The Monetary System, Part 1: Spending & Debt (2026)

Framework evergreen; figures dated. Data as of August 2026. The way spending, deficits and debt fit together is durable and reads that way here. Every number is a dated benchmark from the source named beneath its table — re-source it before reuse, because fiscal data is revised routinely. Informational only, not investment advice; AI-assisted and human-reviewed (see Section 12). Update cadence: annual.

Every government does three things with money: it spends, it taxes, and it borrows the difference. Do that year after year and the borrowed differences pile into a debt — and across the rich world and China alike, that pile has been growing for a quarter-century. This guide explains what governments actually spend on, why deficits turn into debt, and how the debt burden has moved since 2000 across nine big economies — the United States, China, Japan, Germany, the United Kingdom, France, Poland, Canada and Australia. It is the free, big-picture primer; when you want the company- and project-level data that sits underneath the macro picture, that is what Metal Pilot is for.

The headline: the debt build is real and broad, but it is not one story — a country’s path depends on what it spends on, whether it runs a primary deficit, and whether its economy grows faster than its borrowing costs.

This is Part 1 of “The Monetary System” — a three-part series on how government money works. The parts: Part 1 · Spending & Debt (this guide), Part 2 · Bonds & Yields , and Part 3 · Interest Rates & Inflation .

TL;DR & Key Takeaways

• Spending is dominated by a few big blocks. In the US, health programmes, pensions (Social Security), defence and interest together make up roughly three-quarters of federal outlays; in Europe, social protection alone runs near a fifth of GDP. The budget is mostly promises to people, not discretionary choices.
• Deficits are the flow; debt is the stock. A deficit adds to the debt each year; the debt is the accumulated total. A country can shrink its debt ratio while still running a deficit — if the economy grows faster than the debt.
• Two shocks did most of the work. The 2008 financial crisis and the 2020 pandemic are the step-changes: debt ratios jumped at both, and rarely fully retraced.
• The paths diverge sharply. Debt burdens now range from Japan’s world-record ~207% of GDP, and the US, UK and France near or above 100%, down to Australia’s ~51% and Poland’s ~59%. US and French debt roughly doubled since 2000, China’s climbed from ~23% to ~99%, while Germany cut its ratio back from its 2010 peak. Debt destiny is a policy choice, not fate.
• The real problem is off the balance sheet. Headline debt understates the position: unfunded pension and health promises — US Social Security and Medicare, Europe’s ageing costs, China’s local-government debt — are the larger, slower fiscal risk.
• One number for the quarter-century. The Metal Pilot fiscal-health score (Section 9) rolls growth, real pay, tax and the budget balance into one running index: China ends near 223, Japan near 39, and the surplus economies pull away from the perpetual borrowers.

New to the topic? Read straight through from Section 1. Here for the payoff? Jump to the fiscal-health score in Section 9. Every specialised term is defined in the Vocabulary (Section 11).

1. The budget identity: spending, deficits, debt

Public finance rests on one identity. In any year, a government’s spending minus its revenue (mostly taxes) equals its balance. When spending exceeds revenue, the balance is negative — a deficit — and the government borrows to cover it, usually by issuing bonds. When revenue exceeds spending, it runs a surplus. The running total of all past deficits, minus surpluses, is the debt.

Two distinctions do most of the analytical work, and getting them straight prevents most confusion about fiscal news.

The first is flow versus stock. A deficit is a flow — an amount per year. Debt is a stock — a level at a point in time. “The deficit fell last year” and “the debt rose last year” are usually both true at once: as long as the government runs any deficit, it adds to the debt, so the debt keeps climbing even as the annual shortfall shrinks. Only a genuine surplus reduces the debt in cash terms.

The second is the primary balance versus the headline balance. The primary balance is spending minus revenue excluding interest payments on existing debt. It isolates the part of the budget policymakers control this year from the legacy cost of past borrowing. A government can run a primary surplus — taxing more than it spends on programmes — and still post a headline deficit, because interest on the existing debt swamps the primary surplus. That gap — programme choices versus the legacy cost of past borrowing — runs through the debt chapters that follow.

Throughout this guide, monetary quantities are shown as a percent of GDP rather than in raw currency, because a debt or deficit only means something relative to the size of the economy that has to service it. Unless stated otherwise, “debt” means general government gross debt (central plus state and local, before netting out financial assets), and “deficit” means general government net lending/borrowing — the internationally comparable measures the IMF and OECD publish.

2. The economies underneath: GDP and growth

This whole guide measures debt and spending as a share of GDP — so before going further, it is worth seeing that denominator directly, and understanding what it actually is. A debt or a spending bill only means something next to the economy that has to carry it, and the growth of that economy is what decides, in Section 4, whether its debt compounds or fades.

2.1 What GDP measures — and how it is counted

Gross domestic product is the total market value of all the final goods and services a country produces in a given period, measured in money. “Final” is the load-bearing word: GDP counts the loaf sold to a household but not the flour that went into it — the flour’s value is already in the loaf’s price, and counting both would double-count. “Produced” is the other key word: GDP measures production this period, so it excludes resales of existing assets (a used house or car) and purely financial transfers — buying a share, bond or lottery ticket — that move money without making anything.

Statisticians measure the same total three ways that by construction agree: the production approach (the value added at every firm — sales minus bought-in inputs), the income approach (what that production pays out in wages, profits, rents and taxes), and the expenditure approach (who buys the final output). This guide uses the expenditure lens, a single identity worth carrying in your head:

GDP = C + I + G + NX, where the four elements are:

• C — consumption: household spending on goods and services, from groceries to haircuts. It is the largest slice almost everywhere — roughly two-thirds of US GDP.
• I — investment: business spending on plant, equipment, software and new housing, plus the change in inventories. This is the volatile part that drives most of the business cycle.
• G — government: the state’s own spending on goods and services — teachers’ and soldiers’ salaries, the concrete in a road. Crucially this is the government buying goods and services, not transfer payments: a pension or unemployment cheque is not in G — the state buys nothing with it, just moves money to a household, which shows up in C when spent. That is why Section 3’s biggest blocks (pensions, benefits) are largely transfers, not government purchases.
• NX — net exports: exports minus imports. Imports are subtracted because C, I and G already include spending on foreign-made goods, which are not domestic production.

Those four elements do not carry equal weight, and how they split is one of the sharpest ways to tell economies apart. Table 1 shows the mix for the nine — each element as a nominal amount, with its share of GDP in brackets below.

Table 1. GDP by expenditure component: amount and share of GDP (2024)

Source: shares from World Bank national accounts and national statistics offices — household final consumption (C), gross capital formation (I), general government final consumption (G) and net exports of goods and services (NX = exports − imports), each as a share of GDP, latest available year (2023–24). Each cell shows the approximate nominal amount with the share below it in brackets; nominal amounts are the share applied to 2024 nominal GDP (Table 1a) in current US dollars, so they are indicative and rounded, and the components may not sum exactly to GDP because of rounding and the statistical discrepancy. G here is government consumption; government investment is counted within I (gross capital formation), following national-accounts convention — so the state’s full claim on output is larger than the G column alone.

The contrast is the whole point. The US and UK are consumption economies — roughly two-thirds of output is households spending — and both run trade deficits (negative NX). China is the mirror image: investment is its single largest block, above 40% of GDP, household consumption unusually low for a large economy, and it runs a trade surplus. Germany stands apart for its export machine — a +6% net-export surplus — with the commodity exporters Australia (+2%) and Poland (+4%) also in surplus; France for the sheer size of its government (23% of GDP); and Japan for the highest investment share of any rich economy here (28%). Canada is a smaller US — a consumption economy near trade balance. Same identity, nine different economies.

