Ben S. Bernanke
It would be difficult to overstate the economic, social, and political impact of the Great Depression of the 1930s. At the worst of the slump, in 1933, as many as one in four workers were unemployed, with many more on short hours and reduced pay. In the absence of an adequate social safety net, millions of Americans contended with poverty, hunger, and homelessness. And yet, even as the jobless searched desperately for work, many of the nation’s factories lay idle, with industrial production falling by half [Cb31] and real gross domestic product declining by more than a quarter [Ca6] between 1929 and 1933.1 Depression-era unemployment was not caused by a shortage of productive capacity.
For economists, the coexistence for more than a decade of idle labor and unused capital—want in the midst of plenty—presented a troubling paradox, and a satisfactory explanation of the Depression proved elusive for many years. An important breakthrough came with the work of Friedman and Schwartz (1963), who blamed the economic contraction on a steep decline in the money supply—the result of misguided monetary policies and waves of bank failures that the Federal Reserve failed to arrest. Subsequent research put the Friedman-Schwartz narrative in an international context (Eichengreen, 1992, is the definitive reference). This later work supported the monetarist thesis by documenting that countries (including the United States) that tied their currencies to a flawed international gold standard, and that consequently adopted tighter monetary policies than appropriate, experienced deeper and more extended slumps than countries that left the gold standard earlier. Although no event as complex as the Depression has a single cause, most economists now accept the monetarist cum gold standard story as an essential piece of the
explanation. Indeed, this framework underlies much of my own work on the period
(collected in Bernanke 2024).
However, despite its strengths, the standard monetarist explanation of the Depression has significant gaps. Notably, it provides limited insight into the monetary transmission mechanism, the channels through which the fall in the money supply putatively led to large and persistent declines in US output and employment. Moreover, the monetarist account views a distinctive feature of the Depression—widespread financial distress and disrupted credit markets—through a narrow lens, focusing only on banking panics and their effects on the money stock. But the broader credit crisis of the 1930s was no sideshow. It devastated both bank and nonbank lenders, paralyzed bond and stock markets, and pushed millions
of borrowers—including farmers, homeowners, and businesses—into default, bankruptcy, or foreclosure.
In Bernanke (1983), I argued that the credit crisis of the 1930s was, in fact, the primary mechanism through which monetary contraction debilitated the economy. The prelude to the crisis was a substantial buildup in private debt during the 1920s, reflecting, among other factors, post–World War I optimism about US economic prospects. That optimism was shattered when the collapse in the money supply helped trigger an economic downturn in the early 1930s. Because debt contracts are normally written in nominal (dollar) terms, the declines in prices and nominal incomes that accompanied the downturn made it increasingly difficult for many borrowers to service their debts, resulting in escalating rates of delinquency and default.
Given preexisting vulnerabilities, including structural weaknesses in the US banking system and chaotic financial conditions abroad, deteriorating credit quality in turn posed existential threats to banks and other lenders. In particular, many weaker banks succumbed to depositor runs that forced them to close their doors, while surviving banks girded themselves against possible future withdrawals of deposits by stockpiling safe, liquid assets, calling outstanding loans, and reducing new lending. As borrowers struggled and lenders failed or retrenched, refinancing existing debts or obtaining new credit became extremely difficult, with one consequence being reduced private spending on goods and services.
Declining demand further depressed production and prices, helping to propagate the economic slump and adding to the financial pressures on both borrowers and lenders. Government policies aimed at normalizing credit markets—rescuing delinquent borrowers, resolving failed institutions, strengthening surviving lenders, and implementing financial reforms—took considerable time to have their full effect, helping to explain why full employment was not restored until the United States entered World War II.
These ideas were not new when I first wrote about the Depression but descended from two traditions in economics. The older tradition goes back at least to John Maynard Keynes and Irving Fisher. Fisher (1933) believed that the effects of deflation on the real value of nominal debts—a phenomenon he dubbed debt-deflation—was by far the most important source of the Depression, an argument he made in testimony before the US House Ways and Means Committee in April 1932. Even earlier, Keynes (1931) had predicted that deflation, through its effects on the ability of borrowers to repay, would force the global banking system into insolvency, with dire economic consequences.
