Japan: Fiscal Health Check Aug. 2026 — Debt Down, Interest 15.6% of Tax

General government gross debt fell to 214.5% of GDP in 2024, down 5.8 percentage points from 220.3% in the previous year (IMF WEO, vintage WEO-2026-04, calendar year). Over the same period, general-account interest costs rose to 15.6% of tax revenue (FY2026 supplementary-budget-adjusted budget), from 10.8% in FY2025, an increase of 4.8 percentage points. While the relationship between the denominator (nominal GDP) and numerator (debt outstanding) has improved, higher interest rates are beginning to reach interest costs with a lag—this is the most important change in this issue. The provisional answer to “Is Japan okay?” is conditional: as long as nominal growth remains above the effective interest rate, the debt ratio will fall arithmetically, but the gap (r−g) narrowed from −5.22pt in 2023 to −2.90pt in 2024.

Key Points in This Issue

• Debt burdens (as a share of GDP) fell for the second consecutive year. Both gross and net debt declined within the same vintage, with the decline larger for net debt.
• The interest burden has begun to rise. General-account interest costs are ¥13.0tn (FY2026 budget, budgetary interest-rate assumption of 3.0%), equivalent to 15.6% of tax revenue.
• The holders of government bonds are changing. The BOJ’s share fell 3.8pt in one year, while banks, overseas investors, and public pensions absorbed the bonds.

In a word: While prices and nominal growth are pushing down the debt ratio, higher interest rates are gradually entering through another channel: interest costs.

Fiscal Health Scorecard

Direction compares with the prior period of each source; rows without a health judgement (holder shares, burden ratio) show up/down. Prev. issue = value in the previous issue. Series vintages: 2026-08, WEO-2026-04 (WEO- is the IMF WEO edition, others are the Cabinet Office projection); the prior-period comparison stays within one vintage. Gross debt (% of GDP): rank 1 of G7 (highest first) Net debt (% of GDP): rank 1 of G7 (highest first) Net interest / revenue (G7 rank): rank 6 of G7 (highest first)

The “direction” in the table is mechanically determined by comparison with the previous figure (the previous year, fiscal year, or quarter) and is not an assessment of whether the level is good or bad. The indicators use different bases (general government = IMF; central+local governments = Cabinet Office; general account = Ministry of Finance; flow of funds = market value), so figures cannot be added or subtracted across rows.

Question 1: Is the Debt-to-GDP Ratio Falling?

Within the same vintage, both gross and net debt are declining. In the IMF WEO (WEO-2026-04), gross debt fell from 222.7% in 2021 to 214.5% in 2024, while net debt fell from 157.6% to 141.7%. The decline in net debt (15.9pt) was larger than that in gross debt (8.2pt). The fiscal balance also narrowed from −2.4% in 2023 to −1.7% in 2024. Japan’s general government debt series has been revised in the past and is not continuous with levels often cited previously; therefore, no comparison of levels across vintages is made. Figures for 2025–2027 are IMF projections, not outturns.

On a central+local government basis (Cabinet Office medium- to long-term projections and data tables, vintage 2026-08), the primary balance was 0.0% in FY2025, up from −1.8% in FY2024. The three-fiscal-year average (FY2023–2025) was −1.3%, so the single-year and average pictures differ. The three-year average is a presentation devised for this column; the government has not defined an evaluation period. Honebuto (Basic Policy on Economic and Fiscal Management) 2026 places the “stable decline in central+local government gross debt as a share of GDP” at the core of fiscal management, and states that the PB is “an indicator to be monitored toward a decline in the debt-to-GDP ratio and managed over multiple years in a manner consistent with that stable decline” (as of 2026-07-21). Under this framework, the multi-year direction, rather than the sign in any single year, is the relevant object of analysis.

r−g was −2.90pt in 2024 (r=0.05%, g=2.95%). The sign remains negative, which works automatically to push down the debt ratio. However, it narrowed by 2.32pt from −5.22pt a year earlier. The extremely low r reflects the fact that low-coupon outstanding bonds account for most of the debt stock and that the measure is on a net-interest basis (after deducting interest received). The high g is driven more by prices than by real growth; potential growth was only 0.40% in FY2025 (Cabinet Office medium- to long-term projections). The gap from the debt-stabilising primary balance was +4.7% (pb=−1.55, pb*=−6.21); as long as it is positive, the debt ratio falls. This is an identity, not a standard of fiscal soundness. The key change this period is that it nearly halved from +9.1% in the previous year.