GDP also makes deliberate choices at its edges: it values government services at cost (there is no market price for defence) and imputes owner-occupiers’ shelter, while leaving out unpaid household work, the informal economy and volunteering — and it says nothing about how output is distributed or whether it improves wellbeing. It measures economic size and activity, not welfare or accumulated wealth.

One last distinction runs through this guide. Nominal GDP is measured in current prices; real GDP strips out inflation to show the change in actual volume. Debt-to-GDP ratios use nominal GDP — the debt and the economy in the same current dollars — which is why the nominal growth rate (real growth plus inflation) is the g that matters for debt dynamics. The figures below are nominal, in current US dollars, so cross-country levels also reflect exchange rates.

2.2 The nine economies compared

Table 1a. The nine economies: GDP, population and growth

Source: World Bank national accounts — GDP and GDP per capita in current US dollars, total population (2024), and real GDP growth (2024). GDP is nominal (current-price) US dollars, so cross-country levels also reflect exchange rates. “GDP, 2000→2024” is the multiple of nominal-dollar GDP over the period.

Three things shape the fiscal story. Size: the US economy, at about $29 trillion, is larger than the next three put together; China is second at roughly $19 trillion; then a mid-size cluster — Germany ($4.7tn), Japan ($4.2tn), the UK ($3.7tn), France ($3.2tn), Canada ($2.3tn), Australia ($1.8tn) — with Poland (~$0.9tn) the smallest. Growth is where they diverge: China’s economy multiplied about fifteen-fold in dollar terms since 2000 and Poland’s more than five-fold, while Japan’s dollar GDP actually shrank (×0.8) as a weak yen and near-zero inflation erased its gains — proof that nominal size can go backwards. The commodity economies grew briskly (Australia ×4.2, Canada ×3.1), the US ×2.9, old Europe only ×2.2–2.4. That growth gap is the quiet hero of the debt chapters: for two decades China’s economy grew faster than its borrowing costs, so the debt ratio stayed low even as the raw numbers ballooned — only now, as growth slows, does it climb. Wealth is not size: per person the US (~$86,000) leads, then Australia, Germany, Canada and the UK ($53–65k), France (~$46k), Japan (~$34k), Poland (~$25k), and China (~$13,000) — still middle-income despite its total. China spreads its output across 1.4 billion people, the US 340 million, Japan 124 million, the rest 27–84 million.

The trajectory since 2000 is as telling as the 2024 snapshot. Across six benchmark years, the US pulls steadily ahead in dollar terms while China closes most of the gap from a near-standing start; Japan actually slips back, and the mid-size economies barely move by comparison (Table 1b, Figure 1).

Table 1b. Nominal GDP by economy, 2000–2024 (US$ trillion)

Source: World Bank , nominal GDP (current US$), 2000–2024, rounded to the nearest $0.1 trillion. Dollar levels also reflect exchange-rate moves, which is part of why the European economies dip in some years despite growing in their own currencies.

Figure 1. The US leads, China surged, the rest cluster.

Source: values from Table 1b; nominal GDP in current US$ trillion, World Bank. Column heights are each value ÷ the 30 US$tn axis maximum; each country keeps its colour across all six years, so the read is both economy-against-economy within a year and each economy over time. The values themselves live in Table 1b.

Growth is worth seeing year by year, not just as the 2024 snapshot, because growth is what Section 4 weighs against the interest rate a government pays — the two forces that decide whether a debt ratio rises or falls. Table 1c shows it two ways: real growth, with inflation stripped out, and nominal growth — real growth plus inflation, in each economy’s own currency. It is the nominal figure that governs debt ratios, because debt is repaid in nominal money, not inflation-adjusted units.

Table 1c. GDP growth, real and nominal, 2000–2025 (%)

Source: real GDP growth from the World Bank (annual %, constant prices); nominal growth is that real rate compounded with the World Bank GDP deflator — growth in each country’s own currency, the g relevant to that country’s own-currency debt. Rounded to one decimal; a negative value is a contraction. 2025 figures are World Bank estimates.

Two patterns carry into the debt chapters. First, nominal growth stayed high through almost the whole period — the growth that Section 4 weighs against what governments pay to borrow. Second, the 2020–22 whipsaw is violent in both rows: the pandemic contraction, then an inflationary rebound that briefly pushed nominal growth into double digits in the US and UK before it faded.

The takeaway for the fiscal picture: the same 100%-of-GDP debt burden means something very different on a $29-trillion economy still growing at 3% than on a $4-trillion one that is flat. Size and growth are not background — they are half of what decides the debt trajectory the rest of this guide turns to.

2.3 Wage growth, real and nominal

The income an economy generates is, above all, paid out as wages — compensation of employees is the single largest slice of GDP on the income side, and it is the part of “growth” a household actually feels in its pay. So it is worth seeing the same real-and-nominal split as Table 1c, but for pay. Nominal wage growth — the number on the payslip — feeds the same nominal g that drives debt dynamics and, through income and payroll taxes, the tax base Section 8 returns to; real wage growth — pay after inflation — is what living standards track, and where wages beat or lag prices is a thread Part 2 picks up in its inflation chapter. Table 1d shows both, every year, for the nine.

Table 1d. Wage growth, real and nominal, 2000–2025 (%)

Source: real and nominal average-wage growth for the eight OECD economies computed from OECD Average Annual Wages (average wage per full-time-equivalent employee, national currency; real on constant prices, nominal on current prices), 2000–2025. China is on a narrower, non-comparable basis: the NBS series covers urban non-private units only — a higher-paid subset of employment that overstates the economy-wide average — with nominal growth computed from the NBS average annual wage and real growth deflated by the World Bank China CPI ; the computed nominal and real rates reproduce the NBS-published figures (e.g. 2020 +7.6% nominal, +5.2% real; 2025 +4.3%, +4.2%). Rounded to one decimal; a negative value is a pay cut in that measure. OECD 2025 values are provisional and China’s pre-2015 annual figures are indicative — both to re-verify before reuse; a 2024 change in NBS survey coverage lifts China’s 2024 level slightly.

Two stories run through the table. In the rich economies, real pay only crawled. Compounded across the whole period, real wages rose roughly +35% in the US, +29% in Canada, +26% in the UK, +23% in Australia, +21% in France and +20% in Germany — on the order of 1% a year — while Japan’s went nowhere at all, about −2% over the twenty-six years, the clearest case anywhere of a stagnant-wage economy and the pay-packet side of the flat nominal GDP in Table 1a.

The catch-up economies are a different world. Poland’s real wages rose about +77%, the fastest in the set, as post-2004 EU convergence lifted pay. China is off this scale entirely: on the NBS urban-units basis its wages multiplied many times over — roughly a fourteen-fold nominal rise and about an eight-fold real gain across the period — the wage side of the fifteen-fold GDP expansion in Table 1a, even allowing that the urban-units series flatters the true average. Its pace has since normalised, from double-digit real gains through the 2000s to low single digits by 2024–25.

