The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
DO "VULNERABLE" ECONOMIES NEED DEPOSIT INSURANCE?: LESSONS FROM THE U.S. AGRICULTURAL BOOM AND BUST OF 1920s Charles W. Calomiris Working Paper Series Issues in Financial Regulation Research Department Federal Reserve Bank of Chicago October, 1989 (WP-89-18) Do "Vulnerable" Economies Need Deposit Insurance?: Lessons from the U.S. Agricultural Boom and Bust of the 1920s Charles W. Calomiris Department of Economics Northwestern University and Research Department Federal Reserve Bank of Chicago October 1989 This paper was prepared for the Sequoia Institute Conference on Financial Risk and Financial Regulation in Commodity-Exporting Economies. The author thanks Lee Alston, Jeremy Atack, Herbert Baer, Michael Bordo, Philip Brock, Douglas Evanoff, Kenneth Kuttner, Larry Neal, Hugh Rockoff, Larry Schweikart, Lawrence J. White, Eugene White, and seminar participants at Northwestern, Rutgers, and the University of Illinois--Urbana for helpful comments, and Eric Klusman for excellent technical assistance. I . INTRODUCTION To justify the substantial protection that governments offer banks, regulators frequently refer to banks' unique position as the channel through which payments clear, and through which essential short-term commercial and workingcapital credit are provided to parties whose access to other sources of funds is limited. Banks are "special" because all other industries rely on them to maintain their operations and execute their transactions in a timely, convenient way. In particular, whether one defines the payments system narrowly to include only check clearing, or more broadly to include lines of short-term credit to "information-intensive" borrowers, it would be hard to conceive of a payments system without banks. Thus shocks that threaten the viability of banks, encourage financial disintermediation, and cause disruptive bank failures or suspensions of deposit convertibility can be very costly to society, and these costs may be far greater than the reduced profits, or bankruptcy costs, incurred by banks. Recent research on the peculiar severity of the Great Depression and the vulnerability of agricultural producers to banking disturbances in the 1980s have received particular emphasis as examples of socially costly financial disruption (see Bernanke, 1983; and Calomiris, Hubbard, and Stock, 1986). The externalities generated by banks' special role as check clearing agents and commercial credit suppliers, therefore, may provide a rationale for regulation of banks. Notwithstanding this presumed vulnerability of the payments system and the essential role of banks, critics of current government interventions into banking have argued that the government has gone too far in guaranteeing bank liabilities and consequently has promoted an unacceptable degree of socially undesirable risk taking by banks. For example, from a theoretical perspective, Calomiris and Kahn (1988) argue that demandable-debt banking, and the first-come first-served rule of bank repayment to depositors, were part of an incentive-compatible equilibrium in which informed depositors (often other banks) were rewarded for investing resources in monitoring banks. Insurance removes the reward, and hence the incentive, that encourages such monitoring. Insured banks' incentives to undertake excessively risky projects are magnified by shocks that reduce bank capital. Such shocks increase the banker's potential gain from pursuing long shots (see Calomiris and Kahn, 1989) by increasing the implicit value of the put option inherent in deposit insurance (see also, Karaken and Wallace, 1978). Empirical evidence of excessive risk taking by insured financial intermediaries, especially in response to adverse shocks that reduce bank capital, has been provided by Kane (1988) and Calomiris (1989), among others. Furthermore, critics argue that regulators underestimate the extent to which the financial "safety net" could be provided with little or no government insurance of banks. example, private clearinghouses historically provided 2 For coinsurance among member banks that reduced the incentive for depositors to remove funds from banks during periods of financial uncertainty. Mutual regulation and monitoring ensured that members would not free ride on the group protection (see Cannon, 1910; and Gorton, 1985). Coordination among banks, sometimes even across state lines, was enhanced in and among states that permitted branch banking -- in particular, in the antebellum American South (see Calomiris, 1989).1 With fewer, and better diversified, banks it was easier for banks to respond to crises as a group, again effectively coinsuring by continuing to "make markets" in other banks' deposits and notes. Similarly, in three unit-banking states of the antebellum North, statewide bank liability insurance plans modeled on private clearinghouse coinsurance arrangements managed to protect the payments system and limit (or eliminate) bank failures and suspensions of convertibility without encouraging excessive risk taking by members. These plans gave member banks authority to enact and enforce regulations, and provided the incentive for effective self regulation and monitoring by making member banks fully and mutually liable for the liabilities of any failing banks. These systems managed to maintain the smooth functioning of the payments system within and across states, and saw few, if any, bank failures relative to states that lacked an effective means of bank coordination. A review of the experiences of these antebellum bank insurance 3 success stories, and the very different experiences of other state bank insurance schemes, is provided in Calomiris (1989). The apparent lesson of historical bank clearinghouses, early Southern branch banking, and mutual-guarantee self regulating insurance plans under government sponsonship is that banking coalitions can act to coinsure effectively against many threats to the payments system. The successful operation of private clearinghouses in today's financial markets — for example, the CHIPS network, or the futures and options clearinghouses — indicate that these lessons can be applied successfully in the modern context, as well. Private coinsurance schemes, however, cannot offer unlimited protection against financial collapse in all circumstances. Private insurance is not effective in preventing disintermediation by depositors who question the ability of the coalition to guarantee the losses of its members. Once a shock becomes large enough to threaten the capital of the group of banks as a whole — a small subset of its members — rather than simply coinsurance ceases to be credible. Furthermore, the geographic range of privately coinsuring groups — and consequently the potential for coinsurance -- may be restricted by laws that limit branch banking, and thus impair the ability of bankers in different locations to communicate, monitor one another, and coordinate their behavior. Most financial crises in U.S. history began as small 4 disturbances, relative to aggregate bank capital, which were insurable, in principle, by mutual protection among banks. Reasonable fears of insolvency of a subset of banks, confusion as to which banks had suffered most from the shock, and the absence of a mechanism for mutual protection at the state or national level, however, provided incentives to depositors, who were unable to determine the precise incidence of the disturbance, to withdraw their funds (see Gorton, 1989; and Calomiris, 1989). Lacking effective means to coinsure against such disturbances, the thousands of independent and geographically distant unit banks sometimes were forced to suspend convertibility as a defensive reaction during such economy-wide bank runs. Suspensions of convertibility limited depositors' and noteholders' liquidity, and reduced the desirability of placing funds in banks, reducing the supply of loans and forcing banks to adopt more conservative lending practices than under normal circumstances. Other banks failed due to losses incurred as the result of "fire sales" of assets to meet withdrawal requests (see Calomiris and Schweikart, 1988; and Calomiris, 1989). The relative success of statewide systems of branching banks, or mutual-liability banks, in meeting such crises as the Panic of 1857 suggests that, for an economy as diverse nationally as the United States, a combination of full nationwide branch banking and government-sponsored, privatelymanaged mutual-liability insurance may be sufficient to prevent 5 large sector- or region-specific shocks to bank capital from becoming a threat to aggregate bank capital, and therefore, the payments system.^ One could argue, however, that this approach might not be sufficient for economies with intrinsic vulnerability to large sector-specific shocks that threaten aggregate bank solvency. In the case of the United States, full interstate branching virtually could eliminate the risk to banks from regionally concentrated shocks to the terms-of-trade, which have proved particularly important for the agricultural and oil-producing sectors (see Alston, 1983; Stock, 1985; and Calomiris, Hubbard, and Stock, 1986). In smaller countries with less diversified economies, however, the risk from terms-of-trade shocks is large (see Brock, 1988), but the potential for reducing payments system risk through diversification is more limited because national sovereignty limits the development of full international branch banking. The limitations due to national sovereignty can be viewed as an example of the "timeinconsistency" problem. Banks chartered in country X may decide to leave their local branches in country Y stranded rather than pay for their losses during bad times, and there may be no way for country Y to force them to do so. Furthermore, governments may find it optimal to limit the repatriation of bank profits to support bank branches in other countries. The central question I will address in this paper is 6 whether the governments of such intrinsically risky economies stand ready to bail out banks in the event of a large adverse shock to the country's economic base? The question may be posed in two parts: How great are the advantages of a government's insuring the payments system (whether narrowly or broadly defined) from the strains of such shocks, rather than relying on a privately administered, mutual-guarantee system?; Are the social costs of excessive risk taking by banks, which result from the existence of bank deposit insurance, greater or less than the supposed benefits of insurance?^ The specific historical cases I will discuss are the experiences of agricultural areas of the United States in the 1920s — a period that witnessed a rapid, sharp terms-of-trade reduction for agricultural producers and an unprecedented rate of farm, business, and bank failures in the most affected regions. The 1920s provide a particularly useful context to investigate the role of different regulatory regimes in reducing or magnifying the effects of the shock on the viability of financial intermediaries. Interstate branch banking was not permitted, though some states allowed full, or limited, intra-state branching. Furthermore, some states had enacted deposit insurance prior to the crisis. Finally, the existence in each state of nationally chartered banks (under a common regulatory regime across states) provides a point of comparison across states of the magnitude of the shock to banks in each state, which can provide insight into the comparative 7 performance of the various state-chartered banking regimes, and the contribution of deposit insurance or branch banking in magnifying or lessening the impact on banks. In Calomiris (1989) I presented evidence from the 1920s of much higher initial growth, and subsequent failure rates, for four state-chartered, insured banking systems relative to national banks operating in the same state, which were prohibited by the U.S. Comptroller of the Currency from joining state deposit insurance funds. While this comparison was a useful first step, it is important to establish that the differences between insured state-chartered and uninsured national-chartered bank failure rates are not merely an artefact of different exposure to agricultural risk, due to different locational patterns (rural vs. urban) for state- and national-chartered banks, or more restrictive regulations on national-chartered banks — on real estate loans.^ in particular, stricter limitations Here I look at all eight of the insured systems, and compare the performance of national- and statechartered banks within and across states, taking account of differences in economic and regulatory environment (e.g., the existence of branching and deposit insurance), and using additional indicators of bank performance. No single historical example can provide a conclusive "answer" to the broad question of whether banking system bail outs are socially desirable in price-sensitive economies; but it is only through the accumulation of evidence from examples 8 of the costs of such crises, and the consequences of the decisions to provide insurance or allow branching, that policy makers will be able to make the difficult choices of bank regulatory policy in an informed manner. Section II provides cross-sectional evidence on price, income, and wealth movements, and indicators of financial distress experienced by the various states in the 1920s. Section III measures changes in the size, number, and portfolio structure of national- and state-chartered banks prior to and during the crisis. Section IV evaluates differences in the performance of the state-chartered banking systems in response to the crisis -- specifically, the rates of bank suspension and bank failure, the costs to depositors of failures, and the ability of the banking systems to recover from the crises, under different state regulatory regimes. Section V returns to the central question of whether deposit insurance is desirable for economies with intrinsic vulnerability to large income disturbances.I . II. THE POST-WORLD WAR I AGRICULTURAL CRISIS Typically, wars have been good times for farmers. World War I, like the Napoleonic Wars, and the Crimean War of the 1850s, witnessed a rapid expansion of agricultural income. As for previous wartime booms, however, the end of war brought with it a severe decline in the agricultural terms of trade. Declines in price and income became translated into declines in 9 farm land values, and farmers who had expanded operations with debt financing during the boom found their incomes slipping as their leverage ratios rose, often to levels that were unsustainable. The crisis was quite sector- and region-specific. Indeed, for most sectors (and consequently, the "non-agricultural" states) in the U.S. the 1920s were a "new age" of unprecedented stability and growth. In many states with a heavy reliance on agricultural earnings, however, the period was one of sustained declining income and financial collapse.^ Differences across the states in the degree of agricultural stress reflected different movements in earnings and wealth, as well as differences in farmers' financial vulnerability to those declines. Table 1 provides an index of real gross farm income and its components for 1910-1930. These figures show that the decline in agricultural income affected virtually all producers, though the timing and severity of decline varied across activities, with staple foodstuffs and textile raw materials suffering the worst percentage declines from 1919 to 1921. The uneven sectoral decline within agriculture produced different responses of income and wealth across states. Furthermore, Alston (1983) finds that similar farm wealth and earnings reductions produced far greater rates of farm foreclosure in some states than in others. 10 States that had expanded farm acreage, and farm leverage, during the wartime boom, suffered much higher rates of farm foreclosure, holding declines in wealth and income constant. Foreclosure rates for farms throughout the country during the 1920s and 1930s reached historic highs that have never been exceeded. For 1921-1940 foreclosure rates averaged more than five times the highest average levels for any other decade from 1913 to 1980. While the national average was high during the interwar period, the uneven incidence of foreclosure across states made matters far worse in some states. In Montana, from 1921 to 1923, 28 percent of farmers lost farms or property.^ From 1926 to 1930, foreclosures in Montana relative to owneroperated farms in the state averaged 52.2 per thousand, per year.