In the Japanese case, the period between the real estate price peak (September 19901) and the outbreak of the systemic banking crisis (November 1997) lasted more than seven years.

This was exceptionally long. Table 4.1 shows the corresponding periods for ten systemic banking crises in advanced economies since 1990. In seven episodes, the crisis occurred in the year of the real estate price peak or in the next. In one episode, two years later and in another, three years. None is comparable to the seven in-between years seen in Japan.

Table 4.1 In-between years in major banking crisis episodes since 1990

Estimating the Size of the Problem

In 1991, although the stock price was about 40% below the peak and the land price in large cities had already started to decline, still many Japanese had not realized how devastating the consequences would be. The government declared in October that the current expansion had become the longest in the post-World War II history of Japan, not recognizing that the business cycle had already peaked in February.

The first symptoms of the problem were a series of scandals. It was revealed in April 1991 that persons affiliated with a yakuza organized crime group had taken over the management of a trading firm and built a deep relationship with the CEO of a major bank, and that billions of dollars of loan from the bank to the firm had been syphoned to yakuza. In June, it was revealed that major securities houses had guaranteed investment returns to their major clients and had illegally compensated for their losses. In July and in August, two separate but similar incidents were revealed where bankers had issued false deposit certificates amounting to a billion dollars, which had been used as collateral for borrowing money for speculative purposes.

The nation had the impression that, though bankers had previously been well respected and trusted, some of them had badly been contaminated during the period of frenzy.

By early 1992, bankers and regulators started to realize that the bad loans created by the bubbles and the ensuing bust could be fatal to the Japanese banking system. Some journalists in and out of Japan were also aware of this.

The regulator carefully and narrowly defined bad loans when publishing the amount of them. In April 1992, the Ministry of Finance disclosed that as of March 1992 the amount of loans in arrear held by 21 major banks amounted to seven to eight trillion yen in gross and two to three trillion yen net of the parts covered by collaterals or guarantees. Two to three trillion yen of potential losses looked benign: the 21 banks’ annual profits amounted to 0.94 trillion yen, and they had unrealized gain of 17 trillion yen on securities held.

But the definition covered only a fraction of the problem. They did not include the problem loans held by non-bank lenders affiliated to banks. They did not include doubtful loans not in arrears, including whose interest payments were financed by the lending banks.

At the time, there were no solid statistics of the total size of the problem based on the results of individual loan quality reviews. Regulators had only several rough estimates, which were highly dependent on the assumptions used. Naturally, the estimates failed to incorporate the effects of the subsequent land price declines, but still the numbers were totally ominous. According to Nishino (2003), the Ministry and the Bank of Japan in 1992 both privately estimated the gross amount of non-performing loans in the banking system to be much higher, somewhere between 30 and 50 trillion yen (Table 4.2).2 Similar amount was reported by the Financial Times in May.

Table 4.2 Non-performing loan amounts estimated in 1992

Catch 22

The 30–50 trillion yen was a formidable number, amounting to 6–10% of the annual GDP. But the situation was even worse than the number suggests. In finding ways to resolve the bad loans, bankers and regulators faced with at least four types of Catch 22 style accelerator mechanism.

First, to resolve the bad loans, banks need to foreclose the borrowers and sell collaterals, making the land price decline further. Lower land price, however, increases bad loans and reduces the value of real estate collaterals.

Second, bad loans were not the only source of vulnerability on the Japanese banks’ balance sheet. Japanese banks had large cross-shareholding with their customers. According to the simulation by the Ministry of Finance in the summer of 1992, if the Nikkei stock index had broken the 12,000 yen line, some banks would be in negative net worth, and if it had fallen below 10,000 yen, large number of banks, even without recognizing losses on bad loans.3 The Nikkei index, which reached 38,915 yen at the end of 1989, recorded 14,194 yen on August 19, 1992, drawing near the red line. Banking problem can push down stock prices, and lower stock prices can exacerbate banking problem.

The third one was much more complicated. Loans extended by non-bank lenders amounted almost to 100 trillion yen. Around 80% of the debts of those lenders were financed by banks. Among the non-bank lenders, the seven Specialized Housing Finance Companies (Jusens) were particularly aggressive in making commercial real estate loans. They extended loans amounting 13 trillion yen in total, and the majority of them went sour.

