headspace and move on to do other things. An annuity does that for me. This is not the case for everyone.
Listening to J Pow’s press conference whilst typing this it occurs to me a deccumulator should think about an SWR calculation a bit like an investor should think about the Fed’s dot plot. It’s a sort of forecast the forecaster is at pains to insist you really shouldn’t rely on. A worthwhile exercise to impose a disciplined framework around a constantly evolving decision making process. The output the current best estimate of what an appropriate future rate might be based on all currently available information, all of which is certain to evolve unexpectedly.
@NickH – You twice mention a fear of the annuity provider failing, but isn’t there 100% FSCS protection for annuities (subject to a couple of conditions)?
That should probably be chalked up as a significant benefit over the £85k max per institution protection for a drawdown approach.
@Invariant – yes, there is 100% FSCS protection for annuities.
That’s why the scenario where it fails is the one where everything fails (and hence the nightmare inducing scenario). Sorry, the original comment was probably a
bit flippant.
And yes, the additional security of a state backed guarantee is definitely worthwhile (and a factor in my decision to annuities).
Interestingly, from discussions I’ve had with family/friends, not that many people seem to be aware of the FSCS protection.
Are there any income tax drawbacks/benefits in using an annuity as opposed to drawdown?
Less flexibility ie set income stream in place – possibly incurring more tax liabilities with annuities?
@xxd09. There used to be a slight tax advantage to what were called something like ‘non-compulsory’ annuities as part of the return was deemed return of capital however this link might help
https://www.mandg.com/wealth/adviser-services/tech-matters/investments-and-taxation/purchased-life-annuities/purchased-life-annuities
One issue with annuities I think is that you need to think about the possible need for capital eg helping out children, going private for that knee replacement, rethatching the house ( that one is bit specific)….
Good article and comments – thanks.
Now 4 years into retirement I never thought I would consider annuities but with rates being good in recent times I have just annuitised about 25% of my pension estate. This takes some investment risk and worries off the table and provides an income floor (as @Al Cam and others have previously talked about). One worry hopefully removed by annuitisation is government meddling with pensions?
The rest of my pension remains in a SIPP of 80/20 equities/bonds currently in phased drawdown just taking dividend cash amounting to about 2.5%. The SIPP provides much flexibility for drawdown, to take more tax free cash if required for any life events and hopefully some upside if markets are kind to us. Also future phased annuitisation of the SIPP is possible, something I hadn’t considered before.
Another potential option in the annuity arsenal is an instrument known as a temporary annuity, whereby a pension pot buys an annuity for a set period, say 5 years. At the end of this period, the annuity ends and a lump sum is returned.
Obviously, the key risk here is reinvestment risk (what to do with the lump sum at the end of the period). Plus the return of all the other risks that direct investment entails.
These are difficult to model due to the increased uncertainty, but provide yet another tool in our armoury.
I think that the key point is as Vic laid out in an earlier comment. There is no “one number” (or alternatively, correct solution), except in hindsight. We just have to muddle through as best we can.
I do wish everyone the best in your own personal efforts.
@NickH(#53):
Did you manage to get any quotes for a temporary annuity?
@Vic – “Many solve the problem by just being extremely wealthy and others “solve” it by being completely unaware. I think most of us here fall in-between those camps, hence the interest in it.”
Love this!
@NickH – It’s interesting to hear how you’ve handled it. I haven’t heard of temporary annuities before. Are they different to a shorter term annuity you might buy for a phased annuitisation as mentioned by @BillD?
@TA:
Temporary annuity aka fixed term annuity (FTA), AFAICT. To date, both Alan S and I have been unable to confirm the existence of FTA’s in the UK – hence my Q (#54) to NickH.
If/when you do a post on [UK] Annuities it might interest you to know that in the US there seems to be two distinct classes of annuities: simple/fixed annuities (like SPIA’s, etc), and complex/variable annuities that bundle a simple/fixed annuity with some form of “investment product”. AIUI, the UK only does the former class of annuity.
https://www.lv.com/retirement-planning/help/fixed-term-annuity-calculator?gclsrc=aw.ds&gad_source=1&gbraid=0AAAAAD1ULTo9OcfpBoFIfKvYSENcigVFv&gclid=Cj0KCQjw-e6-BhDmARIsAOxxlxUeCsVcUn4mV4vRS5sS0ZCvB4hbiPBsTvONpPOjzunmL3_iJqMPabsaAoaEEALw_wcB
It’s there and the initial questions worked, it’s a weird one.
@Vic #44
I think you summarise this very well, unknowable, like the weather, a subject of much discussion and study, which is not particularly useful but we develop a feel for these things.
A thunderstorm often follows an extended period of very hot weather….
Correction:
FTA’s can be bought in the UK, see e.g. chatter between snowcat and lategenxer at: https://monevator.com/should-you-build-an-index-linked-gilt-ladder/
Alan S & I could find no evidence that Deferred Annuities (DA) can actually be purchased in the UK – even though they are described here and there.
My bad.
Hi @AlCam.
Yes, I’ve been quoted for a temporary annuity. Only one firm I’ve been in contact with seems to do this.
I’m not trying to promote them, but they are Retirement Line. The product didn’t suit my needs so I didn’t take them up.
As an example:
Single life, fixed term, 5 years, RPI linked, monthly in arrears, £23,750pa. Then after 5 years, a guaranteed maturity amount of £449,400.
It was a bit or a surprise to me too. It seems that the underlying insurance company that provides this is Standard Life, so seems legit enough.