The second, newer tradition on which I drew was the then-evolving literature on the role of asymmetric information in markets, including credit markets (Stiglitz and Weiss 1981). Because lenders have imperfect information about potential borrowers’ creditworthiness, they must incur costs (which I called the costs of credit intermediation) to screen and monitor borrowers. Intermediation costs are lower when potential borrowers are financially strong. For example, a borrower who is able to make a large down payment or offer adequate collateral will have both a greater incentive and a greater ability to avoid defaulting on the loan, reducing the need for costly screening and monitoring by the lender. Moreover, as lending institutions are also borrowers, obtaining funds from depositors and other creditors, the financial health of lenders also affects the costs of extending credit. For example, a well-capitalised bank, being safer, is able to borrow more cheaply; it is also generally more willing to make risky loans, because any losses it sustains are less likely to threaten its solvency.
Debt-deflation erodes the financial strength of both borrowers (the real burden of their debts increases) and lenders (their asset quality deteriorates but their nominal liabilities are unchanged). If the losses to either borrowers or lenders, or both, are severe enough, the cost of intermediating credit may become so high that private credit markets function poorly, if at all. A contribution of my paper was to relate these two traditions and show how together they help us better understand the Great Depression. In later work with Mark Gertler and others (see references in Bernanke 2024, chap. 1), I explored these ideas more formally and
showed how they could be incorporated into standard macroeconomic analysis.
This article revisits the proposition that the credit crisis was a key monetary transmission mechanism during the Depression. I see the evidence for this story today as far stronger than when I advanced it 40 years ago, for two broad reasons.
First, in recent decades a growing literature has focused on the macroeconomic
implications of so-called “financial frictions,” including asymmetric information, the real costs of bankruptcy and illiquidity, the effects of “fire sales” on asset prices, and other departures of financial markets from the frictionless ideal. Indeed, Akerlof (2019) has argued in this journal that the exclusion of financial frictions from the standard Keynesian-monetarist framework would prove to be that framework’s greatest weakness. Not surprisingly, the global financial crisis of 2007–2009 and the subsequent recession accelerated this line of research by demonstrating that credit-market dysfunction can have large macroeconomic consequences, even absent price deflation (Bernanke 2018).
Second, there has been a remarkable improvement in recent decades in the data and techniques used in empirical studies of the economic effects of disrupted credit markets, including during the Depression. Advances include the development of detailed microeconomic datasets that allow comparisons across geographic areas and over time, the inclusion of better controls to help isolate credit effects, and the application of new econometric techniques. Most important, I believe, is the emphasis of recent work on identifying causal relationships. Much of the evidence in my original paper was anecdotal and historical, with the only formal econometric tests based on timing relationships among aggregate variables.
More recent empirical work on the Depression has made substantial strides in identifying and quantifying causal linkages, using clever instruments or natural experiments applied to microeconomic data. This article can draw on only selected examples from this burgeoning empirical literature. Taken as a whole, however, the new empirical research provides significant support for the view that credit-market dysfunction was a key transmission mechanism during the Depression, and in some other episodes (including the 2007–2009 crisis) as well.
The next section provides a closer look at the credit boom of the 1920s and the crisis that followed. I focus on the financial distress of Depression-era borrowers—who are largely ignored in the standard monetarist narrative—as well as the troubles of banks and other lenders. I then discuss the government’s response to the crisis, documenting that contemporary policymakers viewed the normalization of credit flows as a top priority in the fight against the Depression. Finally, I briefly consider some alternative monetary transmission mechanisms that have been proposed and offer a short conclusion.
Historical Perspectives on the Credit Crisis of the 1930s
According to what has become a conventional taxonomy, credit crises have their
roots in periods of heavy borrowing and rapid increases in asset prices that make the
economic system more vulnerable to external shocks or changes in sentiment (for
comprehensive cross-country evidence, see Jordà, Schularick, and Taylor 2013). Not
all credit booms end in crises, but when they do, the increased difficulty of refinancing
old debts or obtaining new credit depresses the willingness of households and businesses to spend, potentially leading to extended recessions. In severe crises, even households and businesses with no immediate need for credit will hunker down, knowing that credit might not be available if needed in the future.