^ Other Northern and Western states with extremely high foreclosure rates (per thousand owner-operated farms) for 19261930 include South Dakota (70.4), North Dakota (58.0), Oklahoma (50.0), Iowa (48.3), Arizona (42.7), and Colorado (42.4). In the South, South Carolina (68.0), Georgia (56.5), Mississippi (47.7), and Louisiana (40.1) had substantially higher rates of foreclosure than the other Southern states. Arkansas (39.7), Nebraska (38.4), Idaho (37.6), and Missouri (34.1) also experienced foreclosure rates substantially above the national average of 27.1 per thousand. Tables 2 and 3 provide a variety of measures of economic conditions for each of the 48 contiguous states during the period 1919-1930. Table 2 contains data on: gross farm income 11 growth from 1919 to 1921; growth in total net income from all sources received by the farm and non-farm populations from 1919 to 1921; the ratio of farmer to non-farmer income in 1920; the percentage difference in the value of crops sold from 1922 to 1925, and from 1925 to 1928; and the percentage change in the state-specific crop price index from 1919 to 1924. Table 3 reports: the change in the value of farm real estate per acre over the periods 1913-1920, 1920-1925, and 1925-1930; the ratio of mortgage debt to farm real estate value in 1920; the farm-to-total population ratio for 1920; and the farm foreclosure rate for 1926 to 1930. The choices of dates for each series in Tables 2 and 3 reflect data availability, as well as the peaks and troughs of the agricultural cycle. While the income, wealth, and price variables in Tables 2 and 3 are all expressed in nominal terms, rather than adjusted for aggregate price level movements, the GNP deflator was roughly constant for the years 1919-1929, with the exception of 1920, according to recent estimates, both by Romer (1989) , and Balke and Gordon (1989) . reproduced in Table 4. These estimates are Moreover, from the standpoint of the sustainability of farms and farmers' ability to repay debt to banks, it is nominal income and wealth that matter, since debt and debt service are set in nominal terms. Tables 2 and 3 indicate that the first decade of the agricultural crisis (1920-1930) can be divided usefully into three stages: the initial shock of 1920-1921, a period of 12 partial recovery from 1922 to 1924, and a subsequent period of decline. Because of differences in crop mix, supply-side variation, and financial vulnerability, the experiences of the various states differed substantially during these three sub periods, as the tables show. No single indicator provides an adequate index of the experience of a particular state during one of these sub periods. First, income and price indicators are extremely sensitive to the specific dates over which they are calculated. For example, 1924 was a relatively good year for Montana and North Dakota, and differs markedly from either 1923 or 1925 in this respect. Second, some income or price movements are perceived as transitory, while others are viewed as more permanent. Aside from the immediate cash-flow effects of such changes, the economic impact of income shocks on farmers' wealth and financial viability depended on perceptions of how permanent these disturbances were. Third, the impact of a wealth or income shock depends on the vulnerability (leverage) of farms — that is, how severe the shock was relative to previous expectations of future income. The highest foreclosure rates occurred in states with a relatively high ratio of farm mortgage debt to farm real estate value, as indicated in Table 3. These considerations suggest that, while the changes in prices and income provide measures of the sources of disturbances, changes in the value of farms and the farm foreclosure rate are more indicative of the likely 13 (anticipated) long-run changes in farm income associated with those shocks.® Finally, in evaluating the impact of agricultural shocks on statewide bank performance, the proportion of state income derived from farming, and the proportion of the labor force employed in farming, are of obvious importance. The geographical isolation of farming communities is also relevant, as it affects the abilities of merchants or bankers in these areas to diversify. The links (explored in section IV, below) between indicators of economic condition reported in Tables 2 and 3, and the threat to banks in a given state, therefore, are subtle. Ideally, in analysing these links, one would want to take account of the perceived permanence of different income shocks, the degree of financial leverage, the rapidity and cumulation of shocks, and the link between the degree of concentration of income in agriculture and the impact on banks from agricultural shocks. The focus of this study is on the role of regulatory regimes in limiting the incidence and costs of financial disruption in the face of a substantial challenge to the financial system. To evaluate the influences of the different state regulatory decisions in propagating adverse shocks, I compare the performance of banks in the states that were substantially affected by the agricultural depression, under different regulatory environments. 14 The sample of states defined to have been substantially affected, whose financial systems are analysed in greatest detail below, includes any state that experienced farm real estate value reduction (per acre) of greater than 20 percent from 1920 to 1930, or an average annual farm foreclosure rate of greater than 20 per thousand from 1926 to 1930. This sample includes states that suffered extreme depression, as well as those with more moderate commercial failure rates and bank failure experiences (discussed in section IV). The states in the sample are listed in Table 5, categorized according to their deposit insurance and branch banking laws. III. BANK MEMBERSHIP AND BALANCE SHEET PATTERNS ACROSS STATES Regulation's Influence on Membership. Location, and Risk National banks were governed by common regulations across different states, and thus their experience provides something of a state-specific benchmark against which to compare the behavior of state-chartered banks across states. Bank entry and asset growth, as well as financing and portfolio decisions of state-chartered banks, can be compared to one another, in absolute terms, as well as relative to the behavior of national banks in the respective states. Of course, national banks were not identical across states and faced different exposure to agricultural risk. While in every state national banks were larger on average, and located 15 more often in cities than their state-chartered counterparts, differences across states in the locational and size patterns of national and state banks were important. In some states, national or state banks operated more in urban locations and were larger on average than in others. Furthermore, national banks in cities that served as regional reserve centers for agricultural areas may have suffered asset depletion due to the impact of agricultural disturbances on correspondents. In what follows, I try as much as possible to control for these differences across states. The years prior to 1920 saw the establishment of deposit insurance systems in several states. Often it has been remarked that incentive problems due to insurance made insured banking systems grow at a "reckless" rate, and caused them to limit the growth of capital, and overextend themselves in the farm loan market (Thies and Gerlowski, 1989/ Calomiris, 1989; White, 1983; FDIC, 1956; ABA, 1933; Robb, 1921); however, no systematic quantitative comparisons of the behavior of the different state-chartered systems have been made before, to my knowledge. It is difficult to distinguish incorrect expectations of persisting prosperity from excess risk taking induced by deposit insurance unless one has a standard against which to measure the behavior of insured banks. When one controls for differences in economic environment, using uninsured state banking systems in other states, and national banks in the same 16 state, one has provided such standards of comparison. The dates for which the different state deposit-insurance systems came into and out of operation are given in Table 5. For three states (Kansas, Texas, and Washington), participation in state-run deposit insurance was voluntary. Numbers and deposits of participating and non-participating state banks in these states are given in Table 10. All state-run insured banking systems were in operation during the boom of 1918-1920, and with the exception of Washington, the state-operated insurance systems were the dominant component of the statechartered systems by 1920. In Texas, state banks that did not belong to the state-run system were privately insured, as required by regulation; while in Washington and Kansas statechartered banks could avoid insurance altogether. In describing the peculiar incentives of insured banks, one should distinguish between voluntary and involuntary state systems. Under voluntary insurance legislation, banks could retain state charters without joining the insured system. Since national charters were a costly means for many banks to avoid the insurance fund, voluntary state insurance was an important additional option. The laws governing withdrawal from a state's insurance plan were extremely important as well. In two of the three voluntary systems (Washington and Kansas), banks opting out of state-run insurance could avoid any form of insurance. These two systems also limited the effectiveness of insurance — 17 and thereby reduced risk subsidization among banks — by allowing member banks to leave the insurance system at any time. Washington's system went further, and provided essentially no protection for large losses, because it allowed banks to opt out at any time without even retaining liability for past losses. In Washington, low initial insurance premia and the ability to leave the voluntary systems seem to have encouraged many banks that were not aggregious risk takers to join, only to opt out once troubles began. In Texas, voluntary withdrawal was not permitted until the insurance law was amended in 1925. Of course, banks could also opt out of any of the compulsory or voluntary state systems by securing a national charter, although this would have entailed costs to banks that relied on activities prohibited by national law, or those with insufficient capital. In Texas, even though all state-chartered banks were required to have some form of insurance, the privately insured banks were unlikely to have had the same opportunities to take advantage of insurance through excessive risk taking. While there is much evidence that supervision and regulation were lax in the state-run plans, historical examples of privately run insurance (see Calomiris, 1989) indicate that excessive risk taking was not a problem because of strong incentives by insurers to provide effective regulation and supervision. Thus Texas' state-chartered banks that chose private rather than state-run insurance are likely to have assumed risks comparable 18 to uninsured banks in other states. Both compulsory and voluntary insurance during this period differed from current U.S. federal deposit insurance in several important respects. Typically, interest rates on insured deposits were restricted by law (except in Nebraska), and capital requirements were much higher than today (typically, 10 percent of deposits for insured banks). While these interest rate ceilings were sometimes hard to enforce because of outright fraud, or the use of discounts as an alternative to interest (see Cooke, 1910), they constrained the availability of funds somewhat, in contrast to FDIC and FSLIC regulations that allow risk-taking member to attract funds by offering unusually high-interest insured CDs. Furthermore, as in virtually all state systems, and the national banking system, stockholders had extended liability equal to the amount of capital in the bank. This was not equivalent to a doubling of the capital stock because collections from assessments on the stockholders of failed banks averaged less than 50 percent of assessments, for all state banking systems from 1921 to 1930.® Finally, the state systems were not insured by the state treasuries, but rather by member banks as a group, through an insurance fund to which banks contributed annual assessments. These assessments had upper bounds annually, which meant that solvent banks' liability was limited. Furthermore, by leaving the system, or forcing repeal of the insurance statute by threatening to do so, solvent banks who belonged to the insured 19 systems in the 1920s were able to avoid much of the liability to depositors of failed banks (more on this below). All these considerations imply that the effective protection of depositors, and the potential for excessive risk taking would have been less under the historical insured systems than under current federal deposit insurance. Thus evidence on incentive problems in these plans provides an a fortiori case for potential excessive risk taking under government-guaranteed insurance of the kind available in the U.S. currently. Evidence on the Effects of Deposit Insurance Tables 6 through 10 present measures of state banking system averages and aggregates, disaggregated by type of bank charter and by state, for the 32 "agricultural-crisis" states for various dates. The indicators include: the number of banks, the proportion located in towns or cities of 2500 or more, average total assets per bank, aggregate total asset growth, and the ratio of capital to assets. As the data for the various state- and national-chartered systems show, not all types of banks were equally likely to join one or another system. Larger minimum capital requirements and more restrictive portfolio regulations for national banks meant that small banks, particularly those that wished to specialize in agricultural credit backed by real estate, would be attracted to the state-chartered systems. 20 Table 7 reports the proportion of banks in each system that were located in towns and cities of greater than 2500 inhabitants, and the average size of banks in each system. While there is considerable variation among states, national banks were always larger on average, and always had a higher proportion of banks located in cities. Historical accounts and economic theory lead one to expect that deposit insurance for state-chartered banks reinforced this propensity for small, rural banks to belong to the state system, and for large, urban banks to join the national system, because the potential benefits of deposit insurance were greater for small rural banks. White (1983, pp. 198-200) finds that the support for deposit insurance regulation was greatest among small bankers operating in agricultural areas in unit banking states with low minimum capital requirements. For large, urban state banks (which opposed deposit insurance legislation) deposit insurance was seen as a burden, a legislated subsidy from large to small banks operating in the periphery.I® Insured deposits typically had interest rate ceilings that kept them from being as competitive in the market for large, sophisticated depositors as in the market for deposits in rural areas.H Capital requirements in the insured systems (typically 10 percent of deposits) were more of an impediment to risk taking for large banks than for smaller banks operating in the periphery. A group of oil men, ranchers, or farmers could organize a small bank to finance 21 their expansion, while placing limited funds of their own at risk.^ Many of the large city banks found advantages to operating in a more disciplined environment, with stockholders and subordinated debtholders keeping watch over conflicts of interest between bank and banker. For urban banks, the expanding opportunities in trust activities and alternatives to standard demand-deposit banking as a means of finance were the wave of the future; for small rural banks, deposit-financed agricultural lending was the way to expand.^ exceptions. There were Some particularly unscrupulous large city banks chose to enter the insured systems, intending to use them as a means to create and exploit conflicts of interest, and finance speculative expansion on a scale that would not have been possible for a rural unit bank.-^ Bank membership and balance sheet patterns indicate that deposit insurance was an important force in determining who joined or left the various systems, and in influencing bank expansion and risk taking during the boom and bust. No single indicator in Tables 6 through 10 provides a "litmus test" of the importance of deposit insurance for adverse selection in bank membership and excessive risk taking, but a combination of factors apparent in the tables indicates that state systems that featured deposit insurance were a special class, during both the era of expansion and that of contraction. During the boom period of 1914-1920 the insured banks grew more rapidly than others. Sixteen state banking systems grew 22 by a factor of greater than 2.5 from 1914 to 1920, as shown in Table 11. Of these 16 systems, 7 were insured (one of the voluntary systems, Washington, is excluded from this list). The compulsory systems ranked first, fourth, fifth, eighth, and eleventh in asset growth over this period. Two of the voluntary-participation state systems (Kansas and Texas) ranked thirteenth and fifteenth. High growth, by itself, does not imply excessive risk taking. As Table 11 shows, high growth was not confined to insured systems, as the experiences of Wyoming, North Carolina, and Idaho demonstrate, in particular. Three factors, however, made the high growth rates of the insured systems unique: they accomplished high growth mainly through increases in the numbers of banks, rather than in assets per bank; growth seems to have been concentrated in relatively sparsely populated regions; and insured banks operated with low capital-to-asset ratios, typically reserved for systems of larger average size. Of the 8 banking systems that averaged less than $400,000 in total assets in 1920, 6 were insured banking systems, with the frontier states, Wyoming and New Mexico, accounting for the remaining two. The West as a region, therefore, was experiencing its era of extraordinary banking growth, comparable to the growth of New England banking in the early national period (1790-1830), or the South from 1820 to 1860.^^ But the insured banks differed in certain respects from the other high-growth Western states. In New Mexico, state-chartered banks operated in more 23 populous areas, on average (see Table 7), and the fragility inherent in such rapid growth and small size were offset, in part, by the unusually high capital-to-asset ratio of banks (12 percent) in 1920, as shown in Table 9. Wyoming's capital-to- asset ratio of 9 percent was higher than any of the insured banks of comparable size, as well. Thus if one uses the combined standard of high growth, small bank size, and low capital-to-asset ratios, the insured banking systems appear especially vulnerable at the peak in 1920. Texas operated with a relatively high capital ratio, because its law required capital as a percentage of deposits of between 10 and 20 percent (depending on deposit size), while other insurance systems required 10 percent. It is important to verify that the high growth and unique vulnerability of the insured state systems relative to other state systems is attributable to different banking responses, rather than different fundamental economic conditions. To this end, additional comparisons of insured systems with other banking systems within and across states are useful. Specifically, I consider three standards of comparison: the relative growth of insured and uninsured state banks in states where insurance was optional; the growth of state-chartered banks across states, relative to the growth of national banks in the same state; and the growth of insured banking relative to uninsured state-chartered banking in adjoining states with similar "economic fundamentals." 