“You can retire at 60.”
That sentence is not a fact. It is an output. Somewhere behind it sits a spreadsheet or a planning tool with a column of numbers, and those numbers were chosen by someone — possibly you, possibly an adviser, possibly a default setting nobody has looked at since the file was created. The output is only as trustworthy as the weakest assumption feeding it.
This article is not about whether you personally can retire at 60. It is about how to read the model that claims to know. The useful question is not “what does it say?” but “what would have to be true for this to hold, and how would I know if it stopped being true?”
Three layers inside every model
Any retirement cash-flow model contains three kinds of input, and they deserve very different levels of trust.
Layer one: verified external facts. These are rules and data published by official sources. They constrain what is possible. They do not determine whether your retirement date works.
Layer two: personal inputs only you can supply. Your spending, your stop-work date, your balances, your State Pension forecast. These are checkable in principle, but often estimated loosely in practice.
Layer three: forward-looking guesses. Investment returns, inflation paths, future tax rules, care costs, how long you live. These are the assumptions most likely to be wrong, and they are usually presented with the most confidence.
The uncomfortable truth is that the model’s answer is dominated by layer three, and layer three is the part nobody can verify in advance.
What the checkable layer actually says
Start with what can be looked up. It is less than you might hope, but it is solid.
State Pension eligibility. You need 10 qualifying years on your National Insurance record to get any new State Pension at all, and the qualifying years on your record affect how much you get. Eligibility for the new State Pension depends on your date of birth: broadly, a man born on or after 6 April 1951 or a woman born on or after 6 April 1953 falls under the new State Pension; those born earlier fall under different rules. These are the constraints. They tell you whether you get a State Pension and roughly how the amount is built. They do not tell you what it will be worth in 2045, because uprating is a future policy decision. Source: GOV.UK, The new State Pension.
Longevity. ONS national life tables for 2022 to 2024 put life expectancy at age 65 in the UK at 21.2 years for females and 18.7 years for males. That is period life expectancy — it describes the average number of years people would live if current mortality rates applied throughout their lives. It is not a prediction for you. ONS also publishes cohort projections, which attempt to account for future improvements in mortality: girls born in the UK in 2024 can expect to live to age 90.2 years on average, and boys to age 86.9 years, on that basis. Source: ONS, Life expectancies.
Here is the trap. A model that runs to age 90 for a 60-year-old is not being conservative. It is being roughly average for one person, and it is almost certainly too short for a couple, because the probability that at least one of two people lives past the average is high. If your model assumes both partners die at 90, it is quietly assuming a joint outcome that is less likely than the individual outcome it was drawn from.
Inflation. The Government has set the Bank of England a target of keeping inflation at 2%. That is a target, not a forecast, and not a guarantee. The Bank’s own framework acknowledges deviation: if CPI inflation is more than 3% or less than 1%, the Governor writes to the Chancellor to explain why and what will be done to return it to target. Source: Bank of England, Inflation and the 2% target.
So when a model uses 2% inflation for thirty years, it is using the target as though it were the outcome. That is a choice, not a fact. A model that cannot show you what happens at 3% or 4% is not stress-tested; it is decorated.
Where the answer is actually decided
The checkable layer tells you the rules of the game. The answer to “can I retire at 60?” is decided in layer three, and specifically by four assumptions that are rarely shown as ranges.
1. The return path, not the average return
Most models use a single average return, often somewhere between 4% and 6% nominal for a balanced portfolio. That number is not the problem. The problem is the order in which returns arrive.
Two portfolios can have identical average returns over thirty years and produce completely different outcomes if one suffers a poor first decade while the other enjoys a strong one. This is sequence-of-returns risk, and it bites hardest in the first ten years of drawdown, when withdrawals are converting a temporary loss into a permanent one.
A model that shows only the average return is not showing you the risk. Ask for the path: what happens if the first five years deliver zero or negative real returns, and withdrawals continue as planned?
2. The withdrawal rate and its flexibility
If the model assumes a fixed real withdrawal — say £40,000 a year, rising with inflation — it is assuming a spending pattern that most real retirees do not follow. Spending tends to decline in real terms through the middle of retirement and rise again if care is needed. A model that holds spending flat and rising may be overstating the required pot, or understating the late-life cost, depending on which end you look at.
More importantly: does the model allow spending to be cut if returns disappoint? A plan with a 4% withdrawal rate and no flexibility is fragile. A plan with a 4% starting rate and a stated rule — “if the portfolio falls below X, we cut discretionary spending by Y” — is a different proposition. The flexibility is the plan.
3. Fees, and how they compound
A 1% annual charge on a £600,000 portfolio is £6,000 a year. Over thirty years, with no growth, that is £180,000. With growth, the opportunity cost is larger, because the fee is levied on a rising balance and the money removed would itself have compounded.
Models often bury fees in a net return assumption. That is convenient and opaque. Ask for the gross return, the fee, and the net return as separate lines. If the person who built the model cannot produce them, the model is not a planning tool; it is a black box.
