≡ Menu

How much wealth do I need in my ISA versus my SIPP to achieve financial independence?

How much wealth do I need in my ISA versus my SIPP to achieve financial independence? post image

This is part three of a series on how to maximise your ISAs and SIPPs to achieve financial independence. Part two explained why the tax advantages of personal pensions make them superior to ISAs later in life. Be sure to also read the first part of the series.

To make the most tax efficient use of your savings, your ISAs need only fund the years between early retirement and your minimum pension age.

Obviously it’d be marvellous if our ISAs piped hot income streams into our lives long after that, but our primary concern is to fund them so we’re unlikely to run out of money before our personal pensions take over.

That’s because there’s no point oversaving into our ISAs, either. That would see you delay financial independence by paying tax that would have instead been clawed back through pension tax relief and added to your growing nest egg.

The ISA/pension balancing act

Should investment returns turn out to be poor, we would expect our ISAs to be running on empty as we dock with our SIPPs.1

We would then discard the ISAs like empty booster rockets and ride on using our SIPPs, and eventually a State Pension slingshot.

To put that plainly:

Phase 1 – You need to be able to draw all your income needs from your ISA / taxable accounts without fear of running out of money,2 until you reach the minimum pension age.

Phase 2 – You need to be able to draw all your income needs from your personal pension, once you can access it, without fear of running out of money until you die.

Lifeboat – It’s quite likely the State Pension will provide some support later in life. The lower your income, the older you are, and/or the sketchier your plan, the more important the State Pension becomes.

We’ll construct the plan so the State Pension is primarily a back-up and, later in the series, we’ll draw upon research that shows how you can adjust your plan to account for it.

How much income and for how long?

How much annual income do you need in retirement? And how many years do you need it to last?

These are the big two questions to answer for each phase of our plan.

Guesstimating your required retirement income is not so hard, especially if you already track your expenses.

Let’s say you’ve decided you’ll need £25,000 per year for the rest of your life. (We’ll assume all calculations are in real terms, so we’re accounting for inflation.)

How much wealth do you need in your ISA to sustain £25,000 in annual income?

It depends on how long you need that income to last (Phase 1) before your pension income becomes available (Phase 2).

There are two basic ways to fund your Phase 1 pre-pension, post-retirement income:

1. The usual sustainable withdrawal rate (SWR) approach – a portfolio of mixed assets in your ISA that you ‘create’ an income from by selling a planned proportion each year.

2. Liability matching – a big pot of cash or bonds3 that won’t grow much or at all after-inflation, but that starts out big enough to take your desired income from each year until you can crack open your pension.

Let’s look at both in turn.

Method #1: Drawing down an ISA portfolio

The infamous 4% rule says we need to build wealth that’s worth 25 times our annual income requirement to become financially independent.

25 times your assets comes from: 1 / 4 x 100 = 25

£25,000 x 25 = £625,000

So we need £625,000 to take an annual, inflation-adjusted income of £25,000 at a 4% sustainable withdrawal rate (SWR).

But the 4% rule applies specifically to 30 year time frames.

What if you only need your ISA to last ten or 20 years until your personal pension comes on stream?

Then your sustainable withdrawal rate (SWR) from your ISA can be higher.

Let’s say you can use an 8% SWR to sustain spending from your ISA for ten years.

We can save much less into our ISA in that scenario:

1 / 8 x 100 = 12.5

£25,000 x 12.5 = £312,500

Now we only need ISA wealth of approx £312,500 to draw an income of £25,000 for ten years. After that, we will look to rely on our personal pension income, if our ISA is exhausted.

Bear in mind that a SWR calculates the maximum amount you can take from your portfolio without running out of money, based on historical returns data. (Terms and conditions apply.)

In most scenarios, you actually end up with a healthy surplus in your account when you use an appropriate SWR, even if your retirement was blighted by economic times of darkness such as the Great Depression, Stagflation, and the World Wars (provided you were on the winning side).

Nonetheless, we want a plan that minimises the chances of being forced back to work against our will.

We’ll therefore use cautious SWRs throughout this series that suit our possible timeframes – no matter if we need our tax shelters to last ten years or 50.

Method #2: Liability matching

The safest way to fund your retirement is to match your future expenses (liabilities) with a treasure chest of near risk-less, guaranteed income.

A ladder of inflation-linked bonds would be ideal.

In the ISA example above, a tranche of your bonds would mature every year, depositing £25,000 of inflation-adjusted income into your account for each of the ten years until you can access your pension.

Alternatively, you could save up enough cash to cover the ten years, remembering to factor in an allowance for inflation.

Liability matching with low risk assets generally requires more capital than investing in an equity heavy portfolio. The more resources you have, the less growth you need, and the less risk you need to take. It’s a trade-off.

My assumptions suggest that it’s likely quicker and safer for everybody to save cash4 to bridge an eight-year gap or shorter, between Phase 1 and Phase 2.

I’ll go into more detail on this later in the series.

Minimum pension age

Our ISAs need to span the gap between our early retirement age and our minimum pension age – the latter being when we can officially smash open our defined contribution pensions like piñatas.

Which will be when exactly?

Unbelievably – ahem – that’s not so easy to pin down.

Currently you can get into your defined contribution personal pension from age 55.

But the 2014 Coalition Government (remember them?) indicated that the minimum pension age would rise to 57 in 2028. Your minimum age would then be set to ten years before your State Pension age, from then on.

Thing is, they didn’t get around to legislating the minimum pension age change. So it’s not yet law. And then Brexit happened. Eyes were taken off the ball. Now no one knows what’s going on.

