THE THREE CHECK-CLEARING SYSTEMS

At the onset of the Great Depression, three check-clearing systems operated in the United States.⁶

In addition to the primary function of facilitating check transactions, clearing houses helped members attain numerous collective goals, such as maintaining confidence in the strength of local banks. Bankers recognized that “no bank can be in an unsound position without hurt to the whole local banking community.”⁹

Another common goal was the provision of liquidity, particularly during periods of panic, when depositors withdrew funds en masse, forcing banks to scramble for cash. In such circumstances, clearing houses eased the pressure on bank balance sheets by issuing clearing house certificates. These substitutes for currency circulated between banks and among their better customers, usually bigger businesses, alleviating the pressure to sell assets at depressed prices.¹¹

At the end of 1929 clearing houses operated in 45 states and the District of Columbia, as indicated by Table 1. Each day, the 349 clearing houses processed transactions totaling billions of dollars.¹²

The Federal Reserve operated the second check-clearing system. Federal Reserve member banks forwarded checks to the nearest Federal Reserve check-processing facility, which cleared checks drawn on member banks within its district and forwarded checks drawn on banks in other districts to the pertinent processing center. Banks that belonged both to the Federal Reserve and to a clearing house employed both clearing systems. Regulations required Federal Reserve members to process local transactions through the clearing house and out-of-town checks through the Federal Reserve System. Clearing house members often settled daily transaction balances by transfers on the books of the Federal Reserve, by telephoning or telegraphing the debits and credits to the Federal Reserve at the end of each business day.¹³

The Federal Reserve cleared checks drawn on nonmember state banks using procedures similar to those which clearing houses used to clear items drawn on country banks. The Federal Reserve shipped batches of checks to banks on which they were drawn, either via private messenger or the postal services, and awaited the return of funds (usually in the form of a draft drawn on an account of a Federal Reserve member or a money center bank). The Federal Reserve did not immediately credit out-of-town checks to the accounts of banks that deposited them. The Federal Reserve imposed a waiting period whose length was based on the typical length of time that it took for a check drawn on a type of bank in a particular location to remit funds. The waiting periods lasted up to eight business days depending on the distance involved and the type of bank on which the draft was drawn. Clearing houses imposed similar waiting times on outside checks.¹⁴

The third system for clearing checks was correspondent banking. The typical situation involved a bank outside a reserve city (called a country bank) that deposited funds in a bank in a reserve city (called a city bank) and received services in return. The bank making the deposit was referred to as the respondent. The bank providing the services was called the correspondent. When a country bank received out-of-town checks from depositors, rather than mailing the checks directly to the banks on which they were drawn, the country bank deposited the checks in its city correspondent. The correspondent then cleared the checks through the most convenient method, either through a clearing house, through the Federal Reserve System, or by directly contacting the institution on which the checks were drawn. Country banks found the services of city correspondents economical because it enabled them to clear all of their checks by making a daily deposit via the United States postal service, rather than employing a staff of clerks to handle the correspondence needed to send each check directly to the bank on which it was drawn. Correspondent clearing also enabled country banks to avoid exchange charges.

In addition to cashing checks, correspondents offered respondents an array of financial services.¹⁵

On the eve of the Great Depression, correspondent networks formed a complex web of interbank relationships anchored by banks in large commercial centers. A country bank often had a correspondent located in a nearby town, which was a member of the Federal Reserve, and therefore, could conveniently supply it with cash and clear many of its checks. A country bank also often possessed at least one correspondent in a financial center, where it held the bulk of its banker's balances, which usually comprised a substantial share of its financial reserves.

Country banks employed the services of correspondents because they provided greater benefits at lower costs than the alternatives of operating in isolation or joining the Federal Reserve System. The correspondent banking system, however, had weaknesses. One was the potential for a chain reaction. Deposits in correspondents counted as a portion, often substantial, of a respondent's reserves. When a correspondent closed, its respondents' reserves disappeared, placing pressure on other sources of reserves, such as vault cash and liquid assets. Thus, the closure of a large correspondent could force its respondents to suspend operations. Another weakness arose from the competing needs of correspondents and respondents. During panics, both needed cash, but correspondents could slow respondents' withdrawals by invoking contractual clauses limiting amounts that could be withdrawn each day, by requiring advanced notice of withdrawals, and by slowing the processing of withdrawal requests. Correspondents could also refuse requests for credit, and in extreme situations, call in loans. Thus during panics, when correspondents faced demands on many fronts, the deposits that they held lacked the liquidity normally associated with bank reserves.

