The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Ju ly /A u g u s t 1 9 9 2 Vol. 7 4 , No. 4 3 T h e E f f e c t o f L e g is la t in g P r o m p t C o r r e c t i v e A c t io n o n t h e B a n k In su ran ce Fund 2 3 T a r g e t i n g M 2 : T h e Is s u e o f M o n e t a r y C o n tro l 36 U n d e rsta n d in g th e T e r m S tr u c tu r e o f I n t e r e s t R a te s : T h e E x p e c t a t io n s T heory 5 1 T h e G r e a t D e p o s it I n s u r a n c e D e b a t e 78 T h e R esp on se o f M ark et In tere st R a t e s to D i s c o u n t R a t e C h a n g e s THE FEDERAL RESERVE BANK of ST. IX)l IIS 1 Federal R eserve Bank of St. Louis Review July/A ugust 1992 In This Issue . . . In the first article in this Review, "The Effects of Legislating Prompt Corrective Action on the Bank Insurance Fund,” R. Alton Gilbert inves tigates w hether recen t legislation is likely to reduce the losses of the Bank Insurance Fund (BIF). The Federal Deposit Insurance Corporation Im provem ent Act of 1991 m andates prom pt corrective action by the fed eral supervisors of insured depository institutions w hen the capital ra tios of these institutions fall to relatively low levels. The m andate for prompt corrective action is intended to reduce the losses of the BIF. Gilbert exam ines w h eth er this legislation is likely to have such an ef fect. The prompt corrective action mandate is based on the assumption that, the longer a bank rem ains in operation with a low capital ratio b e fore it fails, the larger the ratio of BIF loss to total assets. Gilbert shows that the data do not support this assumption. His evidence indicates that th ere is no relationship betw een the length of time banks operated with low capital ratios b efore they fail and the BIF’s loss ratios. These results raise doubt about w h eth er the recen t legislation will reduce the BIF’s losses. * * * In the second article in this issue, “Targeting M2: The Issue of Mone tary Control,” Daniel L. Thornton investigates the controllability of M2. Thornton notes that the existing structure of reserve requirem ents is such that the Fed has direct control over only the M l portion of M2, and he provides evidence that the Fed's ability to control the oth er com ponents of M2 indirectly, say, through in terest rates, has been essential ly nil. Consequently, the Fed can control M2 only through its control over M l. Because M l accounts for only 25 to 30 percent of M2, this means that, at times, M2 control can only be achieved w ith very large and potentially destabilizing changes in M l and reserves. W hile not endorsing such actions, Thornton outlines changes in the Federal Reserve’s system of reserve requirem ents that could enhance significantly the Fed’s ability to control M2. * * * The role of interest rates in the econom y has recently attracted a great deal of attention. One question that comes up frequently w hen in terest rates are discussed is: How are short-term and long-term rates related? The relationship betw een long- and short-term interest rates is called the “term stru ctu re.” In the third article in this issue, “Under standing the Term Structure of Interest Rates: The Expectations T h e ory,” Steven Russell describes the most popular theory of the term structure, the expectations theory. A fter laying out the building blocks of the expectations theory, Russell shows how the expectations of participants in financial m arkets and the JULY/AUGUST 1992 2 decisions they make create linkages betw een the m arket in terest rates on short- and long-term securities. Finally, Russell shows how the expec tations theory can be used to explain tw o im portant em pirical features of the interest rate term structure. * * * Federal deposit insurance is a defining feature of our nation’s financial landscape. For many years, deposit insurance was regarded as a trem en dous success. By protecting individual depositors, it discouraged banking panics, thus contributing greatly to m onetary stability. The painful ex periences of the 1980s have soured this ch eery assessm ent. Recent legis lation has made significant changes in deposit insurance, and many are calling for fu rth er reform s. As we assess the various options for reform , we can recall that fed er al deposit insurance was extrem ely controversial at its inception in the Banking Act of 1933. In the fourth article in this Review, “The Great Deposit Insurance D ebate,” M ark D. Flood re-exam ines the debate that surrounded the adoption of federal deposit insurance, first to see what the issues and argum ents w ere at the tim e and, second, to see how those issues w ere treated in the legislation. Flood finds that the legisla tors of 1933 both understood the difficulties with deposit insurance and incorporated in the legislation num erous provisions designed to mitigate those problems. * * * M arket interest rates sometimes respond to discount rate changes, while other times they do not. Policymakers, of