Private-sector debts did increase rapidly in the 1920s, relative to prewar levels—a development that some observers noted with concern at the time (for example, Persons 1930). Economic optimism motivated much of this borrowing. Indeed, following a relatively brief recession in 1920–1921, America’s economic prospects seemed bright. With Europe rebuilding from World War I, the United States was the world’s dominant economy, with a growing middle class eager to purchase automobiles and other mass-produced consumer durables. Short recessions in 1923 and 1926 did not quench this optimism, as they suggested that temporary downturns would not impair longer-term growth. Indeed, real GDP would grow at a 4.9 percent annual clip between 1921 and 1929 [Ca9].
However, the downside of post–World War I optimism was insufficient caution. Lenders loosened credit standards and ran ad campaigns to encourage borrowing. The period also saw rapid financial innovation, including the popularisation of new forms of credit (installment loans, personal loans), market-based instruments (mortgage securitizations), and quantitative credit evaluation tools. There were episodes of intense if ill-fated speculation, including land-price booms and busts in Florida and elsewhere and the breathtaking ascent of stock prices at the end of the decade.
By the end of the 1920s, the buildup in private debt had made the economy vulnerable to declines in the price level (deflation) and incomes. Indeed, the deflation, when it came, would be severe: Consumer prices fell by about 27 percent between December 1929 and April 1933 [Cb74], consistent with a 25 percent decline in the M1 money stock [Cj42] over the same period. Together with the broader slowing of the economy that began in 1929, falling prices and incomes left many debtors struggling to make payments, which in turn pushed financial institutions closer to insolvency. The worsening credit crisis helped to convert what initially seemed to be a normal downturn into a deep and persistent depression. The rest of this section takes a closer look at how the crisis played out in some key economic sectors: agriculture, the household sector, nonfarm businesses, and banks.
Agriculture
In 1929, about one-quarter of Americans lived on farms and more than 40 percent lived in rural areas [Aa712, Ac414, Aa699]. Developments in agriculture thus had substantial economic and political impact.
American farmers had already experienced a boom and bust in the 1920s. During World War I and its immediate aftermath, demand for US food exports led farm prices to more than double [Da1337], with farmland values rising sharply [Da27]. To ramp up capacity, between 1914 and 1921 farmers in turn more than doubled their outstanding mortgage and nonmortgage debts, from $6.3 billion to $14.1 billion (Historical Statistics 1975 edition, K361, K376). The reliance of farmers on debt-financed expansion continued, with the share of farms with mortgages rising from 35 percent in 1920 to 42 percent in 1930 [Da578]. For farms with mortgages, mortgage debt as a percentage of the value of land and buildings rose from 29 percent in 1920 to 40 percent in 1930, then to 50 percent by 1935 as farm valuations fell [Da579].
However, within a few years after the end of World War I, crop prices began to soften, reflecting the recovery of global food production, increasing protectionism, productivity gains, and other factors. Lower prices for their products pressured financially overextended farmers and, despite the general prosperity of the 1920s, the decade saw a significant rise in farm defaults and foreclosures. Collateral damage included the failure of thousands of small agricultural lenders over the decade (Alston, Grove, and Wheelock 1994).
The advent of the Great Depression further worsened farmers’ finances. Between 1929 and 1932, crop prices fell 58 percent [Da1338]. In 1932, a farmer would have needed to produce 263 bushels of wheat to meet a fixed $100 obligation, compared to 96 bushels in 1929 (Chandler 1970, p. 60). In addition to mortgages, many farmers relied on short-term credit to finance planting and harvesting even as they continued to make payments on debts incurred in the 1920s to expand acreage under tillage and buy new equipment. Farm owners also faced property tax payments, which typically adjusted only slowly to declines in farmland values. A contemporary estimate found that farm income, net of operating expenses, taxes, and interest costs, fell from $4.7 billion in 1929 to only $1.3 billion in 1932 (Forster and Weldon 1934, Table 5). Compounding farmers’ woes, the Dust Bowl ravaged the western plains in the middle of the decade, destroying crops and eroding the topsoil, leading farmland values to drop further (Hornbeck 2012).