(IMF WEO, WEO-2026-04, 2024, general government, % of GDP. Canada 110.0/10.8/−2.1/0.0%, United Kingdom 99.9/92.4/−6.1/5.4%)

Japan ranks first in the G7 for both gross and net debt, in descending order of the figures, but ranks sixth for net interest ÷ revenue at 0.3%; in other words, its interest burden is among the lightest. Japan’s distinctive feature is the coexistence of a heavy stock and a light flow.

In short: Japan has the heaviest debt stock in the G7, but its debt-to-GDP ratio is falling, supported by nominal growth exceeding interest rates. That gap is narrowing.

Question 2: How Much Heavier Has Interest Become Relative to Tax Revenue?

On a general-account basis, interest costs ÷ tax revenue reached 15.6% in the FY2026 budget, sharply higher than 10.8% in FY2025. Interest costs of ¥13.04tn divided by tax revenue of ¥83.73tn produce this figure; interest costs rose by ¥3.9tn from ¥9.1tn in the previous fiscal year. However, this is the budgeted amount based on a 3.0% budgetary interest-rate assumption and differs from market conditions—the average 10-year JGB yield during August 2026 was 2.871%. Actual payments may not match the budgetary calculation.

The 15.6% figure and the IMF-based net interest ÷ revenue figure of 0.3% (2024) differ by an order of magnitude, but they are not inconsistent. The former divides general-account gross interest costs by national tax revenue alone; the latter deducts interest received from interest costs for general government (consolidated, including social-security funds) and divides the result by total revenue, including social-insurance contributions. The population, denominator, and numerator definitions all differ. When presented together, they measure separately “the burden within the national budget” and “the net burden of the government sector as a whole.”

Debt service was ¥31.28tn, or 24.9% of expenditure, of which ¥17.89tn was debt redemption expenses. These are accounting transfers based on the 60-year redemption rule and are excluded from expenditure in international comparisons. The overall financing picture is measured by gross financing needs (GFN). In FY2026, total issuance of ¥183.81tn ÷ nominal GDP of ¥689.30tn = 26.7%, down from 28.3% in FY2025. The composition was 73.9% refunding bonds, 17.8% new-financing bonds, and the remainder FILP bonds and others. Market-absorbed issuance was ¥168.50tn. The 20% benchmark used by the IMF in 2013 was abolished in 2022; no threshold judgment is made.

The speed at which higher interest rates reach interest costs is determined by maturity. The average remaining maturity of the debt stock was 9.3 years (FY2026, versus 9.4 years in the previous fiscal year), while that of the flow was 7.0 years (versus 7.2 years), meaning the flow shortened slightly. A debt stock maturity of about nine years means that even when market rates rise, only slightly more than one-tenth is replaced each year on average. Conversely, once rates rise, they continue to have an effect for a long time at roughly the same pace.

In short: The interest burden is determined not by “today’s interest rate” but by the average rate on bonds issued in the past. Higher rates do not arrive all at once; they take effect little by little each year, but for a long time.

Question 3: How Are Government-Bond Holders Changing?

Banks, overseas investors, and public pensions are jointly absorbing the amount reduced by the BOJ. The year-on-year changes in the flow of funds statistics (market value, excluding Treasury discount bills) are shown below.