Each of the Jusens were sponsored and owned by multiple banks. On average, they borrowed 30% of their debts from sponsoring banks, another 30% from non-sponsoring banks, and the remaining 40% from agricultural cooperatives.

Sponsoring banks could have been better off by resorting to a bankruptcy procedure, but once one Jusen should fail, then non-sponsoring banks and agricultural cooperatives, feeling betrayed by sponsoring banks, would start to run on other Jusens. Run could potentially affect other non-bank lenders.

Relationship between Jusens, banks, and agricultural cooperatives were entangled like spaghetti and no bankers or regulators had enough information to disentangle it.

The combination of the three booster mechanisms meant that once resolution of bad loans was started, bad loans would increase, the value of collaterals would decline, and the latent gains on equities held by banks may disappear or turn into latent losses.

The total size of the banking sector’s own capital was as small as 30 trillion yen. The resolution of bad loans would expose that the whole banking sector, not just a handful of reckless banks, was severely undercapitalized or may have even been in negative net worth. To quickly resolve bad loans and keep the banking system and the economy alive, tens of trillions of yen of public support would have been needed.

But the national government’s budget for fiscal year 1992 was 56 trillion yen, excluding the government’s own debt service expenditure. Devoting the whole budget for SMEs (0.2 trillion yen) would have been totally insufficient. Even the whole social security budget (13 trillion yen) could have been dwarfed.

Here comes the fourth and the most important dilemma. The regulators needed public support for new resolution powers and an enormous amount of funding to avoid a systemic crisis. The series of bankers’ scandals revealed in 1991 were surely not helpful in securing such support. Most likely, such political support can be acquired only after a systemic crisis. Of course regulators could have tried to convince the public that the amount of money was truly needed to avoid a catastrophe, but depositors’ confidence in the banking system would then be lost half way through the persuasion and a catastrophe would be triggered before the law and budget were passed.

This kind of self-fulfilling prophecy was not without a precedent in the country. In 1927, the government submitted a bill to resolve banks’ accumulated bad loans. The Diet deliberations revealed how bad the balance sheets of major banks were and the revelation caused the first wave of bank run before the bill was approved. Then a deterioration of the condition of one major bank resulted in the second wave of run, and the government tried to resort to an emergency imperial decree to bailout the bank. The public opinion got agitated, and the Privy Council rejected the proposal. The third and by far more devastating nationwide bank run ensued, and the government had to declare a three-week moratorium. Then an emergency session of the Diet approved a bailout bill in five days.4 In 1992, the Ministry of Finance studied this episode, known as the Showa Financial Crisis, in depth and considered it as a guide to the crisis it was facing.

The Finance Ministry’s Choice

The Ministry of Finance decided not to repeat the Showa Financial Crisis and to take necessary time to resolve bad loans. It hoped that banks’ annual profits over years and unrealized capital gains on securities held by banks would eventually prove to suffice for the resolution of bad loans. The Ministry planned to request strong resolution tools and enough budgets only if the scenario should turn wrong and depositors should realize that their deposits were at risk.

An editorial which appeared on the Nikkei newspaper on May 29, 1992 describes a supervisory practice at the time: “Every year in April or in May, meetings are informally held at the Banking Bureau of the Ministry of Finance to give bankers administrative guidance how banks should account and publish the financial results for the accounting year ended in March. We see a few bankers sit in front of the desk of the Director,5 and the conversation there de facto determines how the bank will report its annual financial results.”

The editorial goes on to read between the lines of the financial reports published by major banks in May: “One can guess what was written in the clinical chart the Ministry prepared. The patients, ailing from the aftermath of the exuberance during the bubble economy, would not survive surgeries needed to cut off the diseased organs. The only available option is non-surgical, internal therapy of taking nutrition and waiting bad blood to be gradually purified. Or more explicitly, it must be that the Ministry instructed bankers not to throw away the affiliated non-bank lenders in trouble, but to support them with forbearance in interest rate and repayment schedules and with additional lending, and to make up the costs by annual income flows.”

A high official in charge of banking supervision at the Ministry made the following remark to me during a private conversation in early 1992: “What I tell bankers these days is just this. Don’t panic. Be patient.”