As farm prices fell and the debt crisis worsened, the foreclosure rate per 1000 farms rose from 14.7 in 1929 to 38.8 in 1933, the latter figure corresponding to more than 200,000 foreclosures within the year. For comparison, decadal averages between 1950 and 1980 were well under 2 foreclosures per year per 1000 farms (Alston 1983, Table 1). Moreover, the foreclosure data for the 1930s almost certainly understate the financial distress in agriculture, as these data refer only to legally completed foreclosure actions and exclude cases in which the farmer ceded the property without contest or filed for bankruptcy. In addition, some potential foreclosures were avoided or delayed by lender forbearance, government credit programs, and state-ordered foreclosure moratoria. A more comprehensive indicator of farm distress is the rate of delinquency. At the beginning of 1933, a remarkable 52 percent of farm mortgage debt was in arrears, often by several years (Bean 1934, p. 4). In part, high delinquency rates reflected the fact that many farmers, unable to meet the down-payment requirements of standard mortgage loans, had taken out second mortgages, usually at high interest rates.
The usual terms of farm mortgages were themselves a source of stress. Except for a limited amount of longer-term credit provided through federal programs, most farm mortgages had a short term—typically five years or less—and were non-amortising, requiring a balloon payment at maturity. When farmland values were rising in the early 1920s, farmers typically had little difficulty refinancing when their mortgage contracts expired, but when farm values began to fall, refinancing the principal amount became increasingly difficult.
Crushing farm debts and tax burdens led to unrest, political agitation, and occasional violence. In the Midwest, foreclosures and tax sales provoked what amounted to a farmers’ strike. Farmers destroyed crops and livestock, picketed crop deliveries, disrupted auctions of foreclosed farms, and harassed or threatened officials attempting to enforce foreclosure judgments.
By closing off access to credit and reducing discretionary income, the agricultural debt crisis hurt the broader economy by limiting the ability of farmers to invest in their farms and to buy consumer goods. For example, after rapid increases in mechanisation in the 1920s, net investment by farmers in tractors and motorised trucks was negligible in the early 1930s [Da623, Da627]. Hausman, Rhode, and Wieland (2019) have recently examined the effects of the crisis on farmers’ consumption spending. Following an insight of Temin and Wigmore (1990), these authors looked at the period after the 1933 dollar devaluation, which pushed up crop prices (as discussed further below).
As predicted by theory, they found that higher crop prices in a particular county or state were associated with increased local auto purchases (an indicator of overall consumption), with purchases increasing more in areas in which local crop prices had risen the most and in which agricultural debt burdens were particularly large. Hausman, Rhode, and Wieland calculated that increased consumption spending by farmers in 1933 contributed significantly to the rise in aggregate demand that followed the devaluation.
Households
In the Depression era, like today, home mortgages made up the bulk of nonfarm
household liabilities. Overall, the 1920s were a banner decade for residential real
estate, reflecting rising affluence, rural-to-urban migration, increasing rates of automobile
ownership (which promoted the growth of suburbs), and limited housing construction during World War I. Housing starts rose from 247,000 in 1920 to a peak of 937,000 in 1925 [Dc510]. Starts then moderated, declining to 509,000 in 1929, a drop that contributed to the recession that began in August. Starts collapsed to 93,000 by 1933, crushing employment in construction and ancillary industries, including building materials, furniture, and transportation. House prices fell about 30 percent between 1925 and 1933 [Dc826]. Because of the concurrent fall in the general price level, the decline in house prices was not large in real terms, but was much more significant relative to fixed mortgage obligations.
In earlier times, residential housing had typically been acquired for cash. By the 1920s, in contrast, obtaining a mortgage was relatively easy for middle-class families. The most active mortgage lenders were the so-called buildings and loans (B&Ls), the antecedents of savings and loans. In 1929 there were more than 12,000 B&Ls making close to 40 percent of new residential mortgage loans (Fishback et al. 2020). Commercial banks were not a major source of residential mortgage credit at the time [Dc983-Dc989], making less than 15 percent of new mortgage loans in the peak lending year of 1928. Until the late 1920s, nationally chartered banks faced restrictions on their holdings of real estate. However, some state-chartered banks, including the notorious Bank of United States (whose failure in December 1930 triggered a panic) loaded up on mortgages in the 1920s, which would prove disastrous when real estate prices plummeted. Other sources of mortgage credit included mutual savings banks, life insurance companies, and individuals.