course, would like to know why. In the final article in this Review, "The Response of Market Interest Rates to Discount Rate Changes,” Michael Dueker finds em piri cal evidence to suggest that the response of the three-m onth Treasury bill rate to a discount rate change varies with the magnitude of the dis count change, the Federal Reserve's operating procedure and the unem ployment rate. The latter factor, says the author, indicates that the m arket has com e to expect active policy steps from the Fed to cou nter act high unemployment. D ueker also investigates w h eth er the m arket can anticipate discount rate changes. His evidence suggests that the timing of discount rate changes is not easily predicted, so anticipations of discount rate changes do not appear to have m uch of an effect on m arket interest rates. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis * * * 3 R. Alton Gilbert R. Alton Gilbert is an assistant vice president at the Federal Reserve Bank of St. Louis. Richard I. Jako provided research assistance. The Effects of Legislating Prom pt Corrective Action on the Bank Insurance Fund T„e FEDERAL DEPOSIT Insurance Corpora tion Im provem ent Act of 1991 (hereafter, FDICIA) authorized m ore federal governm ent funds for the Federal Deposit Insurance Corpo ration and made m ajor changes in the supervi sion and regulation of depository institutions. One section of FDICIA requires supervisors to take prompt corrective action w hen an institu tion's capital ratio falls below the required lev el.1 Banks that are classified as well-capitalized or adequately capitalized are subject to the few est constraints on their activities (see table 1). Supervisors are required to impose limits on the activities of banks with relatively low capital ratios and to close them promptly if their capital ratios fall below some critical level. Some exam ples of the constraints on poorly capitalized banks include limits on their asset grow th, divi dends and various insider transactions. As FDICIA states, the purpose of prompt co r rective action is “to resolve the problem s of in ’ The legislation applies to the supervisors of commercial banks and thrift institutions. This paper refers exclusively to commercial banks and the effects of their failure on the Bank Insurance Fund. The Federal Deposit Insurance Cor poration (FDIC) insures the deposits of banks and savings and loan associations but maintains a separate fund for sured depository institutions at the least possible long-term loss to the deposit insurance fund.” The legislation is based on the assumption that losses to the Bank Insurance Fund (BIF) would have been low er in recen t years if supervisors had acted as required by FDICIA. This paper in vestigates w h eth er the evidence is consistent with the assumptions that underlie the case for this legislation. T H E CASE F O R LEGISLATING P R O M P T C O R R E C T IV E ACTION A few years ago, as part of a program to re form the supervision and regulation of depository institutions, several econom ists began promoting proposals for prompt corrective action (PCA) by supervisors.2 The report on financial reform by the Treasury D epartm ent in February 1991 in cluded a version of these early proposals.3 The General Accounting Office recom m ended a su- banks. Banks pay their premiums into the Bank Insurance Fund which then covers any losses when a bank fails. 2Brookings Institution (1989) and Shadow Financial Regula tory Committee (1989). d e pa rtm e nt of the Treasury (1991), pp. 39-41. JULY/AUGUST 1992 4 Table 1 Supervisory Actions Applicable to Depository Institutions under Provisions of the FDICIA for Prompt Corrective Action1___________________________________ Capital Category Mandatory Actions Well capitalized or adequately capitalized May not make any capital distribution or pay a management fee to a con trolling person that would leave the institution undercapitalized. Discretionary Actions None Undercapitalized Mandatory Actions Subject to provision applicable to well capitalized and adequately capital ized institutions. Subject to increased monitoring. Must submit an acceptable capital restoration plan within 45 days and im plement that plan. Growth of total assets must be restricted. Prior approval from the appropriate agency is required prior to acquisi tions, branching, and new lines of business. Discretionary Actions Subject to any discretionary actions applicable to significantly under capitalized institutions if the appropriate agency determines that those ac tions are necessary to carry out the purposes of PCA. Significantly undercapitalized http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Mandatory Actions Subject to all provisions applicable to undercapitalized institutions. Bonuses and raises to senior executive officers must be restricted. Subject to at least one of the discretionary actions for significantly under capitalized institutions. Discretionary Actions Actions the institution is presumed subject to unless the appropriate agen cy determines that such actions would not further the purposes of PCA: Must raise additional capital or arrange to be merged with another in stitution. Transactions with affiliates must be restricted by requiring compliance with section 23A of the Federal Reserve Act as if exemptions of that section did not apply. Interest rates paid on deposits must be restricted to prevailing rates in the region. Other discretionary actions: Severe restriction on asset growth or reduction of total assets may be required. Institution or its subsidiaries may be required to terminate, reduce, or alter any activity determined to pose excessive risk. May be required to hold a new election of its board of directors. 