24 For two of the three states with voluntary systems (Kansas and Texas), it is clear that the growth differences between national- and state-chartered banking from 1914 to 1920 (Table 8) were due to the disproportionate growth of state-run insured banking, as Table 10 reveals. These two voluntary systems grew rapidly during the boom period 1914-1920 relative to other state banking, both in number of banks and total deposits. Texas' system — The which did not allow voluntary withdrawal by member banks, and therefore, provided more anticipated insurance protection than the Washington or Kansas system -was the fastest growing of the three voluntary insurance systems, relative to national or uninsured banks in the state. In Texas, the deposits in banks of all types in the state grew by 271 percent from the end of 1914 to the end of 1919, while those in insured banks grew by 402 percent. The total number of banks in the state increased by 11 percent, while the number of insured banks increased 25 percent, from 1914 to 1920. In Kansas, total deposit growth was 131 percent for 1914-1919, while insured-banking deposit growth was 173 percent. In the third voluntary-insurance state, Washington, the state-chartered system as a whole grew slowly compared to the national system, and the insured system never accounted for more than 41 percent of state-chartered deposits (FDIC, 1956, p. 50). Several features of the Washington experience made it a special case. First, Washington's free-exit provision provided virtually no protection, and hence no encouragement 25 for excessive expansion. Second, its was the last insurance system to be established (in 1917), and there was less time for banks to join prior to 1920. Third, Washington's banking growth during this period was concentrated more in the large urban banks. Its national banking system was third among the sample of 32 states in average asset size of banks in 1920, and experienced above average growth in assets from 1914 to 1920. The lack of a rural/agricultural boom in Washington — farm land prices grew a modest 38 percent from 1914 to 1920 — further limited any perceived advantages to small rural banks of membership in the insured system. A second standard for comparing growth during the boom -one that controls for state-specific economic conditions -- is the relative growth of state and national systems across states with and without insurance. Typically, state banking systems grew faster than national systems, but in some states this difference is especially pronounced; in others it is actually reversed. A rough comparison is provided in Table 12 -- a four-byfour matrix, that arranges states according to the quartile growth rates of their national- and state-chartered banking systems for 1914-1920. Only two state-chartered systems ranked two or more quartiles higher in growth of assets than the quartile rank of their state's national banks: Mississippi and Nebraska. These were two of the five compulsory insurance states. 26 A more formal approach to comparing state-chartered banking growth to national bank growth, across states, is presented in Table 13. Using cross-sectional data for the sample of 32 states, I regress state-chartered bank asset growth for 1914-1920 on: national-chartered asset growth, the percentage rise in farm land value per acre, the ratio of farm population to total population, and dummy variables for the presence of insurance. In one version I separate the voluntary insurance states -- Kansas and Texas -- from the compulsory insurance systems. Washington is excluded from the set of insured states altogether. I also add a dummy variable (which interacts with the growth of national banking) for states that contained especially important "reserve centers". National asset growth is included as a measure of statespecific opportunities for expansion, holding regulation constant. The growth in the value of farm real estate is included to control for different expectations of long-run profitability from agricultural loans (which should have a disproportionate effect on state banks). The reserve center dummy is included to control for peculiarities in the growth of national-chartered banks due to interstate influences through correspondent relations. The regression results confirm that insurance was associated with very high relative rates of growth of statechartered banks, and that national banks in reserve-center states grew more than national banks elsewhere. 27 As predicted, the effect of compulsory insurance is stronger than that of voluntary insurance, because voluntary plans provide less cross-subsidization, and because (in Kansas) withdrawal was allowed for by law. Even in Kansas and Texas, however, the effects of insurance dummies were important (accounting for an additional 33 percent of asset growth from 1914 to 1920), although the few degrees of freedom, and consequent high coefficient standard errors, limit the power of hypothesis tests. Finally, comparisons among state banking systems in the same regions also support the conclusion that insured banking growth was unusually high, and that insured states were more vulnerable, during the boom. First, consider the states in the Western region adjoining the Western insured states. These include: Arkansas, Colorado, Idaho, Iowa, Missouri, Minnesota, Montana, New Mexico, and Wyoming. How do these states compare, in growth, bank size, and capitalization, to the insured states of Kansas, Nebraska, North Dakota, Oklahoma, South Dakota, and Texas? Data on the ratio of state-chartered bank assets in 1914 relative to 1920, average state-chartered bank size, and capitalization are reported in Table 14. A similar comparison is performed in Table 14 between Mississippi statistics and those of the uninsured states in the deep South in our sample: Alabama, Georgia, and South Carolina (Louisiana is excluded, because of the special role of New Orleans as a financial center.). 28 These data reveal that the nine uninsured state-chartered systems of adjoining Western states were larger on average, grew less, and had higher capital than their counterparts in the insured systems. On average, uninsured Western asset ratios (1914/1920) were 0.42, compared to 0.35 for the insured group. The average size of the uninsured group was $448,000 in total assets, while the insured banks averaged $334,000. The historic vulnerability of small banks explains why, ceteris paribus, their depositors required them to maintain higher than average ratios of capital to deposits.^ But in this sample, capital averaged 9 percent of assets for the uninsured group, and 8 percent for the insured. When Texas is excluded -- the insured state with a high legally mandated capital-to-deposit ratio that exceeded "market-determined" bank leverage in other states -- the difference becomes even greater. The comparison between Mississippi and its neighbors is similar. The asset ratio averaged 0.45 in the uninsured states, as compared to 0.34 in Mississippi. The ratio of capital to assets for the uninsured states was 9 percent, as compared to 6.6 percent in Mississippi. Average bank size in Mississippi was greater than that of the other states (664 as compared to 538), but this size difference is partly attributable to the much higher growth in assets in Mississippi, which tripled from 1914 to 1920, compared to the other states whose aggregate assets roughly doubled over the same period. Also, Mississippi's state-chartered banks 29 included older, relatively large branching banks (10 banks with 24 branches in 1920) that were allowed to continue operating, even though new branching was not allowed. Finally, as discussed in section IV, many of Mississippi's rural banks had failed during the boll weevil crisis of 1912-1913, and the state banking regulators were notoriously restrictive in granting entry by new banks. Insured Banking: From Boom to Crisis Having established, using several standards of comparison, that deposit insurance was associated with high growth and greater bank vulnerability (small size and low capital) during the boom, I now turn to evaluate the effects of insurance on the membership and balance sheet responses of state banking systems to the crisis. As several authors (ABA, 1933; FDIC, 1956; White, 1983; and Calomiris, 1989) have documented, the insurance plans did not provide effective protection to the states' payments systems or to bank depositors. Reimbursements to depositors were neither timely nor complete, and exit from the insured systems relieved solvent banks of the responsibility of covering insolvent banks' liabilities. Here I quantify the role of deposit insurance, and the vulnerability it entailed, in preventing state-chartered banking systems in states with insurance from responding to the crisis as well as other state systems. As one would expect, failures and assessments rose during 30 the collapse, and there was widespread defection of relatively healthy insured banks to alternative systems, as shown in Table 15. In all cases, there was a net transfer of banks from the insured state systems to the national system. Table 15 reports data on changes of charter across the two systems within each state from 1921 to 1930. From 1921 to 1930 the 48 contiguous states as a group experienced a total of 346 net conversions from state- to national-chartered banking. All 8 states with deposit insurance had positive net conversions over this period, and as a group they accounted for 278 of the 346 net conversions -- an average of 43 per state. At the same time, neighboring states in the West experienced virtually no net conversions to national banking. As a group, Montana, Iowa, Colorado, Idaho, Wyoming, Oregon, Arizona, Arkansas, Minnesota, and New Mexico had only 5 net conversions in all. Furthermore, few other states witnessed a substantial number of conversions to national banking. Only 8 states other than those with insurance plans had net conversions of greater than 5: Alabama, California, Illinois, Minnesota, Missouri, Oregon, Virginia, and Wisconsin. This group of states -- unlike the insured states -- did not suffer a collapse of state banking during this period. Alabama's state system showed essentially flat total assets over the period; California's, Wisconsin's, and Virginia's state systems experienced substantial intensive growth; in Illinois only a small percentage of banks converted, and total state banking assets grew substantially relative to 31 that of national banks; in Minnesota, the percentage that converted was also small; and Oregon was not an "agricultural crisis" state. Thus the insured state-chartered systems were virtually the only cases of national banking gaining at the expense of state banking in response to the agricultural crisis. In the states with voluntary state-run insurance participation, there was widespread movement to the other state systems, as well. from 1924 to 1926. In Kansas and Texas banks switched en masse, The timing of the demise of the Texas system reflects the fact that withdrawal was not allowed by law until 1925. In Kansas, the failure of the largest bank in the insured system in 1923, and a court ruling in 1926 that absolved withdrawing banks from liabilities for prior bank failures (above the amount of securities already deposited in the state fund), explain the timing of withdrawal. In Washington, one bank failure -- again, that of the largest bank in the state, which accounted for one-fifth of insured deposits — prompted all other insured banks to leave the system.^-® Thus while deposit insurance produced abnormally high growth during the boom, it caused abnormally low statechartered growth during the crisis. Table 16 reports regression results analogous to those of Table 13, but for the periods 1920-1926 and 1920-1930. The average annual rate of business failure from 1921 to either 1925 or 1929, relative to the average rate for the four years prior to 1921, is included 32 in the regressions to capture better the financial distress banks faced in each state. The regressions are run for two sub-periods because prior to 1930 Nebraska's insurance fund chose not to close many insolvent banks, thus contaminating the measure of solvent bank deposits. For this reason I exclude Nebraska from the dummy for insured states in the 1920-1926 regression. Results for the 1920-1930 regression are reported with and without including Nebraska in the dummy banking variable. The regressions show that the presence of insurance was associated with lower growth during the decline. Growth for the insured systems from 1920 to 1926 was 27 percent lower (as a fraction of the 1920 level) than in uninsured state systems. Not surprisingly, the difference in growth is lessened if one chooses a longer period (1920-1930) to gauge recovery from the crisis. When the postponed collapse of Nebraska is included for the 1920-1930 sample, there is an increase from a 19 percent to a 24 percent slowdown in decadal growth. variables generally have the predicted signs — Other failure rates and farm population concentration are associated with lower growth, and controlling for omitted variables by including national bank growth is important for the 1920-1926 period. Land value changes add little after taking these other control variables into account. Insured banks were not the only ones that saw a decline in growth during the crisis. Many states experienced a 33 substantial decline as agricultural earnings fell and bankruptcies rose. Interestingly however, there was substantial variation in the rate of recovery from the crisis across states, and across banks within states. As Table 8 shows, the state systems of Arizona, Idaho, and Wyoming saw high growth rates in the period 1927-1929 that essentially reversed the negative growth of the previous seven years. In all three cases, state-chartered banking growth for the period 1920-1929 exceeded the growth of national-chartered banks in those states . ^ Furthermore, within states the growth of state-chartered banks was not identical across banks. In almost all cases (with the exception of the insured systems) the average size of state-chartered banks increased from 1920 to 1929. In some extreme cases assets per bank doubled (Arizona, Illinois, Michigan, Ohio, and Wyoming). Interestingly, aggregate recovery of banking asset levels and increases in average bank size are positively related during this period (as exemplified by the experiences of Arizona and Wyoming, in particular). This suggests that as small, rural banks failed they were not likely to be replaced by similar institutions, but rather by larger banks. (1985) White finds that the merger wave in banking from 1919 to 1933 was partly the result of the desire to move away from a system of small, fragile unit banks. While several factors could account for variations across states in the extent of 34 consolidation (e.g., a reduction in the perceived desirability of rural farm loans and a change in emphasis toward industry located in cities and towns, in which larger unit banks operate), in part this variation may reflect different regulations across states, in particular laws governing branch banking In states that allowed branch banking, it should have been easier to acquire small rural banks that failed or replace them with new branches, because the cost (including risk) of establishing branches was lower than that of opening a bank.^l Chapman (1934, chapter XI) provides evidence of relatively high growth of branching banks for the nation as a whole during the 1920s. A thorough analysis of the relative growth of branching banks and unit banks during the 1920s in states that permitted branching would require a panel data study at the level of individual banks, which is beyond the scope of this paper. Instead, using available data, I examine the growth in the number of branching banks and their branches at the state level and link it to total banking growth, in number and total assets. I also report results using disaggregated data (at the individual-bank level) for a few states where this is feasible. Branch Banking and Banking System Recovery Table 17 summarizes the data on the growth of the number of total banks, branching banks, and branches of national and state banks for 1924 (the earliest available data) and 1928 35 (the last disaggregated data available for the 1920s), categorized according to state banking laws on branching, for the 32 "agricultural-crisis" states. The state-bank regulatory regimes are divided into four groups: full free entry for branching banks statewide, full free entry with locational limitations on branches, limited (or zero) entry of new branching banks but continuation of existing branching, and a final category in which prohibitions kept branching from developing. Wyoming is an exceptional case in that it allowed statewide branch banking, but no banks opened branches during the 1920s. Apparently, in Wyoming's mainly livestock-producing hinterlands opportunities for diversification through branching were limited (as contrasted to Arizona's economy in which cotton, livestock, and copper all offered substantial opportunities in rural areas), and profitable opportunities for bank entry were concentrated in the major cities. As Woods (1985, p. 101) points out, banking outside of major cities was confined mainly to very small banks organized in rural areas to provide financing for expansion to local groups of insider entrepreneurs. Of Wyoming's 113 non-national banks in 1920, 31 had a deposit base of under $100,000 (see Woods, 1985, p. 96) . National banks often were permitted to maintain branches in existence at the time of their conversion to national charters, which explains why national banks operated branches in some states. In no states prior to 1927, however, did 36 national banks maintain significant branching systems. Upon passage of the McFadden Act (February 1927), limited national bank branching was allowed in states that permitted branch banking. Even under the McFadden Act national banks were still restricted to establishing branches within the "city, town or village" of their main office. Thus there is little cross- sectional variation across states in national bank branching during the 1920s. Several interesting patterns are visible in Table 17. Because of switching between national and state charters, it is best to focus on branching within the states for national and state banks as a whole. Of the 18 states that allowed branches to exist, only 3 saw a reduction in the number of total branch banking facilities from 1924 to 1928. These reductions all occurred in states that prohibited the establishment of new branches, but allowed existing branches to be maintained (Georgia, Minnesota, and Washington). In all three cases, the reduction is accounted for by the departure (failure or closing) of a single b a n k . ^ 2 in all the other states that allowed branching to continue, but prohibited the establishment of new branches, the number of branching facilities remained the same. In states that allowed new branching, branching facilities uniformly increased at a rapid rate, often as the total number of banking facilities declined, and branching thus came to comprise a much larger fraction of total banking facilities. 