Zombie Banks and Forbearance
Why Did Japan Refuse to Confront Its Banking Crisis?
After the bubble burst, Japanese banks were holding enormous portfolios of loans secured by real estate that was rapidly losing value. By any honest accounting, many of Japan's major banks were insolvent or dangerously undercapitalized by the early 1990s. The economically rational — and eventually necessary — response was to acknowledge the losses, write off the bad loans, recapitalize the banks with government funds, and begin the process of recovery. Japan did not do this. Instead, regulators allowed banks to continue carrying impaired loans at inflated book values, banks continued lending to insolvent companies to keep those companies' loans technically performing, and the bad loan problem compounded for nearly a decade before being partially addressed. The result was what economists call a "zombie banking sector" — banks and companies technically alive but economically dead, absorbing capital and impeding the reallocation that recovery required.
Zombie banks and forbearance: The policy of allowing banks to carry impaired loans at book value rather than marking them to realistic market values, thereby enabling banks to appear solvent while actually being undercapitalized, and creating the incentive to continue lending to insolvent borrowers (zombies) to avoid recognizing losses.
Key Takeaways
- By the early 1990s, Japan's major banks held an estimated ¥50–100 trillion in bad loans — the widely cited range broadens because regulatory classifications of "bad" were loose.
- Forbearance policy — allowing banks to carry impaired loans without writing them down — was chosen to avoid the immediate political cost of visible bank failures and government bailouts.
- Banks under forbearance had strong incentive to continue lending to insolvent borrowers: new loans kept old loans technically performing, avoiding loss recognition; stopping credit would force default and require writedowns.
- Zombie companies — sustained by continued bank lending despite inability to service debt from operations — absorbed capital, employees, and market share from potentially productive entrants.
- The bad loan problem was not seriously addressed until 1997–98, when major bank failures (Hokkaido Takushoku Bank, Long-Term Credit Bank of Japan) forced government action.
- The delay between bubble peak (1990–91) and meaningful banking sector recapitalization (1998–2003) was approximately seven to ten years — the core of Japan's lost decade.
- The comparison with the US Savings and Loan crisis (1989–95), which was resolved more rapidly through the Resolution Trust Corporation, and with the US 2008–09 TARP recapitalization, demonstrates that faster resolution produces better economic outcomes.
The Scale of the Problem
Estimating the true scale of Japan's bad loan problem requires navigating a fog of regulatory classification that was itself part of the problem.
Japanese banks in the early 1990s classified loans into categories based on formal delinquency criteria. A loan was "nonperforming" if principal or interest was 90 days or more overdue. This standard allowed banks to classify large volumes of impaired loans as performing, because banks were extending new credit to borrowers specifically to enable interest payments on old loans.
Various estimates from the late 1990s, when the problem could no longer be denied, put the total bad loan stock at ¥50–100 trillion, with the IMF and some independent analysts suggesting the true figure was at the upper end of this range or beyond it. For context, Japan's annual GDP in 1995 was approximately ¥495 trillion; the bad loans represented 10–20 percent of GDP, comparable to some of the worst banking crisis metrics in developing economy history.
The problem was concentrated in the real estate sector. Banks had lent against land collateral valued at 1989–91 prices; when those prices fell 50–80 percent, the collateral coverage deteriorated dramatically. Construction companies, real estate developers, and affiliated non-bank financial intermediaries (jusen) held the largest concentrations of bad debt.
Why Forbearance Was Chosen
The decision to permit forbearance — to allow banks to carry impaired loans without writing them down — reflected political and institutional factors that made rapid recognition of losses unattractive.
The political cost of visible bank failures. Acknowledging that major Japanese banks were insolvent would require the government to provide recapitalization funds — directly from taxpayer money. The Bubble generation had seen its savings reduced in value; asking taxpayers to fund bank bailouts was politically toxic. The Ministry of Finance preferred a path that allowed the problem to remain off the visible political agenda.
The risk of bank runs. Publicizing the scale of bank insolvency, even without formal failures, might have triggered depositor withdrawals that converted potential insolvency into actual insolvency. Forbearance maintained the fiction of bank soundness that kept depositors calm — at the cost of allowing the underlying problem to grow.