We don’t know whether the rise in the minimum age will take place as mooted. But many industry insiders say the change is still coming and can be legislated whenever the government likes, so it’s best to assume the worst.

If you were born in 1972, you will be 55 in 2027, so you should be fine, right? You can tap your pension in 2027 before the minimum age hikes to 57 in 2028.

Not so fast. There has been talk of tapering the change in. It could be you’re still caught out, even if you’re 55 a few years before 2028.

It’s a mess.

We’ll play it safe by assuming that our minimum pension age is set to ten years before our State Pension age for the purposes of the upcoming and unbelievably exciting case studies we’ve got planned for this series.

These case studies will also show how to calculate how long your ISAs will need to last (roughly), given your current circumstances.

How much do you need in your personal pension?

By the time you retire, your portfolio – when combined across all accounts – should be funded to last the rest of your life. How long might that be? If you’re age 60 or less today then you have at least a 10% chance of living to age 98 according to UK life expectancy data – unless you have good reason to think otherwise. There’s an even greater chance that one of you could survive if you’re part of a couple.

SWRs tend to reduce over longer periods of time, but the curve flattens out. Multiple research papers point5 to a 3% SWR being suitable for retirements of forty years and over – which likely accounts for the majority of people on the FI fast track.

Carrying on the £25,000 retirement income example, the wealth needed to sustain lifetime spending for over 40 years at a 3% SWR is:

1 / 3 x 100 = 33.333

£25,000 x 33.333 = £833,325.

We’ve established that the ISA portion of this wealth target needs to be £312,500 to ensure it doesn’t run out before the pensions come on stream ten years later.

Therefore, your personal pensions need to be funded to the tune of:

£833,325 – £312,500 = £520,825 by the time you pull the trigger.

My thanks to Monevator readers Aleph, Naeclue, and Oxdoc whose dogged persistence corrected my mistaken assumptions when this article was originally published.

Other income streams – So you’ve got other income streams like buy-to-let property and defined benefit pensions to tap into? Lovely. Just deduct those additional income streams from your assumed retirement income when they’re available. Your portfolio will only need to cover the remainder. We’ll cover the State Pension and DB pensions that become available further down the track at a later point in the series.

In the next post, we’ll cover how to choose a credible SWR that matches your personal timeframe and accounts for a low interest rate world, non-US investment returns and the implications that has for your asset allocation in retirement.

Take it steady,

The Accumulator

  1. Self Invested Personal Pensions. []
  2. We mean without fear on a practicable level. Ultimately there is no absolute safety. []
  3. We’re talking a ladder of individual bonds (not a fund) with staggered maturity dates. Each tranche of maturing bonds delivers a payload of capital to match your income needs per year you need to fund. Inflation-linked government bonds are best. A Purchase Life Annuity could also conceivably fit the bill. []
  4. A lack of suitable bonds makes it hard to build an inflation-linked bond ladder in the UK. []
  5. We’ll cover the research in more detail in the next episode in the series. []

Receive my articles for free in your inbox. Type your email and press submit:

{ 109 comments… add one }
  • 51 The Investor January 31, 2020, 9:28 am

    @Aleph — I’m confused as to why you’re confused. 🙂 You don’t seem to be engaging with the fundamental question here, which is that you cannot access your SIPP until a particular age, and the early retiree is trying to retire a number of years before that access date.

    Your example of retiring at 18 with £833K in an ISA is irrelevant to this question — ISA money is always accessible.

    Let’s say you have a 3% SWR and you’ve taken your ‘pot agnostic’ approach to building wealth so you reach 47 with £833K across your two platforms, ISA and SIPP.

    You have £75K in your ISA and you have £758K in your SIPP.

    You begin withdrawing 3% a year (c. £25K) from your ‘unified pot’ — which *in reality* means you have to take it from your ISA, because you *have* to.

    The market is wobbles up and down for a few years, and overall portfolio returns are flat.

    After three years-ish of taking out £25K-ish a year, you have no money left in your ISA.

    “Never mind,” you think. “It’s one unified pot so I’ll withdraw money from my SIPP!”

    You call your SIPP provider. They say no, you cannot access your SIPP for another seven years.

    You have no accessible money at this point. You cannot withdraw 3%. You have to go back to work.

    That is what this series is about. 🙂

    Someone might say “well only £75K in the ISA and flat returns is unrealistic, someone would have at least £150K in their ISA…” then fine, that’s what this series about.

    There’s loads of grey here and the question is as valuable as any very fuzzy answer.

    But it’s simply not true you can treat it all as one-pot.

    To be even more blunt, your theory — it’s all one-pot — implies I don’t need to use ISAs at all to retire early. This is silly.

    I reach 47 and have ‘my number’ in my SIPP. Again, I can’t access it for 10 years. I can’t retire.

    That’s the fundamental issue, not managing max drawdown or whatnot.

  • 52 Naeclue January 31, 2020, 10:07 am

    @Aleph, I complete agree with your analysis, with the proviso that the pension pot gross balance is revised upwards due to tax. If you accept that 3% is a SWR at any age, then you need 833k. You could hold the whole lot in an ISA and you would be fine. You could hold 100k (adjusted) in a SIPP and 733k in an ISA and you would be fine. 433k + 400k would be fine. The split only becomes a problem if the ISA is depleted before you can draw the SIPP.

    @TA’s response illustrates the irrationality of a fixed SWR and is the source of the discrepancy of your 2 different approaches. He gave an example where a bad sequence of returns did not leave sufficient money in the SIPP to continue drawing at 3%, but that is a feature of the assumed 3% SWR and is why a fixed 3% SWR is illogical.