The important point is that these ratios are on a market-value basis. Over the same period, total outstanding bonds fell from ¥1,055tn to ¥1,014tn—not because issuance declined, but because higher interest rates reduced the market value of outstanding bonds. Since sectors holding more long-term bonds experience larger market-value declines, the fall in the holding ratio combines bonds actually sold or redeemed with valuation declines. The BOJ’s face-value holdings of government bonds were ¥519.9tn (BOJ Statement of Transactions, August 2026), down 9.1% year on year; under the BOJ’s reduction plan, holdings are expected to be around ¥480tn at end-March 2027 (announced 2026-06-16; monthly purchases reduced by around ¥200bn each quarter, with monthly purchases of around ¥2tn from April 2027 onward).

BOJ current-account balances were ¥424.3tn. Applying the average uncollateralised overnight call rate of 0.98% produces an annualised remuneration cost of approximately ¥4.2tn (rough estimate; balances not eligible for remuneration and the tiered structure are not considered). There are two ways to interpret this. From the perspective of the consolidated government, long-term government bonds are being financed with short-term BOJ current-account balances, so a policy-rate increase immediately becomes a burden. On the other hand, interest received by the BOJ on government bonds ultimately returns to the government as transfers to the national treasury, so it is offset at the consolidated-government level. As remuneration costs rise, transfers to the treasury are compressed; the two views are opposite sides of the same phenomenon.

In short: Domestic banks and overseas investors are currently filling the gap left by the BOJ. However, part of the movement in “shares” is an apparent change caused by price declines, so it must be read together with face-value data.

Question 4: Do the Conditions of Yen Denomination and Domestic Ownership Still Hold?

They do, but overseas participation has clearly increased from three years ago. The 12-month cumulative current-account surplus from August 2025 through July 2026 was ¥35.8tn, up ¥6.7tn from ¥29.1tn in the same period a year earlier. The breakdown was primary income of ¥42.1tn and a trade surplus of ¥1.4tn; the surplus is driven mainly by interest and dividends from overseas assets. Japan’s net international investment position (NIIP) was ¥561.8tn at end-2025 (Ministry of Finance), ranking third in the international comparison in the same source. Japan as a whole is a net exporter of capital, and most government bonds are denominated in yen and absorbed domestically.

At the same time, the overseas holding ratio for government bonds rose to 8.1% (2026-Q1, market value), up 1.8pt from 6.3% in the same quarter a year earlier. The condition of predominantly domestic ownership has not broken down, but overseas investors are becoming more important as marginal buyers. Overseas investors are more price-sensitive than domestic banks and can adjust holdings more readily in response to changes in interest-rate differentials and currency-hedging costs. Sovereign ratings were Fitch A / Moody’s A1 / S&P A+ (as of 2026-04).

In short: The “defensive” conditions of own-currency denomination and domestic ownership still work, but overseas investors account for a larger share of new government-bond purchases, making price movements more susceptible to external factors than before.

This Issue’s Theme: Final Tax Revenue, Pension Reserves, and Social Security

FY2025 general-account final tax revenue was ¥84.22tn (announced 2026-07-31), slightly above the ¥83.73tn tax-revenue figure in the FY2026 supplementary-budget-adjusted budget. In other words, the FY2026 budget does not assume tax revenue above the previous fiscal year’s outturn. Calculated on an outturn basis, FY2025 interest costs ÷ tax revenue were ¥9.1tn ÷ ¥84.2tn = 10.8%. The FY2026 figure of 15.6% is a budget figure based on a 3.0% budgetary interest-rate assumption, so this level is not confirmed until the outturn is available. The distinction between “final accounts” and “budget” is essential when comparing the two.