Like farm mortgages, residential mortgages in the 1920s typically were not amortising
and thus had to be refinanced at the end of their terms. Down payments on first mortgages were usually in the range of 40–50 percent of the house price. About three-quarters of home mortgage borrowers, unable to cover the down payment, took out a second mortgage as well, usually at a high interest rate (Postel-Vinay 2017).
When the incomes of many homeowners fell in the early 1930s, residential foreclosures
rose sharply, from 68,000 in 1926 to roughly 240,000 per year from 1932 through 1935 [Dc1255]. Available evidence suggests that delinquency rates were correspondingly high. A survey of 22 cities, reported in Hart (1938, p. 164), found that, as of January 1, 1934, rates of homeowner default on interest or principal ranged from 21 percent to 62 percent across the cities surveyed. For the nation as a whole, by 1933 probably between one-third and one-half of American homeowners with mortgages were in default. Renters also faced significant financial stress, and major cities saw sharp rises in eviction rates, which in turn led to rent strikes, mass protests, and vigilante action to block evictions. Declining rent collections in turn affected the ability of landlords to make their own mortgage payments. Indeed, the survey of mortgage defaults cited by Hart found that default rates on mortgages on rental properties were as high or slightly higher than those on homeowners’ mortgages.
The high rates of mortgage delinquencies and foreclosures, together with the advent of government programs discussed later, virtually eliminated private mortgage lending after 1932. New mortgage loans by private financial institutions in 1933 were about one-fifth what they were in 1928, as banks and mutual savings banks cut lending severely and life insurance companies largely left the mortgage lending business [Dc983, Dc984-9]. Buildings and loans saw their portfolios become stuffed with foreclosed properties, reducing their capacity to make new loans (Fishback et al. 2020). Building and loans ultimately were hit harder even than banks: by 1941, more than half of the B&Ls operating in 1929 had failed. The process of liquidation was slow and painful, with many troubled B&Ls not being wound up until late in the decade (Rose 2014). The B&Ls had begun as cooperatives, and many of their shareholders, often middle-class homeowners, lost their entire stake when their B&L failed.
Another source of mortgage funding, which came into its own in the 1920s, was real estate bonds, an early form of securitisation. These bonds helped finance both residential construction and commercial property development. However, with the mortgage crisis, the real estate bond market collapsed. The issuance of new bonds had been $684 million in 1928, near the peak reached in 1925. By 1932, however, issuance was nil as bond values fell sharply (Johnson 1936). The chaos created by home foreclosures, delinquencies, and the failures of lending institutions left a long trail of legal uncertainties about ownership and property rights that took many years to resolve, further delaying recovery in the housing sector (Field 1992).
An additional factor affecting consumer finances was the rapid growth in consumer credit during the 1920s. According to Olney (1999, Table 1), between 1921 and 1929 household nonmortgage debt, including installment debt for mass-market durable goods (automobiles, furniture, appliances) as well as personal loans, rose from 5.5 percent to 9.3 percent of household income. To reduce default risk, installment lenders typically required high down payments and short terms. The penalty for a missed payment was repossession of the item, with no credit for built-up equity. Consumers thus had a strong incentive to keep current with their payments, often prioritising them over even mortgage payments. The need to service installment debts further restricted consumers’ discretionary spending and access to new credit—enough, according to Olney’s calculations, to account for a sharp drop in consumption spending in 1930 that was greater than could be explained by declines in income and wealth alone (Temin 1976).
Businesses
The nonfarm business sector also borrowed heavily during the 1920s. Long-term corporate debt increased from $31.8 billion in 1926, the first year for which the Historical Statistics includes data, to $50.3 billion [Ch518] in 1930, remaining near the latter figure for the rest of the decade.
Data on the debts of noncorporate businesses are limited. Nevertheless, it seems clear that smaller businesses faced the greatest difficulties in servicing debt and obtaining credit, given that small- and medium-sized firms of the Depression era were generally heavily dependent on banks for working capital, while the largest firms were often net creditors of banks (Richardson and Del Angel 2024). While large firms increased their holdings of liquid assets during the Depression, small firms became very illiquid, affecting their ability to carry on normal business and to service debt obligations (Hunter 1982).
