What ‘Before the Vote’ gets right and wrong about the subvention…

The first chapter of John Doyle’s Before the Vote: Planning for a united Ireland examines the Northern Ireland subvention, the gap between taxes raised in Northern Ireland and the public spending allocated to it. Doyle starts from the ONS figure of £14.2 billion for 2022/23 and concludes that a united Ireland would inherit a deficit of about £1.5 billion, which he calls the “remaining subvention”. He is right that much of the published figure would not transfer, and his critics agree on several of his adjustments. But most of the reduction rests on two assumptions about how negotiations with the UK would end; several of the smaller adjustments appear to contain errors, and his rebuttal of the main alternative estimate misstates its figures.

What the subvention measures

The ONS allocates UK revenue and spending to each region. Some of the spending allocated to Northern Ireland occurs elsewhere, including £1.57 billion as a share of UK defence, £3.53 billion as a share of interest on the national debt, and £711 million spent outside the UK. Doyle argues that a united Ireland would not inherit these costs at the same level, and that some taxes are undercounted in Northern Ireland because companies pay them through head offices in London.

On this ground, he and his critics largely agree. John FitzGerald and Edgar Morgenroth, whose 2024 note for the IIEA Doyle criticises, also scale defence and foreign spending to the Republic’s levels and add corporation tax for profits earned in Northern Ireland. On their method, the 2019 subvention of £10.7 billion falls to £9.6 billion before any change to pay or welfare. The disagreement is about the items that remain.

Two assumptions carry the result

Doyle’s Table 1.2 removes £12.7 billion from the published deficit. Of that, £9.3 billion comes from two items: £5.7 billion for pensions, on the assumption that the UK would keep paying pensions earned before unity, and £3.5 billion for debt interest, on the ground that a united Ireland would have no legal liability for UK debt. Doyle calls the pensions question “a political judgement” and argues both points at length. Pensions are already paid across the border under the UK–Ireland social security convention, he says, and the UK, as the continuing state, would remain liable for all its debt.

FitzGerald and Morgenroth take the opposite view. They argue that state pensions are paid from current contributions, that the UK paying pensions while a united Ireland collected the contributions is “most improbable”, and that the Irish side would be in “a very weak negotiating position” after a vote for unity. On debt, the UK would remain liable to its creditors, and Ireland’s share of the cost would have to be negotiated, although the UK could seek compensation elsewhere in a settlement. On pensions, the dispute is who would pay benefits financed by current contributions. The social security convention protects individuals’ entitlements and contribution records, but it does not settle how the two governments would divide the cost of Northern Ireland’s existing pensioners after unity. Doyle’s £1.5 billion depends on the UK continuing to meet existing pension costs and on a settlement that leaves Ireland without a share of UK debt-servicing costs. Neither outcome has been agreed.

Doyle also gives a ceiling. If the UK refused to pay any pensions, he says, the remaining deficit would be “a maximum £5.9 billion”. On the unchanged 2022/23 basis used in Table 1.2, the figure would be £1.49 billion plus £5.72 billion, or about £7.2 billion. Doyle’s £5.9 billion depends on a separate assumption that pension costs would fall to £4.4 billion within two or three years as public servants pay higher contributions. The fall of about £1.3 billion in that projection appears to rest on the Treasury measure of public service pension costs discussed below, which the regional figures do not include.

The pension deduction

The pension deduction is itself too large. Doyle’s £5.72 billion combines an estimate of £1.7 billion for public service pensions with “just over £4 billion” for state pensions, taken from the Treasury’s Public Expenditure Statistical Analyses 2023. The Treasury’s line for “old age” spending in Northern Ireland in 2021/22 is £4,014 million, and adding Doyle’s £1.7 billion gives his total of £5,722 million, suggesting this line underpins his figure. His note says the state pension cost is “provided directly” in the Treasury report, where the old-age total of £4,014 million sits above a pensions line of £3,386 million. Of that £4,014 million, £628 million is personal social services, the care provided to older people; the pensions sub-line is £3,386 million. Care services are not pensions, and nothing in Doyle’s argument suggests the UK would pay for them after unity. The Treasury figures are also for 2021/22, a year earlier than the deficit they are subtracted from. For 2022/23 the Treasury gives £3,654 million for pensions alone, £360 million less than Doyle’s figure. Keeping all his other assumptions, the remaining deficit would be about £1.85 billion. Doyle’s argument concerns pensions earned through contributions, and if he limits the state-benefit part of his deduction to the State Pension while leaving his £1.7 billion for public service pensions unchanged, the Department for Communities’ accounts show £2,730 million for 2022/23. On that basis, the remaining deficit would be about £2.77 billion.