Under the medium- to long-term projections (vintage 2026-08, central+local governments), the PB in the Growth Strategy Realisation Case 1 rises from −0.2% in FY2026 to +0.6% in FY2028 and +1.0% in FY2035, while debt and other government-bond balances as a share of GDP decline from 187.6% in FY2026 to 172.4% in FY2035. Under the Current Projection Case, the PB is +0.3% in FY2035 and the balance ratio is 182.9%, a 10.5pt difference between the cases. The main source of divergence is nominal growth rather than the PB: nominal growth in FY2035 is 3.3% in Case 1 versus 2.0% in the Current Projection Case. The interest-rate assumption is 2.70% in FY2026 in all cases; from FY2027 onward it rises from 3.0% to 3.4% in Cases 1 and 2, while declining from 3.0% to 2.7% in the Current Projection Case (assumptions published in 2026-07). Additional fiscal spending is mechanically set at ¥10tn in real terms each fiscal year from FY2027 onward. Even within the same projections, it is important not to overlook that interest rates are also set higher on the higher-growth path.

On the pension-reserve side, GPIF’s operating assets were ¥317.76tn at end-FY2026 Q1, with an annualised return of 4.95% since FY2001 and cumulative investment gains of ¥221.02tn. The asset allocation was almost evenly divided: domestic bonds 25.59%, domestic equities 24.48%, foreign bonds 24.60%, and foreign equities 25.33%. Quarterly returns fluctuate substantially with market movements and should not be read as a measure of fiscal health. In the flow of funds data, public pensions’ government-bond holding ratio was 7.3% (2026-Q1), up 1.3pt year on year, making them one of the recipients of the BOJ’s reduction.

Social-security benefits amounted to ¥138.30tn (FY2024, ILO standard), or 21.53% of GDP. This is the scale of benefits for the system as a whole, including insurance-premium financing. By contrast, general-account social-security expenditure was ¥39.06tn in FY2026, or 31.2% of total expenditure of ¥125.42tn. The two figures cover different populations and cannot be compared directly. General-account spending fell from ¥50.2tn in FY2021 to ¥36.2tn in FY2023, then rose to ¥39.5tn in FY2025 and ¥39.1tn in FY2026. The expiry and subsequent increase of COVID-related measures created the fluctuations. Social-security expenditure (¥39.1tn) and debt service (¥31.3tn) continue to account for around 56% of expenditure.

In short: Tax revenue remains near a record high, but the FY2026 budget does not assume further growth above it, leaving higher interest costs and social-security spending competing within the same expenditure envelope. Pension reserves have accumulated when viewed over the long-term annual rate of return, and their role as government-bond buyers is also increasing.

Sensitivity: Mechanical Calculation with Assumptions Held Constant

The following is not a forecast. It shows debt ratios when shocks are applied mechanically, holding the 2024 base year fixed (d0=214.5, r=0.05%, g=2.95%, PB=−1.55%, annual refinancing ratio 0.165).

The conditions for each scenario are as follows. Extending the current trend requires the net-interest effective rate of 0.05% to remain in place for five years—namely, that low-coupon outstanding bonds and the current ownership structure do not change substantially. An interest-rate increase of 1pt with growth unchanged corresponds to a rise in the real interest rate; the debt ratio after five years is 4.7pt higher than under the current-trend scenario. If interest rates and nominal growth both rise by 1pt, the real interest rate is unchanged and the ratio falls instead—this is the channel through which inflation reduces the debt ratio.

Factors not included in the table should also be made explicit. The impact of higher interest rates on growth and tax revenue, changes in transfers from the BOJ to the treasury, and residual factors such as advance-funded bonds, foreign exchange, and equity contributions (the “other factors” in the medium- to long-term projections) are not incorporated. The portion of bridge bonds managed separately is excluded from the PB and balances in the medium- to long-term projections, so the scope differs from the IMF series.

Q&A for Non-Specialists

Q1. What does “fiscal collapse” mean? A. It is not a single event; there are at least four forms: (1) default on yen-denominated debt, which is generally unlikely for debt in a country’s own currency; (2) erosion of the real value of debt through inflation; (3) a decline in external purchasing power through yen depreciation; and (4) financial repression, in which policy maintains low interest rates and suppresses holders’ real returns. Japan’s debt is yen-denominated, and the overseas holding ratio remains only 8.1% (2026-Q1, market value). Thus, the conditional assessment is that adjustment is more likely to occur through forms 2–4 than through form 1.