5 Table 1 (continued) Supervisory Actions Applicable to Depository Institutions under Provisions of the FDICIA for Prompt Corrective Action1 Capital Category Discretionary Actions Significantly undercapitalized (continued) Other discretionary actions (continued) Dismissal of any director or senior executive officer and their replace ment by new officers subject to agency approval may be required. May be prohibited from accepting deposits from correspondent deposi tory institutions. Controlling bank holding company may be prohibited from paying divi dends without prior Federal Reserve approval. May be required to divest or liquidate any subsidiary in danger of be coming insolvent and posing a significant risk to the institution. Any controlling company may be required to divest or liquidate any nondepository institution affiliate in danger of becoming insolvent and posing a significant risk to the institution. May be required to take any other actions that the appropriate agency determines would better carry out the purposes of PCA. Critically undercapitalized Mandatory Actions Must be placed in receivership within 90 days unless the appropriate agency and the FDIC concur that other action would better achieve the purposes of PCA. Must be placed in receivership if it continues to be critically undercapital ized, unless specific statutory requirements are met. After 60 days, must be prohibited from paying principal or interest on subordinated debt without prior approval of the FDIC. Activities must be restricted. At a minimum, may not do the following without the prior written approval of the FDIC: Enter into any material transaction other than in the usual course of business. Extend credit for any highly leveraged transaction. Make any material change in accounting methods. Engage in any “ covered transactions” as defined in section 23A of the Federal Reserve Act, which concerns affiliate transactions. Pay excessive compensation or bonuses. Pay interest on new or renewed liabilities at a rate that would cause the weighted average cost of funds to significantly exceed the prevailing rate in the institution’s market area. Discretionary Actions Additional restrictions (other than those mandated) may be placed on activities. 'This description of the mandatory and discretionary supervisory actions under PCA is derived from a proposal by the Board of Governors of the Federal Reserve System in July 1992 to implement the PCA provisions of FDICIA. Other regulations to be adopted by supervisors will make distinctions among institutions based on their capital category, including regulations on brokered deposits and interbank deposits. JULY/AUGUST 1992 6 pervisory system in w hich supervisors would be required to act based on certain indicators of the perform ance and behavior of depository in stitutions, as well as capital ratios.4 Proponents of legislating PCA, including the Treasury and others, have based their case for PCA largely on the incentive for banks to assume risk, not on evidence of the behavior of poorly capitalized banks. The recen t behavior of savings and loan associations provided most of the evi dence that depository institutions assumed greater risk as th eir capital ratios declined.5 The following quote illustrates the thinking of PCA advocates: As banks approach the point of economic in solvency, they have less and less to lose from pursuing aggressive, high-risk investment strate gies in an attempt to return to profitability. The supervisory free rein given undercapitalized thrifts during the 1980s is widely recognized as a leading factor contributing to the cost of resolving insolvent thrifts. Some argue that com mercial bank supervision has been far from per fect, too. In this view, banks are allowed to carry assets on their books at unrealistically optimis tic values and are not appropriately restrained from high-risk behavior and irresponsible divi dend policy.6 EVID ENCE ON T H E UNDERLYIN G ASSUM PTIO N S The direct method of determ ining w hether PCA legislation will reduce the BIF’s losses is to enact the legislation, then observe BIF losses for several years. W aiting several years to form an opinion about the effectiveness of PCA legisla tion, how ever, does not seem the best way. If PCA legislation turns out to be ineffective, we will have wasted valuable time during which m ore effective reform s could have been doing their job. This paper takes an indirect approach, specifying the assum ptions that underlie PCA legislation and determ ining w hether the b e 4U.S. General Accounting Office (1991), pp. 59-71. 