I

The too-big-to-fail problem in banking is the perceived need to bail out a troubled large financial institution in order to avoid the consequences of it failing. It was a major part of the response to the financial crisis of 2007–9, when US regulators injected capital into AIG, provided guarantees to Citigroup, assisted J. P. Morgan with the purchase of Bear Stearns, and injected capital into the nine largest banks. It was also a significant part of the response to the banking crises of the 1980s when regulators bailed out Continental Illinois bank in 1984 and provided forbearance to the large money center banks on recognizing losses from loans to less-developed countries (LDC).Footnote ¹

The Continental Illinois bailout is particularly significant because it brought the too-big-to-fail problem to the public's attention.Footnote ² With nearly $41b in assets it was by far the largest bank that had failed up to that point in time. What is not well known, however, is that it was preceded by a sequence of too-big-to-fail bank bailouts in the 1970s and early 1980s, starting with the bailout of the $1.2b Bank of the Commonwealth in 1972.

The first purpose of this article is to analyze Commonwealth and argue that it was the first too-big-to-fail bailout of the modern era.Footnote ³ There is little existing analysis of Commonwealth or its bailout. Irvine Sprague describes it in his book Bailout (Reference Sprague1986), which covers the bailouts he participated in as a director of the Federal Deposit Insurance Corporation (FDIC), but the academic literature has done little more than to mention it. This article expands on Sprague's description by analyzing Call Report data, annual reports and other contemporary sources to show how Commonwealth took risk and how it failed.

The second purpose is to identify patterns in early too-big-to-fail policy. The article documents that too-big-to-fail considerations motivated bailouts well before Continental Illinois and were applied to surprisingly small banks. It shows that at the time state banking restrictions hindered the ability of regulators to arrange the sale of a failing bank, and that these bailouts were legally justified by successively broad interpretation of FDIC legal powers. Finally, it describes how regulators were often reluctant to bail out the bank, but unable to commit to not doing so. Our reading of these early decisions is that the bailouts of the 1980s and the 2000s were not revolutionary actions but instead a predictable consequence of structural changes in banking and past behavior by financial regulators.

II

When a bank is declared insolvent by its chartering agency, it is turned over to the FDIC. For the period covered in this article (1971–84), the FDIC had three options for dealing with a failing bank: it could pay the insured depositors and liquidate the bank; it could facilitate the purchase of the bank by providing financial assistance to the acquiring bank; or it could provide assistance to the failing bank to keep it open.Footnote ⁴

Under the 1950 Federal Deposit Insurance Act, the FDIC was required to pay off closed banks unless it was less costly for them to sell the bank.Footnote ⁵ However, the FDIC was allowed to provide open-bank assistance if the FDIC Board of Directors found that the bank was in danger of failing and was ‘essential to provide adequate banking service in the community’. This part of the FDI Act along with the decision to invoke it is referred to as the ‘Essentiality Doctrine’.

All but one of the early bailouts were done under the Essentiality Doctrine. Nevertheless, FDIC interventions were only one component of the bailouts. The other component was Federal Reserve discount window lending, which kept a troubled bank open and liquid until an assisted acquisition or bailout was arranged. Furthermore, this lending provided uninsured creditors with time to remove their funds before the bank was resolved by the FDIC. Typically, the only bank liability holders who lose money during a bank failure are the shareholders, particularly those of smaller banks.Footnote ⁶

Despite having the power to provide open-bank assistance since 1951, the FDIC did not invoke the Essentiality Doctrine until 1971 when it was used to bail out Unity Bank. Unity Bank was a small ($11m) bank that was African-American owned and served an African-American neighborhood in Boston. Poor management got the bank into trouble and the FDIC decided to intervene because of concerns about racial unrest if it failed (Sprague Reference Sprague1986).

Unity Bank was not too big to fail, but its bailout is important because of the precedent it set for establishing the Essentiality Doctrine. In the opinion of the FDIC, the wording of the Act gave it the ability to both set the terms of any assistance and define what exactly constitutes a community. Thus, the FDIC concluded that Unity Bank was essential to the African-American community of Boston because there was no other minority-owned bank nearby (Sprague Reference Sprague1986, p. 43).

Including Unity Bank and Continental Illinois, the Essentiality Doctrine was invoked six times through 1984. It was used in 1976 to bailout the $493m Farmers State Bank of Delaware because the bank was half owned by the state of Delaware.Footnote ⁷ It was used two other times to bail out too-big-to-fail banks.Footnote ⁸ The following section describes these three too-big-to-fail bailouts, the bailout of Franklin National Bank, the near bailout of Seafirst, and the related forbearance used during the LDC debt crisis, which was an indirect bailout of the large money center banks. We discuss the Bank of the Commonwealth bailout in more detail than the others because we consider it the first too-big-to-fail bailout and it sets the pattern for the too-big-to-fail bailouts that followed.