Although the closure of a large correspondent could precipitate suspensions among respondent banks, an accounting convention may have made the depression era correspondent-clearing system particularly susceptible to panics. The seventh edition of a popular banker's manual described the arrangement in these terms.

In other words, a country (a.k.a. respondent) bank considered checks mailed to its correspondent as a portion of its financial reserves and immediately indicated so on its balance sheet. The correspondent bank itself carried the float. Academic and professional writings from the era warned of dangers inherent in this practice.

Yet, regulatory authorities in most states approved of this long-standing practice because its convenience seemed to offset potential disadvantages, such as allowing customers to write checks against uncollected funds. Moreover, “rural states adopted liberal bank incorporation laws that reduced minimum capital requirements to a fraction of the federal threshold and so spurred the formation of state-chartered banks in smaller cities and towns.”¹⁸

This common practice linked correspondent clearing and bank reserves in a dangerous manner: Reserves in the correspondent system were inflated by a multiple of the volume of checks in transit. Each check in transit was counted as part of the reserve of every bank through which it passed, until the check was redeemed, and the funds flowed backwards through the system, and the cycle unwound.

Seasonal flows of funds from crop-growing regions exacerbated the situation. Agricultural states possessed large numbers of small rural banks that relied on correspondents to clear checks. Transactions peaked during the harvest and planting seasons, when it would not be unusual for an institution in a small town to process checks equal to its required reserves within each week. If those checks took a week to clear through correspondents (also a typical period), then the bank's possessed no actual reserves, because the “float” exceeded the “reserves” that they carried on their books.

For these reasons, the connection between clearing and reserves placed the correspondent system in a perilous position. A large portion of the reserves of the correspondent system consisted of bankers' balances. Much of those balances consisted of checks in transit. A substantial share of the reserves of the correspondent system, therefore, consisted of fictitious reserves equal to a multiple of the volume of checks in transit. These reserves appeared to expand (but in reality did not change) whenever the volume of checks in transit increased. An event, such as a banking panic, that forced banks to convert reserves into cash, would reveal that the much of the reserves in any correspondent chain consisted of fictitious figures rather than real resources.

Precise data on the quantity of fictitious reserves do not exist, but extant sources yield closely related statistics. One is the ratio of checks in transit to bankers' balances for state commercial banks. Figure 1 displays the ratio for 1 July of each year from 1913 to 1940. The ratio peaked in 1930, when float approached 60 percent of bankers' balances. This peak was nearly twice the proportion prevailing during the preceding decade. The peak was more than four times the proportion prevailing following the trough of the contraction.

FIGURE 1 CHECKS IN TRANSIT AND BANKERS' BALANCES, 1913–1939

The ratio can be used to estimate levels of apparent, actual, and fictitious reserves. The estimation requires knowledge of the relationship between checks in transit and fictitious reserves. That relationship depended upon the speed at which checks passed through the correspondent system and the number of corresponding banks that simultaneously counted checks as reserves on their balance sheets. Those parameters remain unobserved. Safe assumptions, however, facilitate a reasonable estimate: fictitious reserves amounted to approximately 150 percent of checks in transit from banks that cleared via correspondents.¹⁹

This formula yields estimates of apparent (C + B), actual (A), and fictitious (1.5*T) reserves. Figure 2 plots the ratio of actual to apparent reserves on June 30 of each year from 1913 through 1939. The data cover all commercial banks, including those that cleared via correspondents, clearing houses, and the Federal Reserve. The inclusion of data on banks that cleared clearing houses and the Federal Reserve biases the estimate upward. This bias implies that for typical country banks, the ratio of actual to apparent reserves was lower than the figure suggests. The figure indicates that the share of actual reserves fell (and the share of fictitious reserves rose) during the 1920s, as the volume of check transactions increased, and checks in the process of collection became an ever larger entry on banks' balance sheets. For example, on 30 June 1913, state commercial banks listed a total of $195,000,000 of checks in the process of collection. Seventeen years later, on 30 June 1930, state commercial banks listed a total of $1,851,000,000 checks in the process of collection.²⁰

