Japan Nationalized Time
Four stages of a long financial crisis—from trade-surplus recycling and property inflation to zombie banking and the sovereign balance sheet.
When I worked in banking, Japanese banks seemed able to lend dollars at spreads we could not match. They arrived with strong ratings, implicit support and enormous balance-sheet ambitions. Japan's export machine was generating foreign claims, but the banks were not simply relending a pile of export dollars. They also funded themselves in Eurodollar and wholesale markets. The important point was the price of risk: it was extraordinarily low.
That experience is where this story begins. Japan's crisis was not a single event. It was a slow transfer of losses from one balance sheet to another. Rising land values supported more credit. More credit supported higher land values. When the loop broke, the assets fell immediately but the accounting recognition did not.
The losses did not disappear. They moved—from property owners to banks, then toward the government and the Bank of Japan.
A crisis measured in decades
Stage 1 — The surplus and the balance sheet
Japan's merchandise trade surplus rose from roughly $20 billion in 1983 to $82 billion in 1986. Those dollar receipts became part of a larger accumulation of foreign claims. At the same time, private credit and bank assets expanded rapidly relative to GDP.
The graphs do not prove that the trade surplus caused the property bubble. They show the financial setting in which it formed: plentiful savings, strong external accounts, a rising yen after the Plaza Accord, falling domestic rates and banks determined to grow.
| Year | Trade surplus | Private credit / GDP | Bank assets / GDP |
|---|---|---|---|
| 1983 | $20.4bn | about 129% | about 189% |
| 1986 | $82.0bn | 140.6% | 206.4% |
| 1989 | $64.4bn | 167.7% | 228.4% |
| 1990 | $52.9bn | 170.5% | 226.7% |
Stage 2 — The assets fell; the credit remained
Equity and land prices reversed before the Kobe earthquake. Falling collateral should have forced loan write-downs, foreclosure and restructuring. Instead, credit relative to GDP stayed extraordinarily high. Banks extended or renewed loans to borrowers that could no longer support them economically, preserving accounting values while delaying recognition.
| Year | Private credit / GDP | What the balance sheet says |
|---|---|---|
| 1990 | 170.5% | Bubble-era leverage is still present |
| 1993 | 174.9% | Credit has not cleared with asset prices |
| 1998 | 188.3% | Ratio reaches a late high |
| 2005 | 170.5% | Only slowly returns to its 1990 level |
Kobe was a shock, not the origin
The earthquake struck on January 17, 1995, several years after the equity and property reversals. Industrial production fell sharply in January, recovered by March and exceeded its December level in April. Kobe destroyed physical capital, and Japan rebuilt it. The property collapse destroyed financial capital, and the banking system pretended much of it was still there.
Stage 3 — Recognition arrived a decade late
The public non-performing-loan series begins in 1998. Reported NPLs peak at 8.4 percent of gross loans in 2001—roughly a decade after the underlying property losses began. That late peak should not be read as a new wave of bad lending in 2001. It is evidence of delayed recognition.
| Year | NPLs / gross loans | Interpretation |
|---|---|---|
| 1998 | 5.4% | Public series begins after losses are embedded |
| 2001 | 8.4% | Measured peak arrives long after the crash |
| 2003 | 5.2% | Recognition and disposal accelerate |
| 2005 | 1.8% | Reported cleanup largely complete |
Near-zero rates reduced the carrying cost of bad loans. Weak borrowers could survive without restructuring. Immediate liquidation was avoided, but labor and capital remained trapped in firms that could not earn their way out. Time became the hidden subsidy.
Stage 4 — The sovereign balance sheet
Private losses migrated toward government debt and the Bank of Japan. Extremely low yields made the accumulated debt appear manageable. Yield-curve control then suppressed the visible financing cost even as the stock of debt and the central bank's bond holdings grew.
That bargain is now less comfortable. Holding yields down can weaken the yen and add to imported inflation. Allowing yields to rise reveals the fiscal cost concealed by zero rates and reduces the value of long-duration bonds. Japan is not in default, but the policy corridor has narrowed.
| Checkpoint | Value | Why it matters |
|---|---|---|
| FY2025 debt-service budget | ¥28.22tn | The zero-rate dividend is fading |
| FY2026 debt-service budget | about ¥31.3tn | Interest expense rises before all debt reprices |
| Jan. 2026 ten-year yield | 2.24% | A material break from the YCC era |
| Aug. 2026 ten-year yield | 2.94% | Duration risk becomes visible |
| Sep. 2026 USD/JPY | roughly 154–160 | The currency remains historically weak |
Where did the losses go?
Property owners and leveraged firms accumulated the original risk.
Impaired collateral moved onto bank balance sheets and remained there.
Recapitalization and fiscal borrowing financed the delay.
Bond purchases concentrated duration risk on the central-bank balance sheet.
A quiet echo from China
China is not Japan, and this article is not a forecast that it must follow the same path. Its capital controls, state ownership, development stage and political structure are different. Yet the resemblance is worth watching: an enormous trade surplus, property-sector impairment, banks operating inside an implicit state guarantee and lending rates that have moved slowly downward as domestic demand weakened.
The lesson from Japan is not that a large trade surplus automatically creates a financial crisis. It is that abundant savings and administrative control can postpone recognition. If falling rates preserve borrowers and collateral values without forcing restructuring, the apparent stability may be a transfer rather than a cure.
Japan avoided liquidation by postponing the recognition of loss. For three decades, near-zero interest rates made delay appear inexpensive. Now time is charging interest.
Run the complete analysis in RainbowStats or download the script. Data series: Japan merchandise trade balance, private credit/GDP, bank assets/GDP, industrial production, non-performing loans, yen/dollar, and ten-year JGB yield. Historical interpretation is consistent with the Bank of Japan's review of the NPL problem and the Asian Development Bank's study of Japan's 1991–2005 banking crisis. The China comparison is deliberately limited to similarities in external surplus, property stress and the direction of lending rates; it is not an assertion of identical outcomes.