III

The second use of the Essentiality Doctrine was in 1972 to bail out the struggling Bank of the Commonwealth in Detroit. This bank was at the center of a network of banking partnerships run by Donald Parsons in Michigan. This network, known as the Comprehensive Oriented Management Activities Company (COMAC), was a management consulting firm that had the banks as clients (Gies Reference Gies1975). It was used as a way around Michigan regulations that prohibited bank holding companies and restricted branching. Through COMAC, Parsons and his partners directed the operations of their banks without violating bank holding company restrictions.

Commonwealth was a conservatively run, low profit institution that held more treasuries than loans and funded itself with demand and savings deposits. Parsons took it over in 1964 and changed its strategy. He invested in high-yield, long-term municipal securities with the hope that rates would drop and thus deliver a large capital gain. Commonwealth's holdings of state and local obligations increased from 6 percent of assets in 1964 to 23 percent in 1968. It lengthened the average maturity of its municipal portfolio from 12.9 to 23.4 years between 1965 and 1968.Footnote ⁹ In contrast, the two largest banks in Michigan (National Bank of Detroit and Detroit Bank & Trust Company) both had municipal portfolios with an average maturity of under ten years at this time.Footnote ¹⁰ Partly due to these decisions, Commonwealth grew rapidly and nearly tripled in size from 1963 to 1969 ($557 m in assets in 1963 to $1.49b in 1969).Footnote ¹¹ Furthermore, it funded its growth with high-interest certificates of deposit and interbank borrowings.

Commonwealth reached its zenith in 1969 when it made a $12.5 m profit. Unfortunately, it was vulnerable to changes in macroeconomic conditions. In an attempt to control rising inflation, the Federal Reserve increased the federal funds rate by over 250 basis points in 1969, and then, in response to the recession that started in December, it reduced the federal funds rate in early 1970. Unfortunately, long-term rates stayed high due to increased inflation expectations. This not only reduced the market value of Commonwealth's long-term securities but also hurt its funding. Many customers did not roll over their time deposits because they wished to invest in higher yielding alternatives not subject to Regulation Q caps on deposit rates.Footnote ¹² Commonwealth then increased its borrowing in the federal funds market, which raised its cost of funding.

As Commonwealth's funding problems continued, it tried to open a branch in the Bahamas to access the Eurodollar market. However, the Federal Reserve denied the application citing that ‘a serious deterioration has occurred in the financial condition of your bank’ (Sprague Reference Sprague1986, p. 62). In addition to the declining value of the municipal securities, many loans that Commonwealth made during its growth performed poorly, likely due to the economy. For example, Commonwealth's loan-loss provisions expense increased from $401 thousand to $12.9 m in 1970. Furthermore, Commonwealth could not sell its municipal portfolio to obtain funding because it would have had to recognize losses from these sales, and without positive income to offset these losses, its capital would decline. If Commonwealth sold its entire municipal portfolio in 1970 it would have had to recognize $50 m in losses, which equaled 73 percent of its capital.Footnote ¹³

The Fed's rejection of the application exasperated Commonwealth's funding problem, so it used the discount window. Commonwealth's ‘other borrowing’ category on its balance sheet – which includes discount window loans – increased from $58 m to $255 m over the second half of 1970. The Fed then placed conditions on the loans that included Parsons resigning and the bank consenting to a cease and desist order (Sprague Reference Sprague1986, p. 63). It lost $9.8 m in 1970 and $2.7 m in 1971.

Unlike Unity, Commonwealth was a large institution, with total assets of around $1.25b at the end of 1971. The FDIC preferred to arrange a merger, but bank concentration in Detroit along with state banking rules limited the pool of acquirers. At the time, Michigan law prevented out-of-state banks from acquiring Michigan banks, so any acquirer would have had to come from within the state. However, the three largest banks in Detroit already controlled 77 percent of deposits, which Sprague believed would make Detroit too concentrated if one of them added Commonwealth's 10 percent share (Sprague Reference Sprague1986, p. 69).