Other terms and amounts are available on request.
I think that these instruments are an interesting addition to the toolbox.
Hope that this helps.
Ps purchase price for above was £475k presumably this could differ depending on gilt yields and the other various factors.
@TA (#40)
Thanks for the kind words about the papers. Yes, use the data by all means. Given the expense of some datasets, I have a lot of respect for the macrohistory team making their results freely available and wanted to put something back (also, generating the gilt returns took about 6 months effort – I want it to get used!).
The differences in the mean asset returns in the different datasets are interesting.
I should also add that anyone who wants to play with UK data, the retirement calculator at https://www.2020financial.co.uk/pension-drawdown-calculator/ uses historic UK data (returns for stocks and bonds and inflation are from are the Barclay’s data set, i.e. the returns are similar to macrohistory, with actual indices substituted when available, for cash they use the BoE base rate which gives a substantial boost to the outcomes compared to using 3-month bills or savings accounts).
@Al Cam (#41)
One of the many problems with the SWR strategy is that the ‘time to first failure’ is highly sensitive to small changes in the withdrawal rate adopted.
Completely agree about false precision.
@Vic Mackey
My career was spent modelling physical systems (physicist/engineer) and I only started on financial data about a decade ago (in the run up to finalising retirement). I was surprised that no-one had thought much about errors (beyond a hand-waving sort of way) and have regularly seen (and occasionally been guilty of) outcomes quoted to 3 (or more) sig. figs.
Agree completely that flexibility is key. Pure SWR is not a robust approach because there is no feedback from the system (i.e., portfolio value). However, fully variable methods, e.g., a simple percentage of portfolio or natural yield (the only modelling of the latter I have seen is at https://finalytiq.co.uk/natural-yield-totally-bonkers-retirement-income-strategy/) may provide too much income volatility for those without a strong guaranteed income floor.
In one (slightly contrived sense), there is a touch of the uncertainty principle to withdrawal strategies in that one can have either:
Certainty in income but no idea how long it will last (SWR)
or
No certainty in income, but certainty as to how long it will last (pure percentage of portfolio).
Hybrid withdrawal strategies (e.g., Carlson’s endowment formula, mix of SWR/percentage of portfolio, portfolio smoothing, vanguard dynamic, Guyton Klinger, etc.) reduce the certainty in income and increase the certainty in the time over which it is available compared to the SWR strategy (or, depending on your POV, increase the certainty in income and reduce the certainty in the time over which it will last compared to percentage of portfolio strategies).
@xxd09 (#50). As far as I am aware, annuity income is taxed as general income, so it is usually not worth using the TFLS in a pension to purchase one (although a purchased life annuity, i.e., bought with funds outside of a pension, are only taxed on the interest – although I gather the payout rates are poorer – not much of a market).
Apologies for yet another very long post!
I found this analysis article by Standard Life on combining drawdown with phased annuitisation which may be of interest
https://www.standardlife.co.uk/about/press-releases/annuitising-in-phases
Looking at their cumulative income numbers, the gains of the approach seem marginal to me unless I’ve missed something. However it does provide more flexibility if you don’t want to hand the whole of the pension capital over to an annuity company at the start. Annuitisation is a decision that can’t be reversed, after 30 days anyway. Making an irreversible decision with your whole war chest at the start of the game doesn’t feel right?
Thanks, fascinating article.
Not least I confess because I’ve reached the point where I need to figure this out for real.
What strikes me is
– that whatever complex calculations you use, for an individual case there’s a lot of uncertainty.
– annuity rates should be a rough guide to a sustainable withdrawal rate
– an inflation proofed state pension, or two in our case, should probably be incorporated into the overall calculation
– what is known as a variable withdrawal rate feels much more sensible to me, in my simple mind that’s a fixed percentage of total market value of my SIPP which yes, does result in a variable income but can never run out.
@PC #63
“– what is known as a variable withdrawal rate feels much more sensible to me, in my simple mind that’s a fixed percentage of total market value of my SIPP which yes, does result in a variable income but can never run out.”
I like this idea but perhaps a hybrid version with some mechanism of providing a spending floor for essentials (food, utilities, council tax, small home maintenance etc). That could be a linker ladder, annuity or state pension.
I’d imagine most of the big spending events take place within the first 10 years of drawdown. One could easily establish pots for these prior to stepping away from paid work. If you did that and took steps to reduce monthly expenditure, paid house off, updated cars etc there’s realistically a decade or two’s discretionary spending to mainly be concerned with. Depending on your age at retirement.
@PC #63
“annuity rates should be a rough guide to a sustainable withdrawal rate”.
I’ve seen this posted on a number of occasions by posters on this site over the years and I feel it’s somewhat misunderstood by those writing it. It holds true only if you are invested in a “hedging asset” which would be a combination of inflation linked bonds and fixed income type assets. ie the type of assets an annuity provider would be invested in.
One can’t sit largely in equities and point to annuity rates as a suggested withdrawal rate as much as we would like to unfortunately.
@NickH (#60):
Thanks very much for your reply and the additional info.
Unless I am mistaken the temporary annuity (TA) you describe [at #60] is not the same as the FTA’s described in the link I gave at #59 above. In essence, with the FTA’s you pay [in advance] for the years of cover you require – nothing more and nothing less. With your TA you hand over much more than is required and trust them to return the quoted surplus at a pre-determined date in the future. That is quite a different set up. Interesting, and thanks again.
@Alan S (#61):
Re ‘time to first failure’: agree.