5Barth, Bartholomew and Labich (1989) and Garcia (1988). d e p a rtm e n t of the Treasury (1991), pp. X-1 to X-2. 7Several studies examine the incentive for poorly capitalized institutions with deposit insurance to assume risk. See Buser, Chen and Kane (1981), Chirinko and Guill (1991) and Keeley and Furlong (1990). http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis havior of banks b efore FDICIA’s passage sup ports these assumptions. The case for PCA legis lation rests on the assumption that, in recent years, depository institutions assumed greater risk as th eir capital ratios declined. As poorly capitalized institutions assumed greater risk and failed, they added to the losses of the deposit insurance funds. Advocates of PCA legislation also assume that constraints on bank behavior mandated by PCA legislation will constrain the risk assumed by poorly capitalized institutions. The evidence that savings and loan associa tions assumed greater risk as their capital ratios declined, o f course, does not necessarily indicate that PCA legislation will reduce the BIF's losses. Commercial bank supervisors may simply have been m ore effective than the supervisors of sav ings and loan associations in constraining the risk assumed by poorly capitalized institutions.7 Recent studies examine w h eth er poorly capitalized banks have violated the types of con straints that will be imposed under PCA. Gilbert (1991) reported that the behavior of most of the banks with capital ratios below the minimum required level in 1985-89 did not violate such constraints.8 Large m ajorities of the banks reduced their assets while undercapitalized, refrained from paying dividends, and restrained loans to insiders. Recent studies of the “capital cru n ch ” report a positive association betw een the lagged capital ratios of banks and the grow th rates of their assets in the cu rren t period. These results are consistent with the view that supervisors effectively constrained the asset grow th of poorly capitalized banks.9 French (1991) found that, through reports by banks and exam inations, supervisors w ere able to detect the w eakness of most failed banks several years b efore failure. In addition, the in cidence of paying dividends was low er at poorly capitalized banks than at other banks, and the incidence of capital injections was higher. Horne (1991) presented additional evidence on the as sociation betw een capital ratios and dividends. 8Gilbert (1991) does not report observations on the banks that reduced their assets while undercapitalized. About 53 percent reduced their assets by more than 10 percent while undercapitalized, and about 22 percent reduced their assets by more than 25 percent. 9Bernanke and Lown (1991) and Peek and Rosengren (1992a, b). 7 Some banks paid dividends while their earnings w ere negative and capital ratios w ere below re quired levels, but the proportion of banks pay ing dividends is positively related to their capital ratios.10 These studies are consistent with the view that, in recent years, supervisors of com mercial banks influenced the behavior of most undercapitalized banks in ways that will be re quired under PCA legislation. T he exceptional cases may be eliminated by PCA legislation. One argum ent for PCA legislation is that the sanctions to be imposed on poorly capitalized banks will induce oth er banks to maintain their capital ratios above minimum required levels, to reduce the chance that they will be subject to the sanctions. The evidence, how ever, implies that most poorly capitalized banks w ere subject to the sanctions prior to PCA legislation. That legislation, therefore, is not incentive for banks to raise their capital ratios. T H E E F F E C T S O F C A PIT A L R A T IO S B E F O R E F A IL U R E ON B IF LO SS R A T IO S Even if PCA legislation has a limited impact on the behavior of banks while undercapitalized, it may achieve its basic objective of reducing BIF losses by reducing the length of time banks r e m ain poorly capitalized. The length of time a bank operates with a low capital ratio may in fluence the risk it assumes because it takes time for some non-m arketable bank assets to m ature 10Horne (1991) reported the results of an equation for predict ing the ratio of dividends to assets. In that model, profit rates and capital ratios have positive coefficients. "T h is paper does not consider all the possible effects of PCA legislation on BIF losses. It is possible that closing banks with low but positive capital ratios will increase BIF losses, for the following reasons: First, some banks eventu ally would recover with no