The FDIC eventually ruled that Commonwealth was essential because of its ‘service to the black community in Detroit, its contribution to commercial bank competition in Detroit and the upper Great Lakes region, and the effect its closing might have had on public confidence in the nation's banking system’.Footnote ¹⁴ The FDIC lent the bank up to $60 m to replenish the bank's capital (Sprague Reference Sprague1986, p. 73). The FDIC then purchased Commonwealth's municipal securities at market value, which was the main contributor to Commonwealth's $25 m loss in 1972. Lastly, the bailout agreement required Commonwealth to reduce the par value of its stock by $37.6 m and to use this to absorb the losses from the securities. While the FDIC assistance kept Commonwealth open, the bank continued to struggle after the bailout and was eventually acquired by Comerica Bank in 1983.

The bailout of Commonwealth was a significant step in the establishment of too big to fail as a de facto policy.Footnote ¹⁵ For example, Sprague says that the Federal Reserve brought up the ‘domino theory’ of a large bank failure during its Commonwealth discussions (Sprague Reference Sprague1986, p. 68). This theory, that if one large bank failed others would follow, surfaced again and again over the next 40 years as various big banks got into trouble. It was Commonwealth, not Continental Illinois, that was the first too-big-to-fail bank bailout.Footnote ¹⁶

The second too-big-to-fail bailout was that of Franklin National Bank in 1973. With nearly $5b in assets in 1973, it would have been the largest bank to fail in the history of the United States up until that point.Footnote ¹⁷ Franklin's troubles were tied to the decision by the United States in 1971 to abandon the Bretton Woods system and switch to floating exchange rates. Under this new regime, Franklin bet that the dollar would rise by taking a large short position in many foreign currencies. However, the dollar depreciated, so Franklin lost a substantial amount of money (over $30 m in the first quarter of 1974 alone). Furthermore, the bank was involved in fraudulent behavior and borrowed in the interbank market. A run on the bank started soon after, so the Federal Reserve Bank of New York allowed Franklin to borrow $1.7b over a five-month period to stay liquid until a merger could be arranged.

Franklin's large debt to the Federal Reserve Bank of New York and its messy foreign exchange portfolio made it difficult for the FDIC to arrange a merger. These obstacles were removed by the FDIC agreeing to pay off the debt in three years with payments coming from the liquidation of Franklin's assets and the New York Fed assuming Franklin's foreign exchange portfolio with the condition that Franklin pay the New York Fed $15.65 m as compensation for estimated losses. As a result, the FDIC was able to sell what remained of Franklin.

The third too-big-to-fail bailout was that of First Pennsylvania Bank of Philadelphia (First Penn). Founded in 1782, it was the nation's oldest bank and in 1980 it was the 23rd largest with assets exceeding $8b in 1980. First Penn aggressively expanded its loan portfolio during the years 1966–76 and nearly tripled in size, though many of these loans turned into non-performing ones. First Penn then followed the Commonwealth strategy and bet incorrectly on long-term interest rates. These actions put the bank in poor financial shape.

The FDIC hoped to avoid a bailout by arranging a merger with another bank. However, that proved to be impossible because of First Penn's size and Pennsylvania's restrictions on out-of-state acquisitions. The only Pennsylvania bank large enough to safely absorb First Penn was Mellon Bank of Pittsburgh, and the FDIC believed that such a combination would adversely affect competition (FDIC 1998, p. 518). Thus, the FDIC could provide assistance to First Penn or let it fail.

Once the FDIC decided to bail it out, the criteria to declare First Penn essential were primarily based on the bank's size and the negative impact that its failure would have on financial stability (FDIC 1998, p. 520). The FDIC and 27 leading banks agreed to provide subordinated notes totaling $500 m and purchase $20 m in stock warrants. The Federal Reserve Bank of Philadelphia also provided a $1b line of credit through its discount window during this process.

In 1983, the fourth too-big-to-fail bailout nearly occurred. Seafirst Corporation was a $9.6b holding company and the largest bank in the Northwest. Seafirst had purchased $400 m of energy loan participations from Oklahoma City-based Penn Square bank, and originated about $800 m of its own energy loans.Footnote ¹⁸ A downturn in the energy market contributed to an operating loss of $91.4 m in 1982 (Sprague Reference Sprague1986, p. 141). Furthermore, Seafirst was going to report larger losses in the first quarter of 1983. Seafirst's counterparties knew that it was weak and started a run in early 1983.Footnote ¹⁹

Seafirst was too weak to survive on its own, so three actions were taken. The first was to provide liquidity through lending by the San Francisco Fed and a 15-bank consortium organized by the New York Fed (Brimmer Reference Brimmer1984). The second was to find a healthy bank to acquire it. This step was difficult since Washington did not allow out-of-state banks to acquire Washington banks, and the only banks large enough to acquire Seafirst were located out of state. The third was for the FDIC to develop a contingency plan to provide aid in case Seafirst could not find an acquirer.