Assuming you can access credit (or some other form of [family, say] dosh to tide you over) it is possible for a failed scenario to subsequently recover and then go on to succeed or, fail again, provided you can keep some money in play!
FWIW, IMO physical scientists often struggle with economists ‘approach’ to maths!
Vic Mackey #63
Quite right to pick me up on my sweeping unqualified comment about annuity rates. What was in my head was exactly what you say it only applies if you are investing in the way an annuity or DB pension provider would. Last year I worked at Pension Insurance Corporation and saw more than I ever wanted to know about evaluating and balancing investments v liabilities .
Thanks @TA for the super piece and thanks @All for the many excellent comments.
Special thanks to @Alan S here for his really insightful expert input in the comments.
@Vic Mackey #44 first para. sums it up nicely
I’d also highlight his observation than the 91 unique 30 year periods over last 120 years is a very small sample size statistically.
This could also apply, of course, to 30 year real stock market returns over 120 years.
Something to always bear in mind when you see purportedly ‘long run’ data.
The temporary (5 year) annuity ref’d at #60 above might be ideal for me.
I’m just over 10 years off from legacy FSS DB normal scheme pension age at 60, but 15 years off from current CARES DB normal pension age at 65.
This will be my floor.
But I will take a hit (~20%) on the CARES DB for taking it at 60, rather than at 65, if I take both DB pensions at the same time.
A 5 year temporary annuity funded from SIPP could bridge the gap, allowing me to take the CARES DB at 65 without reduction.
What though are the tax implications, if any, of the guaranteed annuity maturity amount? Is it paid out within the SIPP?
One wouldn’t want to turn a tex sheltered capital sum in a SIPP into a taxable receipt outside the SIPP.
This is a great thread.
I’m minded that the SWR is a classic example of the “map is not the territory” metaphor. It’s a crude model in many ways but in identifying its shortcomings we learn a good deal about how to “solve” the withdrawal problem.
@Alan – Great comment. And thank you for the nod re: using your bond data. I can’t believe it took you 6 months! I mean I can – it was obviously a huge and complex task. But of course, it’s so easy to underestimate how difficult a piece of work can be. Naturally I imagined you’d knocked it out in a morning of mathematical wizardry 😉
Cheers for the link to the calculator. I didn’t know it existed!
@DH – I’m rocking 150+ years worth of data here! Massive breakthrough in sample size technology 😉
One of the features of the SWR method is that it leaves you with a fund at the end of your retirement (ie on death).
From a purely optimal consumption perspective, this may not be ideal (eg if you have no one to leave it too, or if a future government decides on a 100% tax rate on unused funds).
There isn’t really a way of providing for an event (death) at an unknown future point in time.
Buying an annuity is a trade off. You get a higher initial yield than under the SWR, but no lump sum at the end. Of course, if you die earlier than “expected” then you lose out on both methods (annuity or drawdown).
This is just another way of looking at the situation. Obviously everyone has their own circumstances to consider, which is why it is a complex but interesting area.
One final post from me (thank heavens!).
The FCA have an annual study on retirement income (link below)
https://www.fca.org.uk/data/retirement-income-market-data-2023-24
The report contains a number of interesting insights. Relevant to the SWR discussion, is that apparently 60% of people with funds in excess of £250k have an annual withdrawal rate in excess of 4% (you have to work with the underlying data to get this %age).
This may be done for a number of good reasons eg need for one-off payments, or having more than one source of income or unexpected ill health etc.
However, it does seem to me that if the higher rate of withdrawal is being done without undue consideration, then we may indeed be heading for a future of impoverished pensioners.
The one time I approached a Financial Adviser (in relation to another matter), he suggested that I could generate income at a rate which would equate to 10%. I know that this is only one data point (so anecdotal, even if real to me), but if this is the sort of “advice” being bandied around by FAs, then I think that “pension freedoms” have the potential to inflict a lot of misery.
@TA – excellent as always.
My approach is that in 8 years I’ll have an income floor comfortably covering 100% of necessary expenses for my wife and I. To get there, I’ve got 8 years in short fixed income. With Royal London Short Term Money currently yielding 6%, it’s a no-brainer (until it isn’t…)
So I’ve got three pots – my SP & DB floor; an ISA to generate dividends for my personal spending containing a mixture income funds, infrastructure ITs and a small amount of accumulating Global tracker for some growth; and my residual SIPP (80:20, mostly Global tracker) for holidays, emergencies, that new Aston Martin, etc and helping the kids with house deposits and legacy, etc.
My take on SWR is that it’s one of a number of guidance tools that I’ll use for my SIPP:
3.5% – am I exceeding this over a three year rolling period?
VPW – am I exceeding this?
4% – am I exceeding this?
This should allow me to match asset types to spending need rather than time horizon. Pots are supposedly sub-optimal, but very comforting.
Really thoughtful post TA and even more thoughtful comments. A few thoughts from me.
First, one of the biggest issue for me with SWRs is that they are heavily determined by inflation and the driving force when it comes to the SAFEMAX is usually periods of high inflation. As Alan S points out, the inflation data we have is patchy. Pre-1948 we have to rely on alternative measures of inflation. CPI itself was only introduced in the late 90s and we don’t know how inaccurate the back-calculated measure is. RPI has methodological issues as well. Ultimately, we know that the data we have is ‘wrong’ and we have little idea how ‘wrong’ it is. So caution very much advised.