The absence of a resolution mechanism. Japan in the early 1990s lacked adequate institutional mechanisms for bank resolution. The US had the FDIC's resolution framework, developed through the S&L crisis. Japan had no equivalent. Building the institutional capacity for bank resolution while simultaneously managing a major crisis would have been extremely difficult.
The belief that time would heal the problem. The Ministry of Finance and many bank managements genuinely believed in the early 1990s that real estate prices would recover, that good bank management would gradually work through the problem, and that the bad loan problem would be manageable if given time. This belief proved wrong; real estate prices continued falling and the problem compounded.
The Zombie Mechanism
The term "zombie" in the economic context was introduced by economists Caballero, Hoshi, and Kashyap in a 2008 paper specifically analyzing Japan's crisis. Their analysis identified the mechanism by which forbearance created a self-sustaining zombie sector.
The incentive structure for banks under forbearance was perverse but internally consistent. Consider a bank with a ¥1 billion loan to a real estate developer whose collateral is now worth ¥400 million. The bank faces a choice:
Option A: Recognize the loss. Write down the loan to ¥400 million, take a ¥600 million loss, reduce regulatory capital accordingly. This may push the bank below minimum capital requirements, requiring recapitalization or regulatory intervention. Management careers are damaged.
Option B: Continue lending. Extend a new ¥50 million credit line to the developer, enabling the developer to make interest payments on the existing loan. The original loan remains "performing." No loss is recognized. The bank's reported capital is unaffected. Management's position is secure — for now.
Option B is clearly the individually rational choice for bank management facing personal career risk from Option A. The aggregate consequence is that ¥50 million of new credit is directed to a company that cannot generate the returns to justify it, while potentially more productive use of that capital is foreclosed.
Multiply this decision by thousands of bank managers and tens of thousands of impaired borrowers, and the aggregate is a massive misallocation of capital — bank credit flowing to zombie companies that cannot service it productively, rather than to new entrants that could create genuine economic value.
The Jusen and the First Phase of Recognition
The jusen — non-bank mortgage lending companies that had borrowed from banks to fund real estate lending during the bubble — were among the first institutions to be formally recognized as insolvent. By 1995, it was undeniable that the jusen held approximately ¥13 trillion in bad loans that they could not recover.
The government's 1996 decision to use ¥685 billion in public funds to resolve the jusen was extremely controversial. Public protests against using taxpayer money to bail out the housing finance industry (which was associated in the public mind with speculative excess) were widespread. The political controversy deterred the government from pursuing more aggressive bank recapitalization in the near term — the public reaction to the jusen resolution made the Ministry of Finance even more reluctant to undertake the much larger bank recapitalization that was actually needed.
The 1997–98 Banking Crisis
The fiction of bank soundness collapsed in 1997–98 as actual bank failures occurred. Sanyo Securities (a mid-size securities firm) failed in November 1997, triggering a breakdown in the interbank call money market. Hokkaido Takushoku Bank — a major regional bank and member of the Fuyo keiretsu — failed in November 1997, the first major Japanese bank failure in the postwar period. Yamaichi Securities, one of Japan's "Big Four" securities firms, failed in the same month. In 1998, the Long-Term Credit Bank of Japan (LTCB) and Nippon Credit Bank — two of Japan's major long-term credit lenders — were nationalized.
These failures forced meaningful government action. In October 1998, the Financial Revitalization Law and the Early Strengthening Law provided ¥60 trillion in government funds for bank recapitalization and resolution — approximately 12 percent of Japan's annual GDP. This was the belated but necessary recognition that the forbearance approach had failed.
The Financial Services Agency (FSA), established in 1998 to replace the Ministry of Finance's banking supervision function, brought more rigorous loan classification standards and began the process of forcing banks to recognize and write off bad loans. The gap between the Ministry of Finance's approach (forbearance, discretionary classification) and the FSA's approach (standards-based, market-consistent classification) was itself a significant institutional change.
The Caballero-Hoshi-Kashyap Analysis
Economists Ricardo Caballero, Takeo Hoshi, and Anil Kashyap, in their 2008 paper "Zombie Lending and Depressed Restructuring in Japan," provided the definitive academic analysis of the zombie mechanism. Their key findings:
Zombie firms received subsidized credit. Using data on Japanese firm-level lending, they identified firms that could only service their existing debt with continued new credit extensions — the technical definition of a zombie. They found that these firms received credit at below-market rates, effectively subsidized by the banking system.
