Who really leaves the country in debt?

What the data actually shows: Day-to-day spending — public sector current expenditure — averaged 33.7% of GDP over FY1997-98 to FY2006-07, ranging from 31.9% to 35.6%. The 37-40% range describes TOTAL managed expenditure over the same period (mean 37.1%, range 34.6% to 39.7%), which also includes capital investment and depreciation.

• current-expenditure-pct-gdp, FY1997-98 to FY2006-07: mean 33.69%, minimum 31.87% (FY1999-00), maximum 35.63% (FY2005-06).
• tme-pct-gdp over the same ten years: mean 37.12%, minimum 34.60%, maximum 39.73%.
• The gap between the two measures is 3-4 percentage points of GDP in every year of the period.

The numbers in the claim are real numbers from the right period — they are just attached to the wrong measure. Quoting total spending as though it were day-to-day spending overstates day-to-day spending by roughly 3-4 points of GDP throughout this series, which is larger than most of the differences these arguments turn on.

Sources: obr.uk

What the data actually shows: This holds for one period and not the other. The Conservative period ending in 2024 did end with a historically high tax burden: national account taxes were 34.9% of GDP in FY2023-24, the third highest of the fifty years since FY1976-77, and receipts were 39.3%, the sixth highest. The Labour period ending in 2010 did not: national account taxes were 32.2% of GDP in FY2009-10 (32nd of 50) and receipts 35.8% (34th of 50), both well below the early-1980s levels.

• national-account-taxes-pct-gdp, ranked over FY1976-77 to FY2025-26: FY2025-26 = 35.99% (1st), FY2022-23 = 35.20% (2nd), FY2023-24 = 34.92% (3rd), FY1981-82 = 33.97% (6th), FY2009-10 = 32.15% (32nd).
• receipts-pct-gdp, same ranking: FY1981-82 = 40.81% (1st), FY1976-77 = 40.21% (3rd), FY2025-26 = 40.14% (4th), FY2023-24 = 39.28% (6th), FY2009-10 = 35.78% (34th).
• The Conservative period that ended in 1997 also ended with a low tax burden: FY1996-97 receipts were 32.30% of GDP, 47th of 50.

The symmetry the claim asserts is not there, and the reason is mechanical rather than political: a period ending in a deep recession ends with revenues collapsed as a share of GDP, whichever party is in office. 2010 was such a year and 2024 was not. Note also that the two revenue measures rank differently — receipts include non-tax income and are higher in the 1970s and early 1980s than the tax-only measure, so a claim about "the tax burden" should say which it means.

Sources: obr.uk

What the data actually shows: Over financial years 1976-77 to 2025-26 the average current budget deficit was 1.23% of GDP in Labour-held years and 2.48% in Conservative-held years — roughly double. The gap survives every filter tested: excluding crisis years it is 0.90% against 1.88%; using crisis years only it is 1.44% against 3.16%; on medians it is 1.19% against 2.04%.

• Majority-period attribution, outturn only, FY1976-77 to FY2025-26: Labour 18 years, mean current budget deficit 1.23% of GDP (median 1.19%); Conservative 32 years, mean 2.48% (median 2.04%).
• Excluding years overlapping a recession, financial crisis, pandemic, energy or geopolitical shock: Labour 7 years, mean 0.90%; Conservative 17 years, mean 1.88%.
• Crisis years only: Labour 11 years, mean 1.44%; Conservative 15 years, mean 3.16%.
• On headline net borrowing the same comparison gives Labour 3.23% and Conservative 4.08% of GDP.

The claim is not supported, but the finding must not be inverted into a causal one. The Conservative set of years contains the early-1980s recession, the early-1990s recession, the post-2008 consolidation and the pandemic; the Labour set contains the IMF-crisis aftermath and the financial crisis. Two different collections of years produced two different averages. That is a description, not an effect. Party averages in these checks describe two different sets of financial years. They are not estimates of what either party caused: governments inherit a debt stock, a tax base and a spending baseline they did not set, they take office at different points in the economic cycle, and the recessions, financial crises and pandemics in this period were not distributed evenly between them.