One indication of concern about credit availability for smaller businesses in the 1930s was the proliferation of both private and public surveys of lenders and businesses. One large survey, US Bureau of the Census (1935), sent questionnaires to 16,500 smaller manufacturers, collectively representing 97 percent of manufacturing firms in 1929, receiving a 46 percent response rate. The survey found that 71 percent of the respondents were or had been regular bank borrowers, and that 45 percent of those reported that they now had difficulty in obtaining working capital. Further, of the firms able to obtain working capital, 40 percent reported that they had no available source for long-term credit. To determine whether the problem was a lack of creditworthy borrowers, the Census also analysed the financial conditions of reporting firms, ranking them by net-worth-to-debt ratios and Dun and Bradstreet ratings. This exercise revealed that at least 20 percent of the manufacturers reporting difficulty in obtaining credit from customary sources were in excellent financial shape. As the survey was conducted after the stabilisation of the banking system, its results suggest a persistent effect of banking problems on credit availability for smaller firms.
Even with better access to banks, larger firms were not necessarily free of credit problems. The issuance of new equity was difficult and expensive; by 1934, new stock issues had fallen to a minimal $35 million [Cj837]. The corporate bond market continued to function, albeit also at low volumes, with issuance falling from $4.2 billion in 1927 to $570 million in 1933 (Hickman 1953). The reluctance of investors to lend to all but the strongest firms was evident in the widening of credit spreads. In a study of the pricing of high-yield bonds (rated B or lower), which at their 1940 peak made up more than 40 percent of rated bonds, Basile et al. (2015) documented that the cost of capital for firms of low to intermediate credit quality (as measured by the interest-rate spread between high-yield and government bonds) rose substantially early in the Depression and remained high for the rest of the decade.
In further evidence that the corporate bond market became inaccessible for all but the financially strongest corporations, Benmelech, Frydman, and Papanikolaou (2019) used data for 1928–1933 for about 1,000 large firms, separating their sample into firms with long-term debt maturing in 1930–1934 and those without maturing debt in that period. The maturity dates of long-term debt were presumably determined well before the Depression, so splitting the sample in this way provides a natural experiment. If firms with maturing debt have difficulty refinancing, or find doing so prohibitively costly, they might choose instead to reduce costs, including their wage bills. Benmelech, Frydman, and Papanikolaou found quite large differences in employment between firms with maturing debt and those without, implying that even relatively sizable firms found bond issuance difficult and costly.
Banks
Although almost all types of lenders were adversely affected by the Depression, research on lending institutions, following the lead of Friedman and Schwartz (1963), has mostly focused on the banking system and banking panics.
Bank failures had long been common in the United States, but the banking panics of the early 1930s were especially intense, with roughly one-third of all US banks suspending operations between 1929 and 1933. Bank failures or suspensions, which dissipated the local knowledge and personal relationships that facilitated many loans, doubtlessly affected credit availability. Those effects would have been particularly devastating in remote and rural areas served by few banks, or for smaller borrowers dependent on a particular bank. In addition, the freezing or loss of deposits in suspended banks hampered the ability of businesses to make payrolls and of households to service debts. But importantly, bank failures did not capture the full impact of the crisis on economic activity. The altered behavior of surviving banks in the face of heightened run risk was at least as important.
Banks that survived the initial panics understood that some triggers of depositor runs were out of their control. For example, expectations of dollar devaluation likely contributed to the nationwide banking panic of 1933 (Wigmore 1987), while Friedman and Schwartz (1963) point to the confidence-destroying effects of Britain’s departure from the gold standard in 1931 and similar events. But if national or international developments were the main source of depositor runs, then panics would have tended to be national in scope. Detailed studies of panic episodes have found instead that most panics were local or regional in nature and could be traced either to well-publicized problems of individual banks or banking groups, such as Caldwell and Company or the Bank of United States in 1930 (Wicker 1996), or to adverse conditions facing all banks in an area. Even the nationwide 1933 banking panic began with the highly visible collapse of prominent banking concerns in Detroit.
Rather than macroeconomic events, the evidence is strong that, at least before 1933, the condition of each bank’s own balance sheet, or the balance sheets of its close affiliates, was the primary determinant of the bank’s run risk (Calomiris 2007). This link is supported by careful studies using detailed data on individual banks to understand how features of banks and their portfolios affected the risk of depositor runs in the 1930s. For example, Calomiris and Mason (2003b) found that the composition of bank assets and structural features (like size) that affected banks’ ability to diversify effectively predicted failures in the early 1930s. Similarly, using micro-level data, White (1984) identified bank-specific loan losses as a key source of failures in the panic of 1930, and Postel-Vinay (2016) found that large real estate holdings (which in the absence of secondary markets were highly illiquid) were an important predictor of bank failures in Chicago in the 1932 panic. Using a different identification strategy, Mitchener (2005) found that variations in pre-Depression regulatory oversight, and thus in the quality of banks’ lending and portfolio management, also helped predict failure rates in the 1930s.