FIGURE 2 ACTUAL RESERVES AS A FRACTION OF APPARENT RESERVES

Figure 3 examines the implications of these patterns by plotting apparent and actual reserves on 30 June of each year as a fraction of apparent reserves held by banks in 1913. The figure illuminates an important point. During the 17 years following the founding of the Federal Reserve, apparent reserves doubled, while actual reserves fell by half.

FIGURE 3 ACTUAL AND APPARENT RESERVES OVER TIME

Figure 4 reinforces this message by plotting apparent and actual reserves as a function of total assets. The figure shows that in 1913, actual reserves amounted to 14 percent of total assets, and in 1917, actual reserves peaked at 16 percent of total assets. Actual reserves declined steadily thereafter, as banks substituted checks in collection for cash in the vault. When the banking panics began, actual reserves amounted to roughly 3 percent of all assets. Note that apparent reserves also fell during the Roaring '20s, as banks worried less about liquidity and more about profits.

FIGURE 4 ACTUAL AND APPARENT RESERVES RELATIVE TO TOTAL ASSETS

SOURCES OF DATA

Where does this evidence come from? Information about the three payment systems comes from the professional and academic literature of the era, information available in bankers' handbooks and manuals, and contemporary sources widely available at the time. For example, data on clearing houses, reserve cities, reserve requirements, and correspondent linkages comes from Rand McNally's Bankers Directory. Data on banks' balance sheets come from the compendium All Bank Statistics.²³

The remainder of this essay is based on a new and unique source. From 1929 through 1933 the Federal Reserve Board's Division of Bank Operations recorded information about changes in a bank's status on three forms. Form St. 6386a reported bank consolidations. Form St. 6386b reported bank suspensions. Form St. 6386c reported all other bank changes. These forms provide uniform and comprehensive information about changes of status for all banks operating in the United States—national and state, member and nonmember, public and private. The complete series of forms survives in the National Archives of the United States.²⁴

As part of its ongoing data collecting endeavors, the Federal Reserve Board developed a lexicon for classifying changes in bank status and the causes of bank suspensions. In this lexicon, a suspension was a bank that closed its doors to depositors and ceased conducting normal banking business for at least one business day. Some, but not all, suspended banks reopened for business. A liquidation was a permanent suspension. A liquidating bank closed its doors to the public, surrendered its charter, and repaid depositors, usually under the auspices of a court appointed officer known as a receiver. A voluntary liquidation was a category of closure in which banks ceased operations and rapidly arranged to repay depositors the full value of their deposits. Voluntary liquidations did not require the services of receivers and were not classified as suspensions. A consolidation (or merger) was the corporate union of two or more banks into one bank that continued operations as a single business entity and under a single charter. The categories of bank distress were typically construed to be temporary suspensions, terminal suspensions (i.e., liquidations), voluntary liquidations, and consolidations due to financial difficulties.

The Federal Reserve attributed most bank suspensions to one of five common causes. The first was slow, doubtful, or worthless paper. The term worthless paper indicated an asset with little or no value. The term doubtful paper meant an asset unlikely to yield book value. The term slow paper meant an asset likely to yield full value in time, but whose repayment lagged or which could not be converted to full cash value at short notice. The second common cause of suspension was heavy withdrawals, the typical example being a bank run. The third was failure of a banking correspondent. Correspondents were banks with ongoing relationships facilitated by deposits of funds. A typical example is a county bank (the respondent) that kept its reserve deposits within and cleared its checks through a national bank in a reserve city (the correspondent). The fourth common cause was mismanagement. The fifth was defalcation, a monetary deficiency in the accounts of a bank due to fraud or breach of trust.²⁵