Sources: obr.uk

What the data actually shows: Public sector net investment averaged 1.56% of GDP over financial years 1997-98 to 2009-10. Even restricting to FY2001-02 to FY2009-10, after the first-term spending restraint, the average is 2.00%. The 2.2% figure is not reached on any reasonable window inside the period.

• public-net-investment-pct-gdp, FY1997-98 to FY2009-10 by year: 0.63, 0.63, 0.56, 0.50, 1.16, 1.53, 1.70, 2.04, 1.85, 1.81, 1.82, 3.05, 3.04. Mean 1.56%.
• FY2001-02 to FY2009-10 mean: 2.00%. FY2004-05 to FY2009-10 mean: 2.27%.

The period average is dragged down by four years at or below 0.63% of GDP at the start and pulled up by two crisis years above 3% at the end, so a single average is a poor summary of it either way. A 2.2% figure is only obtainable from the last six years of the period, which is a defensible window to quote but not the one the claim states.

Sources: obr.uk

What the data actually shows: Public sector net debt was 36.4% of GDP at the end of the last financial year before Labour took office (FY1996-97) and 64.4% at the end of its last full year in office (FY2009-10). The direction and rough magnitude of the claim hold; the endpoints are 36.4% and 64.4%, not 38% and 65%. Almost all of the increase came in two financial years: debt was 36.5% in FY1997-98 and 35.5% eleven years later in FY2007-08, then rose 28.9 percentage points across FY2008-09 and FY2009-10 alone.

• psnd-pct-gdp: FY1996-97 = 36.36%, FY1997-98 = 36.46%, FY2007-08 = 35.47%, FY2008-09 = 50.24%, FY2009-10 = 64.39% (OBR Public Finances Databank, July 2026, outturn).
• Change FY1997-98 to FY2007-08 = −1.0 percentage points; change FY2007-08 to FY2009-10 = +28.9 percentage points.
• In cash terms over the same period the debt stock rose from £360bn (FY1997-98) to £1,028bn (FY2009-10), while nominal GDP rose from £970bn to £1,567bn (psnd-gbp, nominal-gdp-fy-gbp).

The choice of start year matters and is worth stating: 36.4% (FY1996-97) is what was inherited, 36.5% (FY1997-98) is the first year in office. Both round to 36%, not 38%. The flat eleven-year stretch followed by a two-year jump is the single most important feature of this series and is lost entirely by quoting only the endpoints — which is exactly the error the attribution module's headline warning describes.

Sources: obr.uk

What the data actually shows: Public sector net debt was 64.4% of GDP at end-March 2010 and 94.2% at end-March 2024 on the current data vintage — a rise of 29.8 percentage points, not to 98%. On the measure that excludes the Bank of England, debt was 85.8% at end-March 2024. The 98% figure does not appear anywhere in the 2024 data on the current vintage; the ONS series reached 98.1% in September 2020, at the height of the pandemic.

• psnd-pct-gdp: FY2009-10 = 64.39%, FY2023-24 = 94.20%, FY2024-25 = 93.40% (OBR Databank, July 2026, outturn).
• psnd-ex-boe-pct-gdp: FY2023-24 = 85.80%, FY2024-25 = 87.90% — 8.4 points lower than the headline measure in FY2023-24.
• ONS series HF6X (net debt excluding public sector banks, % of GDP): highest monthly reading since 2020 is 98.1% in September 2020; the annual figure for 2024 is 94.8%.

The gap between the claim and the data is probably a vintage effect rather than an error at the time: a debt-to-GDP ratio falls when ONS revises nominal GDP upwards even though no fiscal number has changed, and the June 2026 Quarterly National Accounts did revise the denominator. Any 2010-2024 debt comparison should also state whether it includes the Bank of England, because since quantitative easing began the two measures have diverged by up to about 10 percentage points and the choice changes the answer by more than the differences usually being argued about.