The prescription which was implicit in banks’ financial reports was made public in August 1992. The Ministry of Finance released a statement titled “Financial sector policies for the coming period.” In its preamble, the Ministry stated:

Banks are currently in a difficult condition they have never experienced since the start of the country’s rapid economic growth after the World War II. . . . We cannot deny that, to overcome the impacts of the crash on banks, we need to continue rigorous and serious adjustment efforts for a significant period of time. . . . However, it is our firm belief that it is not inevitable that the problems will make the financial system dysfunction and put undue burden on the national economy. The problems can be overcome by avoiding simple pessimism and calmly laying many layers of steady and serious endeavors. . . .

Following the preamble, the statement laid out policy measures to prevent banks from beefing up their profits by realizing latent gains, to make banks come up with credible plans to deal with their non-bank affiliates, to create a market for liquidated collateral real estates, to enhance disclosure by banks, and to encourage banks to streamline their operations.

Many commentators argue that the policy statement showed that the Ministry was complacent and overlooked the seriousness of the problem. Perhaps these commentators overlooked the pathetic overtone of the statement and why the Ministry had to make repeated calls to be calm and steady.

The Bank of Japan and the Prime Minister

There were three attempts to challenge this line of thought, twice by the Bank of Japan and once by the prime minister.

The first attempt was made by the Bank of Japan. In May 1990, the Bank established a department responsible for financial stability. Upon the instruction by Governor Mieno to prepare for future bank failures, the head and the number three of the department visited authorities in the United States and Europe to learn from their experience.

In January 1991, the Bank’s cross-department team reached a conclusion. An insolvent bank should be resolved, not by liquidation but by purchase and assumption approach (P&A) ,6 as is the case with most failed banks in the United States. The Japanese Deposit Insurance Law, after its amendments in 1986, had indeed been equipped with the approach. Depositors should be protected, but the bank’s management should be replaced, and shareholders should bear losses. If needed, the Deposit Insurance Corporation should provide financial assistance to support purchase and assumption operation, and the Bank of Japan should lend necessary amount of money.7

But the Ministry of Finance was not supportive of the principle that an insolvent bank should be resolved.8 If the land price and the stock price should recover, banks with negative net worth would become solvent again. Why should we resolve them now, shock the nation, and prevent possible future recovery of asset prices? And if the asset prices should continue to deteriorate, then the whole banking sector would become insolvent. Can the recommended purchase and assumption approach resolve the whole banking system?

The second attempt was by the Prime Minister Miyazawa Kiichi himself. He was particularly worried about the stock price plunge in August 1992 and even considered closing the Tokyo Stock Exchange. He was in frequent conversation with Governor Mieno of the Bank of Japan. According to Mieno (2007), he said to Miyazawa “It is already time to inject public funds, but the Ministry of Finance bureaucracy will never propose that to you. You yourself therefore need to instruct,” and Miyazawa responded to him, “Well, is that so? I see, perhaps there is no other way than doing it myself.” Mieno considered that Miyazawa saw what Mieno meant.

Miyazawa spoke at a seminar for ruling party members held at the end of August 1992:

The growth rate of money supply declined from annual 10 percent to almost zero. Some argue that this is due to weak corporate financing needs, but the truth is the capacity of banks to lend is impaired by bad loans and other factors. The weak stock market reflects this. This is a factor unique to the current recession, which we have not experienced before.

It is the responsibility of the government and the central bank to take necessary measures if the market economy ceases to function properly.

We should create a mechanism to liquidate real estate collaterals held by banks by the end of the year as an element of the forthcoming economic policy package. The most desirable outcome is that financial institutions work together contributing their ideas and funds, but, if needed, I would not shy away from public support.

This is not to save banks. Everyone would suffer if the blood ceases to circulate round the body of the national economy. I would not hesitate from doing what is needed for the whole national economy, on the condition that bankers behave in line with their public mission and disclose the amount of bad loans. The government and the central bank will never idly overlook future risk of financial instability.9

These were remarks given by the man of highest political power in the country, whose party had just won the upper house election the month before. But they were immediately reacted by criticism. Vice minister of finance, or the head of the civil service at the Ministry, denied in public any use of public money. Business leaders categorically rejected public support to banks. The chairman of the Bankers’ Association stated that they will help themselves. Miyazawa abandoned the idea.