37 Moreover, the recovery of total bank asset levels was higher for systems that permitted growth in branch banking. Arizona, Kentucky, Louisiana, Michigan, North Carolina, Ohio, Tennessee, and Virginia all saw relatively high rates of asset recovery in the late 1920s relative to other states. These were also the states that experienced the largest increases in the average size of banks. South Carolina was the only exception to the rule, with negative asset growth in both banking systems during this period. Clearly, this exception "proves the rule," as South Carolina witnessed a more than doubling of its branch banking facilities from 1924 to 1929, even though the combined growth of unit and branching banks was negative. More formally, in Table 18, I regress bank asset growth from 1920 to 1926 and 1926 to 1930 on the same regressors used in Table 16, with the addition of branching dummies for cityrestricted and out-of-city branching. Out-of-city branching includes the statewide branching systems, and Ohio, that allowed limited out-of-city branching. I also report regressions using the change in average bank size as the dependent variable. While the few degrees of freedom in the regressions recommends a cautious interpretation of the results, the branching indicator variables were both relatively large and statistically significant. Indeed, branching indicators have a larger, more significant, and more persistent effect on total asset growth than deposit-insurance 38 indicators in the regression. These results indicate that, from the standpoint of long-run banking recovery, the distinction between unit and branch banking was more important than the distinction between insured and uninsured banking. Deposit insurance mainly caused a retreat from the statechartered systems until the insurance fund was dissolved, after which the state systems gradually recovered as well as other unit banking systems. In contrast, the effect of branching on banking growth and average bank size increase with time. These comparisons actually understate the difference in growth between branching banks and independent unit banks, because many unit banks operated as members of bank "chains." The Federal Reserve, which collected data on "chain" banks, distinguished chains from other banking conglomerates. Chains were defined as groups of corporately independent banks "under centralized control. "^3 As was recognized at the time, chains sometimes served as a "second best" substitute for branches in states where branching was prohibited. While banks in chains were separate corporate entities, they imitated to a lesser degree some of the advantageous features of branch banks. First, chains of banks could reduce individual bank risk by coordinating their response to crises and coinsuring as a group. Second, chains pooled resources and staffs to reduce overhead expenses and improve account management procedures (see Chapman, 1934, pp. 322-363). The potential for chains to allow member banks to diversify seems to have been more 39 limited, as the high failure rates of chains relative to branching banks indicates.^4 As Table 19 shows, the freedom to branch was inversely related to the prevalence of chain banking. Table 19 reports the number and proportion of chain banks in the state- and national-chartered systems for our sample of 32 agriculturalcrisis states. States with branching restrictions saw much higher incidence of chain banking, and that incidence increased with the extent of the branching prohibition. Summary of Findings The evidence on the aggregate growth, average size, and membership patterns of banks during the 1920s indicates that state banking systems can be usefully grouped into three categories: states where deposit insurance made the system more fragile, magnified the expansion in response to the agricultural boom, and worsened the contraction during the bust; other unit banking states with less drastic swings in aggregate growth; and branch-banking systems (restricted, or statewide) that managed to respond most successfully to the challenges brought by the declining terms of trade in agriculture. IV. BANK FAILURE COSTS AND THE ROLE OF REGULATION Aggregate data on numbers of banks and their assets over time do not distinguish voluntary exits by banks from bank 40 failures. In particular, it is conceivable that the decline in insured banking was primarily the result of voluntary exit in response to rising assessments once a few banks failed, in conjunction with laws that permitted banks to switch charters. If this were the case, skeptics of the failings of the insured systems might argue that the prohibition of voluntary exit would have been sufficient to make the systems viable. Evidence on bank failures, and their costs, therefore, provides a complement to the results reported in section III. A study of bank failure propensities permits one to distinguish exits from failures, and supplies further direct evidence on the extent of risk taking during the boom across different regulatory regimes. ABA (1933), Calomiris (1989), and Thies and Gerlowski (1989) provide evidence that insured banks were more likely to fail than national banks in the same state, uninsured state banks in the same state, and uninsured state banks in other states. While the within-state comparisons made by these authors of the failure propensities of insured-state and uninsured-state banks in Kansas (summarized below) is compelling, the other evidence is less so. Differences in states' product specialization, and differences across states in the relative agricultural risk exposure of national and state banks (due to other regulatory differences between national and state banks) must be controlled for if one wants to isolate the role of deposit insurance regulation in 41 promoting risk taking.^5 Furthermore, the definitions of "bank failure" may differ across these studies in ways that are not always clear. In analysing bank failures, I restrict attention to involuntary liquidations of banks. Sometimes banks suspended operations, and reopened; other times banks were acquired by other institutions; and, finally, banks could choose to close while still solvent. The withdrawal of a bank, therefore, may reflect very different events with different social costs. Suspensions, consolidations, and voluntary closings may have had social costs as well -- consolidations and closings may have reduced the supply of banking services in some areas, and suspensions were disruptive to the payments system. I focus on failures because they offer a clearer index of the costs of the crisis across states — depositor losses — forced permanent departure of banks and and provide a clearer measure of the risk taking of banks, since closings, acquisitions, and suspensions often had explanations other than bank insolvency.^® I also focus on average failure rates for several years, rather than perform a year-by-year comparison across states. Differences in state closure rules (in particular, the long delay in closing insolvent banks by the Nebraska Guaranty Fund)^7 argue for this approach. I examine three dimensions of the failure "performance" of banking systems: the rate of bank failure, the severity of bank failure -- measured as the ratio of claims on failed banks to 42 their remaining resources (excluding payments by insurance funds) -- and the efficiency of the bank-liquidation process, with emphasis on the roles of deposit insurance and branching regulations. Bank Failure Rates Table 20 provides data on average annual bank failure rates, by state and type of banking system, for various sub periods from 1917 to 1929, for the sample of 32 agriculturalcrisis states.2® These data echo the substantial variation in economic fundamentals and banking system responses across states and types of banks that were visible in earlier tables. Clearly, the cotton belt and the grain-producing states suffered disproportionately during the 1920s. Table 20 shows the pitfalls of using the difference between state and national bank failure rates for a given state (as in Calomiris, 1989) as a measure of the role of deposit insurance. While it is true that the difference between state and national bank annual failure rates for 1921-1929 are greater for insured states than for uninsured states on average (1.4 percent for compulsoryinsurance states, as compared to 1.0 percent for states without deposit insurance), this difference is positive for most states, presumably because of their smaller size and more liberal real-estate lending regulations. Similarly, state comparisons across state-chartered systems reveal several cases where uninsured systems fared 43 worse than insured. The difference in annual failure rates between uninsured (2.26 percent) and insured (2.92 percent) state-chartered banking systems for 1921-1929 on average is 0.68 percent, but by varying the definition of region -- a control used in Thies and Gerlowski (1989) -- one could easily conclude from such simple comparisons that insured statechartered banks had lower failure experiences than uninsured state-chartered banks. For example, one could define Texas and Oklahoma to be in the same region as Arizona and New Mexico or define Mississippi to be in the same region as Alabama, Georgia, Louisiana, and South Carolina. Regional distinctions, of course, are intended as rough proxies for economic environments under which banking systems operate. Thus, rather than experiment with different definitions of economic regions, I include measures of economic environment directly in weighted-least-squares regressions to capture the marginal effects of deposit insurance on bankfailure propensities. i refrain from reporting these results because I found that, depending on the precise mix of control variables one uses, the measured impact of deposit insurance (and the control variables) varied greatly and were typically positive and insignificant. In other words, given the few degrees of freedom available, regression results seem unable to deliver much information on the contribution of deposit insurance to bank failure propensities. The only robust findings from this analysis were the strong positive 44 association between commercial failure rates and bank failure rates, and the strong negative relation between average bank size and bank failure rates. Perhaps the best evidence of excess failure rates for insured banks remains the simple comparison of failure propensities of insured and uninsured state-chartered banks operating in Kansas.^1 Kansas provides a unique "controlled experiment" because it was the only state with a large number both of insured and uninsured state-chartered banks. The annual failure rate for insured banks in Kansas from 1921 to 1924 (prior to the mass conversions of banks to uninsured charters) is 1.90, compared to an annual failure rate of 0.67 percent for uninsured banks. Bank Failure Severity for Insured and Uninsured Systems It would be a mistake to place too much emphasis on bank failure rates as indicators of the costliness of financial crises. Bank failures are discrete events; particularly severe financial crises force many banks to cross the threshold of failure. For this reason, bank system performance may be better gauged by the overall losses of depositors, rather than the propensity to fail, which may show relatively little variation. Data exist with which to perform cross-state, cross-system comparisons of insolvent banks' asset shortfalls in the 1920s and thereby measure the average severity of bank failures across states. Complete data for insured banking systems are 45 provided in FDIC (1956), but data for the rest of U.S. banking systems are only available for banks whose liquidations were completed by 1930 (see Data Appendix). As Table 21 shows, for some state-chartered systems only a small percentage of liquidations that occurred during the 1920s were processed in time to be included in this sample. The ratio of repayments to total unsecured deposit claims from the limited sample in each state is likely to be a biased indicator of the total sample; for example, banks with higher losses might take longer to liquidate. Notwithstanding this problem, there is little room for doubt that insolvent insured banks suffered worse asset depreciation in the 1920s than state-chartered banks in other states. The rates of shortfall for insured state banks, were among the highest in the country, as shown in Table 21. Regional comparisons are particularly telling. Consider the difference between North and South Dakota's low ratios of repayments from assets to total claims (17.2 and 24.1 percent, respectively) and their neighbors' ratios: Montana (51.9), Idaho (47.4), Wyoming (53.7), Colorado (68.1), and Minnesota (48.2) . A comparison of insured banking in Nebraska (35.4) with Iowa (53.6), Missouri (52.6), Colorado, and Wyoming is similarly revealing. Kansas, Oklahoma, Texas, and Mississippi showed ratios more similar to the average experience of their neighbors. Significantly, two of these were voluntary-insurance states, 46 and Texas' exceptionally high required capital ratio may have played a role here, as well. Oklahoma's compulsory insurance system lasted only until 1923, and thus should have had a relatively small influence on failure propensity for the 1920s as a whole. As current critics of deposit insurance emphasize, much of the losses that occur in an insured system reflect responses by banks to adverse shocks that reduce bank capital and magnify the incentives for risk taking (see Kane, 1988; and Calomiris and Kahn, 1989). By closing its system early in the 1920s Oklahoma may have avoided this magnification of risk taking. Mississippi had the lowest rate of asset shortfall of the five compulsory-insurance states, as well as the lowest rate of bank failure for that group by far for the period 1921-1929. Mississippi's special experience may reflect, in part, the circumstances of its creation. The Mississippi deposit insurance law was passed in response to the state banking crisis of 1912-1913, induced by the destruction of cotton crops in those years by the boll weevil. The relatively low failure rate and degree of asset shortfall in Mississippi during the 1920s may indicate simply that many of the most vulnerable banks in that state had collapsed prior to the period of deposit insurance coverage, leaving surviving banks that on average were less likely to use deposit insurance protection to promote high-risk agricultural expansion. That is, larger, more urban banks were more likely to survive the attack of the 47 boll weevil. Entry by new banks seeking to take advantage of deposit insurance was notoriously difficult in Mississippi, as well, due to the strict chartering standards set by the state's regulators.^2 Thus Mississippi seems to have avoided the failure rates of the other compulsory systems mainly because its insurance system was enacted after a major agricultural depression, and because its regulators prevented the entry of small rural unit banks that were so common in the other insured states. This view is consistent with the relatively large average size of banks in 1918 and 1920 in Mississippi, relative to its neighbors, or relative to other insured banking systems (see Table 7). Inefficient Bank Liquidation Procedures in Insured States A final interesting difference between insured and uninsured banking was the efficiency of the liquidation procedures. Delays in winding up the operations of banks impose costs of illiquidity and forgone interest on depositors, apart from the ultimate larger losses due to asset shortfalls. Delays in closing banks, or in final liquidation of closed banks, also may afford insolvent bankers greater opportunities for risk taking or fraudulent behavior. On average, for the United States as a whole during the 1920s, it took 3 years and 11 months for state-bank liquidations to be completed, and for national banks it took 4 years and 2 months. In the five compulsory-insurance states 48 time delays for insured state banks were much longer than for state banks in other states, and much longer than for national banks in those states (see Table 22). In Nebraska, state- chartered banks that were liquidated before 1930 took an average of 6 years and 4 months to be liquidated, compared to 4 years and 9 months for national banks• In North and South Dakota state-bank liquidation delays averaged 6 years and 3 months, and 5 years and 7 months, respectively, compared to 4 years, and 4 years and 8 months, for national banks in the respective states. In Oklahoma, delays averaged 5 years, compared to 3 years and 8 months for national banks. Voluntary-insurance systems showed roughly comparable average delays to national banks operating within the same states, as did virtually all other agricultural-crisis states.