How should we think about life expectancy when modelling retirement?

I've been looking at a few real retirement plans recently, shared with me by people who have a financial adviser. Two of them assumed the money only needed to last until age 87 or 88. Most of the others assumed both partners would live to exactly 100.

For context, I'm one of the founders of Towpath. We're building a new financial advice firm, and one of my tasks this year has been to build our cashflow modelling engine. One of the first questions I had to answer was what age to run each projection to. But it quickly became clear that was the wrong question.

Picking an age

Most cashflow models I've seen pick an age and assume everyone lives exactly that long. Often it's 100, for both partners in a couple. The logic is simple enough: if the money lasts to 100, it'll last for any shorter life too.

Some plans use average life expectancy instead, like the two that ran to 87 and 88. The FCA found this in its 2024 review of retirement income advice, and flagged it as a concern. One firm had told a customer there was "a 95% chance the fund would last until 'average' life expectancy".¹ This sounds reassuring, until you remember that half of people live longer than average.

The natural response to this is to plan far into the future. 100 seems far enough. But I don't think 87 was a problem because it was too low. It was a problem because it was one number. It ignores every other way a 20- or 30-year retirement could play out.

What the numbers say

The best data comes from the Office for National Statistics (ONS), which publishes how likely someone is to reach each age, based on their sex and the year they were born.

As an example, let's take a couple I'll call Anne and David. They're both 65, retiring this year, and most of their guaranteed income will be David's: a final salary pension plus his State Pension.

If you asked someone in the street what UK life expectancy is, they'd probably say about 80, but that's the figure at birth. By 65, Anne and David have already made it through the risks of childhood and working life, and the odds look quite different. David has a 50% chance of reaching 86, and Anne of reaching 89.

Source: ONS 2022-based cohort life tables, UK, for a man and a woman born in 1961. Ages over 100 extended by Towpath.

Put the two of them together and it's even more likely that one of them lives a long time. There's a 64% chance at least one of them reaches 90, and about a 33% chance one of them reaches 95.

But the chance that they both reach 100 or more is under 0.2%, about 1 in 600.

Source: Towpath calculation from ONS 2022-based cohort life tables, UK, treating their lifespans as independent. Ages over 100 extended by Towpath.

Why one age is too simple

That's the trouble with assuming both partners live to 100. It's treated as the worst case, but really it's just one possible future out of thousands, and one of the least likely. It gets two things wrong.

It makes the trade-off for them. Planning to 100 assumes Anne and David will both need an income for another 35 years. It's sensible to be ready for that, but it's far from the most likely outcome. And the more that's held back for it, the less they have to spend or give away now. By picking 100, the adviser has made that choice for them. Anne and David never even see that there was a choice to make.

Holding back for a long life might well be the right call. But it should be their call, made knowing what the trade-off looks like.

And it misses the futures that actually hurt. For Anne and David, the danger isn't them both living to a great age. It's David dying early.

When one partner dies, the household's income often falls much further than its costs. Anne would inherit little or none of David's State Pension, and his final salary pension would drop by half. But the bills don't halve. (For couples who aren't married, it can be even harder. The first death can be treated very differently for tax and inheritance, and the survivor may not get a spouse's pension at all.)

So if David dies at 75 and Anne lives to 95, she faces 20 years on a much smaller income, with most of the same bills. Modelling both to 100 never tests that, because nobody dies early. Many advisers would run a separate scenario where David dies first. But that's still one age picked by hand, with no sense of how likely it is.

Testing a plan against thousands of lives

Rather than picking an age, a better approach is to test each plan against thousands of possible futures. (You might know this as Monte Carlo simulation, but that makes it sound more complicated than it is.)

In each future, Anne and David are each given an age at death, drawn at random from the ONS odds for someone of their sex and age. In some futures David dies at 72. In others they both reach their late 90s. Investment returns and inflation vary between futures too, but I'll stick to lifespan here.

Then we count how many of those futures run out of money. So the question stops being "does the plan work if they both live to 100?" and becomes "in how many of their possible futures does it work?"

Suppose we built a plan to last until they're both 100, which feels like the safe option. If we ran that against their real range of lifespans, it would run out of money about 10% of the time.

That sounds odd. Surely a shorter life needs less money? But if we look inside those failing futures, most of them are where David dies early. Much of their income dies with him, and Anne is left with most of the same bills. That's the future planning to 100 never tests.

The averages aren't the person

The ONS figures are for the general population and don't take into account the individual. Someone with a serious health condition will probably have less time. Someone whose parents both lived to 98 might have more. And people who get financial advice tend to be wealthier, and wealthier people tend to live longer.

The temptation is to fix this by picking a different age. Mum lived to 98, so plan to 102. But that's the same mistake again: one number standing in for a whole range of possibilities. We're not actuaries or doctors, so I don't think it's our place to adjust the odds on gut feel. The better answer is to start from the ONS figures, be open that the person in front of you might not fit them, and talk it through. The odds can be updated at each annual review too, once they're a year older.

Care makes this harder still. The longer someone lives, the more likely they are to need it. But that's one for another post.

The hard part is talking about it

All of this only helps if you can actually have the conversation, and it's not an easy one.

Asking someone how long they expect to live is uncomfortable. Most people would rather not think about it, and advisers don't want to seem morbid. That discomfort can show up in ways that are easy to misread. A customer who keeps changing the subject isn't necessarily being difficult. They might just be deflecting from the thing the conversation is really about.

People also tend to underestimate their own chances. The Institute for Fiscal Studies found people in their early 60s underestimate their chance of reaching 75 by more than 25 percentage points. So "I won't be around that long" is worth gently challenging.

It's harder again with couples, talking through what happens if one of them dies first while they're both sat there. I think it helps to talk about the plan, not the person. "The plan needs to work for whoever's left" is a lot easier to hear than "when you die".

Once the range of possibilities is on the table, the trade-off from earlier becomes a real choice. Anne and David might want to spend more in their late 60s and 70s, while they're most likely to be healthy enough to enjoy it, knowing they may need to cut back if they both live a long time. Or they might rather hold more back. Either is fine, as long as they're making an informed decision.

It also opens up questions that picking a single age quietly hides. If living a long time is the worry, an annuity is worth discussing, because it hands that risk to an insurer. If the worry is David dying first, a joint-life annuity keeps paying an income to whoever is left.

So how should we think about life expectancy?

For me, it starts with accepting that nobody knows how long they'll live, and planning for the whole range of ages they might reach, not just one.

Planning to 100 feels prudent, but it's too basic. It silently builds the whole plan around one potential future of many, and fails to spot the scenarios most likely to cause trouble. I suspect part of the reason it's so common is that most of the popular cashflow tools run a plan to one age that you choose.

How long the money needs to last is one of the biggest unknowns in anyone's retirement, and it shapes almost every other decision in the plan: how much to spend, how much to give away, how much to hold back. Pick a single age and those trade-offs get made for the customer, often without them knowing. I'd much rather an adviser can talk about the odds and the implications, and help customers make an informed decision.

If you're a financial adviser, I'd love to hear how you approach this. What age do your plans run to, and how did you settle on it? How do you talk to a couple about the chance that one of them dies early? And do you adjust for someone's health or family history? Send me a message on LinkedIn.