Doyle’s £1.7 billion for public service pensions needs a larger correction. His note says it is “a per capita breakdown of the UK figure”. The UK figure that fits is the Treasury’s resource-accounts measure for 2021/22, £58.9 billion, and Northern Ireland’s population share of it is about £1.7 billion. That measure counts the annual growth in pension liabilities, and it matches Doyle’s description of a cost that “peaked in 2022” and is expected to fall “by over 75 per cent”. The regional fiscal balance does not count changes in pension liabilities. Its spending figures come from the Treasury’s Country and Regional Analysis, which records Northern Ireland’s public service pension spending as pensions paid less contributions received: £224 million in 2022/23, made up of £149 million for Northern Ireland’s own schemes and £74 million as its share of UK-wide schemes. Doyle describes the cost as the Treasury’s “top-up” to cover the gap between contributions collected and pensions paid, and on that description the deduction should be £224 million. The remaining deficit would then be about £2.97 billion, or about £3.33 billion or £4.26 billion when combined with the two corrections to the state pension figure above. Under the settlement Doyle proposes, in which the UK pays pensions already earned, the deduction would instead depend on the gross pensions paid to Northern Ireland pensioners and on which state collected contributions from serving staff. The published regional data show neither.

The smaller adjustments

Several of the remaining entries need correcting. Doyle cites FitzGerald and Morgenroth’s 2019 paper for an estimate that corporation tax is undercounted by about £500 million a year. Their 2024 note, which updates that paper, puts the undercount at £281 million for 2019 and £285 million for 2021, with no year above £370 million, all calculated at UK tax rates. Doyle argues elsewhere in the book that a united Ireland “would be able to offer a lower … corporation tax rate”, and he estimates that cutting Northern Ireland’s rate to the Republic’s could cost “almost £650 million” in the short term. That loss would follow from a policy choice, not from unity itself, but his tables count the revenue gain from moving to the Republic’s income tax and PRSI system while leaving out the short-term loss from its lower corporation tax rate.

Doyle takes from Mike Tomlinson an estimate that moving Northern Ireland to the Republic’s income tax and PRSI system would raise “€769 million” more a year. In Tomlinson’s article, the figure is in euros, for 2019 and adjusted for purchasing power. Table 1.2, in sterling and using 2022/23 data, enters it unconverted as £769 million. Separately, the table’s total of adjustments is printed as £12,773 million, although the items add to £12,733 million. That is a misprint, and the remaining deficit of £1,487 million is calculated correctly.

Doyle’s cost of raising pensions to the Republic’s rates assumes a pension population that does not yet exist. He prices it at €426 million a year: about €22 a week for each of the 325,350 people receiving State Pension, which is the gap for the new State Pension of £241.30, and €15 a week for the 68,990 receiving Pension Credit. His note accepts that those on the older State Pension are paid “at much lower rates”, but expects “the great bulk of pensioners” to be on the new rates by the earliest likely date of unity. That may prove right, but it is a forecast. The Department for Communities’ Benefits Statistics Summary for November 2024 records only 124,510 recipients of the new State Pension, so about 200,000, or 62%, were on the older one. For a pensioner on the full basic rate of £184.90 with no additional State Pension, the gap to the Republic’s €299.30 is about €87 a week at Doyle’s exchange rate, although many on the older pension also receive an additional State Pension. Doyle’s deficit tables use 2022/23 data, while his equalisation cost assumes a later pension population, so the two do not describe the same year.