Sources: obr.uk · ons.gov.uk

What the data actually shows: Two full-year NET BORROWING surpluses were run in that window — FY1999-00 (−1.07% of GDP) and FY2000-01 (−1.46%) — with FY1998-99 essentially balanced (a deficit of 0.01% of GDP) and FY2001-02 back in deficit (0.49%). On the CURRENT BUDGET, which excludes borrowing to invest, there were surpluses in all four years from FY1998-99 to FY2001-02.

• psnb-pct-gdp: FY1998-99 = +0.01, FY1999-00 = −1.07, FY2000-01 = −1.46, FY2001-02 = +0.49 (negative = surplus).
• current-budget-deficit-pct-gdp: FY1998-99 = −0.62, FY1999-00 = −1.64, FY2000-01 = −1.96, FY2001-02 = −0.67 (negative = surplus).
• Public sector net investment over the same four years was 0.63%, 0.56%, 0.50% and 1.16% of GDP — the lowest readings in the whole 1976-2026 series apart from the late 1980s.

The claim is true or false depending entirely on which deficit measure is meant, and it almost never says. This is the clearest illustration in the whole set of why the current budget and net borrowing have to be distinguished: on one measure it is two years, on the other four. The current budget was also the operative measure of the fiscal rule in force at the time, which is context, not a reason to prefer it.

Sources: obr.uk

What the data actually shows: Net investment rose from 0.63% of GDP in FY1997-98 to a peak of 3.05% in FY2008-09 and 3.04% in FY2009-10. The starting figure is right; the peak on the current data vintage is just above 3%, not above 3.3%.

• public-net-investment-pct-gdp: FY1997-98 = 0.63%, FY2008-09 = 3.05%, FY2009-10 = 3.04% (OBR Databank, July 2026, outturn).
• Gross investment in FY2009-10 was 5.60% of GDP; the difference is depreciation at 2.55% (public-gross-investment-pct-gdp, public-sector-depreciation-pct-gdp).

A ratio to GDP moves when GDP is revised, and FY2009-10 was published contemporaneously against a smaller nominal GDP estimate than the one now in use, so a figure above 3.3% was plausibly correct at the time. The direction and the near-fivefold increase are not in dispute; only the peak level is. Note that net investment more than doubled in the two crisis years alone, so it is not a smooth upward trend.

Sources: obr.uk

What the data actually shows: Net investment fell from 2.51% of GDP in FY2010-11 to a low of 1.45% in FY2013-14 — below 1.5% in one of the five financial years from FY2010-11 to FY2014-15, not throughout. The five-year average was 1.93%.

• public-net-investment-pct-gdp: FY2010-11 = 2.51%, FY2011-12 = 1.87%, FY2012-13 = 1.90%, FY2013-14 = 1.45%, FY2014-15 = 1.93%, FY2015-16 = 1.67%.
• FY2013-14 is the lowest reading anywhere between FY2002-03 and FY2025-26.

The claim identifies a real trough and dates it correctly, but generalises one year into a five-year characterisation. The fall from the FY2009-10 peak is the substantive point and is large: 3.04% to 1.45% of GDP, a halving in four years.

Sources: obr.uk

What the data actually shows: Net investment recovered to an average of 2.32% of GDP over FY2021-22 to FY2025-26, reaching 2.63% in FY2025-26. No single year in the period sits at 2.1%: the readings are 3.40% (FY2020-21, inflated by pandemic support), 2.20%, 1.77%, 2.44%, 2.56% and 2.63%.

• public-net-investment-pct-gdp: FY2019-20 = 1.87%, FY2020-21 = 3.40%, FY2021-22 = 2.20%, FY2022-23 = 1.77%, FY2023-24 = 2.44%, FY2024-25 = 2.56%, FY2025-26 = 2.63%.
• Mean FY2021-22 to FY2025-26 = 2.32%; mean FY2020-21 to FY2025-26 = 2.50%.