^3 What can explain this phenomenon? Others have noted that deposit insurance systems redeemed the losses of depositors slowly and partially, owing in part to the limited resources of the funds (see ABA, 1933; and FDIC, 1956), but these data reveal that even the liquidation of failed banks was more protracted in the insured systems than otherwise. One explanation for the inordinate delays is political. Perhaps solvent banks and bank regulators sought to delay insolvent bank asset liquidation to limit the rate of increase of the obligations of the guarantee funds. The evidence of delayed closure of banks, especially in Nebraska, is consistent with this interpretation of delayed liquidation. 49 This is akin to the FSLIC's recent policy of delaying the closure of insolvent savings and loans, purportedly at the behest of members of Congress or Savings and Loan owners themselves.^4 state politicians of the 1920s may have acted similarly, and clearly solvent banks had a motive in encouraging delays, as this would have given them an opportunity to switch charters in anticipation of increasing obligations and assessments. Whether political motives or other factors explain delays in closures and liquidations must await further historical research into the process of bank liquidation in these states. The Unusual Survivability of Branching Banks In section III, I established that branch banking flourished in response to the crisis of the 1920s in states that allowed branching. While it is likely that the physical costs of entry of branches was lower than unit banks in many cases, another dimension of the advantage to branching -- one that was particularly noted by contemporaries in the 1920s -was that branching banks suffered lower risk of failure. References to this phenomenon were quite common, Cartinhour, 1931). (e.g., The Congressional hearings of 1930 on "Branch, Chain, and Group" provided data that allow some quantification of the lower risks of branch banking in the U.S. during the 1920s. From 1921 to 1929, only 37 branching banks operating 75 branches were liquidated. More than two-thirds of these banks operated a single branch, and no more than 6 of 50 them operated three or more branches.^5 In 1924 714 banks operated 2,293 branches. Thus only 112 of the 3,007 branch banking facilities in existence in the middle of the decade, or roughly 4 percent of branching facilities, failed over the entire decade. Of course, national comparisons can be misleading. California and other states that were relatively prosperous during this period account for a large percentage of branching facilities. In 1924, the 32 agricultural-crisis states contained 1,312 of the 3,007 branch banking facilities. But even if all branching failures had been concentrated in these states during the 1920s, the annual rate of branch-banking facility failure would be only 0.85 percent. This is a very low rate of failure compared to those of state systems on the whole. Only four state-chartered systems had lower failure rates than 0.85 — Nevada — Illinois, Michigan, Ohio, and and none of these states was among those most affected by the crisis; for example, they all had below-median farm foreclosure rates (see Table 3). In some cases, specific within-state comparisons are possible. In the states that prohibited new branching from 1924 to 1928, but allowed branching banks to continue to operate branching (Alabama, Arkansas, Indiana, Minnesota, Nebraska, Washington, and Wisconsin), branch-bank failures can be derived from the difference between the number of branches in operation in 1928 and the number in operation in 51 1924. in this sample of seven states, 28 branching banks operated 58 branches in 1924; and 26 branching banks operated 53 branches in 1928, for a remarkably low annual failure rate (for all facilities) of 0.02 percent. Finally, for other states, branch-bank failure experiences can be gleaned from data on bank "disappearances," using The Bankers Encyclopedia to trace the presence or absence of banks from 1920 to 1929. In all cases, a careful review of entries revealed whether disappearances were due to acquisitions or to closings. I traced the entries for the branching banks of three states over this period: Arizona, Mississippi, and South Carolina. These states were chosen because they experienced high rates of bank failure, had a small number of branching banks (making data collection easier), and because branching banks in these states operated branches mainly outside their home city. In Mississippi, all 24 branches in operation in 1920 were located outside their banks' home cities. The same was true of Arizona's 20 branches in operation in 1920. In South Carolina, 13 out of 15 branches operated outside the home city. These banks, therefore, provide a useful measure of the potential advantages of statewide branching during a crisis. Arizona permitted statewide branching throughout the period. In Arizona in 1920, 8 banks operated 20 branches. By 1929, 2 of these (each operating one branch) had been acquired by larger branching banks. One of the branching banks (operating one branch) failed. In the interim, three new 52 branching banks had entered, which explains the stability in the total number of branch banks (see Table 17). The average annual failure rate for total branching facilities, therefore, was 1.6 percent for 1921-1929, compared to 4.3 percent for the state-chartered banks as a whole. Mississippi had allowed branching outside home cities, but later prohibited branching, except for the establishment of limited agency facilities within home cities. Nevertheless, the existing statewide branches were permitted to continue operating. During the 1920s none of the 10 branching banks operating 24 branches failed, while the average annual failure rate for state-chartered banks as a whole was 1.4 percent. In South Carolina from 1920 to 1929, 4 out of 8 branching banks in operation in 1920 closed, but all of these were banks that operated a single branch, and 2 of the 4 operated branches within their home city. Thus of the 23 towns or cities in which branch banking facilities were located, 19 retained their branch-banking facilities. This is important since the lack of available banking facilities in thinly populated areas (where virtually all branches were located in Arizona, Mississippi, and South Carolina) increases transactions costs in those locations and can inhibit the flow of capital to worthy enterprises located there. The overall failure rate of existing branching facilities in South Carolina was 2.9 percent, compared to a rate of 4.9 percent for all statechartered banks. 53 Entry into branch banking was especially strong in South Carolina, and entrants apparently learned the importance of establishing multiple branches. Two new entrants -- The Peoples Bank of South Carolina and the South Carolina Savings Bank -- entered during the 1920s and established 18 and 9 branches, respectively, operating outside their home cities. The lessons of the high survival rates of branching banks during the crisis apparently were not lost on bankers. As Table 17 shows, and these examples confirm, in states where branching was allowed, it flourished and increasingly took the form of multi-branch banks, where that was allowed. Four of the eight states that had enacted deposit insurance legislation prior to the 1920s passed laws in the aftermath of the crises of the 1920s and 1930s allowing branching. By 1939, North Dakota had provided for limited branching, and Mississippi had reversed its previous prohibition on new branches to allow limited branching as well. South Dakota and Washington permitted full statewide branching. For the U.S. as a whole, 19 states allowed full branch banking, and 17 allowed limited branching, compared to 12 statewide and 6 limited branching systems in operation in 1924.37 Unfortunately, policymakers in many agricultural unit banking states did not change their approach to branch banking following the debacle of the 1920s and 1930s. Thus the 1980s saw a repeat of the same patterns of high unit bank failure rate in states hit by the agricultural crisis of 1980-1985. 54 As in the earlier period, branching banks weathered the storm far better than unit banks. In California, for example, despite relatively high farm-loan delinquency rates and loan chargeoffs, only one bank failed during the crisis (see Calomiris, Hubbard, and Stock, 1986, p. 469). V. LESSONS FOR POLICY IN PRICE-SENSITIVE LDCs It has been widely known that deposit insurance systems enacted in the 1920s failed ex post to offer sufficient and timely protection to depositors, or the payments system more generally. In this paper I have shown that deposit insurance created costs as well. It provided incentives for excess risk taking by banks and hampered banking system recovery from the agricultural crisis because of the costs to solvent banks of remaining in the insured banking system. The excessive growth of the halycon days of 1914-1920 was matched by the excessive failures of banks and declines in banking operations of insured states in response to the crisis. Voluntary insurance systems provided less coverage than insured systems. In the extreme case of Washington's free-exit policy, there was virtually no insurance protection. The positive aspect of the failed voluntary plans, however, was that the limits on depositor protection also limited the cross subsidization of risk among banks. This explains the differences between the observed responses to voluntary and compulsory insurance. 55 During the boom voluntary-insurance systems grew less than compulsory-insurance systems, but more than did unit-banking state systems without insurance plans. Voluntary-insurance states were also intermediate cases with respect to failure rates and liquidation delays. Branching banks suffered much lower risks of failures and enjoyed disproportionately high rates of growth and entry during the 1920s, relative to unit banks, in states that permitted branching, especially those that allowed branching outside the home city of the bank. Thus from the standpoint of desirability of outcomes during the 1920s the various regulatory regimes could be ranked (in descending order) as follows: full statewide branching, limited branching, uninsured unit banking, voluntary-insurance unit banking, and compulsory-insurance unit banking.®® The contrast between the effects of branch-banking regulations and deposit-insurance regulations is ironic, since the two regulatory choices were viewed as alternative solutions to the problem of providing banking-system stability, without sacrificing banking services in remote areas, during the years of active bank regulatory reform after the Panic of 1907 (White, 1982 and 1983). The history of the 1920s reveals that branching and deposit guarantee in fact had opposite effects with respect to generating banking stability. From this perspective, an added cost of deposit insurance legislation was that it was incorrectly perceived as an alternative to branch banking, and thereby helped to perpetuate 56 unit banking. Proponents of deposit insurance, however, might argue that it was the manner in which deposit insurance was implemented, not insurance per se, that caused systemic collapse in the 1920s. They might argue that higher capital requirements, better supervision, risk-based insurance premia, and government financing of the insurance fund might have produced a better result. Clearly with high enough capital requirements and sufficiently strict entry barriers (as in Mississippi), the moral hazard and adverse selection problems of deposit insurance will be reduced and may disappear, but at the expense of higher financing costs to banks, and less entry of banking into peripheral areas. Elsewhere (Calomiris, 1989) I have argued that a more successful, efficient, and historically proven reform of deposit insurance would be to move to a mutual-guarantee system of self-regulating branching banks, in which the governments' main role would be to define membership criteria for mutually insuring groups of banks. Such systems were extraordinarily successful in dealing with financial panics during the pre-Civil War years in the U.S., while providing access to affordable loans in peripheral areas. Unlike virtually all government deposit insurance regulators, banks regulated and monitored one another effectively, discovered and corrected unsound banking practices quickly, and kept the payments system operating smoothly in the face of financial crises.^ Evidence of similarly successful self 57 regulating systems in other countries is provided in Bordo and Schwartz (1989) and Pope (1989). A possible objection to this approach is the difficulty of banks as a group to insure themselves against very large disturbances because of limited aggregate banking capital. In such circumstances, a systemic collapse could occur. Furthermore, given this possibility it might be impossible for the government to commit credibly to allowing the banking system to fail. Knowledge of this implicit guarantee may provide incentives for risk taking. I have two responses to this objection. First, if an economy is prone to shocks of this magnitude it may be that deposit insurance is inadvisable altogether. Why should not banking capital in aggregate be allowed to fall drastically at a time when the investment opportunities of an economy have been so devastated? In the presence of free entry and branch banking, one would expect new banks or branches to arise to take the place of failed ones, as in Arizona and South Carolina in the 1920s. Furthermore it seems inadvisable for an economy so devastated by a terms-of-trade shock to attempt a bankingsystem rescue, particularly in a developing economy that relies on indirect (often financial) taxation to finance such bail outs . It might be more advisable to act in advance to subsidize new industries with an eye toward diversifying the economy, rather than focus on banking system solvency as a panacea. The fundamental problem of such an economy, after 58 all, is not its financial system, but its economic base. My second response to the supposed need for governmentfinanced and government-regulated deposit insurance is an empirical one. In most cases during the 1920s aggregate banking capital within each state would have been sufficient to repay losses to depositors of failing institutions -- and thus mutual-guarantee, self-regulating systems operating even at the state level would have been feasible. Table 23 reports total deposits of suspended banks (deposits of failed banks are not available) aggregated by state from 1921 to 1929 for national and state banks in the 13 states with the largest total bank failure rates and provides estimates of the total shortfall of assets in failed banks of each type. A rough indication of the rate of asset shortfall for national- and state-chartered banks that failed in these states can be derived from Table 21, although as already noted these data are imperfect indicators. To obtain an estimate of total asset shortfalls I multiply the total deposits of suspended banks by the shortfall rate from Table 21 (one minus the repayment rate) and multiply this product by the ratio of failed banks to suspended banks. As Table 23 shows, in many states, the number of bank suspensions far exceeded the number of bank failures, because banks were sometimes reopened or acquired rather than being placed in receivership. A comparison of the average size of failed banks (estimated using data on completed liquidations) and average size of suspended 59 banks revealed that larger banks were more likely to avoid liquidation after suspension. Thus I adjusted for the average size difference between suspended and liquidated banks in estimating the total asset shortfalls. To summarize, the estimated shortfall of assets (the difference between depositor claims and receipts from asset liquidation) is given by the product of four terms: total deposits of suspended banks, the ratio of liquidations to suspensions, the shortfall ratio (estimated using data from completed liquidations), and the ratio of the average size of liquidated banks to the average size of suspending banks (again, estimated using data from completed liquidations). These estimates appear in Table 23 for state- and national-chartered banks. The level of bank capital plus surplus (bank equity) of solvent banks in 1930 is provided for comparison. Only 3 of the 13 states show a ratio of shortfall to bank equity approaching or above unity: Nebraska, North Dakota, and South Dakota. For all other states banks as a whole would have had sufficient capital to support failing banks. The national banking failures in all the states could have been absorbed by surviving national banks, and statechartered systems could have covered losses of failed banks in every state except Nebraska, North Dakota, and South Dakota. These three state systems, however, suffered bank losses several times the size of remaining state banks' equity. Significantly, these were the only states that had compulsory 60 insurance for most of the 1920s (this criterion excludes Oklahoma), and that also allowed substantial entry by new banks (this criterion excludes Mississippi). These states had foreclosure rate and land depreciation experiences comparable to several other states (see Table 3) — notably Montana, Georgia, and South Carolina -- but none of these other states' state-chartered systems approached the banking losses relative to remaining equity of the three long-lived, compulsoryinsurance systems. This evidence supports the evidence from balance sheet data and the evidence on failure rates and failure severity reported above. Moreover, it supports the argument that, absent compulsory deposit insurance (and free entry), the fundamental disturbances experienced in these states would have had very different consequences for their banking systems. If statewide branch banking had been permitted within these states, bank failures would have been even lower, and the entry of banking capital during the 1920s would have been higher. Moreover, in a mutual-liability, self-regulating system of banks (like that of three states in the pre-Civil War era) bank risk, taking would have been substantially circumscribed by self-imposed regulations and vigorous supervision of other banks. Of course, no degree of regulatory wisdom could, or should, have made the 1920s a profitable time for banks in agricultural regions affected by drastic declines in prices and 61 land values. inevitable. In the face of these shocks, some failures were What regulation could have done, but did not do, was make the system as a whole less susceptible to shocks and more resilient in its response to failures. 