When determining the causes of failures, the Federal Reserve Board sought to gather information about suspensions from the man on the spot who knew the facts of the issue at hand. The Board gathered information from examiners, receivers, correspondents, state banking departments, court officers, the bank's own management, and local and national publications. These sources, now no longer extant, provided the Federal Reserve Board with an array of information, now unavailable to economic researchers, such as the health of a bank's assets on the date of suspension, the deposits lost by the bank in the period preceding suspension, the lawsuits (or criminal charges) pending against bank management, and the links that the failed bank had to other financial institutions. Federal Reserve agents at that time could determine whether a bank experienced a run, failed to maintain cash flow, or feared impending insolvency; whether a bank's correspondent(s) closed; or whether a bank's management embezzled money. Modern scholars have difficulty detecting these phenomena in the extant evidence. Phenomena such as correspondent closures and bank runs remain unobservable.

The data demonstrate that for the contraction as a whole, asset problems were the primary cause of about half of all bank liquidations (i.e., terminal suspensions) and a contributing cause of another one-quarter. The share of liquidations due primarily to problems on the asset side of the balance sheet fell until the summer of 1931, rose in 1932, and fell again in 1933. The share of liquidations due primarily to withdrawals varied in the opposite chronological pattern. The share rose during the second half of 1930 and fell during the later half of 1931.²⁶

For temporary suspensions, the pattern was different. Temporary suspensions typically occurred when banks lacked enough cash on hand to satisfy depositors' demands. Heavy withdrawals were the primary cause of more than a half of all temporary suspensions. Temporary suspensions due to bank runs and heavy withdrawals rose during 1930 (particularly during the last three months) and peaked in 1931. The closure of counterparties caused a sixth of all temporary suspensions. The share of temporary suspensions due to the closure of correspondents peaked during the fall of 1930.²⁷

As this essay focuses on suspensions due to the closure of correspondents, it is worth discussing the nature of these events and how Federal Reserve agents determined them to be the cause of a suspension. The typical case involved a respondent bank that received notice that its correspondent had suspended operations. Respondents located near their correspondent usually received these notices via messengers. Respondents located far from their correspondents usually received these notices via telegram. The respondent's manager ordered the staff to close the bank's doors and summoned the board of directors. The directors met and voted to keep the doors closed on the following morning. The Federal Reserve classified events of this type, where a respondent bank suspended operations immediately after the receipt of information about the closure of a correspondent, as a suspension due to the closure of correspondent. Federal Reserve agents also classified as a suspension due to the closure of a correspondent a bank that closed soon (for example, several days) after learning of its correspondent's closure, if no other factors appeared to precipitate its suspension.²⁸

Related events received different classifications. For example, Federal Reserve agents classified as a suspension due to heavy withdrawals a bank whose correspondent closed, but which remained in operation for several days thereafter, during which depositors withdrew sums so substantial that the bank had to suspend operations. In such cases, comments written by the Federal Reserve agents often attributed the run on the bank to news of the correspondent's closure.²⁹

I emphasize that I make no judgments about the reasons why particular banks failed. Those judgments were made by contemporary experts possessing far more information about each event than is available to scholars today. This essay merely reports the experts' conclusions.

CORRESPONDENT NETWORKS AND THE INITIAL PANIC OF THE DEPRESSION

The correlation between suspensions and the closure of correspondents was particularly pronounced during the initial banking panic of the depression. Figure 5 plots the number of suspensions each week due to the closure of correspondents (and for sake of comparison also plots total changes due to financial distress). The typical week witnessed few, if any, suspensions from correspondents. The weekly mode and median were zero. The weekly number rose during July of 1929, when a Mediterranean fruit fly epidemic produced a banking panic in Florida, and the suspension of banks in Tampa, which served as the principal correspondents for banks in central Florida, forced the banks to suspend operations throughout the region, but remained near zero until November 1930, when it spiked sharply upwards.

FIGURE 5 SUSPENSIONS DUE TO CLOSURES OF CORRESPONDENTS, JANUARY 1929 THROUGH MARCH 1933

In a two week period, more than 120 banks suspended operations due to the closure of correspondents. The event is an obvious outlier. Never before or since have so many financial institutions fallen in a counterparty cascade. It does not seem surprising that this event, the worst chain reaction in United States financial history, marked the onset of the worst period of bank distress and most catastrophic financial contraction in United States financial history. In the immediate aftermath, bank runs radiated from the locus of the counterparty cascade, forcing hundreds of banks to close their doors to depositors.