The shocks read as sharply in pay as in output. The 2009 crisis dented nominal pay but low inflation cushioned real wages; 2020 threw up a statistical quirk — US real pay jumps +6.0% not because anyone got a raise but because low-wage jobs vanished first, lifting the average; and 2022 is the broad real pay cut, as inflation outran wages almost everywhere (real pay fell about −2.9% in the US, −2.2% in Australia, −2.1% in Germany and −2.0% in the UK) — the mirror image of the nominal-growth spike in Table 1c, and the reason Part 2 turns next to what inflation does to a fixed income. For the fiscal picture the link is direct: nominal wage growth is most of what fills the income- and payroll-tax base Section 8 finds nearly maxed out, and a quarter-century of near-flat real pay is part of why the pension and health promises of Section 3 feel heavier with each passing year.

3. Where the money goes

Before debt, spending. Governments do not spend on a thousand small things in equal measure; a handful of functions dominate every big-economy budget, and — with one important exception — the mix is remarkably similar across them, because it is driven by the same forces of ageing populations and universal programmes.

Start with the scale of the two flows. In 2024 total general government spending ranged from about 38% of GDP in the US to 57% in France (Germany ~50%, the UK ~45%, China ~33%), with revenue a few points below in each — roughly 31% (US), 51% (France), 47% (Germany), 40% (UK) and 26% (China). The gap between those two, country by country, is the deficit that Table 5 tracks: France raising 51% against 57% spent is its ~−5.8%, the US’s 31% against 38% its ~−7.3%. This section is about the larger of those two flows — where the spending goes.

Put the nine economies side by side on a common basis — general government spending by function, as a share of the total — and the pattern is clear. Two or three blocks account for well over half of everything a government spends, and most of the differences between countries are differences of degree. The one difference of kind is China, which channels a far larger share into investment than any of the others.

Table 2. Government spending by major function: share of total (latest year)

Source: general government expenditure by function (COFOG), as a share of total general government expenditure, latest available year (2023–24). France, Germany and Poland from Eurostat COFOG (2024); the UK on the same basis from ONS /OECD; the US, Japan, Canada and Australia from OECD Government at a Glance / IMF GFS; China from its Ministry of Finance general public budget , broadly but not strictly COFOG-comparable. Shares are rounded and sum to 100; the four added economies (Japan, Poland, Canada, Australia) are approximate and should be re-verified. Pensions is the COFOG “old age” sub-division; social protection (excl. pensions) is the rest of that division — unemployment, family, sickness and disability, and income support. “Other” pools public order and safety, housing, environmental protection, recreation and — for China — public security and science & technology.

Figure 2. Health leads in the US; pensions in Japan, Poland and Europe; investment in China.

Source: figures as in Table 2; the seven main functions are shown and “Other” is omitted. Each country keeps its colour across every function; the shares are printed in Table 2.

A modern budget is really six big commitments. Here is each in turn — what it is, how large it is, and where the money inside it actually goes.

3.1 Pensions and social protection

Pensions and the rest of social protection are the biggest blocks across the rich world, and together they dominate the budget — around 41% of all spending in both France and Germany, and 34% in the UK. Pensions alone run to roughly a quarter of spending in France (23%, ~€390bn) and a fifth in Germany (20%, ~€430bn), the fingerprint of older populations and generous public schemes; the rest of social protection — unemployment, family and children, sickness and disability, and income support — adds another 18–21% in those countries. Both are smaller in the US (12% pensions, 7% other), where more provision runs privately and the public safety net is dominated by federal Social Security (about $1.5 trillion a year on the OASDI basis — old-age, survivors and disability — wider than the old-age row in Table 2), and smallest in China (9% and 5%), a thinner welfare state that its slowing economy and rapidly ageing population are now straining.

Almost all of this is cash: pensions and most benefits are transfers written straight to households, not goods or services the state buys — which is what makes the block so hard to trim, since cutting it means sending a named person a smaller cheque. And it grows automatically as populations age. That pressure is simple demographic arithmetic, and it is visible across the last quarter-century: people keep living longer while the age at which they can start drawing a pension has barely moved — and in France sits below every neighbour’s — so each retiree is paid for more years (Table 2a).

Table 2a. Living longer, retiring no later: life expectancy and pension age, 2000–2025

Source: life expectancy at birth (years) from the World Bank ; the “2024/25” life-expectancy figure is the latest available year (2023). Statutory (normal) pension age is the men’s age in force that year, rounded to the nearest year, from OECD Pensions at a Glance 2025 and national reforms; multi-year reforms raised the age in monthly steps by birth cohort, so transition-year values are approximate, and women’s ages were lower in the UK (until 2018) and China. The “2024/25” pension-age figure is the current statutory target: France 64 (phasing to 2030), China 63 (men, phasing from 2025), with the UK and Germany heading to 67 (by 2028 and 2031). Life expectancy at birth understates time in retirement — those who reach pension age live well beyond the average — so pensions are drawn in practice for roughly two decades. Note the COVID dip in the US (78.8 in 2019 to 77.0 in 2020).

3.2 Health

Health is the fastest-growing group, and its weight flips the ranking. In the US it is the single largest block — around a quarter of all government spending (~$2.5 trillion a year) — because Medicare and Medicaid are huge public programmes even in a system with heavy private provision. Medicare is the federal health-insurance scheme for people aged 65 and over (plus some younger people with disabilities); Medicaid is a joint federal–state programme covering low-income households, including much of long-term nursing care. Between them they insure well over a third of Americans. The UK’s NHS pushes health to ~19% of spending; in France and Germany it runs 15–16%; China spends the least (~7%). Ageing and rising treatment costs push it up structurally everywhere.

Unlike a pension, health is bought — care delivered by hospitals, doctors and drug-makers — rather than handed over as cash, which is part of why its cost is so hard to contain. The US’s ~$2.5 trillion government health bill runs through a handful of big programmes (Figure 2a): Medicare (for the over-65s) and Medicaid (for low-income households) are more than 70% of it between them, with veterans’ and military health and a long tail of other public programmes making up the rest — and almost all of it is paid out to providers for care delivered, not sent to households as cash.

Figure 2a. Where the US health dollar goes: two insurance giants.

Source: US government (federal, state and local) health spending, latest available year (FY2024), approximate composition. Medicare (~$0.9tn) and Medicaid (federal + state, ~$0.9tn) from CMS and KFF ; veterans’ and military health from the VA and Department of Defense; “other public programmes” pools CHIP, ACA premium subsidies, and federal, state and local public-health spending. Shares are rounded and indicative.

3.3 Education

Education is a larger share in the US (~13%) and China (~15%) than in much of Europe (8–10%) — in the US because it is funded mostly by states and localities and so shows up in general-government totals, and in China because it is a stated national priority. It is, above all, a payroll: around 80% of US public-school current spending goes to staff salaries and benefits — teachers, aides and administrators — with the rest split between buildings, equipment, transport and materials. That is why education spending tracks teacher numbers and pay far more than any other lever — you cannot cut it much without cutting people. It is also the one big block that shrinks as a share over time in ageing societies, quietly giving way to pensions and health.

3.4 Economic affairs and investment

Economic affairs — the investment block — is where China stands apart. Roads, rail, energy, industrial subsidies and state-directed investment absorb around 28% of Chinese government spending, against roughly 10–11% across the rich economies. In nominal terms that is a vast sum — out of a general public budget of about 28.6 trillion yuan (~$4 trillion) — flowing into high-speed rail, power grids, ports and factories rather than into transfers to households. In the West the same block is smaller and more mixed: transport and infrastructure, energy, agricultural support, R&D and subsidies to business. It is the fingerprint of the investment-led growth model, and the single clearest way the Chinese budget differs from a Western one — the state building things rather than paying people.