    If you start with 833k in an ISA, then you can take 25k per year. If the ISA falls in value to £600k after a year, you can still 25k, but if you left it one year before going into drawdown you would only have about 625k in the ISA, so could not take 25k as that would be in excess of 3%. The fixed 3% SWR could force you into a lower withdrawal amount if you delay drawing for a year!

    This is why I use a variable 3% rule, which I will increase as our life expectancy decreases.

  • 53 John B January 31, 2020, 10:08 am

    Remember that that £25k is not a hard limit. There is so much optional expenditure you could curtail or postpone (new cars, sofas, holidays, charitable donations, house repairs), and you could borrow money with secured or unsecured loans, interest free credit cards etc, that you could easily eke out an ISA for 2 extra years if the market goes against you.

    Charity is an interesting one, as you should always give that under gift aid when actually paying tax. When I FIREd I had to write to the National Trust etc withdrawing my gift aid declarations.

  • 54 Naeclue January 31, 2020, 10:18 am

    @TI. The difference between @Aleph’s and @TA’s approach is explained by @TA’s additional requirement that there must be at least 833k in the pension pot in 10 years time, otherwise you would be drawing down above 3%.

  • 55 The Investor January 31, 2020, 10:36 am

    @Naeclue — Ah, I see (I’m skimming these comments this week). However I think that’s not entirely accurate because he’s not (from what I’ve skimmed but maybe I missed an earlier comment) addressing the accessibility part of the picture.

    Me and @TA have a gentleman’s disagreement about the wisdom of applying SWRs to short-run ISA portfolios anyway (I’d favour the lump of cash approach) but I can see why he disagrees with me.

    I think as I say the question here is the most interesting part, for all the interesting stuff it spins out.

  • 56 Aleph January 31, 2020, 10:37 am

    @ The Investor. Hi, actually I did engage with that in my first post. You need enough in your ISA that it doesnt run out in 10 years – i.e. 325k as in the article. That doesnt change the TOTAL pot needed to support a 3% perpetual withdrawal rate, which is 833k, as also in the article. My diasagreement is not that some minimum money is needed in the ISA, that’s obvious. I just dont think the split changes the total pot you need for a perpetual withdrawal rate. I’m not sure how else to illustrate this because it seems really obvious to me.

    The hypothetical drawdown to 512k total happens whether you have your money split or not. Starting with a 325/512 split at age 48 and finding yourself at 512k left in your SIPP at age 58 and zero in ISA is obviously equivalent to starting with 833k in an ISA and seeing it decline to 512k by age 58. You have 512k left and full access to it at age 58 in both cases. The article is saying those scenarios are not equivalent because it claims that you need to top yourself back up to 833k at age 58 in the split case, but you dont in the all ISA case. I dont follow why I need to be richer at age 58 depending on how I historically split my money between age 48 and 58!

  • 57 The Investor January 31, 2020, 10:39 am

    Remember that that £25k is not a hard limit.

    It’s a hard limit per @TA’s mental exercise here, which is to try to find the optimal way to distribute assets (20-30 years in advance) between the two tax shelters to NOT have to go back to work / run out of money / employ other methods.

    I’m the “keep doing some work forever” guy (and also the “live off dividends don’t spend capital guy”) so my personal mileage is very different.

    But for the purposes of @TA’s conundrum, I don’t think saying you won’t buy a sofa and you’ll whack £50K on credit card (… whole article there… 🙂 ) is a solution.

    (Agree with a wider point that it’s useful to remember there’s probably some flexibility and there’s definitely a need to flex with uncertainty.)

  • 58 The Investor January 31, 2020, 10:40 am

    @Aleph — Ah, thanks for the clarification. I missed that early comment. Your position much more sense with that context. 🙂

  • 59 Aleph January 31, 2020, 10:44 am

    Yep, missed an earlier comment. 😉

  • 60 Jonathan January 31, 2020, 1:02 pm

    Actually the scenario ought to be the perfect one for the modellers trying to design an SWR strategy to attempt.

    The perfect SWR allows you to extract the set amount of money each year, with the capital reaching exactly zero at the point of death. (Planning to leave an inheritance is irrelevant, that can be a separate pot of money). The problem is that you don’t know in advance the date of death to plan for, or the returns on capital and in particular their fluctuation over the time period. In this case the “date of death” is defined, it is the point when the SIPP is planned to take over, so you can model just the strategy to deal with returns uncertainty.

    The usual approach always seems to me a bit of a cop-out, essentially it is to use a model to estimate the spread of final remaining sums and then adjust the withdrawal rate so under most scenarios there is money remaining and the lower end of the spread where the money left goes negative is reduced to some arbitrary “acceptable” probability, typically 5%.

    But actually what would be useful for the modelling to come up with is a set strategy that effectively reduces the final spread and allows withdrawal to more closely resemble the perfect SWR. For example a two-stage SWR where a set core amount is always taken and a second discretionary amount can be held over under some pre-defined conditions (but recognising there is a limit to how long some discretionary expenditure can be delayed, e.g. house repairs). Another might be keeping a portion of the fund in cash (or something cash-like) with a rule about how this is replenished when investments grow but run down when they fall.

    Of course TA may already have got lots of ideas about this but is simply saving them for a future article …

  • 61 Naeclue January 31, 2020, 6:23 pm

    @Harps, “Naeclue’s calculations can be further refined by using the Marriage Allowance”

    I don’t that is true. Transfer of the marriage allowance is quite restrictive and AFAIK, cannot be transferred to a spouse earning over £50k. This link seems to confirm that.