The archival evidence enables us to describe this banking crisis in detail. Before mid-October, the pattern of bank suspensions resembled the pattern of failures throughout the 1920s. Banks failed at a steady rate. The principal cause was problems with asset quality. The pattern changed dramatically in November 1930, when the rate of suspension rose suddenly. Contagion through correspondent chains caused the initial increase. Thereafter, runs (and fear of runs) forced scores of banks to close their doors, and adverse circumstances pushed many weak banks into insolvency.

Comments written on the St. 6386 forms tell the tale. On 7 November the Bank of Tennessee (Nashville), a subsidiary of Caldwell and Company, closed due to “depreciation in the value of securities” and irregularities that left it with “bills payable of $2,887,100.00” and debts “on real estate of $260,079.20” on a deposit base of $10,000,000.³⁰

On 17 November Armageddon arrived. The National Bank of Kentucky (Louisville) suspended operations because of “heavy withdrawals and affiliation with the Caldwell Chain.”³²

The American Exchange Trust Company (Little Rock, AR) also suspended operations due to “heavy withdrawals due to rumors caused by failure of Caldwell and Company [in] Nashville.”³⁵

One respondent of the American Exchange Trust Company remained in operation for a month. The Citizens Bank and Trust Company (Harrison, AR) endured by calling on the resources of the A. T. Hudsputh Chain, for which it was the principal bank. But on 17 December, when those resources ran thin and the loss of funds on deposit in the American Exchange Trust Company appeared irreversible, the Citizens Bank threw in the towel.³⁹

The archival evidence cited attributes the suspension of the Caldwell conglomerate's principal banks to either: financial difficulties directly attributed to Caldwell's demise, or runs because of the banks known connection to Caldwell and Company. The archival evidence attributes the suspension of nearly 100 additional banks to: the severing of correspondent links to institutions, such as the Bank of Tennessee, controlled by the Caldwell conglomerate, runs due to known affiliations with the Caldwell organization, or runs due to geographic proximity to Caldwell controlled institutions or geographic proximity to banks undergoing runs.

Similar events occurred in Illinois, where correspondent chains without connection to Caldwell collapsed. On 8 November Quincy-Ricker National Bank and Trust Company (Quincy, IL) suspended operations due to the collapse its largest borrower, the “Smith and Ricker Land and Cattle Company, of Kansas City, Missouri.”⁴¹

The effects of these suspensions spread across state lines. On 20 November the Hannibal Trust Company (Hannibal, MO) suspended operations due to “heavy withdrawals due to closing of a number of banks in their section at Quincy, Illinois.”⁴⁵

During December 1930 similar dominoes of collapsing correspondent networks radiated out from Sioux City, Iowa and Clarksdale and Tupelo, Mississippi. On 6 December both the Sioux City National Bank and the First National Bank in Sioux City suspended operations due to “slow, doubtful, or worthless paper.”⁴⁷

On 24 December the Peoples Bank and Trust Company (Tupelo, MS) suspended operations due to “excessive bills payable” and “slow, doubtful, and worthless assets.”⁵²

CORRESPONDENT-CLEARING AND BANK CLOSURES DURING 1931 AND 1932

The previous section showed the banking crisis in the fall of 1930 began with a counter-party cascade. This section shows correspondent networks played roles later in the contraction. An example is an event that Milton Friedman and Anna Schwartz named the Second Banking Crisis. During this surge of suspensions in the spring of 1931, many of the banks that failed belonged to banking groups.⁵⁵

Group failures were most prominent in the city of Chicago, where between 6 and 10 June, 25 banks failed. Nineteen of those banks belonged to groups. Eleven belonged to the John Bain Group. Seven belonged to the Foreman Group. One belonged to the Ralph E. Ballou and E. L. Wagner Group. Additional banks in all three groups failed in the days preceding and following the panic.⁵⁶

Each of these banking groups was a collection of independently chartered institutions operating under the auspices of a corporate conglomerate, which owned the majority of the capital stock in each of the banks, and by appointing directors and hiring managers, controlled the bank's affairs. The leading bank of each of group served as the correspondent for the subordinate institutions. The groups were relatively recent innovations. They had grown during the 1920s, as correspondent banks in Chicago took control over respondent banks.