The rebuttal of the IIEA note

Doyle describes the IIEA estimate as adding “almost an additional £10 billion per annum” to the subvention, for a total of “£20 billion per annum”, and later as “the figure of €20 billion over 20 years”. The note’s main figures are for 2019, because the authors treat 2020 and 2021 as distorted by the pandemic: an existing subvention of £10.7 billion (€12.1 billion), a base of £9.6 billion after their adjustments, and £18.1 billion (€20.6 billion) once Northern Ireland’s public pay and welfare are raised to the Republic’s levels. Doyle’s “£20 billion” matches the note’s total for 2021, £20.5 billion, and his “almost an additional £10 billion” is close to its 2019 cost of raising pay and welfare, £9.1 billion, which the note adds to its adjusted base and not to the published subvention. The note contains no 20-year projection. The phrase appears in press coverage, including the Irish Times report that Doyle also cites.

Doyle’s charge that the note “uncritically uses the published subvention figure” does not survive a reading of it. The note adjusts defence, foreign spending, the EU contribution and corporation tax. Its base case does include both debt interest and pensions, as Doyle says, but it also shows the effect of removing each of them. His claim that it seems to have compared private and occupational pensions is also unsupported, since the note says it used the ratio of average pension payments and cites Department for Communities benefit statistics.

Three of his criticisms do hold. The note adds the full cost of raising public pay without deducting the tax and pension contributions it would generate, which Doyle estimates would reduce the £4.2 billion cost by more than £2.2 billion. It raises all welfare payments by the ratio of average pension payments, a crude method that the authors themselves call “a crude estimate”. And it excludes the wider economic effects of unity, which its authors acknowledge.

The year-one figure

Doyle’s Table 1.3 builds a year-one deficit of €3.0 billion on the £1.5 billion: €152 million for raising public pay over 15 years, €115 million for the gradual transfer of pensions and €1 billion of new spending. The arithmetic is correct, but the figure carries every assumption above. It also rests on data for 2022/23, which Doyle calls the most recent available, and he expects the deficit to fall. The ONS tables published on 26 May 2026, before the book went to press, show that the published regional deficit had instead increased. On the same basis as Doyle’s £14.22 billion, Northern Ireland’s net fiscal deficit was £14.04 billion in 2022/23 after revision, £14.64 billion in 2023/24 and £15.72 billion in 2024/25.

Doyle shows that the published subvention is a poor guide to the cost of unity, and that point should now be common ground. His £1.5 billion is a different matter. It assumes the UK would continue paying existing pensions and that Ireland would take on no share of UK debt-servicing costs, leaves out the revenue cost of the lower corporation tax rate he proposes, deducts more for public service pensions than the published deficit contains, and contains currency and accounting inconsistencies, some of which make the remaining deficit appear smaller. The two outcomes that matter most are whether the UK keeps paying existing pensions and whether Ireland takes on a share of UK debt costs. Voters would need to see the deficit under each combination of the two to know how much of the answer depends on the UK.

Sources:

John Doyle, Before the Vote: Planning for a united Ireland, Dublin: DCU Press, 2026 (open access): pp. 9 (“a political judgement”); 12 (2022/23 data “the most recent available”); 13 (Table 1.1); 14–15, 22 (defence £1,567 million; outside the UK £711 million); 16–19 (pensions, £5,722 million; public service pensions £1.7 billion; cost “peaked in 2022”; Treasury “top-up” to fall “by over 75 per cent”; UK–Ireland convention); 19–22 (debt interest, £3,533 million); 24 (accounting adjustments; “€821 million”); 25–6 (corporation tax, £500 million); 26–7 (Tomlinson, “€769 million”); 27 (Table 1.2); 28 (£1.5 billion; “a maximum £5.9 billion”); 28–31 (critique of the IIEA estimate); 30 and 34, note 3 (pension equalisation, €426 million, including €15 a week for 68,990 Pension Credit recipients; older pensions “at much lower rates”; “the great bulk of pensioners” on new rates by unity); 32 (pension cost falling to £4.4 billion); 34, note 1 (“a per capita breakdown of the UK figure”); 33–4 (Table 1.3); 54 (cost of a lower corporation tax rate, “almost £650 million”); 162 (“a lower … corporation tax rate”).