Understanding the link between credit losses and asset illiquidity on the one hand and the risk of depositor runs on the other, surviving banks tried to protect themselves by “scrambling for liquidity,” shifting their holdings as quickly as possible toward safer, more liquid assets, like government securities, while reducing new lending and calling or failing to renew existing loans.4 In general, the value-added of the banking system is its ability to convert short-term liquid funding into long-term, often illiquid loans and investments. In the early 1930s, most banks could no longer fulfill that function.
How important were bank failures and the retrenchment of surviving banks for credit availability and economic activity in the Depression? As mentioned earlier, a strength of the more recent research is the use of detailed microeconomic datasets that allow for sharper comparisons across time and space. For example, using bank-level data, Calomiris and Mason (2003a) found that factors that predicted bank fragility also predicted declines in economic activity at the state and county levels in 1931 and 1932. Mitchener and Richardson (2025), also using bank-level data as well and the retrenchment of surviving banks for credit availability and economic activity in the Depression? As mentioned earlier, a strength of the more recent research is the use of detailed microeconomic datasets that allow for sharper comparisons across time and space. For example, using bank-level data, Calomiris and Mason (2003a) found that factors that predicted bank fragility also predicted declines in economic activity at the state and county levels in 1931 and 1932. Mitchener and Richardson (2025), also using bank-level data as well as controls for local economic activity, found that the reactions of surviving banks to depositor withdrawals during panic periods accounted for 39 percent of the decline in bank lending between July 1929 and December 1932.
Several researchers have taken advantage of the pyramidal structure of the
Depression-era banking system to achieve identification of credit supply effects. At the time, the large so-called money-center banks in New York or Chicago both provided credit locally and acted as correspondents to smaller banks around the country, meaning that they stood ready to accept small-bank deposits during periods of excess liquidity and to lend to their correspondents when they needed extra cash. Money-center banks, with the help of the Federal Reserve, were used to dealing with the cash inflows and outflows associated with the annual agricultural cycle. However, they were not well prepared to deal with cash outflows caused by banking panics in the interior of the country in the 1930s, when regional and rural correspondents tried to reinforce their liquidity buffers. The cash outflows from New York and Chicago banks arising from regional panics were presumably not closely related to economic developments in New York or Chicago. Nevertheless, money center banks reacted to these liquidity drains and the associated increase in the risk of runs by reducing local business loans and shifting their portfolios toward more-liquid assets (Mitchener and Richardson 2013). Likewise, Calomiris and Wilson (2004) found that unexpected outflows of deposits at individual, publicly traded banks in New York City led the money-center banks to de-risk their portfolios, including by reducing local lending.
Several papers, following a strategy due to Rajan and Zingales (1998), identify the effects of Depression-era bank distress on lending and economic activity by considering differences among firms or industries in the extent of their historical dependence on external finance, as opposed to cash reserves built up internally. These papers differ in their details but agree that bank suspensions and the risk-averse behavior of surviving banks had substantial effects on lending, output, and employment (Lee and Mezzanotti 2017; Mladjan 2019; Gorton, Laarits, and Muir 2023). Using a 1934 Federal Reserve survey of both banks and Chambers of Commerce in various localities, as well as county-level measures of bank distress, Carlson and Rose (2015) found that bank failures were a dominant source of business concerns about credit availability, with funding constraints and deposit outflows also playing a role.
Although government intervention stabilised the US banking system in 1933, bank lending remained stagnant, creating a drag on the recovery. In particular, local bank-borrower relationships were rebuilt only gradually after 1933, slowing the recovery of lending and economic activity in rural and isolated areas (Cohen, Hachem, and Richardson 2021). The lengthy legal process associated with rehabilitating and opening suspended banks also helps explain the slow recovery of bank lending (Anari, Kolari, and Mason 2005).