Later, in 1998–2001, Miyazawa served as finance minister. At the peak of the banking crisis, a congressman referred to his speech and asked him whether the crisis could have been alleviated if he, as the prime minister, had had the courage to resolve bad loans as proposed in his speech. Miyazawa responded, “I failed to enlist public opinion in support of my proposal. You asked me what if I had implemented the proposal, but I believe this is a clear case showing that any major surgery cannot be carried out without a support of the public opinion.”10

The Ministry of Finance chose to wait till the condition was ripe. The prime minister demonstrated that the condition was not ripe and perhaps fortified the public allergy against the use of public funds by a premature proposal abruptly made without necessary political groundwork or a plan on the steps to be taken to reach the goal.

But the Bank of Japan did not give up. It made the third attempt: The Bank, after a deliberation at its policy board, submitted in May 1993 a report to the Ministry of Finance as a formal Bank of Japan proposal. The report assumed the amount of bad loans to be between 30 and 50 trillion yen. It included a detailed analysis of the banking problems and proposals for bank resolution and use of public funds, operationalizing the principles it proposed in 1991. However, the conversation between those who focused on what needs to be done and those who focused on what can be done was not easy.

If the Bank of Japan line of thought had prevailed, the four Catch 22 accelerator mechanisms would have been promptly activated, and perhaps Japan would have had a systemic banking crisis in 1992 or 1993. The period between the commercial real estate price peak in 1990 and the crisis would then have been two to three years, in line with the other crisis cases shown in Table 4.1. Which would have been better for the nation, a slow crisis which happened or a fast crisis which did not materialize?

After reviewing the whole developments of the financial cycle, we will come back to this question in Chapter 6.

Orderly Resolution Without Bailout11

The Ministry’s strategy survived five more years. Strong balance-of-payment and net-external-asset positions of the country, trust in banks and the regulator built over decades of no bank failure, and accounting standards highly restrictive in provisioning against credit losses must have worked to sustain the strategy. But the strategy would not have survived without the concerted efforts by relevant players to attain orderly resolution overcoming the limitations in the toolbox.

In early 1994, the Ministry confirmed its view that latent gain on shares held by banks and annual profits would suffice to finance costs needed to resolve bad loans if enough time was given (Ministry of Finance 1994). But bad loans grew as land prices continued to decline, and it became evident that at least smaller and weaker links in the system could not survive. In 1995, the Ministry moved somewhat closer to the 1992 Bank of Japan view: It publicly confirmed that the total amount of bad loans in the system was around 40 trillion yen, that there could be cases that banks’ own efforts might not suffice, and that issues including public involvement in resolution would need to be discussed (Ministry of Finance 1995).

The Japanese deposit insurance system started in 1971 with the only available option to pay out deposits with limited coverage. The purchase and assumption approach (P&A) was added in 1986, but the amount of financial assistance was capped at the payout cost, and the Ministry of Finance had to persuade other banks and local governments to contribute to fill the gap. During the in-between years, the Ministry faced the need to resolve 29 deposit taking institutions (banks, shinkin banks,12 and credit cooperatives), but each case required joint tightrope acts by the Ministry, the Bank of Japan, the Deposit Insurance Corporation, and other stakeholders.

Initially, the Ministry succeeded in persuading existing bank to acquire failed banks, but the persuasion had become more and more difficult. Starting from 1994, some cases were resolved by establishing new recipient banks. In 1996, the national Diet passed the bill proposed by the Ministry to lift the limit on financial assistance by the Deposit Insurance Corporation on purchase and assumption. The deposit insurance fee was raised by seven times to finance the costs. It was also made possible for the Deposit Insurance Cooperation to borrow with government guarantee, but this option was allowed only to finance costs to resolve credit cooperatives, or the smallest members in the system.

This was the maximum the Ministry could attain before the outbreak of the systemic crisis. In 1996, the Ministry proposed to use 0.7 trillion yen of taxpayers’ money to resolve seven Jusens . The money was needed to avoid incurring losses to politically mighty agricultural cooperatives, which lent to Jusens. The proposal was approved by the Diet, but the use of taxpayers’ money to resolve reckless non-bank lenders attracted strong social anger.

