Fiscal risks from public service pensions

Central government pension schemes are unfunded statutory defined benefit (DB) pension schemes. They cover most government employees and provide pension benefits based on salary and length of service. Most public service schemes operate on a ‘pay-as-you-go’ basis, meaning there is no fund of assets which is invested. Instead, contributions from current employers and employees are paid into schemes and used to (partially) cover benefits to current retirees. As total contributions and the amounts paid to current pensioners may differ, an annual balancing payment is made by the Treasury to schemes to cover any shortfall, while any surpluses are returned. These ‘pay-as-you-go’ schemes, which had liabilities estimated at £1.4 trillion as of the end of 2024-25, are not included as liabilities for the purposes of PSND or PSNFL but are on the balance sheet for PSNW.^a Other DB pension schemes, like the Local Government Pension Scheme, are examples of funded pension schemes, and are recorded in PSNFL with around £550 billion of both assets and liabilities in 2024-25.

Unfunded pension liabilities represent the second-largest government liability after gilts but are not included in PSNFL.^b This statistical treatment is in line with the European System of Accounts 2010 (ESA10) followed by the ONS for most public finance statistics, where obligations under unfunded schemes are considered to be contingent and are therefore not recorded on the public sector financial balance sheet in the core UK National Accounts publications. Treating unfunded pension obligations as contingent liabilities, while those of funded schemes are treated as concrete liabilities, could be seen as counterintuitive as the Government’s legal obligation to members of funded and unfunded pension schemes is essentially the same. ESA10 justifies this by observing that unfunded schemes are similar to the contribution-based social security schemes (state pension schemes) that operate in many countries. This is an example of a limitation of balance sheet analysis in the public sector context where, in addition to unfunded pensions, other future spending streams – many of which are near-inevitable commitments such as the state pension, health, and education – do not appear on the balance sheet as liabilities, while on the other side, the most valuable asset a government has – the ability to raise taxes – is also not counted as such.

Because of the ‘contingent’ nature of unfunded pension liabilities, the associated flows only impact aggregates such as public sector net borrowing (PSNB) when they are actually paid rather than when the obligation to pay is incurred. The net impact on PSNB of these schemes is therefore the net cashflow of benefits paid to retirees minus contributions from employees. This treatment can lead to counterintuitive financial impacts. If the Government hires more employees, as has happened in recent years, then the extra contributions from those employees will produce a positive medium-term impact on PSNB under a cash treatment. If an accruals treatment were used, then borrowing would be negatively impacted due to the value of the increased pension rights earned by those workers (which historically exceeds their and their employers’ contributions).

The impact of this cash (rather than accruals) treatment of unfunded pensions on the public finances can be determined by estimating the accrued borrowing impact. The ONS publishes a reconciliation between its Public Sector Finances data and borrowing using accounting guidance under the IMF’s Government Finance Statistics manual, which includes an estimate of this impact. Chart A illustrates this reconciliation on a stocks and flows basis and shows that borrowing would have been, on average, 1.6 per cent of GDP higher under an accruals methodology relative to the current cash treatment, while unfunded pensions would have increased balance sheet liabilities by an average of 50 per cent of GDP between 2013-14 and 2023-24.^c The sharp increase in both stocks and flows in 2016-17 was driven by a change in the ONS discount rate from 5 per cent to 4 per cent. The temporary spike in the stock in 2020
was driven by the impact of the sharp fall in nominal GDP during the pandemic. The recent fall in the stock impact has been driven by the sharp increases in nominal GDP due to high inflation, which has only been partially reflected in increased liabilities.

Chart A: Reconciliation of ESA10 and GFS flows and stocks of unfunded pensions

SCAPE is the process that is followed to set employer contribution levels for unfunded public service pension schemes.^d For funded DB pension schemes, contribution levels are set to meet the cost of expected benefits through valuations. These valuations use a discount rate set with reference to the return expected from the assets held by the scheme to determine the appropriate contribution level into the schemes. This will reflect both an adjustment to any mismatch between assets and liabilities that has already built up, and the cost of future benefit promises. As unfunded schemes do not have assets to pay pension benefits, they are instead funded through contributions by government as the employer and from the employees themselves. As a result of this lack of assets, a different process is needed to calculate contribution rates – this is SCAPE.

As a part of SCAPE, a discount rate (known as the ‘SCAPE rate’) is applied to the schemes’ expected future pension payments so that the cost of pension promises being built up can be expressed as a present-day cost. The choice of discount rate, which is set by the Treasury, can have a significant impact on employer contribution levels.

Since 2011, the SCAPE rate has been based on the OBR’s expectations for long-term GDP growth. The Treasury has stated that discounting public sector pensions using the expected long-run growth rate of nominal GDP would ensure that contribution levels reflected future affordability constraints. However, one risk to this approach is that if GDP growth is persistently lower than projected, liabilities will be underestimated, and contributions will be too low. This could lead to a change to the SCAPE rate and an increase in contribution levels. For example, the decision to change the SCAPE rate from CPI+2.4 per cent to CPI+1.7 per cent in March 2023, following a change to the OBR’s long-term growth assumption, increased employer contributions by an average of £5.6 billion a year from 2024-25 to 2028-29. This increased pension scheme receipts and led to a subsequent fall in net unfunded pension spending of
£23.6 billion a year on average over the same period. However, this had no impact on PSNB as the Treasury exactly compensated departments for the cost of the increased employer contributions.

These recent developments illustrate that the impact of changes such as this on the public finances depends on Treasury decisions on how to fund any such increase in contributions. Because these contributions are essentially an accounting movement inside the public sector (from public sector employers to public sector pension schemes) they do not automatically increase total government spending or borrowing. If the overall departmental spending envelope
were fixed, then an increase in pension contributions would reduce the proportion of a department’s spending allocation available to deliver public services. However, in practice the Treasury has usually compensated departments whose employment costs are centrally funded through departmental expenditure for changes in contribution rates due to SCAPE, leaving their overall spending power (and the amount of public sector spending overall) unchanged. This
therefore results in no impact on borrowing or public spending, despite a change in discount rate reflecting a change in outlook for the affordability of unfunded pensions.

Ultimately, the real test of the affordability of these pensions is the likely trajectory of gross payments over time set against the tax base that will finance the payments. In the long-term projections in the 2024 Fiscal risks and sustainability report, we estimated that annual payments out of the schemes would fall from 1.9 per cent of GDP in 2023-24 to 1.4 per cent of GDP in 2073-74. This is based on the assumption that contributions, which are linked to average earnings, rise more quickly than payments, which are assumed to be uprated by CPI inflation. This suggests that if these assumptions hold, then these schemes do not pose a significant fiscal risk in themselves, but they do make up a significant share of the Government’s overall liabilities which are projected to continue to rise over the next 50 years.