The direction is right and the recovery is real, but the level quoted is low and the series is not flat — FY2022-23 dipped to 1.77% before rising again. A single "recovered to X" figure is a poor description of a series moving between 1.8% and 2.6%.

Sources: obr.uk

What the data actually shows: Of the 682 PFI and PF2 projects still live at 31 March 2022 with a recorded financial-close date, 581 (85%) reached financial close in financial years 1997-98 to 2009-10, accounting for £43.6bn of £52.2bn (84%) of the total capital value. That is a clear concentration in the 1997-2010 period. But PFI was introduced by the Conservative government in 1992, continued under the Coalition (88 projects, £7.7bn, from FY2010-11), and "heavily" is a judgement the data cannot settle — what they settle is when the surviving PFI estate was signed.

• HM Treasury PFI/PF2 current projects as at 31 March 2022: 683 project rows, 682 with a financial-close date, £52.2bn total capital value.
• By financial year of financial close: FY1992-93 to FY1996-97, 13 projects, £0.9bn (1.7% of value); FY1997-98 to FY2009-10, 581 projects, £43.6bn (83.5%); FY2010-11 onwards, 88 projects, £7.7bn (14.8%).
• Annual unitary charge payments across all live projects rose from £0.15bn in FY1997-98 to £9.77bn in FY2021-22, with about £151bn still to be paid from FY2022-23 onwards and the last contract expiring in 2048.

SURVIVORSHIP LIMITATION, and it cuts against the claim: HM Treasury publishes only projects still live at the snapshot date, so pre-1997 deals — older, and more likely to have reached contract expiry by 2022 — are under-counted relative to later ones. The 1.7% figure for the 1992-1997 period is therefore a floor, not an estimate of how much PFI that government signed. The National Audit Office's "PFI and PF2" report is the standard reference for the programme's overall scale and for the value-for-money question, which these data do not address at all.

Sources: gov.uk · nao.org.uk

What the data actually shows: At Budget 2018 (29 October 2018) the government announced it would no longer use PF2, PFI's successor, for new projects, having found the model "inflexible and overly complex"; the same passage records that PF2 had not been used since 2016. This was a policy decision to stop using the model, not a legal abolition, and it did not affect existing contracts: 682 projects were still live at 31 March 2022, with about £151bn of unitary charge payments still to be made and the final contract expiring in 2048.

• HM Treasury, Budget 2018: "The Budget announces government will no longer use PF2 for new projects." The surrounding text records that the government "found the model to be inflexible and overly complex" and that PF2 "has not been used since 2016".
• HM Treasury PFI/PF2 data as at 31 March 2022: one project reached financial close in FY2018-19 (£33m capital value) and none thereafter in the dataset.
• Remaining unitary charge payments from FY2022-23 onwards total about £151bn across live projects (pfi-unitary-charge-payments-gbp).

The Budget document's own wording is about PF2 specifically. "Abolished" is reasonable shorthand for the announcement but obscures two things worth keeping: the commitment was about future use rather than a statutory change, and the fiscal consequences of PFI run for another two decades regardless. The payments are not an extra cost sitting outside the spending aggregates — they are inside departmental budgets and total managed expenditure.

Sources: gov.uk · gov.uk

What the data actually shows: The arithmetic supports the description of what happened, but not an attribution of the decision. Between FY2009-10 and FY2013-14 net borrowing fell by 4.53 percentage points of GDP and public sector net investment fell by 1.59 points — about 35% of the total reduction. In proportional terms capital was cut far harder than day-to-day spending: net investment fell 52% as a share of GDP over those four years while day-to-day spending fell 5%.

• psnb-pct-gdp: FY2009-10 = 10.19%, FY2013-14 = 5.65% (−4.53 points).
• public-net-investment-pct-gdp: FY2009-10 = 3.04%, FY2013-14 = 1.45% (−1.59 points, a 52% fall as a share of GDP).
• current-expenditure-pct-gdp: FY2009-10 = 40.37%, FY2013-14 = 38.23% (−2.14 points, a 5.3% fall as a share of GDP).
• current-budget-deficit-pct-gdp fell from 7.14% to 4.20% over the same four years (−2.94 points).