losses to BIF. It is difficult to es timate the size of this effect with data for periods before FDICIA, since a change in the closure rule may change the behavior of other parties. Shareholders of the banks that ultimately recover may realize that their banks have good prospects and inject capital more quickly than they would have in the past. Second, some theoretical models indicate that an increase in the capital threshold at which banks are closed causes banks with certain characteristics to assume greater risk. See Levonian (1991). before the proceeds can be reinvested in higherrisk categories. By shortening the time banks are permitted to operate with low capital ratios, supervisors will limit their opportunities to act on incentives to assume greater risk.11 This ar gument rests on the assumption that there is a positive association betw een the length of time banks w ere poorly capitalized b efore failure and the BIF losses resulting from their failure. Measuring Capital Ratios B efore Failure To test the hypothesis that ratios of BIF losses to total assets are positively related to the length of time banks w ere poorly capitalized prior to their failure, one must specify the fol lowing: first, a m easure of capital, second, a criterion fo r classifying banks as poorly capital ized, and third, the lag betw een changes in cap ital ratios and changes in risk assumed by poorly capitalized banks.12 The paper uses tw o m easures of capital: equi ty and an alternative m easure, which adjusts eq uity for the m arket value of securities and for nonperform ing loans. The criterion for an ade quately capitalized bank is specified initially as a capital-to-asset ratio of 5 percent or more. This level is based on the maximum leverage ratio under the new risk-based capital requirem ents. For banks with relatively poor asset quality, su pervisors may specify a minimum ratio of Tier 1 capital (essentially the same as equity for most banks) to total assets as high as 5 percent. The A few banks are excluded because they did not report total assets one year before failure and because of other problems with missing data. Sixteen banks are excluded from the sample because they were involved in mergers within two years of their failure dates. Six bank holding companies in Texas had all of their bank subsidiaries closed at the same time, for a total of 88 failed banks. BIF losses attributed to at least some of these banks reflect problems at their affiliates. These 88 banks are excluded from the sample to avoid problems in relating BIF losses to the characteristics of individual failed banks. Thirty-nine banks were in existence less than three years when they failed. Since new banks tend to have relatively high capital ratios and rapid asset growth, these banks might distort the analysis as outliers in some comparisons. These 39 banks are retained in the sample. Effects of deleting these banks are noted where the difference would affect the description of the data. 12See Bovenzi and Murton (1988) for a description of loss es timates and an analysis of the determinants of FDIC losses from individual bank failures. The sample in this paper ex cludes savings banks insured by the BIF. Since savings banks hold different types of assets than commercial banks, the determinants of BIF losses for failed savings banks are likely to be different than for failed commercial banks. Thus, the sample includes only failed commercial banks. JULY/AUGUST 1992 8 analysis in this paper is modified to consider other capital ratios as w ell.13 Advocates of PCA legislation do not specify how quickly they assume poorly capitalized in stitutions increase their risk after their capital ratios decline. Rather than picking an arbitrary lag, we divide banks into three groups based on the length of time their equity capital ratios w ere below 5 percent b efore failure (table 2). Banks in group one had equity capital ratios b e low 5 percent for five or m ore consecutive quarters b efore failure. The choice of this peri od reflects seasonal patterns in bank accounting practices and capital injections. (Capital in jec tions and accounting entries that recognize loans as losses tend to be clustered in the fourth quarter.) A bank with a relatively low capital ratio for five or m ore quarters would have a relatively low capital ratio in m ore than one calendar year, no m atter w hen in the year a bank is declared a failed bank. Suppose, for instance, that a bank failed in February 1990. If the equity capital ratio of the bank was below 5 percent for five or m ore con secutive quarters, its ratio would have been b e low 5 percent at least as early as the fourth quarter of 1988. Thus, as early as then, the shareholders of the bank exhibited their inabili ty or unwillingness to inject the capital n eces sary to raise the ratio to 5 percent and did not eliminate the capital deficiency in subsequent quarters. Table 2 also includes an interm ediate group of banks that had relatively low equity capital ra tios betw een two and four consecutive quarters before failure (group two). If the groups in table 2 reflect relevant time periods, the argum ents for PCA legislation would imply that the BIF loss ratios would be highest for banks in group 1 and lowest for banks in group 3. A com parison of average ratios of BIF losses to total assets at the failure dates does reflect this pattern, but the d ifferences in the mean BIF loss ratios are not statistically significant. 