These groups magnified the weaknesses of the correspondent clearing system. The subordinate banks considered checks in transit as financial reserves. So did the controlling bank. The group as a whole, therefore, overstated its reserves. It is not surprising, therefore, that when depositors withdrew funds in large quantities, banking groups rapidly ran out of cash.

Table 2 examines the relationship between clearing systems and bank closure for the contraction as a whole. The table indicates the number of banks that suspended operations temporarily and terminally in three groups: first, banks that cleared checks solely through correspondents; second, banks that belonged to clearing houses (but not the Federal Reserve System); and third, banks that belonged to the Federal Reserve System (including those that also belonged to clearing houses). The table shows that the preponderance of banks that suspended operations cleared checks via correspondents. About a fifth belonged to the Federal Reserve System. Only a small fraction belonged to clearing houses.

Across these categories, the percentage of banks that suspended operations temporarily or terminally varied. Of all banks that cleared checks via correspondents, over one-third temporarily suspended operations at some point during the depression, and over one-fourth departed permanently from the banking business. These failure rates were four times higher than those of clearing-house members and two-and-one half times larger than those of Federal Reserve members. The failure of the later two types was concentrated during the three months following Britain's abandonment of the gold standard in the fall of 1931, to which the Federal Reserve reacted by contracting the money supply and raising interest rates. That reaction changed fundamentals in a way that burdened money center banks. So did later periods of contractionary policy. Outside of those periods, banks that cleared checks via correspondents failed at rates much higher than banks that cleared checks through clearing houses or the Federal Reserve.

These raw correlations do not, of course, reveal why banks that cleared via correspondents failed at such high rates. Banks that cleared via correspondents differed from clearing-house and Fed-member banks along many dimensions, such as location, size, investment opportunities, customer base, regulatory environment, access to the discount window, and management expertise. Those dimensions undoubtedly influenced banks' prospects for success (or failure).

DISCUSSION

The correspondent check-clearing system played a hitherto unrecognized role in the collapse of the banking system during the Great Depression. The initial banking crisis began when correspondent systems collapsed. A large literature discusses this event. Friedman and Schwartz attributed the crisis to the loss of depositor confidence after the failure of The Bank of United States.⁵⁷

Two new observations complement McFerrin and Wicker's conjecture. First, soon after Caldwell's collapse, several correspondent chains independent of Caldwell imploded. Second, the simultaneous implosion of independent correspondent networks suggests that in addition to problems particular to Caldwell, some unrecognized malady afflicted correspondent chains in general. Fictitious reserves are a candidate for this general cause.

Correspondent-clearing also played a part in the second banking crisis. That crisis began during the spring of 1931 when runs struck banking groups in Chicago. These groups formed correspondent networks whose leading institution had purchased controlling shares of stock in their respondent banks. Little scholarship exists concerning the cause of this crisis. Friedman and Schwartz attribute its onset to runs on banks in Illinois. Friedman and Schwartz do not explain why depositors drained funds from banks at that time or why this drainage drove the banks out of business. This essay provides a potential explanation. Depositors observed the weakness of correspondent networks during preceding months, and precipitously withdrew funds from correspondent groups in Chicago when fears about their safety arose. Once runs began, these groups lacked reserves sufficient to weather the storm, because much of their reserves consisted of checks in the process of collection, which they double or even triple counted, when a bank within the group served as a correspondent for the others.