WHAT THE DATA CANNOT SETTLE: whose mechanism this was. Spending plans for FY2010-11 were set at the Budget of 24 March 2010, before the general election, so the consolidation path did not begin only in May 2010; establishing how much of the capital reduction was already planned by the outgoing government requires reading the March 2010 Budget's own plans against the June 2010 Budget's, which is a documentary question, not one these outturn series can answer. Note also that Capital DEL — the departmental budget line the phrase "capital budgets" refers to — is a different measure from public sector net investment, and no consistent published Capital DEL series covers 2010-2015 alongside later years (see scripts/economy/ingest-capital-budgets-and-pfi.py), so the decomposition above necessarily uses the national-accounts aggregate instead.

Sources: obr.uk · gov.uk · gov.uk

• The fiscal series establish the first half only: public sector net investment fell to 1.45% of GDP in FY2013-14, its lowest reading between FY2002-03 and FY2025-26, and averaged 1.93% over FY2010-11 to FY2014-15 against 3.04% in FY2009-10.
• They contain no measure of asset condition, deferred maintenance or backlog cost, for any sector, in any year — so the second half of the claim cannot be evaluated from them at all.

THIS CANNOT BE SETTLED BY THE FISCAL DATA and needs a named condition-of-estate source plus a causal argument. The standard UK source for the health estate is the Estates Returns Information Collection (ERIC), published for the NHS estate, which reports backlog maintenance cost by organisation and risk category; the Department for Education runs an equivalent condition data collection for the school estate. Even with those series in hand the causal step is not automatic: a maintenance backlog can grow because capital budgets fell, because the estate aged, because survey coverage or costing methods changed, or because capital was reallocated from maintenance to new build. None of those can be separated using aggregate net investment as a share of GDP. Marked unclear rather than partially supported because no part of the consequence half of the claim has been checked here.

Sources: obr.uk · digital.nhs.uk

What the data actually shows: The descriptive arithmetic can be stated and is striking; the causal claim cannot be. Descriptively: public sector net debt rose from 36.5% of GDP at end-March 1998 to 93.7% at end-March 2026, a rise of 57.2 percentage points, and three financial years account for 39.8 points of it — FY2008-09 (+14.8), FY2009-10 (+14.1) and FY2020-21 (+10.9), about 70% of the total. Over the eleven years FY1997-98 to FY2007-08 the ratio was essentially unchanged (36.5% to 35.5%). This says crisis years dominate the arithmetic. It does not say crises caused the increase, and it does not measure ideology at all.

• psnd-pct-gdp year-on-year changes, FY1998-99 to FY2025-26: the six largest single-year rises are FY2008-09 (+14.8pp), FY2009-10 (+14.1pp), FY2020-21 (+10.9pp), FY2010-11 (+6.2pp), FY2019-20 (+4.9pp) and FY2012-13 (+3.3pp).
• Change over the eleven pre-crisis years FY1997-98 to FY2007-08: −1.0 percentage points.
• Splitting the 1976-2026 record by whether a financial year overlapped a recession, financial crisis, pandemic, energy or geopolitical shock: mean net borrowing in crisis years was 3.63% of GDP in Labour-held years and 5.05% in Conservative-held years; in non-crisis years, 2.59% and 3.23%. Both parties borrowed more in crisis years, and the gap between the parties is similar in both subsets.