13Spong (1990), pp. 64-71, and Keeton (1989) describe the risk-based capital requirements and maximum leverage ratios. 14See Spong (1990), pp. 64-71, for a description of the regu lation of bank dividends in the years covered by this study. In general, banks were prohibited from withdrawing or im pairing their capital through excessive dividend payouts or other means. Member banks (national banks and statechartered banks that are members of the Federal Reserve System) were required to obtain regulatory approval to pay http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Adjustment f o r Changes in Assets in the Last Year The com parisons of the ratios of BIF losses to total assets on the dates of their failure are sub ject to a bias. The longer capital ratios of banks w ere below 5 percent b efore failure, the larger the percentage decline in assets in their last year. Banks with equity capital ratios below 5 percent for five or m ore consecutive quarters had asset declines, on average, o f m ore than 14.5 percent. The average percentage decline in assets was m ore than 11 percent for banks with equity capital ratios below 5 percent for two to four consecutive quarters. The other banks, in contrast, had average asset grow th of about 2.5 percent. These d ifferences appear to reflect the in fluence of supervisors, based on the following assumptions. First, supervisors rate the financial strength of banks largely on the basis of capital ratios derived from the report of condition. Se cond, banks respond to directives from their su pervisors to raise capital ratios by reducing assets. And third, the longer a bank is subject to pressure from its supervisor to raise its capi tal ratio, the larger the percentage decline in its assets. Data on banks that paid dividends in the year ending on their failure date also appear to reflect the influence of supervisors, adding support to the view that supervisors influenced the asset grow th of undercapitalized banks in their last year. Bank regulations restrict dividend pay ments w henever capital is below the required level.14 W hile some undercapitalized banks have violated these regulations, most have foregone dividend payments. Less than 7 percent of the banks with equity capital ratios below 5 percent for five or m ore consecutive quarters b efore failure paid dividends in their last year. The proportion of failed banks that paid dividends in their last year is significantly higher for groups of banks with higher capital ratios in their last year. dividends that exceeded the sum of net profits for a year and retained earnings for the preceding two years. For any banks with federal deposit insurance, dividend payments that could endanger a bank could be restricted under the general enforcement and cease and desist powers of the federal supervisors. See Gilbert (1991), French (1991) and Horne (1991) for additional information on dividend pay ments by poorly capitalized banks. 9 Table 2 Distribution of BIF Loss Ratios by the Length of Time Before Failure That Capital Ratios Were Below 5 Percent, 1985-90 Loss to BIF divided by total assets Group number Characteristics of failed banks Number of banks Total assets as of failure date Total assets one year before failure date Percentage change in total assets in the year ending on failure date Percentage of banks that paid dividends in the year end ing on failure date 1 Equity capital ratio below 5 percent for five or more consecutive quarters before failure 374 0.2736 (0.1365) 0.2196 (0.1171) -14.52 (14.40) 2 Equity capital ratio below 5 percent in the last two quarters before failure and up to four consecutive quarters before failure 302 0.2693 (0.1184) 0.2145 (0.1022) -11.15 (14.07) 25.17 3 Failed banks other than those in groups 1 and 2 178 0.2629 (0.1320) 0.2522 (0.1536) 2.45 (23.47) 44.94 4 Alternative capital ratio below 5 percent for five or more consecutive quarters before failure 546 0.2716 (0.1313) 0.2200 (0.1142) -13.21 (14.54) 11.17 5 Alternative capital ratio below 5 percent in the last two quarters before failure and up to four consecutive quarters before failure 219 0.2752 (0.1226) 0.2247 (0.1078) -8.09 (15.97) 33.79 6 Failed banks other than those in groups 4 and 5 89 0.2456 (0.1320) 0.2649 (0.1807) 12.26 (28.23) 50.56 6.42% NOTE: Standard deviatons are in parentheses under means. t-statistics, in absolute value, for differences between means for groups: 1 and 2 0.438 1 and 3 0.880 2 and 3 0.533 4 and 5 4 and 6 5 and 6 