A second point: this decumulation stuff is hard! And yet increasing numbers of non-financially literate people are going to have to deal with it. We were promised product innovations when pension freedoms came in ten years ago. There have been almost none. I do know that one provider is looking into a mortality pooling / deferred annuity product but there is a huge way to go. In the meantime, more people are going to reach retirement with the ‘hardest problem in finance’ to deal with.
Third, and related to above, this has a real impact on the economy. That uncertainty can lead people to take lower retirement incomes and reduce consumer spending. It also means people take less risk in their asset allocation. All to say, this doesn’t tie in well with the government’s latest ‘growth agenda’. The evidence we have is that, generally speaking, retirees adjust their expenditure to their income. So in the round, lower retirement incomes leads to less spending in the economy. If we can find a way to increase sustainable retirement (through more sophisticated mechanisms I touched on above) this would have a real positive impact on the UK economy.
I would recommend switching from using RPI to average earnings (or wage inflation) as the index in your SWR model (i.e. rather than using RPI as the multiplier each year from the initial drawdown, use wage inflation as the multiplier). That will be a better proxy for a constant standard of living vs. the rest of the population, rather than just a constant cost of living. Since 1950, RPI has averaged 5.3% with a volatility of 4.5%, wage inflation 6.6% with a vol of 4.9%. See what that does to the SWR over a 30-40-50 year horizon.
You could even do a “triple lock” (best of 2.5%, RPI, earnings, each year) and see the SWR under that scenario.
It’s all very good doing complex withdrawal strategies (McClung wastes 379 pages on this) but the real decision is what do you want to use as the inflator. What do you really mean by constant withdrawals in ‘real terms’. That is the biggest assumption and the hardest question to answer. Strangely McClung spends no time on that at all.
@Stroud Green – OK, some shorter retirement lengths for you (UK equities/bonds)
20yrs
100% equities: 3.9
80%: 3.8
60%: 3.6
50%: 3.5
40%: 3.4
25yrs
100%: 3.4
80%: 3.2
60%: 3.1
50%: 3
40%: 2.9
@The Details Man – good to hear from you! The more I think about it, the more annuities seem like an attractive option to solve the complexity. At least at today’s yields.
@ZX – your wage inflation point is well made and I’ve thought about it often in the years since. Pretty much every time I’ve upgraded something! I didn’t factor in an AI subscription when I first starting saving for my retirement that’s for sure.
One touchstone I used was originally a £100000 in a 60/40 portfolio would produce £4000 pa (4% rule)
(Currently £3000 pa is more realistic)
This made me realise from the outset that very large sums were required for a successful retirement far outweighing any house purchase monies
Obviously pensions all things being equal should be prioritised over house purchase for this reason
I have noticed that new posters to any financial blogs are rather taken aback by the large sums involved
For an amateur investor to successfully generate these large financial figures from scratch is rather daunting to say the least.Then to be faced at retirement with managing and living off these hopefully large sums poses more serious issues
Telling my teacher son what sums are required to finance his taken for granted index linked public sector pension always produces a silent moment !
xxd09
@Accumulator #76
Thanks for running those figures. Looking at those SWR’s together with the previous relative failure rate of bond heavy portfolios, it begs the question, are we unduly overly concerned with SoRR?
@TA (#70)
It would be nice if I could have even a single morning of mathematical wizardry – here’s hoping! One thing that contributed to the time taken was that at the start I only had a vague idea of how to do it. It took two major revisions of the method to end up where it did (and there is still a single further step to take that I have been putting off even though I now have the requisite info…). And, I might be retired but I do have other hobbies (and calls on my time) apart from calculating gilt returns as enjoyable as that is!
@NickH (#71). Assuming a historical retirement using the exact MSWR, the amount in portfolio at the end of the planned period ranges from zero to a lot (in real terms potentially many times the initial portfolio). Variable withdrawal methods have the ‘interesting’ property of increasing the amount left in bad retirements and decreasing the amount remaining in good retirements which is probably a useful attribute.
@The Details Man (#74)
One early inflation index excluded the price of beer on the grounds it would only encourage the working classes to drink.
Decumulation is hard since so many things are unknown including longevity, future expenditure, and real asset returns. The first of those can be approximated by assuming long life (e.g., 10% probability), the second can be approximated by assuming that current real expenditure will be continued (the literature is mixed on that – while Blanchett’s ‘smile’ is the most well-known, there are some studies (including for the UK) that suggest expenditure is fairly flat), while for the last one can look to history as TA has done here or use some form of Monte Carlo to give a larger range of outcomes. Each of these approximations requires some thought and knowledge. For example, most people underestimate their longevity and even if they don’t the current value may still turn out to be wrong (e.g., the ONS cohort longevity estimates are just that – part of the process involves a panel of actuaries, medics, and others that try to map future developments in life expectancy – AFAIK, ONS publish three(?) estimates – high, medium, and low).
The collective defined contribution pension (e.g., see https://www.plsa.co.uk/Policy-and-Research/Topics/Collective-Topics or https://www.moneyhelper.org.uk/en/pensions-and-retirement/pensions-basics/collective-defined-contribution-pensions for an overview) might form one solution as a hybrid lying somewhere between a conventional DC and a DB pension. As far as I am aware, there aren’t many of these in operation yet (I think the new royal mail one is an example).