MARKED UNCLEAR BECAUSE THE CLAIM IS CAUSAL AND THESE DATA ARE NOT. Three specific obstacles, none of which the arithmetic above removes. First, crisis-year borrowing is itself a policy choice: the size of the 2008-09 bank interventions and of the 2020-21 furlough scheme were decisions about how much to spend in response to a shock, not amounts the shock imposed. Second, the fiscal position entering a crisis affects how much the crisis costs, so pre-crisis policy is inside the crisis-year numbers. Third, "routine party ideology" is not a measured variable in any of these series and there is no defensible way to construct one from them, so the comparison the claim asks for has no denominator. Attributing a share of the debt rise to crises would also require a counterfactual — what debt would have done without them — which no observational fiscal series can supply. State the decomposition, label it as arithmetic, and stop there. Party averages in these checks describe two different sets of financial years. They are not estimates of what either party caused: governments inherit a debt stock, a tax base and a spending baseline they did not set, they take office at different points in the economic cycle, and the recessions, financial crises and pandemics in this period were not distributed evenly between them.

Sources: obr.uk

• Financial years 1976-77 to 2025-26, majority-period attribution, outturn only: Labour held 18 financial years and ran a net surplus in 2 of them (FY1999-00, −1.07% of GDP; FY2000-01, −1.46%). The Conservatives held 32 and ran a net surplus in 1 (FY1988-89, −0.97%).
• That is 2 of 18 (11%) against 1 of 32 (3%) — the claim holds both as a raw count and as a rate.
• On the CURRENT budget (excluding investment) the counts are closer: Labour 4 surplus years of 18 (FY1998-99 to FY2001-02), Conservative 3 of 32 (FY1988-89, FY1989-90, FY2018-19).

True, but the base rate matters more than the comparison: there were three full-year net surpluses in fifty years, so this is a difference between "very rare" and "very rare". Both parties' surpluses came at cyclical peaks — the late-1980s boom and the 1999-2001 boom — which is a reason to be cautious about reading them as evidence of a governing disposition. FY1998-99 is a borderline case excluded here: net borrowing was +0.01% of GDP, a deficit by a rounding margin. Party averages in these checks describe two different sets of financial years. They are not estimates of what either party caused: governments inherit a debt stock, a tax base and a spending baseline they did not set, they take office at different points in the economic cycle, and the recessions, financial crises and pandemics in this period were not distributed evenly between them.

Sources: obr.uk

• tme-pct-gdp: FY2008-09 = 43.04%, FY2009-10 = 45.97% (the post-war peak until Covid), FY2010-11 = 45.50%.
• Day-to-day spending alone reached 40.37% of GDP in FY2009-10 (current-expenditure-pct-gdp).

Correct as stated, and worth adding that a large part of the rise is arithmetic rather than a spending decision: nominal GDP fell in FY2009-10, so the same cash spending is a larger share of a smaller denominator. This is the "spending share rises in a recession" effect and it operates whoever is in office. The figure refers to total spending, not day-to-day spending.

Sources: obr.uk

• tme-pct-gdp: FY2017-18 = 39.67%, FY2018-19 = 39.09%, FY2019-20 = 39.10%.
• Day-to-day spending fell over the same period from 40.50% of GDP (FY2010-11) to 34.70% (FY2018-19).

Correct. FY2018-19 is also the only year between FY2001-02 and FY2025-26 in which the current budget was in surplus (−0.06% of GDP), which is a more precise way of describing the same position than the total-spending share.

Sources: obr.uk

• tme-pct-gdp: FY2020-21 = 52.27%, the highest reading in the series (which runs from FY1948-49).
• Day-to-day spending reached 46.34% of GDP in the same year; net borrowing was 14.74% of GDP.

Correct. As with FY2009-10, part of the rise is the collapse in the denominator rather than the numerator: real GDP fell sharply in the same year. The series returned to 43.36% the following year without any comparable fall in cash spending.

Sources: obr.uk

• tme-pct-gdp: FY2023-24 = 44.12%, FY2024-25 = 43.97%, FY2025-26 = 44.31%.
• Day-to-day spending in the same years: 39.34%, 39.06%, 39.30% of GDP.

Correct, though the figure sits at the bottom of the quoted band rather than in the middle of it — 44.0-44.3%, never approaching 45%. Total spending has now been within half a point of 44% of GDP for four consecutive years.

Sources: obr.uk