3.5 Defence

Defence is dominated by the US in absolute terms — close to $1 trillion a year (~3.4% of GDP), more than the next several combined — but as a share of GDP the leader is now Poland, at ~4.2%, which has rearmed faster than anyone on NATO’s eastern flank since 2022. China is ~1.7%, and the rest cluster around the NATO 2% guideline: the UK 2.3%, France 2.1% and Australia 2.0%, with Germany 1.9%, Japan (~1.4%) and Canada (~1.3%) still below it (Figure 2b). As a share of spending the US runs ~9% against only ~3% in France and Germany, where the post-war settlement pushed money toward welfare instead — a gap Europe is now, quickly, closing.

Figure 2b. Military spending as a share of GDP, 2024.

Source: SIPRI military expenditure, 2024 (approx. $997bn US, $314bn China, $88bn Germany, $82bn UK, $65bn France, $55bn Japan, $36bn Australia, $35bn Poland, $30bn Canada). SIPRI’s definition is wider than COFOG defence — it includes military pensions and paramilitary forces — so it exceeds the general-government defence figures in Table 2. Poland’s ~4.2% is the highest ratio of the nine.

And a defence budget is not one thing. The US Department of Defense’s roughly $850 billion budget splits into four big pieces (Figure 2c): operations and maintenance — fuel, spares, training, keeping ships, jets and bases running — is the largest at about 39%; personnel (pay and benefits) takes another ~21%; and, unlike a European budget dominated by pay and upkeep, more than a third goes to new capability — procurement (~20%: the aircraft, ships, tanks and missiles themselves) plus research and development (~17%). That tilt toward buying and inventing hardware, not merely paying and maintaining, is what a superpower budget looks like — and the reason NATO presses smaller members, whose budgets are dominated by personnel and upkeep, to put at least a fifth into equipment.

Figure 2c. How the US defence budget divides: upkeep, people, kit, R&D.

Source: US Department of Defense budget by appropriation title, FY2024, approximate shares from CRS and AIP — operations & maintenance ~39%, personnel ~21%, procurement ~20%, RDT&E ~17%, military construction and other ~3% (DoD base ~$850bn). This DoD-budget basis is narrower than the ~$997bn SIPRI figure in Figure 2b, which adds military pensions and other national-defence items.

3.6 General public services and interest

General public services — which includes debt interest — is the block that connects this section to the debt chapters. It runs 8–14% of spending across the nine, and interest — the price paid to bondholders for past borrowing — is the fast-rising part of it: unlike pensions or health it buys nothing new, it services the past, and it climbs with both the debt stock and the interest rate. In the US that interest bill has already passed $1 trillion a year. Sections 4, 5 and 6 take it up in full.

The through-line across all six blocks: most of a modern budget is mandatory — promises already made to pensioners, patients and bondholders — not a blank sheet re-decided each year. That is why deficits are so persistent, and why they accumulate. The one budget that breaks the Western pattern, China’s, does so by spending less on people and more on things — its own kind of commitment, and its own kind of risk.

4. How deficits become debt

A deficit adds to the debt every year, so the debt only stops growing when the economy grows fast enough to outrun it. That race — how fast the economy grows against how fast the debt piles up — decides almost every divergence in the tables around it.

Seen across the last quarter-century, the race is not close, and its outcome is the debt build (Table 3).

Table 3. What outran what: debt vs GDP growth, 2000–2024

Source: derived from figures already in this guide — the nominal GDP multiple is the 2000→2024 column of Table 1a; the debt ratios are the 2000 gross-debt column of Table 4 and the 2024 IMF estimates shown in this table’s final column. “Govt debt grew” is the implied nominal debt-stock multiple (the debt ratio × GDP), which equals the GDP multiple times the change in the debt ratio — indicative and rounded. “Debt” is general government gross debt, % of GDP.

Every government’s debt stock grew — but the debt ratio only held where the economy nearly kept pace. Germany is the tell: its debt grew about ×2.5 while its economy grew ×2.4, the two almost matched, so its burden barely moved (59% → 62% of GDP). Everywhere else the debt outran the economy — US debt swelled ×6.6 against a ×2.9 economy, China’s ×60 against an already-fast ×15.3 — and the ratio climbed. None of them shrank their debt in cash terms; the ratios diverged purely on whether the economy could keep pace with the borrowing.

Why does growth matter so much? Because debt is always measured against the size of the economy, so two forces pull the ratio in opposite directions. Borrowing adds to the debt and pushes the ratio up; growth enlarges the economy and pushes it down, because the same debt becomes a smaller slice of a bigger whole. When an economy grows faster than its debt — as the rich world’s did through the near-zero-rate 2010s, and as China’s booming economy did for two decades — the debt ratio can hold flat or even fall while the government still runs deficits. When growth stalls, those same deficits send the ratio climbing. Interest is the quiet third force: the more a government already owes, the more of each year’s budget goes to servicing old debt rather than anything new — and when borrowing costs rise, that burden grows on its own.

This is why the 2020s worry economists more than the 2010s did. After a decade of ultra-low rates that flattered every government’s position, interest rates rose sharply in 2022–23 just as growth cooled from its post-pandemic spike — so the same debt stock suddenly costs more to carry, and economies can no longer outgrow it as easily. That is exactly why net interest is climbing as a share of spending (Section 3), and why the trajectory of the bond market and central-bank policy — the subjects of the next two guides in this series — now matters so much for the fiscal picture. For the cross-asset consequences of that shift, see the macro-regime guide .

5. The debt build, 2000–2025

Put the nine economies side by side and the pattern is unmistakable: a broadly rising debt burden, punctuated by two step-changes, but with paths that diverge far more than the shared direction suggests. The range is now enormous — from Japan’s ~207% of GDP, the highest in the developed world (and ~229% at its 2020 peak), down to Australia’s ~51% and Poland’s ~59%, both of which entered the century under 40%.

Table 4. General government gross and net debt, % of GDP, 2000–2025

Source: gross debt — general government gross debt, % of GDP, from the IMF World Economic Outlook (France, Germany, China, Japan, Poland, Canada and Australia on the IMF WEO general-government series, IMF DataMapper current vintage; the US and UK the same IMF series via FRED , UK on the Maastricht-comparable general-government basis). Net debt — general government net debt (gross debt minus general-government financial assets), % of GDP, from the IMF Fiscal Monitor (April 2026 vintage); China publishes no general-government net-debt series, so its Net row reads —. Rounded to whole points. The IMF’s US general-government series begins in 2001, so the 2000 US cells (gross 53*, net 34*) carry that first value; 2025 figures are IMF estimates. Gross (WEO) and net (Fiscal Monitor) come from related but distinct IMF products; all gross rows, including the four later-added economies (Japan, Poland, Canada, Australia), are the current IMF DataMapper WEO vintage (read August 2026), so individual cells can still differ by a point from a reader’s live pull.

Gross versus net. The table shows the debt two ways. Gross debt is the headline figure — the full stock of a government’s outstanding liabilities, the number quoted in almost every debt-ceiling or deficit story. Net debt subtracts the financial assets the government itself holds — cash and deposits, loans it has made, and the securities and equity sitting in its own accounts and, in several countries, its public-pension and sovereign-wealth funds. Net debt is the closer measure of what the state would still owe if it cashed those assets in, so it is always the lower of the two, and the gap between them is a rough gauge of how much a government owns set against what it owes.