4. The bridge to State Pension age
If you stop work at 60 and State Pension age is 67 or 68, you have a bridge period of seven or eight years with no State Pension income. The model must fund that bridge from pensions, ISAs, or other assets, and it must decide how.
This is where wrapper decisions and tax assumptions meet. Drawing from a pension before State Pension age may be tax-efficient or not, depending on your other income and the allowances in force at the time. Drawing from an ISA is tax-free but depletes a wrapper you may want later. The model’s answer depends on assumptions about tax rules that do not yet exist for the later years of the plan.
No retrieved source in this article supports specific tax rates or allowances for future years, and none should be invented. The honest position is that the model is using current rules as a placeholder for unknown future rules. That is reasonable, provided it is labelled as such.
A worked illustration, clearly hypothetical
The following is not a client case and not a research finding. It is a hypothetical to show how one assumption moves the answer.
Suppose a couple, both 58, with £700,000 across pensions and ISAs, want to stop work at 60 and spend £45,000 a year in today’s money. A model using a 5% nominal return, 2% inflation, and a 30-year horizon might show the pot surviving to age 90 with room to spare.
Change one assumption: the first five years deliver a 15% real fall, and withdrawals continue. The same model, run with that path, may show the pot exhausted in the late 80s, or require spending to drop to £38,000 for a decade to recover. The average return is unchanged. The answer is not.
Change another: inflation runs at 3.5% rather than 2% for the first decade. The required pot rises materially, because the spending line rises faster than the return line. Again, the average return assumption is untouched.
This is why a single number — “you can retire at 60” — is not a plan. It is one path through a range of possible paths, and the range is the useful information.
What to ask whoever built the model
These questions are not adversarial. They are the minimum standard for treating an output as a decision input.
- Show me the assumptions as a list. Every return, inflation, fee, tax, longevity, and spending figure, on one page, in plain language.
- Show me the ranges, not the points. What happens at the 25th percentile and the 75th, not just the central case?
- Show me the bad first decade. What does the plan look like if the first ten years deliver below-average real returns?
- Show me the joint longevity case. What is the plan’s horizon for two people, not one?
- Show me the fees separately. Gross return, total cost, net return. Three lines.
- Show me the fallback. If returns disappoint, what changes — spending, work, or both — and at what trigger point?
- Show me what would falsify the conclusion. What would have to happen for “retire at 60” to stop being true, and how would we know?
If the model cannot answer question seven, it is not a plan. It is a projection, and projections are not decisions.
The uncomfortable part
The assumptions most likely to decide your retirement date are the ones you cannot verify. Returns, inflation, tax rules, care costs, and how long you live are all guesses. Some are educated guesses. None are facts.
That does not make modelling useless. It makes the model a tool for asking better questions, not for producing a verdict. The verdict — “yes, 60 works” — is a judgement about how much uncertainty you can tolerate and how flexible you are willing to be. No spreadsheet can make that judgement for you.
What the spreadsheet can do is show you the trade-offs. If you want to retire at 60, here is the spending level that works at a 4% return and the spending level that works at 3%. Here is the pot you need if you live to 85 and the pot you need if one of you lives to 95. Here is what happens if fees are 0.5% and what happens if they are 1.2%.
Those are decisions. The single sentence is not.
Where this connects to the wider plan
The cash-flow model is one decision moment among several. It sits alongside the adviser relationship audit — what you are paying, what the suitability report actually says, whether the firm appears on the FCA Register — and the family paperwork that quietly decides what happens to the money: expression of wish forms, lasting powers of attorney, gifting, and care funding.
None of those are substitutes for each other. A model that says “retire at 60” is not a substitute for an expression of wish form that says where the pension goes if you die at 62. A withdrawal plan is not a substitute for an LPA that lets your spouse manage the accounts if you cannot.
If you are at the stage of interrogating the model, the related question — the question you should be asking years before you retire — is worth reading alongside this. It is about the decision that precedes the spreadsheet: what you are actually retiring to, and what the money is for.
FAQ
Does the model need to use the Bank of England’s 2% target for inflation?
No. The 2% figure is the Government’s target for the Bank of England, not a forecast. A model can use it as a central assumption, but it should also show what happens at higher and lower rates. If it only shows 2%, it is not showing you the range of outcomes.
How long should the model run for?
There is no single correct answer. ONS period life expectancy at age 65 was 21.2 years for females and 18.7 years for males in 2022 to 2024, but that is an average, not a maximum. For a couple, the relevant question is the probability that at least one person lives well beyond the average. A model that runs to 90 for both partners may be understating the joint longevity risk.
What if the model says I can retire at 60 but I don’t believe it?
Ask for the assumptions. If they are shown as single point estimates with no ranges, the model is not giving you enough information to decide. If they are shown as ranges and the conclusion holds across most of them, the model is doing its job. If the conclusion only holds at the optimistic end, you have your answer.
Should I use a model at all?
Yes, provided you treat it as a tool for exploring trade-offs rather than a source of truth. The value is in the questions it forces you to answer: what you spend, when you stop, how flexible you are, and what happens if you are wrong.