  • 62 Harps January 31, 2020, 7:38 pm

    @Naeclue. Hmm, I’d wrongly assumed qualification was simply being a basic rate tax payer (i.e. taxable earnings after allowances of £37500). I would still be able to download £66667 and qualify for MA and pay only basic rate tax… Doubt I ever would mind…

  • 63 The Accumulator February 1, 2020, 12:12 pm

    @ Oxdoc, Aleph and Naeclue – you’ve finally made me see the light. Yes, I agree and I’ve been too pessimistic. Thank you for putting me straight. Some re-editing ahead for me!

    @ Jonathan – I recommend Michael McClung’s Living Off Your Money for his analysis of combining guaranteed income ideas with a withdrawal rate. Also, this is a good piece on adjusting strategies to remaining life expectancies:


  • 64 Al Cam February 1, 2020, 2:29 pm

    @ Jonathan:
    Apologies if you already know his work, but Big Ern’s safe withdrawal work covers a lot (and I mean a lot!!) of ground (albeit from a US perspective), see, for example: https://earlyretirementnow.com/safe-withdrawal-rate-series/
    Also, your suggestion re a secondary discretionary amount reminded me of the idea of using a percentage of corpus (a.k.a. unitrust). Some more thoughts on the unitrust idea, and a whole lot of other ideas, can be found at

    Spoiler alert – there is no silver bullet, but I am pretty sure you know that anyway!

    FWIW, I personally favour the floor and upside approach over the so-called safe withdrawal technique.

  • 65 Far_wide February 1, 2020, 3:40 pm

    The position I’ve come to is that the best way to tackle SWR’s in context of the bridging the gap is flexible withdrawals.
    Modelling via portfoliocharts and other tools consistently show that having the capacity to reduce withdrawals in the face of a market turning against you really digs you out of the S**T. The magic numbers for me were being able to reduce my budget by 5% a year up to an absolute limit of 20%.
    You still of course have to then save enough to allow for a budget that can then be reduced if so required, but I think most people here probably do or intend to have that flex.

  • 66 Far_wide February 1, 2020, 3:54 pm

    Btw, thanks to Aleph and co. for persisting, as I was bamboozled as to why I would need a huge extra chunk of cash! I’m FIRE’d and have at least 25 years to bridge across to my pension, so the risk of running out of cash prior to that is a very real one to me.
    (As above, my current view of tackling that is to allow flexible withdrawals and a conservative SWR in the first place (2.75% at present).

  • 67 ZXSpectrum48k February 1, 2020, 5:17 pm

    I think all this discussion about SWRs and clever approaches rather misses the elephant in the room. What is your deflator? How much uncertainty is there in that deflator assumption? I love a bit of data mining and a complex model as much as the any nerd (possibly more) but you have to get a handle on the error bars on each assumption in turn. If the deflator has an error bar of at least +/- 0.5% to possibly +/- 2% that will swamp most other assumptions.

  • 68 Al Cam February 1, 2020, 5:24 pm

    Big Ern (see link above at #64) has quite a few interesting things to say about flexibility.

  • 69 Al Cam February 1, 2020, 5:28 pm

    I hear you!
    In essence, the so-called safe withdrawal rate (SWR) is unknown, and indeed unknowable, in advance.

  • 70 Far_wide February 2, 2020, 12:36 pm

    @al_cam – thanks, I’m already a big fan of ERN. Great stuff.

    @ZXSpectrum , your comments always come across as so sceptical about FIRE/SWR’s, no matter how conservatively or flexibly they are employed.

    Don’t get me wrong, there is room for plenty of cynicism. But in challenging uncertainty around how Governments calculate real prices across different time periods? Aren’t we getting to the nth degree of academia here?

    I’m sure one could find all sorts of variables to make anything less than a 0.5% SWR seemingly unpalatable. But, at some point, we have to get on with our lives, and accept that there’s always a (vanishingly small) chance we might end up working a supermarket till to make ends meet. This is the price we pay for being able to live the dream of very early retirement.

    The alternative choice we have is to work until the grave, “just in case”.

    Out of interest, for a typical reader of this site employing a very boring passive 70/30 global portfolio and wanting to retire at, say, 40, what in your view would an appropriate permanent withdrawal rate be? (permanent as for that long, SWR/PWR pretty much amounts to the same thing).
    Or do you believe it’s just impossible to do so with a sufficient level of confidence?

  • 71 Zoe February 2, 2020, 12:57 pm

    Thanks so much for this series. It has made retiring early seem to much more accessible and has made me look into my NHS Pension. I have an NHS Pension which I can access from 67 and looks like I will get 50k a year from this so sorted pension wise. I am 34 now so thats 33 years to go. How can I work out how quickly I can retire on 25k per year lasting until age 67? I think the maximum I can save each year is 30k.

  • 72 Far_wide February 2, 2020, 1:05 pm

    @al_cam , having revisited ERN, I’m with you now re: their warnings on flexibility. Looks like I have some revision to do! Can one ever get to a position where all of the experts will agree that you’re doing the right thing? I somehow doubt it!

  • 73 The Accumulator February 2, 2020, 4:36 pm

    @ Zoe – the next couple of posts will enable you to do that for your own situation.

  • 74 ZXSpectrum48k February 2, 2020, 4:55 pm

    @far_wide. I think it’s fairly hard to think you can define an SWR to better than +/-1%. Most of the constituents used to derive the SWR are hard to define to much better than say 0.5% and there are a number of variables.