That gap is narrow where the state holds few financial assets and wide where it holds many. Canada carries ~114% of GDP gross but only ~10% net, because the assets of the Canada and Québec Pension Plans net against the debt; Australia was an outright net creditor in the mid-2000s — its net debt turned negative as the Future Fund and other holdings outgrew its modest borrowing, even with gross debt near 10%; Japan’s net debt (~136%) runs roughly 70 points below its gross (~207%), thanks to the vast securities held by its social-security and government funds. The catch is comparability: what counts as a “financial asset,” and how it is valued, differs enough between countries that net debt is the softer, less standardised number. That is why the headline ratios, the deficits that build them, and the fiscal-health score later in this guide all run on gross debt and the annual flows — and why China, which publishes no net-debt series, appears here on gross alone.

Two shocks carry the story. Before 2008, debt ratios were flat or falling almost everywhere. The 2008 financial crisis changed the level permanently: bank rescues, recession-hit revenues and stimulus pushed US debt from ~65% to ~95% of GDP in three years, UK debt from the low-40s to the mid-70s, and France from ~65% to ~86%. The ratios never fully retraced before the next shock. The 2020 pandemic repeated the pattern at a larger scale: emergency support drove the US above 130% and the UK above 100% in a single year. The table brackets each hinge — 2008 beside 2009, 2019 beside 2020 — so the single-year jumps stand out, with only Germany bucking the pattern.

Behind those debt levels sits the flow that feeds them — the annual deficit. The deficit table shows the same two shocks even more starkly, because a deficit reacts to a crisis instantly while the debt stock only accumulates the damage.

Table 5. General government deficit (−) / surplus (+), % of GDP, 2000–2025

Source: general government net lending (+) / borrowing (−), % of GDP, from the IMF World Economic Outlook; France, Germany, China, Japan, Poland, Canada and Australia are the IMF WEO general-government series (IMF DataMapper current vintage), the US and UK the same IMF series via FRED ; the four added economies’ series were compiled from published IMF WEO values and should be re-verified. Negative = deficit; surpluses are highlighted. The US series begins in 2001; 2025 figures are IMF estimates.

Two features stand out. First, the 2009, 2010 and 2020 columns are where the double-digit deficits sit — the crisis years, when deficits widen automatically as revenues fall and benefits rise (the automatic stabilisers) before any rescue package is voted — while the years in between show deficits that shrink but rarely close. Second, Germany is the outlier again: near balance or surplus in most normal years (the accent-coloured surpluses run right through 2013–2019), which is exactly why its debt ratio fell while others’ rose. And notably, the US is still running a deficit above 7% of GDP in 2024, the UK near 6%, and China above 7% — years after the pandemic ended — which is what keeps their debt stocks climbing even outside a crisis.

Seen year by year, the US shows the mechanism whole: the deficit spikes in every crisis (Figure 3), and the debt it accumulates steps higher at each shock, never fully retracing (Figure 4).

Figure 3. The US deficit spikes in crises but no longer returns to pre-crisis lows.

Source: US row of Table 5, every year 2001–2025, shown as magnitude (every value is a deficit). Column heights read off the y-axis; the year-by-year values are in the table.

Figure 4. US debt climbed almost every year, then stepped up at 2008 and 2020.

Source: US gross-debt row of Table 4, every year 2001–2025 (the IMF US series begins in 2001). Column heights read off the y-axis; the year-by-year values are in the table.

But the level in 2024 is where the economies part ways: the US, France and the UK cluster above or near 100% of GDP, China has climbed to ~90%, and Germany stands apart, having actually reduced its ratio since 2010 through a decade of near-balanced budgets.

6. What the debt costs — and who owns it

A large debt has two consequences the headline ratio doesn’t show: it carries a price — interest, paid in cash every year — and it has owners, the bondholders a government must keep willing to lend. Both decide how dangerous a given debt level really is.

The interest bill. Interest is the rent on past borrowing, and after a decade of near-zero rates it is climbing fast as maturing debt is refinanced at 2022–23’s higher rates. The US paid about $880 billion in net interest in 2024 and crossed $1 trillion in fiscal 2025 — roughly 13% of federal outlays, now more than it spends on defence. Europe pays far less relative to its economy: France’s €59 billion and Germany’s €49 billion are 2.0% and 1.1% of GDP, because both locked in ultra-cheap long-dated debt in the 2010s. The UK is the European outlier at around £100 billion (~3.5% of GDP), because a large slice of its debt is index-linked — the coupon rises with inflation, so the 2022 price spike fed straight into the interest bill. China’s interest is still small against GDP (~1%) but the fastest-growing line in its budget.

Table 6. What the debt costs: interest paid, 2024

Source: US net interest from CBO and the Peterson Foundation — $880bn in 2024, over $1tn in FY2025 (~13% of federal outlays, now above defence); France and Germany “public debt transactions” from Eurostat COFOG (2024); the UK (ONS /OBR) and China (its budget) are approximate. The US 13% is a federal-outlay figure (~9% of all US government spending); the European “% of spending” are shares of total general government expenditure.

Figure 5. The interest burden — highest in the US and UK.

Source: interest as in Table 6 (France and Germany exact from Eurostat; US, UK and China approximate). Shown as a share of GDP for comparability; in cash that is about $880bn (US), ¥1.4tn ≈ $196bn (China), £100bn ≈ $127bn (UK), €59bn ≈ $64bn (France) and €49bn ≈ $53bn (Germany) — over $1.3 trillion between them, buying nothing new.

The form: bonds. Almost all of this debt is marketable government bonds — US Treasuries, UK gilts, French OATs, German Bunds, Chinese government bonds — issued as short-dated bills and longer notes and bonds, auctioned to investors and traded daily. A minority is non-marketable: in the US, roughly a fifth of the total is intragovernmental debt the Treasury owes its own trust funds (Social Security, Medicare), never sold in the market at all.

Who holds it — and why it matters. Ownership shapes the risk more than the headline size does.

Table 7. Who holds the debt

Source: US from CRS “Foreign Holdings of Federal Debt” and the Peterson Foundation — foreign holders ~$9tn, Federal Reserve ~$4.4tn, ~20% intragovernmental; top foreign holders Japan, the UK and China. Germany’s ~26% non-euro-area share from ING ; euro-area holdings from the ECB ; China’s ~5–6% foreign share from CEIC ; UK from the DMO . Shares are approximate and mix “% of total” and “% of publicly held” debt — indicative, not exact.

The pattern splits the nine. Japan is the ultimate insider market — foreigners hold barely 7%, and the Bank of Japan alone owns about half of all JGBs, which is why the world’s highest debt has never triggered a funding crisis; Australia, at the other extreme, sells nearly half its bonds abroad. The US owes about a quarter of its debt abroad — Japan, the UK and China the largest holders — a vulnerability that the dollar’s reserve-currency status turns into a privilege, since the world still wants Treasuries. France and Germany sell around half their bonds to non-residents, and the ECB became a huge holder through years of quantitative easing — central-bank money creation used to buy government bonds, the subject of Part 3 of this series. The UK leans on its own central bank (the Bank of England still owns a quarter to a third of gilts) and on domestic pension and insurance funds. China is the mirror image: barely 5% of its government bonds are foreign-held — the debt is funded by domestic savings channelled through state banks, which insulates it from foreign creditors but concentrates the risk in its own banking system and in the off-balance-sheet local-government vehicles of Section 7. A high foreign share exposes a country to global sentiment and the exchange rate; a central-bank- or domestic-bank-heavy base is more insulated, but chains the debt to monetary policy or to the health of domestic finance.