@xxd09 (#77)
You’re right about large numbers. To generate an income equivalent to the SP at 67 a pot of about £230k would currently be needed (i.e., 12/0.052 with a 5.2% payout for RPI annuity). Assuming a worst case growth of 0% real over 40 years (close enough to the worst historical UK 40-year case), about £5700 would need to be invested per year. This means someone on a median salary would need a total investment (including employer contributions and tax relief) of just over 15% of their salary. I note that the median historical case would give a pot three times that (or require contributions of about one third). Just like decumulation, one major problem is that at the start of the accumulation journey it is not possible to predict how much will be in the pot at the end.
@Vic – I’ve found that focusing on the failures helps me understand what the plausible / reasonable remedies are.
If you want a bombproof solution that cannot fail… well, it doesn’t exist. The logical extension of that path is to become a billionaire and buy a fortified bunker in New Zealand in case the revolution comes.
But if you ignore the more common perils of decumulation then you end up taking the advice of some dude on the internet who says, “Well, the S&P 500 has made 10% a year for the last five years, so you can take that.”
Or, as you rightly said, you can take an interest.
Alan hit the nail on the head when he said you can have:
“Certainty in income but no idea how long it will last
or
No certainty in income, but certainty as to how long it will last”
We can all cut our budget to some extent if we retire at a historically awful time.
So the question becomes: what’s your floor (as Rosario mentioned). And are there good ideas for managing your income (up as well as down) dependent on the market.
Another big question is: what heightened the risk of failure in the past?
And obviously inflation looms large there.
So is there a better defence against dreadful inflation outcomes than a standard equity/bond portfolio?
We know that there are.
Then it’s worth considering other factors that may be salient for you personally though not in the basic model. Life expectancy being an obvious one, other sources of income, sensitivity to other dynamics such as wage growth as mentioned by ZX.
From there, I think it’s possible to synthesise a personal answer to the problem of withdrawal. One that’s alive to the risks and remedies without being hijacked by a particular number.
Or, if that all seems too much, buy an inflation-linked annuity when the time comes and live off that. I’ve seen it done. Seems to work fine.
@Vic, @TA:
There are many risks (inc. SoRR) to a successful retirement. There are also opportunities. Taking a balanced view [for you & yours] is IMO important. FWIW, I think far too much is written about SoRR, too little is written about the other risks, and virtually nothing is ever said about opportunities. That is, the coverage of risks and opportunities is far from balanced.
One upside of SWR, by avoiding most/all failures, is likely capital accumulation in most scenarios. Can ratchet “safely” to say 3% of peak portfolio value, which may be 4%+ of the initial portfolio value. In exchange for some initial “lean” years avoiding SoRR.
@Richard Dixon
I tend to agree with that, it’s at its maximum in the early years and where it can lead to extinction level events.
Focusing as we’ve down here on failure doesn’t show the large potential upside pay off profiles. Then again, they’re no consolation if you find yourself on your arse. The answer, that has been suggested by many on the thread is partial insurance. Whether purchased or by early frugality. The problem with the latter is that comes at the cost of time, and as the likes of Ermine has pointed out, the marginal utility of time increases through life. It does go to show that the problem is as much a psychological and philosophical one as much as it is a financial and statistical one. It’s certainly good to hear others’ take on it.
Great thread.
@Alan – I’m stunned to hear that there could be anything more important to do than calculating gilt returns 😉 I also meant to second your earlier comment about the generosity of the RORE team in making public the Macrohistory database. Incredible really, especially when considering this type of data has become the foundation of significant revenues for others in the field.
@Richard D – Good point. There’s a fair bit in the literature on not going mad in the early “go-go” years of retirement, hopefully finding yourself on one of the regular paths (5.3% median SWR for UK 60/40 portfolio, 30-yr retirement), allowing you to ease up on the budgeting thereafter.
One possibility there, that would make a big difference in the early years, is to engage in some enjoyable PT work that supplements income and eases the transition into retirement. Later on, the State Pension comes into play, providing a margin for error if it hasn’t formed part of the initial calculation.
Your comment also reminds me that the big winners of a naive SWR strategy are likely to be a retiree’s heirs.
@Vic & Al Cam – Your comments about the upside are making me try and tease this out…
I’m not sure but I think the upside is implicit to the whole debate. That is, the existence of the upside explains why – to a large degree – the debate isn’t: annuitise your portfolio, live off that, the end.
The SWR chart shows that the upside is historically more likely. So we’re willing to run the downside risk of the nightmare scenario materialising, in the hope that hanging on to the portfolio will make us / our families better off in the end.
Hence, I think, aside from psychological explanations like loss aversion, it makes sense to understand how to mitigate the risks run, while the opportunity is one we’ve already understood and internalised by dint of holding a volatile portfolio.
That is, beliefs about the opportunity are already contained in content such as “the magic of compound interest” or long-run charts of global returns etc, and aren’t discounted because humans are natural born pessimists. I think maybe we’re natural born hedgers 🙂
@TA:
When describing F&U, the late great Dirk Cotton said:
“The most important decision you will make in retirement planning is how much of your resources to allocate to the upside and floor portfolios”
and
“The correct balance [between the upside and floor portfolios] will depend on how willing you are to risk losing your standard of living for the chance of having an even higher one.”
Which I think captures the main F&U issue/dilemma well.
Re my [explicit] take on risks and opportunities, see: https://simplelivingsomerset.wordpress.com/2024/07/17/is-risk-always-bad-news/ and the subsequent short post too.
@Alan S, TA, xxdo9 and ZX (too many names to type), fascinating thread for the maths and intricate thinking behind drawdown. I retired last year (with less short term money than I initially intended or would have wanted), so here’s a guide to my thinking and subsequent approach…
IMO, while detail, composition, perfecting a set of assumptions etc is endlessly cool, nothing beats availability heuristics, self knowledge, asset allocation (both long and short term), using an individually tailored dynamic macro approach, particularly at the start.