    For me the deflator is the most overlooked assumption. First, there is the issue of whether you should use a cost-of-living adjustment or try to attempt to use a standard of living adjustment. The literature uses cost-of-living because CPI indices are available while standard of living is nebulous. It’s probably a function of CPI and earnings growth but it’s hard to pin down. Moreover, over a typical retirement of 20 years, those two time series hopefully won’t diverge too much.

    The problem is that if you think about longer term retirements, say 40-60 years, that approximation is less justifiable. Now, we can use something other than CPI. We could use UK earnings growth instead of CPI over the same historical periods. Knocks 1-2% of the SWR instantly. Not palatable. Ignore that then. Only optimism is allowed in FIREland.

    So back to CPI. Let’s take a typical TV. If we take US Bureau of Labour Statistics data, then a TV set that cost $1,000 in the CPI index 25 years ago, now costs $22 in the index due to hedonic adjustments (quality improvements). Similarly, while the average car price in the US has risen by 30% over the last decade; in the CPI index the increase is just 2%, again due to quality improvements.

    Now, conspiracy theorists argue this is a government plan to under-report CPI so they can pay less on pensions. It isn’t. It’s just that CPI is constructed to compare what a basket of good cost at time t with something at time t+n. It’s doesn’t capture improvements in quality or changes in standards of living (up or down). My concern here is are SWR adherents really understanding that if they use CPI they may well be locking themselves into a retirement that doesn’t include those improvements. Moreover, they can’t actually buy a TV for $22. The difference comes out of their pocket.

    So you still want CPI as the correct deflator, then which one? CPI or PCE in the US; RPI, CPI, or CPIH in the UK. There’s a good 0.5% between most of them. Plus that is just an average statistic. It’s not your CPI. We spend time on asset allocations and clever rules, but so little time thinking about our own personal CPI index. That could easily deviate from the official time series by a couple of percent per annum over 40-60 years.

    That’s just the deflator. When it comes to the asset data, we don’t have enough and the quality is low. This is especially problematic for early retirees With 100-170 years of data, you only have 2 to 3 independent 50-60 year panel data sets. That isn’t statistically significant. Moreover, this idea that the historic data somehow “stress tests” the portfolio is not true. If you had taken the data as of 1990, you’d never had in your sample anything as strong as bond returns of the last two decades. If you’d take asset data in 2010, you’d never had observed such low equity volatility before (and volatility impacts sequence risk). We’re seeing new phenomena all the time.

    Finally, it doesn’t help that even the concept of a fixed SWR violates the concept of stationarity. It’s fair to say the idea of fixed SWR is basically built on sand.

    And breathe …

  • 75 Vanguardfan February 2, 2020, 5:30 pm

    @zoe. If you are really going to get £50k from your NHS pension, then you’ll be looking at life time allowance issues and should therefore be cautious about putting your investments into a pension (although not excessively cautious, as it can still make sense even if you have exceeded the LTA).. However, it’s worth spending some time really getting to grips with the NHS pension scheme to make sure you understand it’s value and limitations, and where that £50k comes from. If it’s a forecast, what assumptions is it based on? That you continue to work until state pension age? And does it assume certain pay progression and full time working? (Also, if you’re only 34, then you won’t be able to access it unreduced until at least 68, and possibly later). You CAN access the NHS pension early, but it will be reduced to take into account being paid for longer. So have a look at that. There are also options for enhancing your NHS pension – have a look at those (but remember the life time allowance).
    I would say that a LISA and ISA should form the majority of your investments for earlier retirement. Don’t discount pensions completely, as if you are a higher rate tax payer the advantages are considerable, but I think you need a clearer understanding of those implications and the nature of the NHS scheme (TA’s work assumes no DB pension benefits).
    And remember that rules will certainly change over the next 30 years….

  • 76 The Accumulator February 2, 2020, 6:20 pm

    Hi all, I’ve edited the article above in light of Aleph, Naeclue and Oxdoc’s gentle correction of my earlier mistakes. My heartfelt thanks to them. With any luck we’ll crack this nut together.

  • 77 The Accumulator February 2, 2020, 6:50 pm

    @ ZX – I understand that you have great knowledge of the problems of inherent in modelling future uncertainties. Your knowledge is greater than the rest of us put together and your scepticism is heeded. Nonetheless, most of us need to retire at some point… 45, 55, 65, 75 whenever. We need to be able to cope with the uncertainty while acknowledging there are no guarantees and plenty of trade-offs to be made.

    This is where it’d be really helpful if you could post advice that would help steer the community.

    At some point, us mere mortals – like good parents – have got to plump for ‘good enough’ over ‘perfect’. My aim, The Investor’s aim, the majority of people here I think, want to plot a course that gives us a fair shot, without falling prey to the ‘Don’t worry be happy’ platitudes that pass for debate elsewhere.

    My appeal to you is to contribute in that spirit. To recognise that others don’t have your level of income, commensurate ability to fund future commitments, or rarified expertise but are definitely willing to learn and use the best advice out there to plot their own course.

  • 78 Al Cam February 2, 2020, 10:17 pm

    re “can one ever get …”
    See my comment #64 above from “Spoiler alert ..…… ” to “….. technique.”
    my comment #69 above from “In essence ….”

    BTW, I am not claiming any expertise.
    However, my conclusion after a considerable time reading about SWR’s, etc is that (like a lot of other things in life) there actually is no bullet proof “right answer”. In which case it makes sense to me to follow the strategy that best allows you to sleep at night!

  • 79 Far_wide February 3, 2020, 10:47 am

    Hear hear for @accumulator ‘s comment. This said, I really do appreciate the comments ZX, as you clearly have specialist knowledge in the area.