Q2. What are the arithmetic routes to stabilising the fiscal position? A. There are only four in terms of the identity: raise nominal growth, inflate nominal GDP through prices, restrain expenditure, or increase revenue. The decline in the debt ratio in 2024 was driven mainly by nominal growth (g=2.95%), while the PB itself was a deficit of −1.55%. Those supporting fiscal expansion argue that expanding the denominator through growth investment is more reliable. Those prioritising fiscal consolidation argue that a denominator effect dependent on growth will disappear if inflation settles. Both arguments ultimately return to the same issue: how long the sign of r−g remains negative.

Q3. What happens, and when, if interest rates rise? A. Higher market rates do not affect outstanding bonds; they are applied progressively only to refinanced and newly issued bonds. With an average remaining maturity of 9.3 years for the stock, bonds are replaced at an average pace of slightly more than one-tenth per year. The flow maturity is somewhat shorter at 7.0 years, accelerating transmission. At the same time, the market value of outstanding bonds falls, so holding ratios and holders’ valuation gains or losses in the flow of funds statistics move first.

Q4. What happens when the BOJ holds government bonds? A. BOJ holdings are financed by remunerated current-account balances, creating an annual cost of approximately ¥4.2tn based on ¥424.3tn × the 0.98% policy rate (rough estimate). Interest received by the BOJ on government bonds returns as transfers to the national treasury, so one view is that it is offset for the government sector as a whole. Another view is that shortening the liability maturity increases sensitivity to higher interest rates. The BOJ is reducing its holdings (¥519.9tn at face value, down 9.1% year on year), so the importance of this issue is likely to diminish over time.

What Would Change the Assessment?

Three figures and conditions to check before the next issue are as follows.

1. Outturn interest costs ÷ tax revenue: The FY2026 budget figure of 15.6% assumes a 3.0% budgetary interest rate. If final interest costs are below ¥13.0tn, the budget was conservative; if they exceed it, transmission was faster than expected. The tax-revenue side should also be assessed against FY2025 final tax revenue of ¥84.22tn.
2. The absorption structure from flow of funds 2026-Q2 onward: As the BOJ ratio continues to decline from 47.9%, will domestic banks at 13.2% and overseas investors at 8.1% continue to rise? If only the overseas share expands, the pattern of price movements will change. The figures should also be cross-checked against face-value data (BOJ Statement of Transactions and the plan for around ¥480tn at end-March 2027).
3. The sign of r−g: It was −2.90pt in 2024, narrowing by 2.32pt in one year. In the medium- to long-term projections (2026-08), the path assumes an effective interest rate of 2.8% in FY2034 in Case 1 against nominal growth of 3.3%, bringing the gap closer. Whether this gap narrows further or reverses in the next IMF vintage and the next medium- to long-term projections will determine the direction of the gap from the debt-stabilising PB ( +4.7% this time).

Monthly movements in government-bond yields and monthly tax-revenue progress are covered in the monthly fiscal-analysis column (reference: as of end-August 2026, 10-year 2.943%, 30-year 4.092%, 2-year 1.743%).

Glossary

---

This column was automatically generated by AI integrating the IMF World Economic Outlook, Bank of Japan flow-of-funds, balance-of-payments and accounts statistics, Ministry of Finance debt-outstanding and JGB interest rate data, the Cabinet Office medium- to long-term economic and fiscal projections, and figures transcribed from published budget, issuance-plan and GPIF documents, as a fiscal health review. Sensitivity tables are mechanical calculations under fixed assumptions, not forecasts. This is not a recommendation to buy or sell any financial instruments or government bonds. Please make investment decisions at your own responsibility and consult professionals as needed.