62 NOTES 1. For a discussion of the behavior of Southern branching banks during the Panic of 1837 and the Panic of 1857, see Calomiris and Schweikart (1988). 2. Of course, there will still be a need for a properly designed monetary authority and lender of last resort to manage the money supply, but this is separate from the question of insuring banks. 3. Of course, any government transfer program must be financed somehow, and this gives rise to additional costs. Raising revenue, either through direct taxation or money creation, can have important adverse allocative consequences. Indeed, as McKinnon (1973, 1988) and others (see Fry, 1988 for a review) point out, in less-developed economies, the need to raise revenue often places a disproportional burden on the banking system. Reserve requirements and mandated loan subsidies are among the means to tax financial intermediaries. The ease of enforcing these taxes, and their indirect nature presumably have made them a desirable means of raising funds for governments with little power to impose or enforce direct taxation. The fact that governments in less-developed economies need to rely on banks as a source of finance may limit the ability of the government to bail out banks. I return to this point in the concluding section of the paper. 4. As White (1983) has shown, these points are related. Banks wishing to locate in rural areas were more likely to choose a state charter, presumably because of the less restrictive provisions for making loans on agricultural real estate. 5. The sector-specific crisis of the 1920s was followed by the general Depression of the early 1930s. The dust storms of 1934-1935 kept agriculture from sharing in the general recovery of 1933-1937. These factors combined to produce a period of agricultural depression that lasted for some fifteen years. 6. Alston (1983), p. 886. 7. Foreclosure data are provided in Table 3. These data are not identical to those reported in Alston (1983), for two reasons. First, 1928 rather than 1930 is used as a benchmark for the number of farms operating from 1926 to 1930. Second, and more important, Alston only subtracted "croppers" from the total number of farms to estimate the number of farms at risk of foreclosure; my estimates subtract all farms operated by tenants, not only "croppers." This alternative definition is 63 meant to capture better the relevant population of farmers subject to foreclosure risk, assuming that tenant-run farms typically are owned by individuals who operate their own farms, as well. 8. It is also important to note that the threat to the financial viability of farms and farm lenders from a given decline in income or wealth depends in a non-linear fashion on the rapidity of the decline and its persistence. Two consecutive years of drastic price and income reductions may produce far more bankruptcies than a similar decline amidst intervening good years, or a similar overall decline spread over a longer period. This is especially true when a rapid decline follows a boom period -- farm leverage, having first been increased by borrowing during the boom becomes further increased by reductions in farm values during the bust, precisely at a time when the cash-flow necessary to meet debt service requirements is reduced. For evidence on the importance of such non-linearities see Rucker and Alston (1987). 9. See Goldenweiser et al. (1932), vol.5, pp. 205-207. 10. See, for example, the discussion of large urban bank reactions to deposit insurance in cities such as Chicago and Philadelphia in White (1983), pp. 191-197. 11. Specifically, in Oklahoma and Kansas interest rates were limited to 3 percent; in Texas deposits bearing any interest would be exempt from insurance; in Mississippi and South Dakota interest rates were limited to 4 and 5 percent, respectively; and in North Dakota and Washington interest rate limits were set by the Guaranty Boards. Summary tables of these and other regulations are provided in White (1983, pp. 210-11) and Calomiris (1989, p. 18) . 12. Evidence of this phenomenon can be found in numerous historical studies of the operations of bankers under deposit insurance. For example, see Robb (1921). 13. In Oklahoma and Kansas, trusts were not admitted to the insured system, which further discouraged large urban banks from joining. 14. Oklahoma's first bank failure, that of the Columbia Bank and Trust, was a clear case of speculative expansion through loans to the oil firms owned by the banker, W. L. Norton. For details, see Robb (1921), pp. 50-53. 15. See Fenstermaker et al. (1984) and Schweikart (1987) . 64 16. The data in Table 12 are end-of-year, unlike the other tables which are end-of-June. Thus, the peak in Table 12 occurs in 1919, rather than 1920. 17. For the 48 contiguous states the correlation between the ratio of capital to assets and the average size of banks is strongly negative. For state-chartered banks the correlation in 1920 is -0.47 (significantly different from zero at the 99 percent confidence level); for national banks the correlation in 1920 is -0.43 (significant at the 97 percent confidence level). 18. See FDIC (1956), pp. 55-58. 19. Some of this exceptionally high banking growth in these states reflects favorable economic fundamentals. As Table 1 shows, livestock prices rose rapidly in the late 1920s relative to grain prices; thus livestock-dependent states like Wyoming, Idaho, and Arizona should have seen more banking recovery. In regression results that follow, I control for economic environment to isolate the role of regulatory regimes in promoting banking growth. 20. Lee Alston has suggested to me that the increased use of automobiles may have permitted greater bank consolidation, as well, by reducing the need for banks to be located in thinly populated areas. 21. The reduced riskiness of branch, as opposed to unit, banks is established in section IV, where I show that branching banks were less likely to fail during the 1920s than unit banks. See also Cartinhour (1931), Schweikart and Doti (1989, chapter III), and White (1983), pp. 218-219. 22. In Georgia, one of the largest banks in the state, operating 30 branches, failed. According to Cartinhour (1931, p. 307), the cause of this failure was "poor management." 23. The Federal Reserve's agents used their own judgment in determining whether banks under a single holding company operated under centralized control. While they attempted to provide an exhaustive survey of bank practices, sometimes the agents found that "neither the power to exercise such control nor the amount of control actually exercised [could] be determined." See Board of Governors (Dec. 1929), p. 766. 24. Data on chain bank failures have not been collected in a consistent or thorough manner, but evidence reported in Chapman (1934), and other examples of the collapse of large chains indicates that unit banks belonging to chains were not insulated from shocks as were branch banks. The failure rates of branching banks are discussed in section IV. 65 25. Thies and Gerlowski (1989) provide a detailed discussion of the Oklahoma experience, and describe regression results showing that insured state systems had a 0.7 percent higher propensity to fail on average than uninsured state-chartered systems, for the period 1921-1929. A separate regression for national banks found no significant difference for national banks in the insured states. While the authors do control for "time, region, and urbanization" (specific results and explanation of data are not provided), this is insufficient to capture differences across states in fundamental disturbances. 26. The distinction between failures and suspensions is empirically important. Their incidences often differed greatly, and the fraction of suspended banks that reopened differed across states and chartering systems. A crosssectional analysis of these differences remains a topic for future research. Thies and Gerlowski (1989) seem to have used suspensions as their measure of bank failures. 27. See FDIC (1956), p. 69 28. These are calculated using the banks in existence immediately prior to the period of failure as a scale variable (denominator) in the calculation. As noted above, insured banks sought to avoid rising assessments, and converted to national charters (as shown in Table 17). This avoids exaggeration of failure rates due to voluntary exit by banks that arises when the average number of banks in existence over the whole period is used instead as the denominator. Using either measure there is an adverse selection problem to consider in measuring failure rates of different systems within the same state. Early failures in the insured system could lead insured banks to exit to the other available systems, and thereby raise the subsequent observed failure rates for national and uninsured state banks. This means that observed differences in failure propensity provide a fortiori evidence of greater riskiness of insured banks. Empirical evidence, however, indicates that the movement of banks from one system to another did not have an important effect on bank failure rates. For example, compare failure rates for national and state banks reported in Table 21 for the periods 1921-1924 and 1925-1929 for Kansas, North Dakota, Oklahoma, Mississippi, and Texas -- all states with a substantial rate of conversion from state to national charters. In two cases (Mississippi and Oklahoma), national bank failure rates fell in the later period; in the other three cases, they rose slightly. 29. Weighted least squares is the appropriate regression technique in circumstances where aggregate failure rates are compared across different samples. I used the log-odds ratio (the log of p/(1-p)) as the dependent variable in the 66 regressions to control for truncation bias, where p is the probability of failing, measured by the proportion of banks failing. For a more complete description of the weightedleast-squares technique and its applicability to this case, see Maddala (1984), pp. 28-30. 30. The variations I experimented with included: whether to include Mississippi with the other compulsory-insurance states (given its stricter entry requirements, discussed below); whether to pool national- and state-chartered banks, estimate them separately, or allow their coefficients to differ within a pooled regression; whether to include the ratio of capital to assets and the average size of banks in the regressions; and whether to use commercial failures, real-estate loans, and land price declines, by themselves, or interacted with farmpopulation proportion and bank real estate loan holdings, as control variables. 31. See also ABA, 1933; Calomiris, 1989; and Thies and Gerlowski, 1989. The inability to separate unincorporated and incorporated state bank failures in Texas and Washington makes a similar comparison impossible for those states. Also, the small number of uninsured banks in Texas, and the short duration of insurance in Washington make such comparisons less interesting. 32. ABA (1933), p. 22 and Robb (1921), pp. 165-170 argue that Mississippi maintained exceptionally high standards for admission of new banks. For example, ABA (1933) writes that "the banking authorities in Mississippi had full discretion in the matter of granting new charters and used it liberally in refusing permission for unneeded banks or to unqualified promoters to open new institutions." 33. For sources see Data Appendix. 34. For a discussion of the costliness of these delays in liquidating Savings and Loans see Barth et al. (1989). 35. U.S. House of Representatives (1930), volume I, p. 462. 36. Georgia is not part of this group because it prohibited new branch banking only in August 1927. 37. See Chapman and Westerfield (1942), pp. 126-130. 38. Note that I am not arguing that insurance systems should allow voluntary exit. I would argue, however, that since none of the insurance systems succeeded in providing payments system protection, voluntary insurance was superior to compulsory insurance in the 1920s because it engendered less loss. 67 39. Not only does branching lead to a, more stable banking system; it also increases the ability of banks to meet the banking needs of peripheral areas. Using current county-level data from the U.S., Evanoff (1988) shows that branch banks provide a far superior means of servicing remote areas than unit banks. Holding demographic factors constant, branching increases the number of banking offices per square mile by 65 percent. 40. Ideally, such a system would allow branching, as well. In the absence of freedom to branch the large number of unit banks creates a problem, as noted in Calomiris (1989). For mutual guarantee systems to be effective they must be small enough to make interbank monitoring worthwhile to individual banks. Systems of hundreds of mutually liable banks entail trivially small marginal gains to monitoring the behavior of another bank. An alternative would be separate smaller groups of mutually liable unit banks. A second problem that arises in either the branching or unit banking versions of the mutual-guarantee system is the potential for banks to abuse their self-regulatory power to inhibit competition. To prevent this, the government should create more than one group of banks, and define group membership in a manner than encourages inter-group competition. For example, in a unit-banking mutual-guarantee system (where local monopolies may arise) groups should overlap geographically. 68 DATA APPENDIX Income and Price Data Data on indices of gross income by type of farm product reported in Table 1 are taken from Strauss and Bean (1940), p. 31. Data on income — farm and non-farm, gross and net — and farm and non-farm population are taken from Leven (1925), pp. 192-209, 259. The state-specific crop price index is defined as the relative price in 1924 of the bundle of crops sold in 1919. These data are reported in U.S. Department of Commerce (1927), passim. Data on the value of crops sold, by state, was compiled by the Bureau of Agricultural Economics, Department of Agriculture, and reported in U.S. Department of Commerce, Statistical Abstract of the United States, various years. Farm Land Values. Mortgages, and Foreclosures Data on farm real estate values per acre, total real estate value, and amount of farm mortgage debt are provided in Clifton and Crowley (1973). Farm foreclosure data are from Stauber (1931). GNP Deflator Estimates Alternative annual estimates of the GNP deflator, reported in Table 4, are from Balke and Gordon (1989) and Romer (1989) . 69 Branch and Chain Banking Data on state branching regulations, numbers of branch banks and their branches, and banking chains are taken from Board of Governors (1924, 1926, 1927, Feb. 1929, Dec. 1929). Bank Balance Sheet Data Bank balance sheet data, and total numbers of banks, disaggregated by state and by type of charter are taken from Board of Governors (1959), and — for insured banking systems-- from FDIC (1956), pp. 66-67. Locations and Survival of Individual Banks Data for individual banks, and bank locations, are taken from Bankers Encyclopedia Co., various years. Numbers of Bank Liquidations Liquidations of national banks are reported in the Annual Report of the Comptroller of the Currency. State-bank liquidations for each state were published in the Comptroller's Annual Report. as well. The definition of banks employed in Board of Governors (1959) is used to construct state-level series for failed "state" banks. This definition includes trusts and unincorporated banks, as well as narrowly defined state-chartered banking corporations. It is not possible to derive consistent series of narrowly defined state-chartered bank balance sheet or failure data using these sources. 70 Charter Switching Data on bank charter switching are taken from Board of Governors (1937), pp. 1087-1122. Business Failures Business failures and number of solvent enterprises for each state are reported in U.S. Department of Commerce, Statistical Abstract of the United States. Number and Deposits of Suspending Banks The number and deposits of state and national bank suspensions are reported in Board of Governors (1943), pp. 286-291. These are used to derive the average size of suspended banks in Table 23. Data on Liquidated Banks Data used in Tables 21-23 on the number, deposits, losses, and time taken to liquidate banks for which liquidations had been completed by 1930 are reported in Goldenweiser et al. (1932), vol. 5, pp. 191-207. 71 REFERENCES Alston, Lee J., "Farm Foreclosures in the United States During the Interwar Period," Journal of Economic History, vol. 43, December 1983. American Bankers' Association, The Guaranty of Bank Deposits. New York, 1933. Balke, Nathan S. and Gordon, Robert J., "The Estimation of Prewar Gross National Product: Methodology and New Evidence," Journal of Political Economy, vol. 97, February 1989. Bankers Encyclopedia Co., The Bankers Encyclopedia. New York, semi-annual. Barth, James R., Bartholomew, Philip F., and Labich, Carol J., "Moral Hazard and the Thrift Crisis: An Analysis of 1988 Resolutions," Research Paper No. 160, Federal Home Loan Bank Board, May 1989. Bernanke, Ben S., "Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression," American Economic Review, vol. 73, June 1983. Board of Governors of the Federal Reserve System, All Bank Statistics. Washington D.C., 1959. Board of Governors of the Federal Reserve System, Banking and Monetary Statistics. Washington D.C., 1943. Board of Governors of the Federal Reserve System, "Changes in the Number of National and State Banks During 1921-1936," Federal Reserve Bulletin. November 1937. Board of Governors of the Federal Reserve System, "Branch Banking in the United States," Federal Reserve Bulletin. December 1924. Board of Governors of the Federal Reserve System, "Branch Banking in the United States," Federal Reserve Bulletin. June 1926. Board of Governors of the Federal Reserve System, "Branch Banking Developments June 30, 1928," Federal Reserve Bulletin. February 1929. Board of Governors of the Federal Reserve System, "Branch and Chain Banking Developments: 1929," Federal Reserve Bulletin. December 1929. 