    Inflation is a funny one. I’m not going to claim to understand the ins and outs of ZX’s fears here, but what I do understand is that inflation is in reality very personal. From the last 4 years, despite no particular effort, my spending has gone down by about 5% nominal. This is despite Brexit knocking the £ (I live abroad mostly). Of course, that works the other way too, perhaps I will get some unfortunate medical condition requiring regular outlay or some other unforeseen expense. In short, isn’t 1% here or there quickly swallowed up by adjustments we all make to our lives each year?

    Aside from this, I think the reality of FIRE/SWR’s is so much muddier than portrayed. In reality, perhaps ZX is right, and 1% SWR is all we can aspire to, who knows? I think for so many though, FIRE is a springboard really, a confidence booster to live your own dream.

    In reality, anyone retiring at (say) 4o will not have zero other income again. You earn £100 cashback on a product from quidco, perhaps £125 from a bank giving you compensation for some minor hassle. You take a long flight that gets severely delayed and receive 600 EUR for your troubles. Eventually the state pension hits in (which we often don’t count) and/or inheritances are received. Maybe you do some freelance work on the side…..and on and on it goes. Yes there are ‘bad’ things too, but I would argue these are mostly outweighed by the good.

    All of the above said, I hope – hope – that we can find a way that more than a 1% SWR becomes realistic using investment alone. As otherwise, how does any normal person hope to ever retire with the demise of final salary pensions?

  • 80 Al Cam February 3, 2020, 11:39 am

    Chapter 12 of
    provides a review of “Quality change and new items in [UK] consumer price statistics”. The chapter covers, amongst other things, how quality changes are reflected both explicitly and implicitly.

    As it happens, I have actually spent quite a bit of time thinking about and examining our own personal inflation index. My motivation for doing this was similar to what you describe above.
    However, to date I have been really rather surprised by the headline outcome.
    Whilst it is verifiably true that our household weights are very different from the official weights, this has had remarkably little impact on “our CPI” or “our CPIH” vs the official figures over the last few years. This result may of course be particular to our circumstances at this time hence my plan is to continue to monitor how this develops.

  • 81 ZXSpectrum48k February 3, 2020, 11:55 am

    @far_wide. I didn’t say the SWR was 1%. I said the error bounds were probably +/- 1%. So if you think your SWR is 3%, then probably it’s more like 2%-4%. The SWR is a probability distribution, not a point forecast.

    The inflation issue is relevant because in terms of the “layer cake” that TA uses, the adjustment due to choice of inflation metric could easily be greater (up or down) than that caused by fees, taxes, asset allocation etc. Example: the UK 100% equities SWR (data from 1900, 50bp fees) is 3.10%. The UK 60/40 equity/bond SWR is 2.60%. So the difference between moving from an aggressive 100% equity portfolio to a more modest balanced portfolio is 0.5%. Note these all use RPI as the deflator. Now the difference between CPIH (1.4%) and UK RPI (2.2%) is currently 0.8%; the difference over the past decade 1.0%, over three decades 0.6% (CPIH doesn’t exist before 1989). So the choice of inflation metric is as material as the choice of asset allocation. Choose CPIH and you could argue you can add another 0.6%+ to your SWR.

    Now, nobody actually has the basket that is contained in official metrics like CPI, CPIH or RPI. Your own personal basket of goods/services may well be even more variable. I’d argue people need to focus as much on analysing the inflation characteristics of their spending than worrying whether their portfolio is 100/0, 80/20, 50/50 etc. I don’t see much of that though. Instead, inflation is treated as a defined discount rate inside the SWR calc with no uncertainty. Or just ignored.

    @TA. I have no particular issues with your approach. It’s far more balanced and refined than the MMM/TEA nonsense whose attitude is “25x expenses and you’re done”. I’m perfectly fine with not commenting further. I’ve only commented again because @far_side seemed to think I was arguing for a 1% SWR.

  • 82 never give up February 3, 2020, 6:59 pm

    I’m really enjoying the series. Thanks so much for putting it together. My set up is fairly simple probably like many of your readers. I just have a DC pension and ISA’s and that’s it.

    I had one query if that’s ok. In the first part you had 15 years to bridge and 12 years worth of expenses saved and that led to a 25% failure rate. In this third part you have 12.5 years worth of expenses to bridge 10 years. I assume this has a fairly significant positive impact on that failure rate, but by how much? You mention this ensures the ISA doesn’t run out but presumably this isn’t 100% successful?

    If following the liability matching approach to really try to guarantee success I assume £50k or so (made up number) may be needed additionally in the pension pot to help the total grow significantly to cover the lack of growth of such a low risk ISA pot? If I’m skipping ahead to a later post please ignore me. My very slow cogs are turning as I try to mull over my own cautious attitude and previous difficulties balancing my ISA and Pension contributions.

  • 83 The Accumulator February 3, 2020, 9:03 pm

    @ ZX – I love the fact that you comment and I’ve learned much from your insights over many moons. It’s just on occasion you’ll say something that leaves me thinking, “Great, on the one hand I’m buggered, and on the other I’m screwed.” I’m exaggerating for fun, but I suppose it comes down to – after accepting the imponderables and risks – we still need to find a path forward. Sometimes I think you underestimate how authoritative your voice is, and I worry that what you might consider food for thought, I might consider food for paralysis. It’s all very subjective and I’d love to see you write something longer on some of the problems you raise, but I appreciate you’ve got other things on. Anyway, your latest comment is very helpful in that I can understand the issue you’ve laid out and then decide how to incorporate it into my own plan.