0.360 1.724 1.820 0.604 2.506’ 2.916* 3.064* 8.884* 7.023* 6.695* 9.782* 4.406* 0.536 2.271 * 1.962’ 4.110* 8.333* 6.397* 6.521 * 7.203’ 2.046' 't-statistics, in absolute value, for differences in proportions ’ Statistically significant at the 5 percent level. JULY/AUGUST 1992 10 The observations in table 2 are consistent with the view that supervisors forced most banks with persistently low capital ratios before failure to reduce th eir assets and refrain from paying dividends. Supervisors may have been less aw are of the troubles of banks with capital ratios above 5 percent during most or all of their last year, and, th erefore, placed less con straint on their behavior. The higher average BIF loss ratios of the banks undercapitalized for longer periods may reflect sharp declines in assets in their last year, rath er than losses on investm ents in riskier as sets. BIF loss ratios can be adjusted for this bias by dividing the losses to BIF by assets one year b e fo r e failure. Average ratios of BIF losses to to tal assets one year b efore failure for banks in groups 1 and 2 are significantly lo w er than the average BIF loss ratio of those in group 3. After adjusting for the effects of this bias, the evi dence does not indicate a positive association betw een the length of time banks w ere under capitalized b efore failure and BIF loss ratios. An Alternative Capital Measure Advocates of PCA legislation have emphasized the need for im provem ents in m easuring the value of bank capital. Perhaps a positive rela tionship betw een BIF loss ratios and the length of tim e bank capital ratios w ere low b efore failure is evident only with an improved m eas ure of bank capital. Alternative capital m easures often are described as "m arket value” capital, with assets and liabili ties m arked to m arket values.15 Berger, King and O'Brien (1991) indicate the various m ean ings attached to the term "m arket value” and the practical difficulties in deriving accurate m easures of the m arket values for some catego ries of assets and liabilities. The authors sug gest, how ever, the following adjustm ents to the value of bank assets: adjust m arketable assets to m arket values, and adjust the value of loans for anticipated losses on nonperform ing loans. The following calculations yield an alternative capital m easure w hich reflects these adjust ments. The d ifference betw een the book and 15Mondschean (1992) discusses the issues raised by proposals for market value accounting. 16See the appendix for a more thorough discussion of the role of the allowance for loan losses in bank accounting principles. FEDERAL RESERVE BANK OF ST. LOUIS m arket value of securities is subtracted from equity. Adjustments to equity for anticipated loan losses involve com parisons of allowances for loan and lease losses to the values of nonperform ing loans (past due 90 days or longer or nonaccrual). T he allowance for loan losses is ac cumulated earnings of a bank set aside to ab sorb loan losses.16 Evidence in Berger, King and O’Brien indicates that a $3 increase in nonper form ing loans tends to increase loan losses by $1. If a bank’s allowance for loan losses equals or exceeds one-third of its nonperform ing loans, there is no adjustment to its equity for antici pated loan losses. The other banks need larger allowances for loan losses to m eet this standard. Increases in their allowances would com e out of equity. The adjustm ent to equity involves sub tracting one-third of their nonperform ing loans and adding their allowance for loan losses. The results in table 3 add support to use of the three-to-one ratio of nonperform ing loans to the allowance fo r loan losses in deriving the al ternative capital m easure. Table 3 presents this ratio for banks in various size categories, from one qu arter to eight quarters b efore failure. The ratio is around th ree fo r banks o f different size and for different lengths of time prior to failure. Table 3 also has implications for the supervi sory treatm ent of banks as they approach failure. As indicated above, the case for PCA legislation is based on the argum ent that in re cent years supervisors should have done their job differently. For example, supervisors should have forced banks to m ake th eir balance sheets reflect m ore accurately the value of their assets. Supervisors may have allowed troubled banks to show higher equity on their balance sheets than justified by the quality of their assets, by p er mitting their allowance fo r loan losses to lag b e hind the rise in their nonperform ing loans as they approached failure. Additions to the al lowance fo r loan losses (called provisions for loan losses) are bank expenses. Thus, additions to the allowance for loan losses reduce earnings and possibly equity, if earnings are negative. Table 3 shows that, while the ratio of nonper form ing loans to total assets rose as banks ap- 11 Table 3 Average Ratios of Nonperforming Loans to the Allowance for Loan