I’m 49. And started a deaccumulating phase last year. Real life is massively different vis a vis those (static & dumb) drawdown internet sites. That said, I enjoyed this one – “rich, broke or dead”
https://engaging-data.com/will-money-last-retire-early/
My approach is to ignore my long term assets initially (accessible age 60). I also ignore my mid term assets (which however remain available). I just focus on short term assets ($330,000), which I’ve assigned as spending money this decade. It’s mentally freeing and I get to practise ….
So with that said, I split my $330,000 as follows: $260,000 ETFs/shares & $70,000 cash (because … for reasons best articulated by xxd09 on numerous occasions). This split throws off $15,000 a year and I need another $15,000 for a base spend. Anyway, put this into excel and inflate the annual spend, use up cash, sell assets as needed, deflate grossed dividends appropriately, there’s still $200,000 or so left. And voila …. combined with midterm assets, the whole stack grows.
Of course, I’m going to spend more than $30,000!!!!!!! But this is basically the amount I spent in year 1, because I most enjoyed reading books, beaches (locally and sunshine coast, also free). I don’t cook so buy food from a Greek lady at the market, speciality providores or nice cafes. I’m not frugal. I’ve already got diamonds. It’s just not expensive to be me! At least, it wasn’t in year 1.
If I want to attend the Monaco grand prix or travel to the Greek islands, I’ll add it to the spreadsheet. The numbers will change… I’ll swap money for memories… It’ll all be cool.
Anyway, that’s what I do, and how I think! Peace of mind doesn’t require a lot of money. A stable political economy, personalised asset allocation and access to excel works pretty well!
Of course, I’m single, rich in existing possessions, my sole dependent is a psychotic cat! I’ve got a midterm and longterm backup (which affords an emotional cushion). And I’m using a ‘spend it all’ philosophy.
I’ll report back in 20 years if it goes wrong and I’m swallowing cat food with my aggrieved feline.
@all, f/up …. A note on privilege … It’s easy to be flippant on the internet.
Midterm assets in accumulation phase $400k. Longer term assets accumulating in a tax favorable environment are embarassingly high …
I’m surrounded by privilege … loving family, close friends, stable govt across all levels, ie free health care, amazing GP, affordable access to additional medical professionals (with/without private health insurance, which I incidentally have, etc etc)… [just thought it was important to add this].
Cheers
@TA You’re absolutely right, in that we all do our sums and our reckoning in order to try to make the most of what we have with the resources accessible to us.
For some, the point of the game is getting to the stage of FIRE (which may be very difficult for some, or even not possible at all for others), for others its to maximise their pot (for the benefit of families, or perhaps just for the fun of it).
I suspect that there are as many different situations and motivations as there are readers of Monevator.
To some degree, I have decided to bow out of the game at this point by annuitising. Other priorities now press on my time, and as @Ermine has said on many, many occasions, time is the one thing you can’t buy any more of.
However, this has been a great thread and discussion. Thank you all, and I wish you well.
@ZX — one thing I’ve never understood about the triple lock is why the 2.5% option is included. If inflation and wage rises are both *under* 2.5%, why would you need to increase your withdrawal/state pension any more than the max of the other two criteria?
@wodger
The 2.5% criteria gently moves the state pension to be a higher % of average earnings (when it bites). Unfortunately no one has ever stated what % of median earnings is the target level for the state pension.
If we did this we’d have a hard stop criteria for the triple lock. And if agreed by both parties it would take a lot of noise out of the debate, so we could focus on a sensible benefits bill as a % of gdp.
@TA – excellent article – I look forward to the next (hopefully more optimistic!) part of the series. Excellent comments also, I’m always impressed by the thought and modelling put into this whole area by contributors.
In terms of flooring 2xSP provides almost exactly our average expenses for the last 4 years (which includes a fair amount of travel and gifts/charity). SP provision may of course be less generous in the future but its hard for me to imagine a world in which it doesn’t exist. Of course there will be a time where 1 SP is covering 1 persons expenses. But thats why we invest, to provide something over and above the minimum.
I think @TA in #85 is correct, the upside potential is the key. At some point for me, much later in life, partial annuitisation may be a sensible move, but at this stage the potential downside seems very offputting. Anecdotal of course but I attended 3 funerals in successive weeks in January of friends and colleagues, all contemporaries (58-61). I know there are various protections that can be purchased, but the crux is that you take your money which you’ve built up over 30 odd years of fairly aggressive saving, from very moderate means, and you just give it away to a third party in return for certainty of income. I definitely struggle with that concept.
@Al Cam – that’s a great quote!
@London A Long Time Ago – Sounds like you’ve got it sorted. The words “self-knowledge” leapt out of me from your comment. I think that’s key to navigating this phase of life and another thing no model can incorporate.
In a sense, the SWR model presents a stark challenge: “You’ve never done this before, it can go badly wrong, how are you going to ensure it doesn’t?”
@NickH – I fully agree!
@Larsen – Yes, there’s resistance in my head too in regards to annuitising at a relatively young age now the yields are high when I’d previously considered it’d be something I’d do around age 70. I need to sit down and have a good chat with Mrs TA about it and see where we land.
Morningstar did an article about swr in the UK context some time back and that came out as something like 2.8% so broadly consistent. I just use it as a sanity check as other factors apply at the moment.