  • 84 The Accumulator February 3, 2020, 9:12 pm

    @ Never give up – I’ll go into the data and assumptions I’ve used for these examples in the next post. I won’t ruin the surprise just yet 😉
    Liability matching can also be determined using a limited set of assumptions, and again, I’ll go into this a couple of posts down the line, including how to customise the calculations for your own situation.
    You’re right that no SWR guarantees success. It may have been historically successful but that doesn’t mean some future set of circumstances can’t rewrite the rules. WWIII could seriously lower our SWRs or perhaps a super Great Depression, or just choosing the wrong period of history using another country’s returns e.g. Germany post WWI.

  • 85 never give up February 3, 2020, 9:20 pm

    I feared I was jumping ahead. No probs and thanks for the reply. I’m really excited for the next post!

  • 86 Zoe February 3, 2020, 10:09 pm

    Thank you

  • 87 Zoe February 3, 2020, 10:13 pm

    Thanks for your reply. Yes I’m definitely going to have problems with the lifetime allowance. The total reward statement portal also doesn’t have any details of my pension contributions since 2015 either . It’s a bit of a minefield to sort. I’m just using Vanguard ISA to save currently and not putting any extra in pensions.

  • 88 Al Cam February 4, 2020, 10:25 am

    I would like to echo TA’s comment:
    “I’ve learned much from your insights over many moons”
    and add that it would be a great shame if that were to cease

  • 89 Chris February 5, 2020, 9:14 am

    So according to the data could I safely invest all of my SIPP into a Vanguard Life strategy 60/40 Fund and take a SWR of 4% for the next 20 years of retirement? If I could then I’m just going to open a SIPP with Vanguard and consolidate everything into the one fund and have done with it. What do you think?

  • 90 Aleph February 5, 2020, 6:46 pm

    No worries. Happy to have made a tiny contribution to this great series. Also, how awesome to have a civil conversation on the internet and actually agree something! This site rocks.

  • 91 Hare February 6, 2020, 10:43 pm

    Making a requested suggestion to add to the post.

    Pension contributions are tied to earnings. Therefore, a lot of us can only make low contributions to a SIPP/LISA (e.g. £2880 for ‘non earners’ plus £720 government top up). This massively changes the numbers and makes a pension not the golden end destiny. More like an ISA, which at the moment anyone can bung £20k pa into.

    I’m not complaining at the assumption and it’s a super good post and much appreciated. But it is an assumption and it would be good to have that basic thing recognised as income isn’t the same thing as earned income for pension contribution purposes.

    Could I also add a general comment that anyone can put money into a SIPP (if that has caveats, please add them). I discovered this on a financial press something somewhere (it may even have been here from some wise comment by ermine or similar) and did it with a family member via HL. It was great for tax planning – moving money from one family member’s taxable account into another family member’s SIPP who didn’t have the spare cash.

  • 92 W February 7, 2020, 3:05 pm

    The point made by @ZX is important in terms of very early FIRE.

    The nature of work will change dramatically over the next generations. Standard inflation measures can’t tell you whether your portfolio and SWR will generate enough to maintain your current socioeconomic position relative to the population, or over 40 years will you slip down the income scale?

    There are no easy answers, it’s the trade off for not working. But better to be aware of the “relative income” factor.

  • 93 Ecomiser February 8, 2020, 1:14 pm

    To those wondering how I live on £5k a year. I don’t know either, I just do it. I gave up trying to restrict my spending a few years ago but old habits die hard.
    It helps that I’m naturally frugal, live in a wholly-owned band A property in the North of England, don’t have a car or wife or children, or expensive hobbies, and can’t be bothered with foreign holidays. Heating only goes on if I’m feeling cold AND putting on an extra layer isn’t convenient.
    I’m fortunate that I can visit both the sea-side and some of the country’s most spectacular moorland for free with a bus pass and public service buses.
    I’m also lucky that my retirement began in 2009, so I’ve had a wonderful sequence of returns 🙂 but that’s actually irrelevant, since I’m not even spending all my SP.

  • 94 Aidan Williams February 8, 2020, 3:17 pm

    @ZX @W I appreciate this might be relevant to extreme early retirement but conversely, by normal retirement age the evidence suggests we spend less as we age so does planning for life long static spending parity (whatever your choice of parity is) make sense for most people other than the fact that it fits neatly into most retirement calculators/is easy to imagine? For instance, I plan to front load some income in the years prior to NRA and reduce further after 75. This of course complicates any guesstimate of portfolio depletion with the upfront boost very costly to long term performance.

    Things get even more complicated when earning in Pounds and spending in Euros – I use liability matching in Euro denominated savings bonds to shield me from day-to-day FX fluctuations (there’s been a few recently for some reason) but should I baseline my withdrawals in pounds or euros?

    Then of course there is tax to consider – if you are no longer UK resident then bye-bye PCLS, ISA capital gains exemptions and hello dividend tax… and oh dear, accumulation units (what was I thinking).

    The point is we can build as many complications into our retirement planning spreadsheet as we like but the bottom line is any model is still just a guess. And life will definitely throw you a few curve balls at some point. The allure of the 4% “rule” is its simplicity; it’s an easily actionable guess despite all the other factors at play. Although I prefer McClung’s Extended Mortality Updating Failure Percentage (EM/ECM) which adapts between a floor and a cap and of course expected longevity. It makes sense to spend more when there is fat and reduce down when times are lean. I might be really unlucky and live too long but it’s a risk I’m willing to take. 😉


  • 95 ZXSpectrum48k February 8, 2020, 4:22 pm

    @W. Yes. I’m trying to make people aware of the issue of relative income. Take a hypothetical example. Let’s look at 3 generations of early retirees in 1968, 1998, and 2008. Each one decides to use a 3% SWR that happens, given their respective portfolio, to produce a withdrawal amount that is equal to the average UK (pre-tax) wage in their retirement year on the basis that this should give them a decent standard living. As with any SWR prescription, they inflate it by RPI each year until 2018.

