Let's do it as a 30-year transition model from 2027 to 2057. With tightened assumptions, we can see both the annual and cumulative effects.

The starting point is the latest Whole of Government Accounts: £1,400.9bn unfunded gross pension liability, £23.8bn current-service cost, and £60.4bn of benefits paid by unfunded schemes in 2024-25. HM Treasury stresses that the £1.4tn figure is a discounted present-value liability, not a cash bill.

Assumptions

For this illustration:

• Reform begins in 2027.
• Existing accrued pension rights are protected.
• Future accrual is reduced by 20%, 30% or 40%.
• We retain the earlier illustrative assumption that 30% of the eventual liability relates to future service. This is an assumption, not an HMT estimate.
• The effect on actual pension payments builds gradually as people with pre-2027 service retire.
• I use £60.4bn as the 2024-25 benefit-payment baseline, held constant in today's money. This deliberately avoids pretending we know future inflation.
• I assume the proportion of payments affected by the reform rises from 0% in 2027 to 100% by 2047. This is a simplification rather than an actuarial projection.

That last assumption is particularly important: the cash-saving figures below are illustrative, not an official forecast.

1. 20% reduction in future accrual

The eventual liability reduction under our assumption is about £84bn.

2. 30% reduction in future accrual

Eventual liability reduction: approximately £126bn.

3. 40% reduction in future accrual

Eventual liability reduction: approximately £168bn.

4. Put the three together

This is probably the clearest way of looking at it:

*Using the assumed 30% future-service component.

There is an apparent paradox here

You can see that the cumulative cash saving by 2057 can be much larger than the reduction in today's measured pension liability.

That's not an error.

They're measuring different things:

• £168bn is an estimate of the reduction in the present value of future pension obligations.
• £495bn is the sum of the hypothetical annual cash reductions over 30 years.
• Future cash payments occur at different dates and therefore have different values today.
• The accounting liability is discounted.
• Our cash model deliberately doesn't discount those future savings.

So the two numbers should not be added together.

5. But there is an important problem with the cash model

I would not yet regard £24.2bn a year in 2057 as a credible forecast.

It is useful as a scenario, but there is a major missing piece: the age and service distribution of the public-sector workforce.

For example, suppose someone has:

• 25 years of pre-2027 service, and
• 10 years of post-2027 service.

A 40% reduction in future accrual does not reduce their entire pension by 40%.

Only the pension earned during those 10 post-reform years is affected.

Conversely, a person who begins public-sector employment in 2027 could have virtually their entire pension affected.

Therefore the real transition is a cohort problem.

6. What the official figures tell us

The WGA actually gives us a useful warning about this.

In 2024-25:

• unfunded current-service cost: £23.8bn
• interest on unfunded liabilities: £65.6bn
• unfunded benefits paid: £60.4bn
• participant contributions: £16.8bn
• Unfunded liability at year end: £1,400.9bn

That means the pension liability isn't simply:

£1.4tn − £60bn every year.

Interest, new accrual, contributions, benefits and actuarial changes all interact.

Indeed, HMT says that the liability can change substantially because of discount-rate and demographic assumptions without changing the benefits actually received by members.

7. The really interesting comparison

We can now compare three quite different approaches.

Cut existing pensions by 40%

Mechanical accounting effect:

£1,400.9bn -> £840.5bn

Reduction:

£560.4bn

Cash expenditure would also fall relatively quickly because people already receiving pensions would receive less.

Protect existing pensions, reduce future accrual by 40%

Our illustrative result:

£1,400.9bn -> ~£1,232.8bn

Reduction:

~£168bn

But the cash saving builds over decades.

Protect existing pensions, reduce future accrual by 20%

£1,400.9bn -> ~£1,316.9bn

Reduction:

~£84bn

And the cash effect is considerably slower.

8. One further refinement would make this much more realistic

The next calculation I'd suggest is not to assume the 20-year straight-line transition.

Instead, we can construct a simple age/cohort model:

• public-sector employees aged 20-29
• 30-39
• 40-49
• 50-59
• 60+

existing pensioners

Then apply the 20%, 30% and 40% accrual reductions only to service after 2027, allowing each cohort to retire at an assumed age.

That would give us something much more useful:

2027 -> 2032 -> 2037 -> 2042 -> 2047 -> 2052 -> 2057

with, for each scenario:

• annual pension payments
• annual government cash saving
• cumulative cash saving
• pension liability
• reduction in liability
• number/proportion of pensioners affected
• and, importantly, the point at which the annual saving becomes substantial.