7. Trajectory drivers and off-balance-sheet risk

Four forces decide whether a country’s debt ratio rises or falls, and the tables above are their fingerprints. Demographics — an ageing population — mechanically lifts pension and health spending. The primary balance is the policy lever: Germany’s near-balanced budgets versus persistent US and French primary deficits explain most of the divergence in Table 4. Growth and interest rates are the race of Section 4 — China’s fast growth long outran its borrowing costs, masking a rapid debt build. And shocks — 2008, 2020 — ratchet the level up in ways politically almost impossible to reverse.

7.1 The demographic engine

Of those four forces, demographics moves slowest and bites hardest — and it starts with one number. The replacement-level fertility rate is the average children per woman at which a generation exactly replaces itself: about 2.1 in a rich country. Below it, each cohort is smaller than the last, so — before immigration — the working-age base that funds pay-as-you-go pensions and health (Section 3) shrinks relative to the retirees drawing on it. Every one of the nine is now well below replacement, most for a generation (Table 8). The US hovered near replacement through the 2000s before sliding to about 1.6; France held near two into the mid-2010s before falling to 1.6; and China has collapsed to roughly 1.0 — among the world’s lowest — after the one-child policy and a later baby bust. Japan and Poland sit near the bottom at about 1.15, Canada at 1.25; only Australia (1.48) and France (1.61) hold above 1.5. The total fertility rate (TFR) shown here is the children a woman would have at current age-specific birth rates — the tap feeding the bottom of the age structure.

Table 8. Fertility rate vs the replacement line: births per woman, 2000–2024

Source: total fertility rate, births per woman, from the World Bank (SP.DYN.TFRT.IN), benchmark years 2000–2024 (2024 latest available). Replacement level is about 2.1 — the rate at which a generation exactly replaces itself in a low-mortality country; a TFR below it means each cohort is smaller than the last, before migration. Rounded to two decimals; only the US (through the 2000s) and France (into the mid-2010s) sat near or above 2.0 at any point this century, and both are now around 1.6.

Fewer births at the bottom and longer lives at the top (Table 2a) reshape the age structure — and the broad 15–64 “working-age” band hides most of the story, lumping a 20-year-old in with a 64-year-old. Broken into 10-year bands (Table 9, Figure 6), the contrasts sharpen. Japan is in a class of its own — the oldest population on earth, with more people over 80 (10.6%) than under 10 (7.1%) and nearly a quarter past 70. Germany and France come next, top-heavy in the 50–69 bands. China looks young on the 15–64 headline but is ageing fastest — it bulges in the prime 30–59 bands (its 50–59 band, at 16.5%, the single largest anywhere), a vast workforce today that is a vast retirement wave a decade out. Poland bulges in the 40–49 band (16.3%, the post-communist echo). The US, UK, Canada and Australia are flattest and youngest. The budget’s key number is the old-age dependency ratio — people 65+ per 100 of working age: Japan already carries about 51, Germany and France ~37, the UK, Poland and Canada ~30–31, the US and Australia ~28, China ~21 — each set to climb as those mid-life bulges cross 65.

Table 9. Age structure in 10-year bands, 2024 (% of population)

Source: population by 10-year age band, % of total, aggregated from the World Bank ’s 5-year age-group series by sex (SP.POP.{age}.MA/FE.5Y), each weighted by the country’s male/female population share (SP.POP.TOTL.MA/FE.ZS), 2024. Columns sum to ~100% (rounding) and reconcile exactly with the World Bank’s broad 0–14 / 15–64 / 65+ shares. The old-age dependency ratio — people 65+ per 100 aged 15–64 — is 28 (US), 21 (China), 51 (Japan), 37 (Germany), 31 (UK), 36 (France), 31 (Poland), 30 (Canada) and 28 (Australia).

Figure 6. Age pyramids unrolled: Japan skews old, Poland and China bulge in mid-life, the young economies sit flatter.

Source: values from Table 9 (World Bank, 2024). Each country keeps its colour across all nine bands, as in Figures 1 and 2; the exact shares are in Table 9. Read the shape: Japan’s bars climb rightward into old age, Poland’s and China’s pile up in mid-life, and the US, UK, Canada and Australia sit flatter across the young bands.

This is the engine beneath the two biggest spending blocks. A rising over-65 share mechanically lifts pensions and health (Section 3) — the mandatory, hardest-to-cut promises — at the very moment the working-age tax base that funds them stops growing, and the fast-growing 80+ “old-old” weigh most heavily of all on health and long-term care — which is why Japan’s 10.6% over 80 is such a warning. It is slow, compounding, and near-impossible to reverse on any budget timescale: the retirees of 2045 have already been born, while the children who would fund them largely have not — and Japan, already carrying one retiree for every two workers, is simply the country furthest down that road.

7.2 Off the balance sheet

But the headline debt ratio, for all its prominence, understates the true fiscal position, because the largest obligations are not counted as debt at all. They are the unfunded promises: pensions and health care already committed to future retirees, financed pay-as-you-go, with no bonds issued yet. These sit off the balance sheet until they come due.

Table 10. Selected unfunded / implicit liabilities (off-balance-sheet)

Sources: SSA Trustees 2025 report and its infinite-horizon projection ; Cato on the Social Security + Medicare share; European Commission 2024 Ageing Report; IMF Article IV on China’s augmented debt. Non-US figures are qualitative estimates/projections, not measured stocks.

The US case is the best-measured. Social Security’s own trustees put its 75-year shortfall — the gap between promised benefits and dedicated revenue — at $25.1 trillion in present value, and on an infinite horizon at $72.8 trillion; Social Security and Medicare together account for essentially all of the federal government’s long-run unfunded obligations. Europe’s version is the ageing cost the EU projects to 2070: pension and health spending rising as the workforce shrinks, an implicit liability that never appears as a bond. China’s is different in form — local-government financing vehicles carry debt that the IMF’s “augmented” measure counts but the headline number does not, which is why China’s true public burden is materially higher than the ~90% in Table 4.

The point for a reader watching the fiscal picture: the debt ratio is the visible tip. The demographic promises beneath it are larger, harder to reverse, and the reason “the debt problem” outlasts any single budget.

8. The tax ceiling

Deficits and debt are only half the ledger; the other half is the tax that funds them. How much a government already taxes decides how much room it has to close a gap by raising revenue rather than borrowing — so the tax burden is the boundary against which every rising interest bill pushes. Two lenses tell the story, each shown year by year below: tax-to-GDP (Table 11 — how much the state takes as a share of the economy) and the tax wedge (Table 12 — how heavily a single average worker’s pay is taxed, all-in).

Table 11. Tax-to-GDP: total tax revenue as % of GDP, 2000–2023

Source: total tax revenue including social security contributions, % of GDP, from OECD Revenue Statistics — the comparative series for 2000–2022 and Revenue Statistics 2024 for 2023. China is not in the OECD comparative series; the values shown (2007, 2019, 2022, 2023) are from OECD Revenue Statistics in Asia and the Pacific, comparable only from about 2019 (≈20% of GDP), so all other years read —. OECD Revenue Statistics runs about two years behind, so 2024–2025 are not yet published on this comparable basis. Rounded; re-source before reuse.

Table 12. Tax wedge: single average worker, % of total labour cost, 2000–2025

Source: tax wedge for a single worker at the average wage with no children — income tax plus employee and employer social contributions as a share of total labour cost — from OECD Taxing Wages , read from the OECD Data Explorer, every year 2000–2025. China is not covered by OECD Taxing Wages, so the whole row reads —. Rounded.