72 Board of Governors of the Federal Reserve System, "McFadden Bill Now Law," Federal Reserve Bulletin, March 1927. Bordo, Michael D. and Schwartz, Anna J., "The Performance and Stability of Banking Systems Under 'Self-Regulation': Theory and Evidence," Unpublished manuscript, 1989. Brock, Philip L., "The Transmission of Terms of Trade Shocks in Latin America," Unpublished manuscript, Duke University, November 1988. Calomiris, Charles W., "Deposit Insurance: Lessons from the Record," Economic Perspectives. Federal Reserve Bank of Chicago, May/June 1989. Calomiris, Charles W., Hubbard, R. Glenn, and Stock, James H., "The Farm Debt Crisis and Public Policy," Brookings Papers on Economic Activity, vol. 2, 1986. Calomiris, Charles W. and Kahn, Charles M., "The Role of Demandable Debt in Structuring Optimal Banking Arrangements," Unpublished manuscript, Northwestern University, 1988. Calomiris, Charles W. and Kahn, Charles M., "Interbank Monitoring as Seignorage Sharing: The Suffolk System," Northwestern University, 1989. Calomiris, Charles W. and Schweikart, Larry, "Was the South Backward?: North-South Differences in Antebellum Banking During Crisis and Normalcy," Unpublished manuscript, Federal Reserve Bank of Chicago, 1988. Cannon, James G., Clearing Houses. Washington D.C., 1910. Cartinhour, Gaines T., Branch, Group, and Chain Banking. New York, 1931. Chapman, John M., Concentration of Banking. New York, 1934. Chapman John M. and Westerfield, Ray B., Branch Banking; Its Historical and Theoretical Position in America and Abroad. New York, 1942. Clifton, Ivery D. and Crowley, William D., Jr., "Farm Real Estate Historical Series Data: 1850-1970," U.S. Department of Agriculture, Economic Research Service, June 1973. Comptroller of the Currency, Annual Report. Washington D.C., various years. 73 Cooke, Thorton, "The Insurance of Bank Deposits in the West," in George E. Barnett, State Banks and Trust Companies. Washington D.C., 1910. Evanoff, Douglas D., "Branch Banking and Service Accessibility," Journal of Money. Credit and Banking, vol. 20, May 1988. Federal Deposit Insurance Corporation, Annual Report. Washington D.C., 1956. Fenstermaker, J. Van, Filer, J.E., and Herren, R.S., "Money Statistics of New England, 1785-1837," Journal of Economic History. June 1984. Fry, Maxwell, Money. Interest, and Banking in Economic Development. 1988. Goldenweiser, E.A., et al., Bank Suspensions in the United States. 1892-1931. Material prepared for the information of the Federal Reserve System by the Federal Reserve Committee on Branch, Group, and Chain Banking, 1932. Gorton, Gary, "Clearing Houses and the Origin of Central Banking in the U.S.," The Wharton School, 1985. Gorton, Gary, "Self-Regulating Banking Coalitions," The Wharton School, 1989. Kane, Edward J., "How Incentive-Incompatible Deposit-insurance Funds Fail," Unpublished manuscript, Ohio State University, 1988. Kareken, John H. and Wallace, Neil, "Deposit Insurance and Bank Regulation," Journal of Business. July 1978. Leven, Maurice, Income in the Various States: Its Sources and Distribution. 1919. 1920. and 1921. New York, 1925. Maddala, G.S., Limited-Dependent and Qualitative Variables in Econometrics. London, 1984. McKinnon, Ronald I., Money and Capital in Economic Development Washington D.C., 1973. McKinnon, Ronald I., "Financial Liberalization and Economic Development: A Reassessment of Interest-Rate Policies in Asia and Latin America," International Center for Economic Growth, Occasional Paper No. 6, 1988. Pope, David, "Free Banking in Australia Before World War I," Unpublished manuscript, 1989. 74 Robb, Thomas B.f The Guaranty of Bank Deposits, New York, 1921. Romer, Christina D., "The Prewar Business Cycle Reconsidered: New Estimates of Gross National Product, 1869-1908," Journal of Political Economy, vol. 97, February 1989. Rucker, Randal R. and Alston, Lee J., "Farm Failures and Government Intervention: A Case Study of the 1930s," American Economic Review, vol. 77, September 1987. Stock, James H., "Real Estate Mortgages, Foreclosures, and Midwestern Agrarian Unrest, 1865-1920," Journal of Economic History, vol. 43, March 1984. Schweikart, Larry, Banking in the American South. Baton Rouge, 1987. Schweikart, Larry, and Doti, Lynne Pierson, Banking in the American West from Gold Rush to Deregulation. Unpublished Manuscript, University of Dayton, 1989. Stauber, B.R., "The Farm Real Estate Situation," U.S. Department of Agriculture, Circular No. 209, Washington D.C., December 1931. Strauss, Frederick, and Bean, Louis H., "Gross Farm Income and Indices of Farm Production and Prices in the United States, 1869-1937," Technical Bulletin No. 703, U.S. Department of Agriculture, December 1940. Thies, Clifford F., and Gerlowski, Daniel A., "Deposit Insurance: A History of Failure," Cato Journal, vol. 8, Winter 1989. U.S. Department of Commerce, Statistical Abstract of the United States. Washington D.C., various years. U.S. Department of Commerce, United States Census of Agriculture. 1925. Washington D.C., 1927. U.S. Department of Commerce, Fifteenth Census of the United States. Washington D.C., 1932. U.S. House of Representatives, "Branch, Chain, and Group Banking," Hearings before the Committee on Banking and Currency, 71st Congress, Second Session (1930). White, Eugene N., "The Political Economy of Bank Regulation," Journal of Economic History. March 1982. White, Eugene N., The Regulation and Reform of the American Banking System. 1900-1929. Princeton, 1983. 75 White, Eugene N., "The Merger Movement in Banking, 1919-1933," Journal of Economic History, vol. 45, June 1985. Woods, L. Milton, Sometimes the Books Froze: Wyoming's Economy and Its Banks. Boulder, 1985. 76 Table 1 Indices of gross Income, from various groups of farm products and from total farm production, crop years [1909-13=100] Year 12 important crops1 Staple food-stuffs2 Fruits3 Dairy products, chickens, and eggs Textile raw materia1s4 Meat animals slaughter and export of live cattles Meat animals adjusted for for changes in inventory values Total farm production adjusted for changes in inventory Total farm values of production_ meat anijrals 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 99.8 97.6 102.4 101.3 102.4 112.1 143.3 220.3 239.5 269.4 177.5 109.6 132.9 150.2 167.9 167.9 142.4 156.6 147.0 143.7 93.2 97.7 101.0 99.6 131.3 146.5 154.6 222.2 284.2 326.2 252.7 150.7 140.5 127.3 162.0 176.4 176.8 177.3 144.9 159.7 101.6 106.8 108.5 102.4 109.0 117.7 126.0 147.3 189.2 260.7 269.4 183.5 222.3 203.1 222.8 223.6 231.5 220.1 221.0 233.6 100.3 88.7 101.5 103.9 105.5 104.5 117.1 158.6 191.7 223.1 241.6 173.8 167.0 189.4 191.3 211.9 223.8 223.6 234.9 245.0 105.3 96.0 99.5 110.0 77.0 85.9 134.3 201.6 231.9 255.4 136.5 84.7 135.6 179.2 195.4 198.8 145.6 167.6 170.6 161.7 99.7 89.9 95.2 107.7 107.5 104.4 129.0 180.8 242.7 239.0 186.6 116.4 129.1 132.0 135.8 163.2 172.8 158.9 163.6 171.4 96.1 83.0 92.7 110.9 117.3 112.2 133.1 189.5 232.8 219.5 173.6 112.5 126.5 122.5 118.6 147.4 163.5 155.2 164.1 172.6 101.4 95.3 102.1 105.6 106.5 110.1 134.2 194.6 231.5 253.5 204.0 132.6 146.3 160.0 169.6 182.8 178.0 179.5 180.4 184.1 101.5 94.2 102.4 107.7 110.6 113.6 136.9 199.2 231.3 250.5 202.3 132.8 147.0 158.7 166.1 180.0 177.3 180.2 182.4 106.4 1930 1931 1932 1933 1934 1935 1936 1937 87.5 56.7 50.5 74.1 71.2 82.4 110.3 105.4 106.8 65.5 51.9 83.0 73.4 88.0 116.3 121.2 184.4 150.0 102.9 137.3 145.6 160.8 166.5 195.7 205.7 157.1 119.8 112.3 126.0 155.5 160.8 169.7 90.3 62.3 53.1 80.9 87.0 83.0 107.1 111.0 146.5 102.6 68.2 73.8 86.0 114.1 139.2 141.8 147.6 105.8 73.4 76.7 64.8 110.8 130.3 136.2 141.0 100.9 76.7 89.0 95.0 115.8 133.1 139.6 143.0 103.0 79.0 90.7 89.7 116.1 132.0 139.5 ^Contained in the index of crop production of the Business of Agricultural Economies: wheat, corn, oats, barley, rye, buckwheat, flaxseed, hay, potatoes, sweetpotatoes, cotton and cottonseed, tobacco. ^Wheat, rye, potatoes, sweetpotatoes, dry beans, rice. ^Orchard fruits, citrus fruits, grapes. ^Cotton and cottonseed, flaxseed, wool. ^Slaughter of cattle, calves, hogs, sheep, and lambs, and export of live cattle. Sources: See Data Appendix. Table 2 Price and Income Changes Across States 1919-1921 l A gross farm income Alabama Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming -44 -37 -44 -24 -41 -7 -39 -35 -57 -39 -50 -48 -50 -45 -46 -47 -26 -42 -14 -34 -48 -51 -48 -30 -51 -38 -12 -29 -37 -25 -41 -46 -44 -50 -30 -32 -22 -56 -53 -39 -44 -43 -15 -41 -21 -33 -30 -43 Sources: See Oata Appendix. 1919-1921 X A total net farm income -38 -26 -49 58 -19 162 -39 8 -78 -58 -89 -101 -113 -66 -32 -64 24 -10 111 -6 -75 -54 -116 NA -92 NA 222 17 40 35 -40 -39 -49 -76 -32 11 64 -88 -101 -28 -63 14 77 -36 -21 1 1 106 1919-1921 X A total net non-farm inc. 25 14 27 82 56 44 39 51 20 20 45 23 3 39 51 35 48 45 53 27 38 17 42 34 16 53 40 44 47 61 29 69 25 14 29 42 61 -6 -19 42 37 37 43 32 37 43 40 77 1919-1924 t A crop price index -29 -35 -33 -45 -43 -36 -34 -37 -54 -54 -45 -55 -50 -32 -48 -43 -65 -45 -41 -55 -34 -37 -51 -47 -37 -38 -49 -51 -18 -50 -47 -1 -56 -32 -48 -46 -48 -62 -36 -39 -16 -50 -44 -46 -55 -48 -54 -34 1922-1925 X A value of crops sold 2 18 1 20 41 -7 13 8 11 61 11 10 5 9 -10 -23 -60 -20 -35 -23 -23 -25 -14 2 23 0 10 8 51 19 4 21 11 21 17 23 18 6 3 -10 -15 43 1 -5 41 14 26 28 1925-1928 X A value of crops sold -16 44 -10 0 -24 -19 -15 31 0 -22 -4 -14 3 19 -2 -43 -65 -45 -41 -55 -34 -37 -51 47 2 -3 -39 -11 -6 -36 -9 -16 -21 0 -13 -34 -42 -14 -14 -6 27 -16 -30 -7 -25 -22 -26 1 Table 3 Farm Land. Population, and Foreclosure Data 1913-1920 1920-1925 t A value % A value farm real estate per acre farm real estate per acre 77 65 122 67 41 37 39 78 117 72 60 61 113 51 100 98 42 66 40 54 113 118 67 26 79 35 29 30 44 33 123 45 59 66 30 40 30 130 81 100 74 67 50 89 40 54 71 76 -11 -56 -20 10 -31 10 -3 75 -40 -34 -27 -32 -34 -19 -30 -22 2 -5 8 -6 -27 -34 -30 -37 -32 -41 11 24 -31 3 -7 -28 -23 -20 -13 -4 14 -34 -37 -19 -14 -20 -7 -7 -17 -8 -12 -54 Alabama Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming Sources: See Data Appendix. 1925-1930 % A value farm real estate per acre -7 2 -12 -2 -10 2 -1 0 -14 -6 -21 -22 -17 -2 -9 -6 0 -6 -1 -9 -16 -10 -18 -4 -8 -3 -2 1 2 -7 -16 -13 -18 -3 -3 -6 5 -25 -19 -10 -5 -3 -2 -13 -3 -13 -10 -2 1920 ratio of farm mortgage debt to farm value .12 .20 .11 .13 .17 .13 .15 .08 .08 .21 .09 .08 .16 .12 .09 .10 .10 .13 .13 .15 .15 .11 .14 .22 .13 .20 .10 .16 .13 .16 .06 .19 .08 .16 .14 .10 .08 .07 .13 .09 .12 .16 .18 .07 .13 .04 .21 .15 1920 ratio of farm to total popul. .57 .27 .65 .15 .28 .07 .23 .29 .58 .46 .17 .31 .41 .42 .54 .44 .26 .19 .03 .23 .37 .71 .36 .41 .45 .21 .17 .05 .44 .08 .58 .61 .20 .50 .27 .11 .02 .63 .57 .54 .48 .31 .36 .46 .21 .32 .35 .34 1926-1930 average am farm forec (per 1000) 29.5 42.7 39.7 16.3 42.4 5.3 13.7 11.1 56.5 37.6 29.0 23.8 48.3 27.2 20.2 40.1 10.5 16.8 6.5 21.6 36.2 47.7 34.1 52.2 38.4 21.0 7.3 7.2 26.3 13.8 23.4 58.0 16.4 50.1 17.4 6.9 6.0 68.0 70.4 20.5 23.7 13.5 10.6 15.6 20.0 9.0 22.6 26.3 Table 4 GNP Deflator Estimates Balke and Gordon (1989) 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 Sources: 11.36 13.35 15.23 17.58 15.30 14.22 14.63 14.64 14.90 14.98 14.72 14.60 14.64 See Data Appendix. Romer (191 13.06 15.20 15.58 17.75 15.12 14.30 14.69 14.51 14.77 14.84 14.48 14.59 14.60 Table 5 32 State Regulatory Regimes of the 1920s No deposit Insurance Full intrastate branching allowed Arizona North Carolina South Carolina Virginia Wyoming1 Limited new branches Kentucky Louisiana Michigan Ohio Tennessee No new branching, old branches remain Alabama Arkansas Georgia2 Indiana Minnesota Wisconsin No branches allowed Colorado Idaho Illinois Iowa Missouri Montana Nevada New Mexico Compulsory Insurance Nebraska (1911-30) Mississippi (1914-30) North Dakota (1917-29) Oklahoma (1908-23) South Dakota (1916-27) ^Branches authorized by Implication; none 1n existence. 2New branching prohibited In 1927. Sources: See Data Appendix. Voluntary Insurance Washington (1917-21) Kansas (1909-29) Texas (1910-27) Table 6 Number of banks National banks Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming Sources: 1918 1920 1923 90 13 57 124 115 55 463 255 341 212 143 31 99 273 37 130 61 228 10 37 73 146 379 343 51 105 113 518 133 77 131 32 91 18 72 122 97 68 469 258 352 234 132 31 105 294 33 131 126 191 10 43 81 165 369 340 81 125 106 543 149 80 147 38 101 20 83 141 93 81 480 254 358 249 134 38 112 331 30 136 145 188 10 47 87 181 370 348 82 136 98 556 165 87 151 47 106 20 88 143 97 73 505 251 349 266 139 34 119 344 31 132 121 182 11 42 83 184 368 459 84 131 105 561 181 115 155 45 See Data Appendix. State banks 1914 1927 105 15 79 124 83 52 490 233 287 257 142 32 134 277 36 135 74 153 10 29 77 141 340 350 65 98 104 649 167 109 156 30 1929 1914 1918 1920 1923 1927 1929 106 14 73 121 80 43 487 224 265 247 138 33 133 272 35 134 69 158 10 28 73 125 323 307 53 93 99 623 164 106 157 25 267 47 425 206 675 134 1439 664 1410 932 467 217 702 863 282 1337 226 749 21 47 384 619 746 574 329 526 378 1038 274 296 652 72 238 60 389 236 659 136 1434 773 1561 1037 444 218 740 1141 266 1407 277 946 23 74 434 693 778 580 336 517 415 1037 300 281 778 98 251 67 404 262 686 141 1489 798 1564 1100 450 229 739 1177 302 1516 286 1037 23 76 491 718 772 612 379 543 450 1125 331 306 819 113 254 55 403 224 586 109 1416 854 1506 1068 474 232 765 1151 303 1495 242 968 24 59 477 648 745 446 345 556 466 1071 343 274 838 89 251 32 376 175 412 92 1358 827 1222 923 444 200 739 912 290 1304 136 896 25 30 432 390 724 348 216 319 418 852 334 224 810 58 244 34 347 159 362 94 1319 757 1129 830 432 193 718 794 277 1191 129 714 25 30 399 309 703 344 170 303 393 791 321 233 801 62 Table 7 Total Assets per bank, and bank location Proportion of banks in towns of greater than 2,500 people, 1920^ Total Assets per bank 1929 1914 banks 1918 1920 1923 1927 1929 2311 2608 1352 2244 3377 1094 3295 2097 1381 1104 2162 4099 4991 2468 2702 4509 1489 1615 2299 1331 2634 702 2785 1448 2371 871 3033 1771 2347 3306 3413 1711 283 555 164 263 224 162 739 352 314 146 245 581 635 235 238 350 269 155 593 197 232 106 806 95 241 136 240 159 329 449 365 148 368 807 304 397 355 335 996 451 423 269 389 950 988 287 450 493 391 298 817 277 345 165 1162 228 342 271 396 242 478 637 440 242 543 974 456 460 534 487 1322 609 563 326 497 1592 14780 425 664 572 436 335 1030 347 578 248 1645 346 536 395 562 375 613 752 626 300 522 974 404 514 519 334 1610 628 542 285 537 1472 1505 403 552 631 393 319 947 364 565 211 2167 211 424 344 558 304 656 520 634 307 578 1745 445 437 489 436 2085 769 562 304 651 2029 2272 443 632 722 605 365 1063 365 744 262 3095 279 471 272 626 379 819 606 776 431 545 2107 492 477 538 568 2584 806 596 320 826 2184 2555 466 660 741 680 383 1228 466 809 289 3271 292 588 318 714 429 878 703 752 514 National banks_______________________________ Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming 1914 1918 1920 1923 806 1215 607 1069 884 546 1912 960 692 531 900 2075 2054 1220 756 2820 895 694 972 612 921 338 1545 343 908 446 1026 705 1265 1610 1592 630 1224 1299 818 1614 1676 811 2764 1319 1010 839 1474 3677 2826 1682 1281 4276 746 1342 1545 879 1379 499 2484 766 1244 718 1583 1081 2045 2570 2065 1102 1516 1766 1020 1801 2145 1088 3562 1667 1301 977 1824 4119 3784 1979 1843 5507 761 1566 1823 963 2064 563 2912 1096 1818 862 2352 1588 2461 3039 2720 1365 1449 1389 1004 1695 1704 827 3068 1635 1144 870 1957 3416 3634 1785 1956 4162 766 1424 1529 968 2086 528 2470 848 1520 731 2070 1356 2110 2482 2476 1369 ^Branches excluded. Sources: See Data Appendix. 1927 1944 1863 1285 2116 3299 1079 3737 1890 1264 1016 2118 3846 4324 2325 2589 4887 1191 1496 1990 1047 2544 644 2638 1219 2048 733 2459 1567 2381 2958 2979 1442 State National banks .55 .70 .66 .42 .75 .58 .52 .57 .36 .43 .63 .79 .71 .34 .80 .63 .28 .40 .60 .47 .77 .16 .59 .51 .72 .27 .65 .49 .50 .56 .62 .47 State banks .26 .54 .31 .23 .31 .25 .31 .37 .15 .18 .24 .31 .25 .18 .28 .22 .16 .12 .39 .37 .28 .06 .41 .21 .43 .07 .23 .24 .29 .29 .24 .13 Table 8 Asset Growth (% A) Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming National banks______________________________ State banks 1914 1918 1918 1920 1920 1923 1923 1927 1929 1930 1914 1918 1918 1920 1920 1923 1923 1927 1927 1929 1929 1930 53 48 70 48 60 84 46 39 51 74 51 77 46 48 51 53 72 62 59 67 66 67 57 121 118 91 45 61 81 66 46 108 37 51 44 29 23 60 32 24 31 24 26 37 43 32 31 34 17 15 18 20 61 24 18 46 48 31 37 50 33 29 35 53 0 -21 4 -5 -17 -32 -9 -3 -14 -5 11 -26 2 -6 10 -27 -16 -12 -8 -10 -4 -5 -16 2 -14 -18 -6 -14 -6 8 -7 -4 33 1 15 8 65 -7 18 7 -9 13 11 6 34 5 54 20 -5 -12 18 -25 13 -7 -1 10 4 -25 18 34 4 13 21 -30 60 3 17 7 35 -47 -6 11 -21 12 22 -14 57 2 71 -19 -7 -13 26 -18 7 -14 -17 17 -16 -31 30 25 0 33 30 -33 16 86 70 73 55 94 34 49 56 105 51 64 64 61 78 48 78 142 51 121 68 75 50 142 45 96 81 52 59 35 44 122 56 35 56 29 57 51 38 39 33 28 30 76 48 53 68 25 15 23 26 28 89 56 40 60 77 46 54 68 41 29 50 43 -3 -18 -12 -4 -17 -47 16 10 -7 -15 14 -6 6 -7 -17 9 -24 -11 -4 -19 -5 -23 27 -56 -28 -6 3 -23 11 -38 4 -19 9 4 3 -33 -33 10 24 19 -16 -8 14 19 46 -13 10 0 -13 6 17 -49 19 -23 39 3 -30 -55 1 -1 22 4 18 -9 -8 28 2 -1 -3 33 20 -4 -2 -5 23 4 9 -8 0 -6 7 -16 16 28 0 -13 3 3 -2 11 7 5 3 11 -4 27 -2 10 -7 -37 -47 -22 73 26 -24 -26 60 16 69 -26 -9 2 -30 -21 30 -47 14 -50 81 -53 -51 -47 11 -19 39 -29 17 -6 Sources: See Data Appendix 1927 1929 20 31 -3 0 -1 -16 -12 7 1 4 -1 10 15 4 1 -8 17 12 16 23 -2 -3 0 4 -6 13 17 9 -3 9 15 -1 Table 9 Capital as a percent of total assets National banks_____________________ 1918 1920 1923 1927 1929 1914 Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming SOURCES: 14 7 15 8 15 12 9 n 10 11 14 11 8 8 13 10 10 10 15 9 13 11 11 13 17 9 11 14 11 9 9 9 See Data Appendix. 