With regard to Fixed Term Annuities, L&G is definitely in this game in the UK. A friend took 25% PCLS from their SIPP, and took an FTA with L&G with the £750k remainder. I think (but couldn’t be sure) it was for 5 years and with a £nil withdrawal rate he was getting his £1m back (albeit crystallised so no more PCLS) at maturity.
@TA and @all, it’s a very high bar but this has been one of the best article + comments combinations on Monevator. Looking forward to part 2!
Thank you!
@TA (#93):
Dirk Cotton was IMO rather special.
For those who do not know, Dirks website is still accessible at http://www.theretirementcafe.com/
OOI he went on a bit of a journey re the level to floor. His earlier posts favoured flooring his envisaged lifecycle, but laterally he had come around to the view that as low as practicable coupled with an upside designed to minimise the probability of ever having to live from just the flooring.
My experience of F&U to date (eight years and counting) is that I initially over-floored. This is primarily why I am so set against fully annuitising your foreseen flooring need* from the off. Just my take and YMMV.
*which I (and others too AFAICT) significantly over-estimated prior to retirement
It would be interesting to dig back into the research literature and see if anyone has attempted to link your observations to different spending mindsets among retirees.
You’ve prompted me to remember the paper: Understanding retirement journeys: expectations vs reality.
The authors segmented UK retirees and categorised some as free-spenders who were even prepared to go into debt early in retirement.
Two categories were predisposed to save although they split into high and low income cohorts.
Yet another group was just scraping by and very short on options.
My anecdotal experience is of a parent who is a natural saver and adept at managing their budget. They intuitively operate within the limit of their annuity income – whatever that may be. My guess is that if the annuity had been set 10-20% higher or lower they’d act the same way.
I’m curious about how you’d categorise your “money mindset” and how your spending behaviour has changed in the light of your ongoing F&U experience?
@TA since you asked.
I’m 8 years into decumulation. I would categorise myself as cautious.
For the first 4 years I lived off a very cautious drawdown schedule, probably living off a 2% withdrawal rate, with SIPP and outside fund mostly in equities.
This proved to myself that 1) I could live comfortably at a relatively low level of income, and 2) the basic mechanics of SWR would work.
Then a couple of things happened which changed things personally for me.
1) Covid came along which fundamentally altered my outlook on life.
2) the incredible bill run on the US market, which have driven equity valuations to what I regard as extreme levels
3) certain family related events (deaths, illnesses etc) which again affected my outlook.
Using the F&U terminology, I have moved from a cautious risk tolerant mindset ie happy to accept risk for potential upside, to a more F focussed mindset.
8 years in, I know what income I need to have a satisfactory lifestyle, and have taken steps (annuitisation being one) to ensure this.
This is partly due to the life stage that my wife and I have reached (no desire for Antarctic cruises etc, or lambi purchases).
Our lifestyle (pretty modest) is reasonably well mapped out, and the kids will be adequately provided for.
So, the need/desire to allow for upside is greatly diminished.
Of course, this is purely a personal perspective. It greatly helps that my hobbies and interests are almost cost free, and we have little desire to travel widely.
I would say that after 8 years of decumulation, and the events listed earlier, my world view – ambition, if you like – is very much smaller than at the outset.
Luck plays a huge part in one’s life journey. We have been very lucky, and are grateful and mindful of this.
@TA (#99):
I remember that report well. None of the consumer groups identified therein particularly resonate with us.
Summary to Date
Cumulative spend vs baseline* ‘budget’ 2017 to 2024: c. 80%
Spendiest year [to ‘budget’]: 133%
Cheapest year: 66% – not 2020 incidentally
Spending may** have increased since starting DB pension in 2023 some four years earlier than assumed in baseline plan.
Conclusions/ Thoughts:
Things change, often as a result of events unforeseen (and in some cases unforeseeable) in advance.
I am not consciously aware of changing our money mindset or approach to spending since we pulled the plug, but it may well have morphed somewhat.
One thing I have noted is that being retired with the DB turned on seems easier than with it turned off. This may be because for all my working life I was salaried, so am accustomed to a regular monthly pay check. Having said that, living for six plus years without such a pay check was definitely not something I regret – I learnt a lot and often about somewhat unexpected things too.
Hope this is of some use to you?
*determined before pulling the plug
** too early to be definitive and other factors at play too
P.S. totally agree with NickH (#99) re luck!
@The Accumulator 76
Thanks … and such a great thread!
@Alan S (#79):
Could you be good enough to point me at any UK study, from the last decade or so*, (other than IFS #R209 – which we have chatted about before) that suggests UK retired expenditure is fairly flat?
TIA
*the so-called U-shaped model of retirement was very prevalent amongst UK advisers, etc years ago
@Nick H and Al Cam – thank you both! It’s interesting that in stark contrast to the proverbial Lambo buyer you both operated well within budget from the start. I wonder if this is behaviour typical of the financially literate (who are all too aware of withdrawal risk) as opposed to some intrinsic money mindset. It’s interesting too that you both report changed perspectives after relatively short periods of retirement, and Al Cam – as you’ve pointed out before – annual spend can vary dramatically.
Moreover, you both observe, I think, that having a dependable floor is very reassuring. This chimes with my own experience. In my case, my floor is built from the income I’ve continued earning on the side, despite calling time on my career. The floor gives me peace of mind and frees me from fretting about the market rollercoaster.