Two things stand out. First, how wide the range is: France takes about 46% of everything its economy produces, the US about 28%, China roughly 20% (Table 11). These are different social contracts — France funds a large welfare state (Section 3) out of tax, the US funds a smaller one and borrows the rest, and China raises little in broad tax, leaning on land sales, state-enterprise profits and the off-budget borrowing of Section 7. The direction crept upward in Europe (all three European economies took more of GDP in 2022 than in 2000) and stayed low and cyclical in the US.

Second, where the ceiling bites (Table 12). In France and Germany a single average worker loses close to half of their total labour cost to income tax and social contributions — among the highest in the world — against about 30% in the US and UK. A wedge that high is a practical limit: push it much further and take-home pay, hiring and compliance all resist. That is the tax ceiling — a maximum somewhere around 45–50% of GDP and a ~50% wedge, beyond which raising rates yields little and costs a lot.

This is exactly how deficits and debt press on taxes. A country near the ceiling — France, at ~46% of GDP and a ~47% wedge — has almost no headroom: it cannot tax its way out of a rising interest bill (Section 6), so the shortfall comes out as more borrowing. A country that taxes lightly — the US at ~28% — has nominal room to raise revenue but deep political resistance to it, and leans on deficits all the same. Either way, where the ceiling is close, debt is the release valve — which is why the two big deficit economies (Section 5) sit at opposite ends of the tax range yet reach the same result. The tax take is near its ceiling; the gap is on the other side of the ledger, where spending (Section 3) and the interest bill (Section 6) keep growing faster.

9. The Metal Pilot fiscal-health score

Everything so far has moved one variable at a time — growth and pay (Section 2), deficits and debt (Section 5), the tax ceiling (Section 8). But a household or finance minister feels them together, in the same year. This closing section rolls four of those threads into a single running index — the Metal Pilot fiscal-health score — so the quarter-century reads as one number per country you can watch rise and fall.

9.1 How the score works

The design is deliberately simple, and built to be tracked year to year. Every economy starts at 100 in 2000, then earns or loses points each year against four pillars — a good year +4, a bad one −6 — so the score is just 100 plus every year’s points since. Because the pillars score against a fixed benchmark, not the country’s own past, the points are comparable across economies and time.

Each pillar scores the same way: take the year’s reading, measure its distance from a neutral benchmark (the value that earns zero), multiply by a sensitivity, and cap the result so no single pillar or freak year can dominate. The four are then summed into that year’s points. Table 13 lays out the rule and works one year in full.

Table 13. The Metal Pilot fiscal-health score — how each year’s points are earned

Source: computed from this guide’s own series — real GDP growth (Table 1c), real wage growth (Table 1d), the single-worker tax wedge (Table 12), and the general-government fiscal balance (Table 5). Benchmarks, sensitivities and caps are Metal Pilot’s, chosen so a typical year lands in the low single digits and only genuine booms and shocks move the score sharply; they are a transparent house metric, not a rating of any security. The fiscal-balance benchmark is a −3%-of-GDP deficit — roughly the shortfall that holds a debt ratio flat at trend nominal growth, and the EU’s own reference value — so a surplus earns points and a large deficit loses them; the balance (Table 5) is used rather than the debt stock (Table 4) because it is the annual flow that builds the debt and the only debt-side series that runs every year. There is no separate inflation pillar: real wage growth is already pay after inflation, so a standalone inflation term would dock the same price shock twice. China has no comparable tax-wedge series (Table 12), so it is scored on the other three pillars — growth, wages and fiscal balance.

Two design choices are worth stating. Growth is scored against a single 2.5% trend for everyone, so a fast catch-up economy genuinely banks points — the point, not a bug: China’s and Poland’s incomes really did pull away from the rich world’s. And the score folds living standards and fiscal sustainability into one number: the fiscal-balance pillar marks down chronic 5–7% deficits even in a good growth year, and rewards a run of surpluses (Germany 2012–19, Canada and Australia through the 2000s).

9.2 The score, year by year

Table 14 is the index itself. The Score row is each economy’s running total; the Points row beneath is that year’s gain or loss, and the four small signed terms under it are the pillar contributions in Table 13’s order — GDP growth · real wages · tax wedge · fiscal balance (France’s 2021 sub-line +2.6 +2.5 −.4 −1.4 sums to +3.3). Read across to watch fiscal health rise and fall; read the sub-line to see which pillar moved it.

Table 14. The Metal Pilot fiscal-health score, 2000–2025

Source: the Score is the running Metal Pilot fiscal-health index — 100 in 2000, plus every later year’s Points accumulated. Points is that year’s gain or loss (Table 13); the signed sub-line beneath breaks it into its four pillars in the fixed order GDP growth · real wages · tax wedge · fiscal balance (leading zeros dropped, so .4 means 0.4). 2000 is the base year and scores no points. China is scored on three pillars — no tax-wedge series — so its sub-line carries three terms (growth, wages, fiscal balance). Rounded to one decimal.

Two patterns run through the points. First, the shared shocks: 2009 and, even more violently, 2020 dock points from almost every economy at once, as the growth and fiscal pillars slam to their caps together. The recovery is telling: 2021 is no clean mirror-rebound, because growth snapped back while deficits stayed crisis-wide, so the fiscal pillar kept dragging — fiscal health does not V-shape the way output does. 2022 is milder and uneven: with no separate inflation pillar, it shows up only where real pay was actually cut. Between shocks the annual points sit near zero, which is exactly why the crises stand out.

Second, the economies pull apart over time. China gains points in every year but the 2020 shock, its two-decade boom and eight-fold real-wage rise (Table 1d) pinning the growth and wage pillars at their caps, even as widening deficits after 2018 shave the fiscal pillar. Japan is its mirror, losing points in 24 of 25 years — not through crises but a chronic sub-trend grind, the fiscal face of Section 2’s flat wages and shrinking dollar-GDP. Between them the rich world splits along the fiscal pillar: the surplus-runners pull ahead, the chronic-deficit ones drift down.

Figure 7. Twenty-five years of points: China banks 123, Japan bleeds 61.

Source: Table 14 — each bar is an economy’s 2025 score minus the 100 every economy started with in 2000; bar lengths share one scale, set by China’s +122.8.

Read down the final column — Figure 7 draws it — and the quarter-century sorts the nine cleanly. China ends near 223, the clearest statement of a catch-up boom. Behind it a convergence-and-surplus tier: Poland (~136) on two decades of catch-up pay, and Canada (~130) and Australia (~128), lifted by the surpluses and low debt they carried through the 2000s. Germany (~111) is the rich-world standout — the one large old-Europe economy to finish above where it started, its 2012–19 surpluses exactly the result the debt chapters predict. The chronic-deficit economies cluster below 100: the US (~87), dragged under by two decades of 5–7% deficits despite strong growth and wage pillars; France (~86), still the most boxed-in, at the tax ceiling with the slowest real-pay growth (Sections 3 and 8); and the UK (~86). And Japan ends near 39 — no single collapse, just twenty-five years of losing a point or two at a time. The takeaway sharpens the guide: with the deficit in the score, fiscal discipline separates from fiscal drift — the surplus economies pull away from the perpetual borrowers even when their growth looks similar.