To stabilize the banking system, much bigger amount of money was needed, but the public anger on 0.7 trillion yen made any further proposal on the use of public money politically untenable. To protect confidence in the system, preventive tools for banks which were distressed but not failed were indispensable, but the tools available at the time could be used only for the resolution of failed banks. The Jusen incident made it difficult to expand tools.

In November 1997, a medium-sized broker-dealer Sanyo Securities collapsed. Nobody thought it posed a systemic risk—it took no deposits and its operations were simple. However, its collapse resulted in the first default in the history of the Japanese interbank market. Although the size (about $10 million) was almost negligible, a few days later the interbank money market started to freeze up. In a stressed environment, a seemingly non-systemic case can have a systemic signaling effect and trigger a chain reaction.

Ten days later, Hokkaido Takushoku Bank, one of the countrys leading 20 banks, and Yamaichi Securities, one of the top four broker-dealers, both encountered with liquidity shortage. The Ministry of Finance, with the help of the Bank of Japan, found a recipient bank for the former over the weekend and the purchase and assumption was announced in the afternoon of Sunday, November 16.

There was no resolution framework for broker-dealers. Given that Yamaichi had a big derivatives book and significant cross-border exposures, the Bank of Japan decided to provide uncollateralized liquidity support. On November 24, Yamaichi announced that it will wind itself down.

The Ministry of Finance, the Bank of Japan, and the Deposit Insurance Corporation, all very thinly staffed, overcome enormous obstacles despite extremely tight schedule and successfully orchestrated orderly resolutions. No depositors or creditors suffered loss except for creditors to Sanyo Securities. Nevertheless, the succession of big-name resolutions shocked the nation. In addition, the Deposit Insurance Corporation had already run out of its policy reserve by March 1997 and the size of financial assistance needed to achieve the purchase and assumption transaction for the Hokkaido Takushoku Bank, four times as large as the annual premium, was far beyond its financial resources.13

On November 25, due to an unfounded rumor, queues were formed at branches of one regional bank. Early in the morning of November 26, the resolution of another regional bank through purchase and assumption was announced.

The day November 26, 1997, started with dark clouds hanging over the archipelago. The in-between years were coming to an end.


  1. 1.

    As the Japanese banking crisis was primarily triggered by commercial real estate bubbles rather than residential real estate ones, the commercial real estate price peak date is used in measuring the Japanese interval period.

  2. 2.

    Teramura (2011), who was the director general of the Banking Bureau of the Ministry of Finance at the time, confirms the report by Nishino (2003) that the Bank of Japan informed its estimate of above 40 trillion yen to the Ministry in August 1992. According to Teramura (2011), the Ministry made an own estimate amounting to 42 trillion yen in early 1994.

  3. 3.

    Teramura (2010).

  4. 4.

    See Takahashi and Morigaki (1993).

  5. 5.

    At Japanese government ministries, the whole members of a division usually work in a single room. At the time, bankers and accredited reporters could freely enter Finance Ministry offices.

  6. 6.

    Purchase and assumption approach is a resolution method in which another bank purchases failed bank’s assets and assumes its obligations with financial assistance provided by the resolution authority. In the first two cases of its application in Japan, failed banks’ whole assets were purchased by sound banks, but starting from the third case, sound banks purchased only sound assets and bad loans were transferred to third parties.

  7. 7.

    Shirakawa (2018).

  8. 8.


  9. 9.

    August 31, 1992 morning editions of Asahi, Nikkei, and Mainichi (newspapers), and Iokibe et al. (2006).

  10. 10.

    Response by Finance Minister Miyazawa Kiichi to the question by Councilor Sato Michio at the Special Committee on Financial Issues and Revitalization of the Economy at the 143rd session of the House of Councilors (Upper House) on October 7, 2018.

  11. 11.

    For bank failures and the evolution of the resolution approaches during the in-between years, see Nakaso (2001) and Deposit Insurance Corporation of Japan (2007).

  12. 12.

    Both shinkin banks and credit cooperatives are non-profit mutual organizations which do banking business with individuals and SMEs in their region of operation. Larger SMEs are eligible to become members of Shinkin banks but not credit cooperatives.

  13. 13.

    The deposit insurance premium rate was raised in the fiscal year 1996 to the level seven times as high as the previous level, but the annual premium received after the raise was about 0.46 trillion yen, while assistance to the recipient bank of the Hokkaido Takushoku amounted to 1.8 trillion yen.