    So the retirees start by withdrawing £895, £8,853, and £22,672, in 1968, 1988 and 2008, respectively. How much are they all drawing by 2018? Well the 1968 retiree is drawing £15,242, the 1988 retiree £23,316 and the 2008 retiree £29,717. The average wage in 2018 is £27,040. So the 1968 retiree is now only drawing 56%, the 1988 retiree 86% but the 2008 retiree 110% of the UK average wage, respectively. The reason is obvious: between 1968 and 2018, earnings growth was 1.2% above RPI; between 1988 and 2018 only 0.5% above RPI; but between 2008 and 2018, earnings growth was 1% below RPI.

    So a 1968 retiree would have found that their standard of living, when compared to the working population, had fallen hugely behind that of the working population over their 50 year retirement. The RPI cost of living adjustment wasn’t anywhere near large enough. Conversely, for the 2008 retiree is was too conservative.

    Now this assumes you are targeting earnings rather than cost-of-living i.e. you care about your relative income vs. the broader population. You might not care. Many will though. In fact, most studies suggest relative prosperity is more important to wellbeing than absolute prosperity, at least beyond a certain floor. Equally, while the US tends to use CPI as it’s deflator for social security, many other countries use earnings growth. The UK state pension uses a triple lock: the higher of RPI, earnings and 2.5%.

    The use of RPI as the deflator in UK SWRs is just an assumption and not necessarily a good one. The variations caused by choosing a different deflator are substantial. In particular, FIRE optimists should never rerun the historical SWR analysis from say 1945, replacing RPI with earnings growth. They would not like the SWR at all!

  • 96 The Accumulator February 8, 2020, 7:01 pm

    @ ZX – that’s fascinating and the first time I can remember seeing this issue even mentioned, never mind confronted. Would be interesting to speculate on the future of earnings growth in light of disempowerment of trade unions, apparent squeeze on middle class earnings, especially in the US over the last 40 – 50 years, effect of globalisation and so on. Futile but interesting 😉

    This next bit comes from a highly subjective perspective. I offer it up in the spirit of debate. My take on FI isn’t purely based on calculating that I can live on, say, £25K today, hence I model how much I need to live on £25K for the rest of my life.

    I see FI as a chance to live life on new terms. I want to focus on quality of life not standard of living. I want to walk away from the hedonic treadmill. For example, I don’t want to upgrade to an 8K TV. I want to wild camp in the hills, or learn how to grow my own food, or learn carpentry skills [insert your own vision of the Good Life here].

    You can buy a TV for £20 today. You just have to accept it’s not going to be HD. I can well imagine a 1968 retiree wondering why people need half the crap that people deem ‘essential’ today. My baby-boomer parents seem quite happy without a smart phone, though I feel like my right-arm has been amputated when I leave the house without it.

    90% of the time I spend thinking about FI is on reimagining life, not the numbers. I’ve been conditioning myself to value time over possessions, overthrow status anxiety, experiment with resilience over convenience, establish a set of values based on a life swimming away from the mainstream.

    All very subjective but relevant as we’re ostensibly talking about early retirement versus traditional retirement.

    For those who’d like to model using historic cost-of-living metrics, what’s your go-to source?

  • 97 The Investor February 8, 2020, 7:43 pm

    @TA: “That’s fascinating and the first time I can remember seeing this issue even mentioned, never mind confronted.”

    I knew you were not really ‘multi-tasking’ the last time we spoke about this. 😉

    From my 2008 ‘one number’ replace-your-salary article:

    Won’t you need more than your salary in 10 years time, due to inflation? Good point. You’ll need an income that is keeping up with the inflation rate to maintain your purchasing power – or better yet, increasing with wage inflation, to keep up with the neighbors.


    The article isn’t at all focused on the point, but it is another reason why I want to target income/natural yield. But this will await my own article. 🙂

  • 98 The Accumulator February 8, 2020, 8:59 pm

    @ Aiden – thank you for the link to the ‘spending less as we age’ article. It’s an important point. There’s a certain swings and roundabouts aspect to all of this.

    @ The Investor – ‘Thanks’ to you too. The key word was: “first time I can *remember* seeing this issue even mentioned…”

    If only you wrote more memorable articles 😉

    Get on with your natural yield article. I can’t wait to troll it.

  • 99 The Accumulator February 9, 2020, 9:00 am
  • 100 ZXSpectrum48k February 9, 2020, 11:23 am

    @TA. I have huge sympathy with the idea that average wage growth will lag inflation or asset returns over the next few decades. There are two issues with that though. First, historically that just hasn’t been the case. Since 1900, earnings growth has outperformed RPI by about 1%. Second, it’s inconsistent to use historic data for asset returns, RPI and then junk it for earnings. What if asset returns are related to earnings growth? You have a few periods, such as late Gilded age (before WW1) and the last decade, where you saw a major divergence between the real economy and asset prices but that was the exception, not the norm.

    As you’ve already linked too, the ONS has data on some of these things. There is also this site https://www.measuringworth.com/datasets/ukearncpi/. I can’t vouch for the data but the timeseries look reasonable.

Leave a Comment