10 7 10 5 8 7 6 8 7 7 9 7 6 7 9 7 8 6 9 7 8 8 7 6 9 6 8 9 7 6 6 5 8 5 8 5 7 6 5 7 6 7 7 6 5 5 7 6 8 6 8 7 7 7 6 6 8 5 6 7 6 5 6 5 9 6 9 5 8 8 6 8 7 8 7 7 6 7 8 8 8 7 9 8 8 8 7 8 10 6 8 9 8 6 7 6 7 5 7 5 7 6 5 7 7 7 7 7 5 6 6 7 6 6 7 7 8 6 6 6 7 6 7 8 8 6 6 6 7 5 7 5 7 6 6 7 6 7 6 7 5 6 6 6 5 6 7 6 8 6 7 6 8 5 7 8 8 7 6 5 State banks 1914 1918 17 7 20 12 20 17 10 13 10 14 17 12 8 10 16 12 14 14 14 19 13 14 8 16 16 12 16 21 15 12 9 16 12 6 12 8 12 9 8 10 8 8 11 8 6 9 8 9 10 8 9 13 9 10 6 8 11 7 9 14 10 9 8 10 1920 1923 1927 1929 9 6 8 8 10 8 7 8 7 8 9 6 5 7 7 7 9 8 7 12 7 8 5 7 8 6 8 11 11 7 6 9 9 9 10 8 11 10 7 9 7 9 9 7 6 8 7 8 10 9 8 12 9 10 10 7 9 8 12 7 7 8 7 9 8 6 5 6 6 8 7 8 7 11 8 9 5 8 10 9 9 11 10 9 6 7 9 6 9 7 12 6 7 7 7 9 9 6 5 6 6 8 7 8 6 8 8 9 6 8 9 8 9 10 10 8 6 6 6 10 11 7 9 13 12 9 7 10 Table 10 Voluntary Insurance Systems Kansas_________________________________Texas_______________________________ Kashi ngton 1917 1919 1920 1922 1924 1926 1928 1917 1919 1920 1922 1924 1926 1928 Participating Deposits JL Not participatinq Deposits JL Participating Deposits JL 577 649 683 698 651 399 794 430 427 409 369 371 547 39 828 907 990 936 8% 34 152 205 191 180 195 79 219 Sources: See Data Appendix. 46 41 41 34 37 748 - 73 88 81 62 75 154 3 Not participating » Deposits - 204 321 266 252 302 3 - 12 15 14 11 21 226 - Participating Oeposits JL Not participating Deposits JL 46 104 116 239 191 190 - - - - - - - 40 80 75 109 123 107 - - - - - - - - Table 11 High-Growth States: Insured vs. Uninsured Assets 1914/Assets 1920 National Bank State Bank Assets per bank in 1920 State Bank National Bank 1920, capital/total assets National Bank State Bank Arkansas Colorado .Idaho Iowa Minnesota Missouri "Montana New Mexico jtyoming Average .406 .522 .341 .507 .509 .490 .495 .501 .314 .454 .379 .450 .316 .503 .406 .540 .489 .352 .315 .418 1020 1069 1088 1301 1979 5507 761 963 1365 1673 456 460 487 562 425 572 436 347 300 448 .084 .081 .059 .057 .054 .063 .077 .073 .048 .066 .085 .083 .088 .104 .069 .072 .091 .119 .090 .090 Kansas Mississippi Nebraska «North Dakota Oklahoma “South Dakota Texas Average .463 .506 .537 .485 .309 .400 ^414 .447 .380 .335 .335 .367 .259 .351 .391 .344 977 1843 1566 563 10% 862 1588 1231 326 664 335 248 346 376 375 389 .066 .069 .057 .068 .126 .053 .071 .073 .079 .066 .082 .081 .070 .062 .112 .078 Sources: See Data Appendix. Table 12 Growth Quart!le Comparison Nationa1-chartered State-chartered Lowest 25th percentile growth 25% - 50% (second quartile) Lowest growth quartile Nevada, Illinois, Kentucky Colorado, Georgia,» Indiana, Ohio Nebraska Second quartile Iowa, Missouri, Montana Minnesota, Wisconsin New Mexico, Tennessee Mississippi Third quartile Alabama, Washington Michigan, Virginia Arizona, Kansas, North Dakota, Texas Arkansas, South Carolina Fourth quantile Sources: 50% - 75% (third quartile) See Data Appendix. Highest 25th percentile growth Idaho, Louisiana North Carolina, Oklahoma, South Dakota, Wyoming Table 13 Regression Results: Early Asset Growth of State-Chartered Banks Dependent variable: Growth in total assets of state-chartered banks, 1914 - 1920 Coefficient Independent Variables Intercept National bank growth (Reserve center)x(Nat. bank growth) Growth in land values, 1914-1920 Ratio of farm to non-farm popul. Presence of voluntary insurance Presence of compulsory insurance 0.156 0.682 -0.115 0.526 -0.328 0.327 0.609 Standard Error 0.468 0.147 0.063 0.334 0.655 0.251 0.189 Prob > 0.741 0.000 0.080 0.127 0.621 0.205 0.004 R2 = 0.683 R2 = 0.607 Dependent variable: Growth in total assets of state-chartered banks, 1914 - 1920 Independent Variables Intercept National bank growth (Reserve center)x(Nat. bank growth) Growth in land values, 1914-1920 Ratio of farm to non-farm popul. Presence of vol. or comp, insurance R2 = 0.670 R2 = 0.607 Coefficient 0.101 0.681 -0.132 0.555 -0.283 0.518 Standard Error 0.465 0.147 0.060 0.333 0.654 0.165 Prob > 0.829 0.000 0.038 0.107 0.669 0.004 Table 14 State-Chartered Reqional Comparison: Assets 1914/ Assets 1920 Insured vs. Uninsured Assets per bank in 1920 Capital /total assets in 1920 Arkansas Colorado Iowa Idaho Minnesota Missouri Montana New Mexico Wyoming Average .379 .450 .503 .316 .406 .540 .489 .352 .315 .417 456 460 563 487 425 572 436 347 300 450 ,085 .083 .067 .077 .069 .072 .091 .119 .090 .084 Kansas North Dakota Nebraska Oklahoma South Dakota Texas Average .380 .367 .335 .259 .351 .391 '.347 326 248 335 346 376 374 334 .079 .081 .082 .070 .062 .112 .081 Alabama Georgia South Carolina Average .553 .412 .390 .452 543 534 536 538 .087 .097 .085 .090 Mississippi .335 664 .066 Sources: See Data Appendix. Table 15 Charter Switches From State - to NationalAlabama Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming Sources: See Data Appendix 10 1 7 16 6 0 0 4 7 0 13 3 2 14 7 0 0 1 4 4 19 10 10 1 31 0 0 2 1 7 6 12 2 113 7 11 0 4 4 8 130 0 0 16 27 2 9 1 From National - to State0 1 4 0 1 1 2 6 0 13 3 7 2 2 3 2 1 1 1 0 5 0 4 0 0 0 1 11 1 8 5 0 1 50 0 8 1 1 1 3 8 0 1 3 2 0 1 6 Net Change 10 0 3 16 5 -1 -2 -2 7 -13 10 -4 0 12 4 -2 -1 0 3 4 14 10 6 1 31 0 -1 -9 0 -1 1 12 1 63 7 3 -1 3 3 5 122 0 -1 13 25 2 8 -5 Table 16 Regression Results: Late Asset Growth of State Chartered Banks Dependent Variable: Growth in total assets of state-chartered banks, 1920-1926 Independent Variable Coefficient Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth in land values, 1920-1925 Business failure rate. 1921-1925 Business failure rate, 1917-1920 Presence of deposit insurance (excluding Nebraska) Standard Error Prob > 0.400 0.598 0.213 -0.251 0.269 0.458 0.239 0.098 0.347 0.540 0.391 0.019 0.039 0.477 0.622 -0.048 0.039 0.233 -0.271 0.123 0.036 R2 = 0.537 R2 = 0.426 Dependent Variable: Growth in total assets of state-chartered banks, 1920-1930 Independent Variable Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth in land values, 1920-1930 Business failure rate. 1921-1929 Business failure rate, 1917-1920 Presence of deposit insurance (excluding Nebraska) R2 = 0.405 R2 = 0.262 Coefficient Standard Error Prob > 1.482 0.063 0.141 -0.648 -0.091 0.554 0.225 0.135 0.475 0.659 0.013 0.782 0.308 0.185 0.891 -0.095 0.053 0.088 -0.194 0.171 0.267 Table 16 (continued) Dependent Variable: Growth 1n total assets of state-chartered banks, 1920-1930 Independent Variable Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth 1n land values, 1920-1930 Business failure rate, 1921-1929 Business failure rate, 1917-1920 Presence of deposit Insurance (Including Nebraska) R2 = 0.429 R2 = 0.292 Coefficient Standard Error Prob > 1.467 0.055 0.130 -0.593 -0.065 0.529 0.220 0.133 0.465 0.641 0.010 0.803 0.337 0.214 0.920 -0.094 0.052 0.079 -0.240 0.155 0.134 Table 17 Growth in Branch Banking Total fac. National Banks_____________________ State Banks________________________ All Banks Br. 8ran- Tot. B.B. Br. Tot. B.B. 8r. Tot. B.B. Br. Tot. B.B. & Br. Tot. Banks ches 1924 1928 1924 1928 1924 B.B.& Br. 1928 Arizona North Carolina South Carolina Virginia Wyoming 19 86 84 352 37 0 2 2 7 0 0 3 3 11 0 15 83 66 182 26 0 4 3 9 0 0 6 8 16 0 64 535 347 216 79 6 39 7 24 0 20 64 17 34 0 53 437 247 376 60 8 39 12 30 0 23 73 28 47 0 83 621 431 568 116 26 108 29 76 0 68 520 313 558 86 31 122 66 102 0 Kentucky Louisiana Michigan Ohio Tennessee 145 41 144 363 110 3 1 10 4 2 7 8 23 4 2 155 41 181 338 122 4 1 9 7 7 15 8 48 7 19 483 303 906 947 512 1 33 53 47 19 5 85 309 199 51 480 299 989 960 446 29 42 55 52 20 34 103 374 243 42 628 344 1050 1310 622 16 127 395 254 74 635 340 1170 1298 568 82 154 486 309 88 Alabama Arkansas Georgia Indiana Hi nnesota Mississippi Nebraska Washington Wisconsin 105 88 101 248 345 36 177 114 157 0 0 2 0 3 1 2 1 1 0 0 7 0 11 1 2 2 2 107 79 97 229 285 37 160 111 159 0 0 4 1 2 1 2 1 1 0 0 16 2 6 1 2 2 2 276 400 608 863 1081 346 925 272 839 5 2 19 4 0 10 0 4 6 19 3 46 8 0 24 0 5 7 269 361 394 808 855 313 746 247 817 5 2 15 3 0 10 0 3 6 19 3 21 7 0 24 0 4 7 381 488 709 1111 1426 382 1102 386 996 24 5 74 12 14 36 4 12 16 376 440 491 1037 1140 350 906 358 976 24 5 56 13 8 36 4 10 16 Colorado Idaho Illinois Iowa Kansas Missouri Montana Nevada New Mexico North Dakota Oklahoma South Dakota Texas 141 70 502 347 260 134 93 11 33 165 421 116 573 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 123 46 484 270 250 134 70 10 29 136 333 97 638 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 201 107 1408 1438 1033 1478 155 23 43 523 388 437 1046 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 164 94 1337 1169 864 1231 132 25 29 354 337 315 816 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 342 177 1910 1785 1293 1612 248 34 76 688 809 553 1619 0 0 0 0 0 0 0 0 0 0 0 0 0 287 140 1821 1439 1114 1365 202 35 58 490 670 412 1454 0 0 0 0 0 0 0 0 0 0 0 0 0 Sources: See Data Appendix Table 18 Reqression Results: Late Asset Growth and Bank Size of State-Chartered Bank Dependent Variable: Growth in total assets of state--chartered banks, 1920-1926 Independent variable Coefficient Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth in land values, 1920-1925 Business failure rate. 1921-1925 Business failure rate, 1917-1920 Presence of deposit insurance (excluding Nebraska) Out-of-city branch banking Within-city branch banking Standard Error Prob > 0.544 0.602 0.178 -0.404 0.037 0.450 0.235 0.098 0.346 0.541 0.239 0.018 0.084 0.254 0.946 -0.040 0.038 0.308 -0.190 0.126 0.146 0.179 0.204 0.124 0.132 0.163 0.136 R2 = 0.601 R2 = 0.462 Dependent Variable: Growth in total assets of state-chartered banks, 1920-1930 Independent variable Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth in land values, 1920-1930 Business failure rate. 1921-1929 Business failure rate, 1917-1920 Presence of deposit insurance (excluding Nebraska) Out-of-city branch banking Within-city branch banking R2 = 0.625 R2 = 0.495 Coefficient Standard Error Prob > 1T1 1.539 0.124 0.078 -0.936 -0.386 0.449 0.200 0.115 0.405 0.551 0.002 0.539 0.502 0.030 0.490 -0.072 0.044 0.118 -0.065 0.140 0.647 0.398 0.428 0.150 0.161 0.014 0.014 Table 18 (continued) Regression Results: Late Asset Growth and Bank Size of State-Chartered Bank Dependent Variable: Assets per bank for state-chartered banks In 1926 Independent variable Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth 1n land values, 1920-1925 Business failure rate. 1921-1925 Buslnes failure rate, 1917-1920 0ut-of-c1ty branch banking W1th1n-c1ty branch banking Prob > IT! Coefficient Standard Error 1341.96 0.101 0.084 -1782.05 -160.61 739.24 0.115 0.072 580.74 884.00 0.082 0.385 0.256 0.005 0.857 -40.55 60.60 0.510 593.49 540.64 198.93 257.51 0.007 0.047 R2 = 0.688 R2 = 0.597 Dependent Variable: Assets per bank for state-chartered banks 1n 1930 Independent variable Intercept National bank growth (Reserve center)x(Nat. bank growth) Ratio of farm to non-farm popul. Growth 1n land values, 1920-1930 Business failure rate. 1921-1929 Business failure rate, 1917-1920 0ut-of-c1ty branch banking W1th1n-c1ty branch banking R2 = 0.700 R2 = 0.612 Coefficient Standard Error Prob > IT! 1868.64 0.072 0.100 -2642.12 -375.72 847.17 0.128 0.079 725.33 952.96 0.037 0.577 0.219 -24.39 75.20 0.749 876.91 736.32 244.74 330.63 0.002 0.001 0.697 0.036 Table 19 Chains and Banks in Chain Systems, by States: June 30, 1929 Banks in chain systems Numbers of Chain systems Total Total 230 Total 1,561 National 596 State 965 State--Wide Branch Banking Permitted Total A 49 25 State 6 Arizona California Delaware District of Columbia Maryland North Carolina Rhode Island South Carolina Virginia Wyoming 30 1 5 20 10 10 Branches Restricted as to Location Total 61 337 135 202 z 1 4 11 4 10 5 33 71 4 6 2 19 3 4 3 14 68 12 17 1 9 3 49 111 6 38 10 22 58 3 12 6 27 53 3 25 4 State Kentucky Louisiana Maine Massachusetts Michigan Mississippi New Jersey New York Ohio Pennsylvania Tennessee Establishment of Branches Prohibited by Law Total 141 1,026 365 661 22 63 11 13 11 50 State Alabama Arkansas Table 19 (cont'd) State Colorado Connecticut Florida Georgia Idaho Illinois Indiana Iowa Kansas Minnesota Missouri Montana Nebraska Nevada New Mexico Oregon Texas Utah Washington West Virginia Wisconsin 2 13 8 5 4 6 3 11 1 12 10 34 4 2 9 2 1 6 6 5 11 32 23 23 81 3 92 55 279 26 15 63 14 6 32 37 50 62 13 8 7 20 2 33 15 130 7 4 15 2 4 14 7 12 26 19 15 16 61 1 59 40 149 19 11 48 12 2 18 30 38 36 5 35 14 21 Establishment of Branches Prohibited in Practice Total 20 149 71 78 7 8 5 60 59 30 20 41 10 40 18 20 State New Hampshire North Dakota Oklahoma South Dakota Vermont Sources: See Data Appendix. Table 20 Average Annual Bank Failure Rate: 32 "Aqricultural-Crisis" States 1918-1920 National State AFR1820 AFR1820 Alabama Arkansas Arizona Colorado Georgia Iowa Idaho Illinois Indiana Kansas Kentucky Louisiana Michigan Minnesota Missouri Mississippi Montana North Carolina North Dakota Nebraska New Mexico Nevada Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming .0000000 .0049751 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0011614 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0008985 .0009921 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0000000 .0014124 .0034364 .0000000 .0029240 .0031299 .0004424 .0000000 .0016317 .0017683 .0016617 .0007491 .0015798 .0004617 .0045537 .0021536 .0000000 .0000000 .0000000 .0009592 .0007491 .0000000 .0000000 .0004323 .0017889 .0000000 .0006588 .0000000 .0019685 .0033784 .0023474 .0000000 .0000000 Sources: See Data Appendix. 1921-1924 National State AFR2124 AFR2124 0.002475 0.003012 0.037500 0.007092 0.002688 0.002793 0.043210 0.000521 0.000984 0.004016 0.000000 0.006579 0.000000 0.005287 0.000000 0.008333 0.067241 0.008621 0.035912 0.017287 0.079787 0.000000 0.001351 0.017241 0.000000 0.040441 0.000000 0.005845 0.001515 0.005747 0.003311 0.042553 0.00697 0.01052 0.07090 0.02958 0.03098 0.01055 0.05674 0.00386 0.00533 0.01432 0.00667 0.01638 0.00169 0.01954 0.01022 0.01407 0.10140 0.01884 0.07103 0.01808 0.08882 0.01087 0.00130 0.04820 0.02309 0.05985 0.00389 0.01689 0.00982 0.01961 0.00336 0.06195 1925-1929 National State AFR2529 AFR2529 0.005714 0.013636 0.000000 0.015603 0.019149 0.043228 0.028571 0.004382 0.005645 0.005385 0.000000 0.000000 0.001653 0.031138 0.007463 0.005714 0.032258 0.019277 0.036364 0.016000 0.018182 0.000000 0.001671 0.014727 0.037037 0.048276 0.003704 0.006283 0.001099 0.0107T4 0.005161 0.016216 0.011673 0.029219 0.031111 0.024876 0.058007 0.027816 0.028037 0.004830 0.012164 0.021878 0.011416 0.014679 0.001038 0.039630 0.023275 0.014667 0.019355 0.024000 0.066922 0.042737 0.051163 0.000000 0.005906 0.028866 0.079394 0.058581 0.021166 0.013958 0.009169 0.011321 0.009615 0.050633 1921-1929 National State AFR2129 AFR2129 0.004400 0.009371 0.016667 0.011820 0.011947 0.024519 0.032922 0.002778 0.003500 0.004909 0.000000 0.002924 0.000992 0.019805 0.004085 0.007407 0.041379 0.014049 0.034377 0.015957 0.042553 0.000000 0.001502 0.017561 0.020325 0.040850 0.002268 0.006195 0.001347 0.010217 0.004415 0.026005 0.009739 0.020627 0.043118 0.023749 0.040168 0.018897 0.037037 0.004253 0.009607 0.017778 0.009630 0.015041 0.001353 0.028887 0.017150 0.014349 0.050894 0.021951 0.058651 0.029787 0.055556 0.004831 0.003742 0.031590 0.048666 0.052793 0.013827 0.014716 0.009735 0.014161 0.006919 0.047198 Table 21 1921-1930 Bank Liquidations (as of 1930) # completely liquidated Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming 2 2 3 8 5 17 2 1 14 4 0 1 0 13 2 2 38 13 0 12 4 21 3 25 8 16 1 21 1 4 2 8 National Banks___________ # in process repayment of liquid. ratio2 11 0 15 9 ii 4 31 12 69 10 3 0 4 43 3 10 15 15 0 2 17 38 7 26 16 34 4 22 3 5 6 2 1.00 .50 .87 .60 .51 .47 .76 .77 .69 .79 .NA .69 .NA .58 .88 .66 .34 .44 .NA .51 .71 .45 .66 .43 .51 .51 .93 .58 .90 .84 .30 .70 __ State Banks # completely # in liquidated process 9 4 37 62 120 28 9 6 182 1191 18 16 2 50 641 109 27 3171 0 18 2 3401 0 13 91 16 2421 12 13 81 4 1* 20 15 39 20 96 9 130 11 131 109 130 O1 41 19 8 245 oi 200 28 01 0 19 87 01 42 01 189 01 61 01 41 O1 40 13 repayment ratio2 .59 .91 .36 .68 .44 .49 .63 .88 .54 .531 .NA .41 .72 .48 .521 .53 .52 .351 .NA .70 1.00 .171 .NA .561 .66 .241 .83 .541 .57 .751 .66 .54 ^Insured Banks only. Includes liquidations completed after 1930. 2The repayment ratio is defined as the ratio of deposits repaid from asset liquidation for banks that were completely liquidated. Sources: See Data Appendix. Table 22 Average Time elapsed between closing and Completed Liquidation (for banks liquidated 1921-1930) Years Alabama Arizona Arkansas Colorado Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Mexico North Carolina North Dakota Ohio Oklahoma South Carolina South Dakota Tennessee Texas Virginia Washington Wisconsin Wyoming Sources: 3 4 2 4 3 4 3 7 3 3 NA 7 NA 4 1 3 4 4 NA 4 2 4 3 3 3 4 0 4 4 4 3 4 See Date Appendix National Banks Months 2 10 5 5 9 2 3 6 8 3 NA 2 NA 1 1 6 6 9 NA 5 11 0 5 8 10 8 11 0 8 2 8 10 # banks 2 2 3 8 5 17 2 1 14 4 NA 1 NA 13 2 2 38 13 NA 12 4 21 3 25 8 16 1 21 1 4 2 8 Years 3 3 2 2 3 4 3 4 3 3 3 4 6 5 6 3 4 6 NA 5 7 6 NA 5 3 5 3 3 3 4 3 3 State Banks Months 3 8 10 11 8 6 10 4 7 4 4 3 3 7 0 2 4 4 NA 0 1 3 NA 0 11 7 11 9 11 1 2 4 # banks 8 3 37 60 113 28 8 5 179 117 17 14 2 48 2 109 23 15 IMA 17 1 35 IMA 64 8 22 6 19 4 32 20 10 Table 23 Estimated Asset Shortfalls of Failed Banks Relative to Remaining-Bank Equity in MSevere-Failure*1 States National Banks Deposits of Nunrtber of Liq. Relative to Suspended Banks Suspensions 1921-30 Arizona Colorado Georgia Idaho Iowa Minnesota Montana Nebraska North Dakota Oklahoma South Carolina South Dakota Wyomi ng Sources: 1256 11003 16538 10601 55984 28338 16287 13695 17438 27364 12153 21109 9154 See Data Appendix .67 .94 .84 .81 .79 .97 .87 .80 .84 .72 .92 .93 .91 Estimated Avg. size Rate of Asset Shortfall of Liq. Bks Rel. to Susp. Shortfall .83 .45 .09 .65 .50 .59 .44 .94 .80 .70 .57 .60 .45 .50 .40 .49 .53 .31 .42 .66 .56 .55 .57 .49 .49 .30 349 1,862 613 2,958 6,855 6,812 4,115 5,767 6,445 7,861 3,123 5,772 1,125 State-Chartered Banks_________________________________ All Banks Total Bank Deposits Liq./ Size Rate of Estimated Total Bank Ratio of Shortfal1 Shortfall Equity Equity of Sus. Susp. Ratio Asset to Equity June 1930 Shortfall June 1930 Banks 1921-30 3,815 13,776 39,064 4,612 35,750 69,387 9,999 26,063 9,210 41,251 11,665 8,477 4,819 15,056 12,187 46,318 9,185 138,995 80,634 31,361 78,093 45,199 38,986 50,970 91,619 7,536 .80 .95 .75 .85 .75 .77 .89 .85 .92 .79 .91 .77 .80 .06 .95 .70 .63 .66 .47 .47 1.04 1.05 .28 .58 1.00 .48 .09 .32 .56 .51 .46 .52 .48 .65 .83 .44 .34 .76 .46 65 3,520 13,618 2,509 31,649 15,174 6,297 44,872 36,240 3,794 9,147 53,615 1,331 8,496 10,273 39,805 4,983 74,935 38,417 9,94/ 27,760 9,695 11,493 17,069 10,848 3,844 .03 .22 .18 .57 .35 .20 .52 .94 2,26 .22 ,43 3,07 .28