The FDIC was ready to assist Seafirst with a $250 m loan, and the papers to do the bailout were prepared, signed by the CEO of Seafirst, and left undated, so that the FDIC could immediately make the loan if need be. However, the bailout proved unnecessary when state lawmakers met in an emergency session and changed the law to allow out-of-state bank acquisitions. California-based Bank of America then purchased Seafirst.

The pattern in these earlier bailouts was repeated with Continental Illinois in 1984. This bailout is well documented, so we only provide a few highlights relevant to the analysis.Footnote ²⁰ Continental Illinois was a nearly $41b bank holding company. It had grown fast, was heavily exposed to less-developed countries and the energy sector, and had other loans that were performing poorly. It relied heavily on borrowing in wholesale markets and had few retail deposits. When its loan troubles became apparent, a run developed and it borrowed $3.6b from the Federal Reserve Bank of Chicago to stay liquid.

A payoff was never seriously considered by the FDIC. Over 2,000 banks had correspondent accounts at Continental Illinois, and there was a fear that other big banks would face funding problems if it failed. In addition, Continental Illinois’ size combined with Illinois’ unit banking laws, which prevented out-of-state banks and limited in-state bank purchases, made it impossible for the FDIC to arrange a merger. Consequently, the FDIC provided a subordinated loan, injected capital, and took on some of Continental Illinois’ assets.Footnote ²¹ In total, the cost of the Continental Illinois bailout was $1.1b dollars (FDIC 1998).

The remaining too-big-to-fail event in banking during this period was the exposure of the large money center banks, both in the US and abroad, to the less-developed country (LDC) debt crisis. The origin of this crisis was high levels of lending to less-developed countries in the 1970s, supported by high commodity prices (FDIC 1997). Problems erupted in 1982 when the global recession, partly caused by the Federal Reserve's efforts to curb inflation, triggered the crisis. Mexico announced that it could no longer meet interest payments and soon many other less-developed countries were delinquent.

Money center banks were heavily exposed. Sachs (Reference Sachs1986) documents that they had lent over 275 percent of their capital to these borrowers as of the end of 1982. The strategy for the United States and other creditor governments was to preserve stability of the financial system by having the IMF lend to the LDCs, while banks continued to lend to ensure interest payments were made and avoid recognizing losses on these loans (Sachs and Huizinga Reference Sachs and Huizinga1987). The forbearance lasted through the 1980s while the money center banks slowly built up loan-loss reserves, gradually charged off the loans, and then finally forgave some of the debt as part of the Brady Plan.

IV

It is clear from the descriptions that the financial regulators prefer to resolve a failing bank by finding another bank to acquire it. Between 1970 and 1984, about 72 percent of commercial bank failures were resolved this way.Footnote ²² Payoffs were only used for the smallest banks. The largest payoff in this period was for Penn Square in 1982, a $484 m bank that was only paid off because fraud and poor accounting at the bank made the FDIC's liability in the case of an assisted acquisition or a bailout hard to estimate.

Assisted acquisitions were not possible for the money center banks affected by the LDC debt crisis. There were too many of them. Instead, forbearance was used to give them and the LDC countries time to work out the problems. While this strategy prevented wide-scale financial instability, it was less successful when applied to the troubled thrift industry, which was exposed to high inflation. For the thrifts, forbearance allowed many weak savings and loan associations (S&Ls) to make risky loans, which later greatly increased the severity of the S&L crisis.Footnote ^23, Footnote ²⁴

For individual banks, the Essentiality Doctrine was basically used for unusual cases where an assisted acquisition could not be done. For Unity and Farmers, the obstacle was their unique positions in their respective communities. For the too-big-to-fail banks (other than Franklin), the obstacle was restrictive state banking laws.

Tables 1, 2, 3 and 4 report the within-state market share of each of these four too-big-to-fail banks. Commonwealth had only a 5 percent market share, but it was still the fifth-largest bank in Michigan and, as discussed earlier, its acquisition would have made the Detroit market particularly concentrated. Concentration was a concern in the cases of First Penn, Seafirst and Continental Illinois too, but there was also a lack of potential acquirers. The only bank larger than First Penn in Pennsylvania was Mellon and acquiring First Penn would have increased Mellon's market share in the state to 26 percent.Footnote ²⁵ Seafirst was the largest bank in the state of Washington with a 37 percent market share and was almost twice the size of the next-largest bank. Continental Illinois had a 25 percent market share in Illinois and combined with the second-largest bank would have had a 47 percent market share.