Re: my own mindset. I am not naturally frugal but became so in order to build wealth. My willingness to spend has reasserted itself since I FIRE-ed, except that it’s far better controlled now. The habits I’ve built in the frugal years have become second nature to some extent and I better understand how to spend on the things I care about and pass on the things I don’t. I think this is the self-knowledge aspect mentioned by London A Long Time Ago.
@Al Cam (#103)
Yes, it was the IFS one I was thinking of. I know you have identified some flaws with that study, but there are some interesting suggestions of cohort differences in that work.
A US study (Roy K, and Carson S., “Spending in Retirement”, J.P Morgan Asset Management) showed the overall decline but also that subdivisions within the overall retired population have different spending profiles, with some experiencing increases in spending with ages. Making a comment on the paper, Pfau said “Spending may decline, so I would not fault anyone for using assumptions of gradual real spending declines such as 10% or even 20% over the retirement period. But pending further research developments, I would avoid moving too far in the reduced spending direction as a baseline assumption.”
Obviously the US case is different largely because of health spending.
@Alan S (#105):
Thanks for the reply.
Just to be clear, I am sure some people (US and UK) will experience spending increases during retirement. However, it is the aggregate scenario that is of general interest. As you say, and as things stand, the pressures for increased spending are probably higher in the US due to health spending – but that could change.
The cohort issues the IFS paper identifies are probably largely attributable to their regression model and that this technique implicitly assumes that the model is complete – ie only the parameters they include are explanatory. This IMO is problematic. For example, the GFC (SoRR if you prefer) effected each cohort differently – but AFAICT SoRR is not modelled.
@TA (#104):
If my experience is anything to go by, things do change – and probably will continue to change. IMO, this is definitely a point worth noting.
Re “Moreover, you both observe, I think, that having a dependable floor is very reassuring.” It is, and in my experience the form of the floor matters too – another thing I did not foresee.
FWIW, I provided further details on my experiences about these points a year or so ago in [some] of my comments to: https://monevator.com/should-you-build-an-index-linked-gilt-ladder/
Alan S (#105):
P.S. thanks for the Pfau quote. It is quite re-assuring. In my baseline model (which in this area dates from at least ten years ago) I assumed a real spending drop of some 12pp across the duration. Albeit, I am already below that level – although I now suspect my earliest years retired spending may, in due course, turn out to be anomalously low. Time will tell.
I know the JPM report and would note that IIRC it uses Chase bank a/c data as its primary source.
@TA, thank you for this thread! Alcam #107, my ears pricked at ” I suspect the form of the floor matters”.
I strongly concur! I’m currently writing up a FIRE chat case study. My cash holdings are too high, but what’s the point of money other than to support you! …. My ETFs/shares dropped $20K from highs this month and I can shrug it off because of my cash bal. …
Further, I know I’m going to grow and change personally over this next chapter!!!!!! My spending could absolutely change, but not the baseline number, because it’s been stable for 5 years ….
I suspect happiness and equilibrium in retirement requires a clear conscience, friendships, some type of furry demanding creature for chaos and a rich imaginative interior world. I’m arranging assets accordingly.
@ Alcam #107, my ears pricked at ” I suspect the form of the floor matters”.
I strongly concur! I’m currently writing up a FIRE chat case study. My cash holdings are too high, but what’s the point of money other than to support you! …. My ETFs/shares dropped $20K from highs this month and I can shrug it off because of my cash bal. …
Further, I know I’m going to grow and change personally over this next chapter!!!!!! My spending could absolutely change, but not the baseline number, because it’s been stable for 5 years ….
I suspect happiness and equilibrium in retirement requires a clear conscience, friendships, some type of furry demanding creature for chaos and a rich imaginative interior world. I’m arranging assets accordingly.
edit last paragraph… “my happiness”
@LLTA (various)
Certainly agree that it is not all about nice tractable numbers. And whilst we all probably need a little chaos in our life – some forms are definitely more welcome than others.
Re form of flooring: all I was really getting at is that being paid monthly from a DB is just so much easier to deal with than having to sort out (approximately annually) which savings cert (or gilt, etc) to cancel and how to roll/re-invest any excess or make up any shortfall. Managing any form of laddered flooring takes some effort, which I found becomes tedious/somewhat wearying after a while. OTOH, it does force you to look at what you might need (ie sort of plan) at a lower level of detail approximately annually – which is no bad thing either. This level of outline planning alone might help explain our possibly anomalous earliest years retired spending. I say this, because I never really planned spending before I retired. And I have already significantly backed-off the detailed planning.
Our next flooring step-ups (which should eliminate all the residual non-pension flooring maintenance tasks I describe above) will be when our SP’s switch on in a few years time.
Intrigued as to what has been “stable for 5 years” and perhaps what bounds you put on stable?
In your post at #87 you mention a feline. Yet your piccie seems to show a dog. What did I miss?
Thanks @TA for an excellent article. As you say the 4% / 3.1% rule is in some respects super cautious in that it takes the worst possible historical retirement point and doesn’t allow any downward adjustment to spending etc. I am looking forward to further articles and hope you might discuss strategies such as Vanguard Dynamic Spending, Variable Percentage Withdrawal and Probability Guardrail Drawdown. All these allow higher withdrawals in good years with the price being often substantially lower withdrawal in bad years. To different extents they factor in a reduction in capital in later years which in the UK with high inheritance tax makes a lot of sense. Thanks again
@Passive Investor .. yes, something that’s changed for me, is the change to charge inheritance tax on my SIPP .. this means my target amount to end up with has changed to just below the inheritance tax threshold.