Zombies crowded out healthy investment. Industries with higher concentrations of zombie firms had lower rates of new firm entry, lower job creation by healthy firms, and lower productivity growth — consistent with the theory that zombies absorbed resources that would otherwise have funded new, more productive activity.
The phenomenon peaked in the 1990s. The incidence of zombie lending was highest in the late 1990s, before the 1997–98 crisis and FSA reforms. It declined after 2000 but remained above pre-bubble levels through the mid-2000s.
Lessons: The Case for Rapid Resolution
Japan's forbearance experience, combined with the alternative experience of the US S&L crisis and the 2008 TARP recapitalization, provides strong evidence that rapid recognition and resolution of banking system bad loans produces better economic outcomes than delayed forbearance.
The US S&L crisis, which involved a similar pattern of deregulation-enabled speculative lending followed by collapse, was largely resolved through the Resolution Trust Corporation (RTC) by the mid-1990s. The RTC acquired failed thrift institutions' assets, sold them at market prices, and closed the institutions. The process was painful, expensive (approximately $130 billion in taxpayer funds), and rapid. The US economy returned to strong growth in the mid-1990s.
The 2008 TARP recapitalization — while controversial and imperfect — provided capital to major banks and required stress testing that established credible minimum capital standards. The US recession of 2008–09 was severe, but the recovery was substantially faster than Japan's 1990s experience.
Common Mistakes in Analyzing Zombie Banks
Treating forbearance as always wrong. Short-term forbearance — allowing a liquidity-stressed but fundamentally sound institution time to recover — can be appropriate. The Japanese error was long-term forbearance for fundamentally insolvent institutions. The distinction requires honest assessment of whether the institution is temporarily illiquid or permanently insolvent.
Assuming rapid resolution is politically easy. Forcing recognition of bank losses, recapitalizing with public funds, and closing insolvent institutions is politically difficult in any democracy. The Japanese political resistance to visible bank bailouts was not irrational — it reflected genuine taxpayer concerns. The lesson is not that the politics are easy but that the long-run costs of forbearance exceed the short-run costs of resolution.
Frequently Asked Questions
How much did the Japanese banking crisis ultimately cost taxpayers? The direct cost of government capital injections and deposit insurance payouts has been estimated at approximately ¥10–15 trillion. Adding the costs from the slower economic growth that resulted from the delayed resolution is substantially larger but harder to quantify precisely.
Did any Japanese bank executives face criminal consequences? Some banking executives faced regulatory sanctions and a few faced criminal charges for specific violations (concealing losses, fraudulent accounting). The general pattern was managerial turnover and reputational damage without criminal prosecution for the systemic forbearance decisions.
Are zombie companies still a problem in Japan today? Academic research suggests that zombie company incidence declined substantially after the early 2000s banking reforms. The Abenomics era (2012–) specifically targeted corporate restructuring and governance reform to reduce remaining zombies. However, some analysts argue that very low interest rates globally — including in Japan — provide ongoing support for marginally viable companies that would otherwise have to restructure.
Related Concepts
- Japan's Lost Decades Overview — the full arc
- The Keiretsu System — how corporate structure perpetuated zombie lending
- Japan's Deflation Trap — the economic consequences
- Japan's Policy Responses — when and how the banking crisis was addressed
Summary
Japan's zombie banking sector was the primary mechanism through which an asset price bubble became a multi-decade economic stagnation. The policy of forbearance — allowing banks to carry impaired loans without writing them down — was chosen for understandable political reasons but created perverse incentives that perpetuated the misallocation of capital for nearly a decade. Banks continued lending to insolvent zombie companies to avoid recognizing losses; zombie companies absorbed resources that would otherwise have funded productive new investment; the economy stagnated. The eventual forced recognition in 1997–98, driven by actual bank failures rather than proactive policy choice, came seven to ten years too late to prevent the lost decade. The comparison with faster-resolved banking crises in other countries provides compelling evidence that speed and transparency of bank resolution is a first-order determinant of post-crisis economic recovery.