10. Summary

Every government spends, taxes, and borrows the gap — and across nine major economies those borrowed gaps have accumulated into debt burdens ranging from Australia’s ~51% and Poland’s ~59% to Japan’s world-record ~207%, with most of the rich world near or above 100% of GDP after the 2008 and 2020 shocks ratcheted the level up. But the shared direction hides different paths: the primary balance and the race between growth and borrowing costs — not fate — decide whether a ratio climbs like America’s and France’s, explodes from a low base like China’s, or falls like Germany’s. Spending is dominated everywhere by a few mandatory blocks — pensions, health, defence and interest — which is why deficits persist, and the visible debt is only the tip: the unfunded pension and health promises beneath it are the larger, slower risk. Carrying the visible pile already costs well over $1.4 trillion a year in interest across the nine — in the US, more than defence — and who holds the bonds shapes how dangerous each debt is. Nor can most simply tax the gap away: France and Germany already take close to half of every worker’s pay and sit near the tax ceiling, so debt becomes the release valve. The fiscal-health score (Section 9) rolls those threads into one index: China ends the quarter-century near 223, Japan near 39, and the surplus economies — Germany, Canada, Australia — pull away from the perpetual borrowers below 100. The next two guides pick up where this ends — how governments finance that debt in the bond market, and how central banks and inflation reshape the balance that decides it; for the cross-asset backdrop, the commodity supercycle guide traces how these forces play out over decades. When you are ready to move from the macro picture to the company- and project-level data underneath, that is what Metal Pilot is built for.

Next in the series → Part 2 · Bonds & Yields — how the debt is financed and priced in the bond market.

11. Vocabulary

12. Sources, methodology & disclaimer

12.1 Sources, methodology & data vintage

Debt and deficit figures (Tables 4 and 5) are general government gross debt and general government net lending/borrowing as a percent of GDP, from the IMF World Economic Outlook, every year 2000–2025. France, Germany and China are read from the IMF DataMapper (current vintage); the US and UK are the same IMF WEO series accessed via FRED (the UK on the Maastricht-comparable general-government basis, IMF-compatible). Because the US/UK and continental series come from slightly different WEO vintages, a handful of cells can differ by a point from the vintage a reader pulls today. The IMF’s US general-government series begins in 2001; the 2000 US cells show that first value, and 2025 figures are IMF estimates. Table 4 additionally carries general government net debt (gross debt minus general-government financial assets) from the IMF Fiscal Monitor (April 2026 vintage) — a related but distinct IMF product from the WEO gross series; China publishes no general-government net-debt series, so its Net row is blank. Table 3 (debt vs GDP growth) adds no new data — it is derived arithmetically from the nominal-GDP multiple of Table 1a and the debt ratios of Table 4 (2000) and the IMF’s 2024 estimates, the debt-stock multiple being the GDP multiple times the change in the debt ratio, so it is indicative and rounded. Spending shares (Table 2) are general government expenditure by function (COFOG) as a share of total expenditure, latest available year: France and Germany from Eurostat COFOG (2024); the UK on the same COFOG basis from ONS/OECD (it no longer reports to Eurostat); the US from OECD Government at a Glance and IMF GFS; and China from its Ministry of Finance general public budget (2024), a national-budget classification that is broadly but not strictly COFOG-comparable — the US and China shares are therefore best read as indicative rather than exactly like-for-like with the ESA-COFOG figures for Europe. The within-function breakdowns in Section 3 (health, defence and education) are US figures: government health spending by programme from CMS and KFF ; the Department of Defense budget by appropriation title from CRS (FY2024); and the staff-compensation share of US public-school current spending from NCES ; other nominal figures are from the cited national budgets. The GDP, population and growth figures in Section 2 (Table 1, Table 1a, Table 1b, Table 1c, Figure 1) are from the World Bank (nominal GDP and GDP per capita in current US dollars, and real growth), 2000–2025. The GDP expenditure-component shares (Table 1) — household consumption, gross capital formation, government consumption and net exports as a share of GDP — are from the World Bank national accounts and national statistics offices, latest available year (2023–24), rounded. The year-by-year growth series (Table 1c) pairs World Bank real GDP growth (constant prices) with the World Bank GDP deflator; nominal growth is the two compounded — growth in each economy’s own currency, the g that governs its own-currency debt dynamics — with 2025 a World Bank estimate. Wage growth (Table 1d) is average-wage growth from OECD Average Annual Wages (national currency; real on constant prices, nominal on current prices) for the eight OECD economies, with China on the narrower NBS urban-non-private-units basis — nominal computed from the NBS series and real deflated by the World Bank China CPI — and therefore not strictly comparable to the others; OECD 2025 values are provisional and China’s pre-2015 figures are indicative. Unfunded-liability figures are from the 2025 SSA Trustees report, the European Commission 2024 Ageing Report, and IMF analysis of China. The demographic figures in Section 7 (Table 8, Table 9, Figure 6) and the life expectancy in Table 2a are from the World Bank : the total fertility rate (SP.DYN.TFRT.IN), life expectancy at birth (SP.DYN.LE00.IN) and, for the 10-year age structure (Table 9, Figure 6), the 5-year age-group series by sex (SP.POP.{age}.MA/FE.5Y) aggregated to 10-year bands and weighted by each country’s male/female population share (SP.POP.TOTL.MA/FE.ZS), which reconciles exactly to the broad 0–14 / 15–64 / 65+ shares (SP.POP.0014/1564/65UP.TO.ZS) and the old-age dependency ratio (SP.POP.DPND.OL), all latest available year (2024). Statutory pension ages (Table 2a) are OECD/national figures, approximate in transition years. The replacement level of ~2.1 is the standard low-mortality benchmark, not a World Bank figure. The tax figures in Section 8 (Tables 11 and 12) are from the OECD : tax-to-GDP (Table 11) from Revenue Statistics comparative tables (total tax receipts incl. social security contributions, 2000–2022) and the tax wedge (Table 12) from Taxing Wages (single worker at the average wage, read from the OECD Data Explorer, every year 2000–2025), with China’s tax-to-GDP from Revenue Statistics in Asia and the Pacific (comparable only from about 2019, hence “—” for earlier years and the OECD-only wedge). Figures are rounded. On the four economies added to this edition — Japan, Poland, Canada and Australia: their GDP, growth, fertility, life-expectancy and age-structure figures are direct World Bank series, and their gross-debt figures (the gross rows of Table 4 and the 2000/2024 debt ratios of Table 3) have been refreshed to the live IMF DataMapper WEO vintage (read August 2026) — this cut Japan’s gross ratios ~20–25 points versus the compiled first edition and lifted Canada’s and Australia’s latest year; their remaining deficit paths (Table 5), tax series (Tables 11–12), spending shares (Table 2), interest bills (Table 6) and the qualitative holders/liabilities rows were compiled from published IMF WEO, OECD, SIPRI and national figures because those data endpoints were unreachable at compile time, and should be treated as best-available estimates to re-verify against the primary sources. Data as of August 2026; comparable cross-country figures are revised as new WEO and national vintages publish, so re-source before reuse.

12.2 Disclaimer & disclosure

This guide is for informational purposes only and is not investment advice, a recommendation, or a solicitation to buy or sell any security. Fiscal figures are estimates as of the stated date and are revised routinely; readers should verify against the primary sources before relying on any number. Nothing here should be taken as a forecast of any government’s finances or any market outcome. This article was prepared with AI assistance: figures were gathered from the named public sources and human-reviewed, but readers should independently confirm before acting. The author holds no position that the content is designed to promote.