Correspondent clearing may have played a broader role in the collapse of the banking system. From 1929 through 1933 banks that cleared via correspondents failed at rates higher than banks that cleared via other channels. The cause of this correlation remains unclear. I argue that the cause may have been correspondent networks inherent vulnerability to cascades accentuated by their practice of treating float as reserve. But, banks that cleared via correspondents differed along many dimensions from banks that cleared checks via the Federal Reserve System or metropolitan clearing houses. When compared to the later, on average, banks that cleared via correspondent possessed less capital, lower deposits, and fewer assets; operated in more rural areas; invested more in agriculture; diversified investments less across industries; and employed smaller staffs with less experience. These factors may be the reason that banks clearing via correspondents failed at higher rates from the summer of 1931 through the winter of 1933. Determining which was the most important requires the collection of additional data and the development of techniques to distinguish the effects of contagion through correspondent networks from other factors influencing banks' fates.

Why did correspondent chains collapse so suddenly and severely early during the depression? Correspondent networks were vulnerable to counter-party cascades. Correspondents and respondents formed lines like dominoes. One bank held deposits for another bank, which in turn, held deposits for a third. These deposits comprised the preponderance of the respondents' reserves (both required and excess). When one domino toppled, the reserves of the next domino disappeared, and it suspended operations also, which forced additional dominoes to fall.

Counter-party cascades of this type have long worried regulators. The systemic risk of correspondent linkages remains high on scholars' research agenda and regulators' list of concerns. Debate revolves around the question: are some banks too big to fail? Could the closure of large institutions drive substantial numbers of their respondents out of business or cause a crisis of confidence that spills over to other banks and financial institutions? This phenomenon manifested itself in 1984, when Continental Illinois, then one of the ten largest banks in the United States, became insolvent.⁶⁰

Effective intervention may have been difficult. Fictitious reserves made depression-era correspondent networks particularly vulnerable to counter-party cascades. The quantity of fictitious reserves peaked just before banking panics began in 1930. In such circumstances, it is not surprising that correspondent chains collapsed so suddenly, that depositor confidence declined so dramatically, that runs struck so many institutions located near the locus of correspondent cascades, or that banks held such sizeable reserves for the next decade.

The evidence presented in this essay only suggests, but does not prove, the existence of sizeable fictitious reserves. Additional research is required. Lack of data will hinder progress. Few sources contain the requisite information. High frequency data concerning checks clearing via correspondents are not extant and may not exist. Few publications of state banking bureaus report the quantities of checks in the process of collection on state-bank balance sheets. Federal Reserve call reports contain figures for checks undergoing collection, but for nationally chartered banks, only two call reports for the period 1929 through 1933 survive, and nationally chartered banks did not clear through the correspondent system. So, data on their checks in the process of collection shed little light on the key issue.

Though difficult, research on fictitious reserves could be consequential. If fictitious reserves were large, then fictitious reserves may have had macroeconomic consequences beyond their effects on bank stability. One possible channel is by inversely linking the volume of check transactions with the level of bank reserves. An increase in the volume of check transactions increased reserves listed on balance sheets by increasing fictitious reserves. This fictitious increase enabled banks to reduce real reserves, or synonymously, this fictitious increased acted like a decrease in the legal reserve requirement. A change in the volume of check transactions, in other words, had an unintended consequence that resembled one of the principal levers of monetary policy. Via this channel, a change in the volume of check transactions could have altered the deposit-reserve ratio, which is a principal component of the money multiplier, and altered the supply of money, thereby influencing the interest rate and price level.

If the quantity of fictitious reserves expanded substantially during the 1920s and contracted during the 1930s, as the preliminary evidence presented in this essay suggests, then fictitious reserves may have accentuated the macroeconomic forces propelling the economic expansion of the Roaring '20s and economic contraction of the early '30s. This possibility raises another issue. Scholastic studies of these events rely on figures for bank reserves, money multipliers, and monetary aggregates constructed without accounting for the fact that banks that cleared via correspondents counted checks in transit like cash reserves, while banks that cleared checks in other ways did not. This means that those statistical series may have systematic biases correlated with both the volume of checks in transit, the size of various banking systems, and economic aggregates. Finally, the problem of fictitious reserves during the contraction of the early 1930s may be one reason that banks held such sizeable excess reserves during the recovery of the later 1930s. Bankers may have realized that the reserve balance on their books did not reflect the resources actually available during panics. To account for that fact, bankers held reserves larger than those legally required.