Table 1.

Ten largest commercial banks in Michigan, 1971

Notes: Only assets held under the commercial bank charter are counted.

Source: Authors’ calculations using Call Report data.

Table 2.

Ten largest commercial banks in Pennsylvania, 1979

Notes: Only assets held under the commercial bank charter are counted.

Source: Authors’ calculations using Call Report data.

Table 3.

Ten largest commercial banks in Washington state, 1982

Notes: Only assets held under the commercial bank charter are counted.

Source: Authors’ calculations using Call Report data.

Table 4.

Ten largest commercial banks in Illinois, 1983

Notes: Only assets held under the commercial bank charter are counted.

Source: Authors’ calculations using Call Report data.

V

Reading Sprague, one can see regulators struggling with the tradeoffs in doing a bailout.Footnote ²⁶ Do the short-term costs of a failure, particularly if a panic ensues, outweigh the long-term costs of increased moral hazard and a decline in market discipline? This tradeoff underlies the well-known time-consistency problem identified by Kydland and Prescott (Reference Kydland and Prescott1977) in which decisions that are ex post optimal need not be ex ante optimal. If policymakers cannot stick with a strategy in the face of short-term costs, then policies that are harmful in the long run may be implemented.

In bank resolution, the questions are how much short-term costs are worth bearing in return for long-term benefits and then whether the regulatory and political institutions can commit to bearing those costs. With large banks, the costs of a failure are viewed to be very high, and this suggests why it can be hard to commit, or even undesirable, to let large banks fail.

The lack of commitment in this period manifested itself through successively broader interpretations of the Essentiality Doctrine. The wording used in the 1951 amendment to define acceptable open-bank assistance is ‘essential to provide adequate banking service in the community’. In practice, these words did not provide much of a constraint. The FDIC used its increasingly broad interpretation of the word ‘community’ to justify the rescue of many of the banks discussed in this article. By the time of the Continental Illinois bailout, Sprague says about the Essentiality Doctrine:

The cost of a lack of commitment is that it shrinks the too-big-to-fail threshold, which increases moral hazard and subsidizes larger banks.Footnote ²⁷ Table 5 lists three size measures for each bailed-out bank: its assets in the year before it failed, its assets relative to GDP, and its assets in 2018 dollars deflated by the growth in banking industry assets. For example, Commonwealth had $1.25b in assets at the end of 1971, which corresponds to a bank with $29.9b in assets in 2018, which would not be considered a large bank today.

Table 5.

Several measures of bank size in year before failure

Notes: Only assets held under the commercial bank charter are counted. Assets in 2018 dollars are calculated by deflating bank assets by the growth in total commercial bank assets.

Source: Authors’ calculations using Call Report data.

We take the Commonwealth bailout as evidence that the lack of commitment lowers the too-big-to-fail threshold. Despite its small size, regulators in 1972 viewed Commonwealth as too big to fail. Sprague says about the Commonwealth bailout decision:

More evidence comes from the failure of US National Bank in San Diego (USNB). Like Commonwealth, this bank failed in the early 1970s with assets of just over $1b. USNB was ultimately acquired in an FDIC-assisted transaction by Crocker Bank, but there were difficulties in arranging this acquisition due to fraudulent behavior by USNB's owner (Spero Reference Spero1980, p. 93). Consequently, a payoff was discussed before the final deal was arranged. A summary of a meeting conducted by regulators on USNB reports that, ‘Comptroller Smith stated that in his view, it was inconceivable that a bank of this size would be permitted to fail.’Footnote ²⁸

VI

The first too-big-to-fail bailout of the modern era was that of the Bank of the Commonwealth in 1972. Commonwealth increased its holdings of long-duration municipal securities to bet that interest rates would drop. When this bet failed, the FDIC used the recently applied Essentiality Doctrine to bail it out because state banking laws and concentration limits made it impossible to find a buyer and, at that time, regulators felt that a $1b bank was too big to fail. The use of a bailout to resolve banks in which the regulator was unable to find a buyer was repeated with Farmers, First Penn and Continental Illinois, and nearly used with Seafirst. In the latter three cases, all were too big to fail and state-level branching restrictions and concentration limits limited the pool of buyers.

By modern standards, Commonwealth was small, so its bailout illustrates an unwillingness on the part of the political system to bear the economic and political costs of bank failures. This bailout set the precedent for the subsequent bailouts that led to the Continental Illinois bailout in 1984, and ultimately, the re-emergent ‘too-big-to-fail’ problem of the 2007–9 financial crisis.