John FitzGerald and Edgar Morgenroth, Northern Ireland Subvention: Possible Unification Effects, Institute of International and European Affairs, April 2024: Table 3 (corporation tax, £281 million in 2019 and £285 million in 2021, at UK rates); Tables 4–5 (defence, international and EU adjustments); section 1 (“a crude estimate”); Table 6 (rerating of pay and welfare: £9,115 million in 2019; £4,246 million pay and £3,856 million welfare in 2021); Table 7 (existing, base and rerated subvention, 2017–2021; £20,502 million in 2021); text after Table 7 (2019 figures preferred because of the pandemic); section 5 (“most improbable”; “a very weak negotiating position”; wider economic effects excluded).

Office for National Statistics, Country and regional public sector finances, UK: financial year ending 2023, 7 June 2024; Country and regional public sector finances, UK: financial year ending 2025, release announcement, 26 May 2026.

Mark Hennessy, “United Ireland could cost €20bn a year for 20 years, says new study”, The Irish Times, 4 April 2024.

HM Treasury, Public Expenditure Statistical Analyses 2023, Table 10.4 (Northern Ireland, 10.2 Old age, 2021/22: £4,014 million, of which personal social services £628 million and pensions £3,386 million). HM Treasury, Public Expenditure Statistical Analyses 2024, Table 10.4, p. 172 (Northern Ireland, 10.2 Old age, 2022/23: pensions £3,654 million, personal social services £715 million). Department for Communities, Benefits Statistics Summary, November 2024, p. 9 (325,350 State Pension claimants; 124,510 new State Pension recipients; average weekly State Pension £206.53). Mike Tomlinson, “Social security in a unified Ireland”, Irish Studies in International Affairs, 33(2), 2022, pp. 228–46, p. 241 and Table 9 (income tax and social insurance yield €769 million higher under the Republic’s tax code, 2019, PPP adjusted €), https://doi.org/10.1353/isia.2022.0011. Department for Communities, Annual Report and Accounts 2022–23, HC 1601, Note 4b, p. 163 (non-voted expenditure, State Pension, 2022–23: £2,730,295,000).

HM Treasury, Public Expenditure Statistical Analyses 2023, paras D.18–D.19 and Table D.1, p. 220 (pay-as-you-go public service pension schemes: net public service pensions, GAAP basis, £58,908 million in 2021/22 and £14,436 million in 2024/25 plans; change in liability £99,607 million and contributions received £40,742 million in 2021/22; changes in liability removed to reach the National Accounts basis). HM Treasury, Country and Regional Analysis 2023 database, December 2023, Northern Ireland, sub-functions 10.2 and 10.3, 2022/23 (Northern Ireland’s own public service pension schemes, segments S099A063–S099A067, £149.3 million; Northern Ireland’s share of UK-wide public service pension schemes, £74.3 million). NISRA, 2021 Mid-Year Population Estimates for Northern Ireland (1,904,600). Office for National Statistics, Population estimates for the UK, England, Wales, Scotland and Northern Ireland: mid-2021 (67,026,292).

Office for National Statistics, Country and regional public sector finances: net fiscal balance tables, editions of 7 June 2024 and 26 May 2026, Northern Ireland sheet (net fiscal balance, North Sea revenues by population share).

Calculations: Table 1.2 items summed (5,722 + 3,533 + 1,367 + 500 + 400 + 265 + 177 + 769 = 12,733); ceiling on the 2022/23 basis (1,487 + 5,722 = 7,209); remaining deficit with the 2022/23 pensions figure (4,014 + 1,708 = 5,722; 4,014 − 3,654 = 360; 1,487 + 360 = 1,847); remaining deficit with the State Pension alone (4,014 − 2,730 = 1,284; 1,487 + 1,284 = 2,771); older State Pension recipients (325,350 − 124,510 = 200,840); basic State Pension at Doyle’s rate (£184.90 × 223/194.32 = €212); population share of UK public service pensions (58,908 × 1,904,600/67,026,292 = 1,674); regional public service pension spending, 2022/23 (149.3 + 74.3 = 223.6); remaining deficit with that figure (1,708 − 224 = 1,484; 1,487 + 1,484 = 2,971); combined with the state pension corrections (2,971 + 360 = 3,331; 2,971 + 1,284 = 4,255).