and Lease Losses and to Total Assets1 Size category of banks (millions of dollars as of failure date) Assets < $25 NPL - ALLL NPL rr TA $25 < Assets < $50 NPL -h ALLL NPL - TA $50 < Assets < $100 NPL -r ALLL NPL ^ TA $100 < Assets NPL * ALLL NPL -r TA Quarters before failure 1 2 3 4 5 6 7 8 2.89 3.07 3.26 3.08 2.93 3.09 2.94 2.95 0.0777 0.0720 0.0677 0.0608 0.0540 0.0504 0.0431 0.0390 2.68 3.19 3.19 3.03 2.83 2.87 2.72 2.82 0.0892 0.0803 0.0703 0.0618 0.0539 0.0487 0.0443 0.0392 3.40 2.81 3.06 3.14 3.18 3.02 3.16 3.13 0.0949 0.0789 0.0717 0.0665 0.0587 0.0552 0.0487 0.0438 3.30 3.21 3.72 3.80 3.41 3.58 3.53 3.55 0.1049 0.0906 0.0808 0.0704 0.0595 0.0526 0.0495 0.0426 NPL — Nonperforming loans (past due 90 days or more plus nonaccrual) ALLL — Allowance for loan and lease losses TA — Total assets 'In total, 836 banks filed reports of condition for the quarter ending one quarter before failure and for the preceding seven quarters. The ratios are calculated as the sum of the item in the numerator divided by the sum of the item in the denominator for a given group of banks. proached failure, their allow ances for loan loss es also rose proportionately. These results are inconsistent with one type of forbearan ce by su pervisors: a general tendency to perm it the al lowance for loan losses to lag behind the rise in nonperform ing loans, to avoid large charges against equity. banks with adjusted capital ratios below 5 per cent for longer periods. Use of the alternative capital m easure d o e s not yield a positive associa tion betw een the length of time banks operated with low capital ratios before failure and BIF loss ratios. Table 2 presents average BIF loss ratios based on this alternative m easure of capital. The ad justm ents to equity reduce the capital ratios for many of the failed banks in their last year. For instance, the num ber of banks with capital ra tios below 5 percent for five or m ore consecu tive quarters before failure rises from 374 with equity as the m easure of capital (group 1) to 546 with the alternative m easure (group 4). Alternative Levels o f Capital Ratios BIF loss ratios adjusted for changes in assets in the last year (BIF losses divided by total as sets one y ear before failure) are lower for Perhaps the difficulty in finding an inverse relationship betw een capital ratios before failure and BIF loss ratios is that all the results in table 2 are based on a 5 percent capital ratio. The relevant ratio for purposes of the hypothesis tested here may be higher or low er than 5 per cent. Table 4 exam ines the relationship betw een capital ratios and BIF loss ratios, for a fixed lag of one year betw een the observation of capital ratios and failure dates. T he hypothesis that poorly capitalized banks assum e relatively high JULY/AUGUST 1992 12 Table 4 Distribution of BIF Loss Ratios by the Ratio of Capital to Assets One Year Before Failure Equity as the measure of capital Group number Range of capital ratio Number of banks BIF loss divided by total assets one year before failure Alternative capital measure Number of banks BIF loss divided by total assets one year before failure 1 0.10 < C/A 30 0.2861 (0.2141) 23 0.2898 (0.2364) 2 0.08 < C/A < 0.10 75 0.2306 (0.1346) 40 0.2523 (0.1575) 3 0.06 < C/A < 0.08 211 0.2214 (0.1116) 109 0.2179 (0.1064) 4 0.04 < C/A < 0.06 214 0.2290 (0.1189) 203 0.2281 (0.1112) 5 0.02 < C/A < 0.04 175 0.2177 (0.1181) 178 0.2344 (0.1300) 6 0.00 < C/A < 0.02 109 0.2129 (0.1120) 154 0.2000 (0.1054) 7 -0.01 < C/A < 0.00 15 0.1842 (0.0796) 54 0.2078 (0.1201) 8 C/A < -0 .0 1 25 0.2458 (0.1034) 93 0.2399 (0.1051) NOTE: Standard deviations are in parentheses under means. risk, which imposes large losses on BIF if they fail, implies higher BIF loss ratios for banks with capital ratios below some critical level b e fore failure. Table 4 indicates that the banks with the highest BIF loss ratios are those with the highest and the lowest capital ratios one year before failure. Among other banks, there is no system atic relationship betw een the capital ra tios of banks one year b efore failure and their BIF loss using either m easure of capital. These ' 7Banks in existence less than three years when they failed account for the relatively high average BIF loss ratio for banks with capital ratios in excess of 10 percent one year prior to failure. Eight of the 30 banks with equity capital ra tios in excess of 10 percent one year prior to failure were in existence less than three years when they failed. Ex cluding these eight banks reduces the average BIF loss ra tio for the remaining 22 banks to 23.72 percent, which is much closer to the average BIF loss ratios for the banks http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis results do not support the hypothesis that banks with capital ratios below some critical capital ra tio have higher BIF loss ratios.17 E x t r e m e C a s e s — A few banks that engaged in extrem e behavior may have imposed large losses on BIF. Thus, PCA legislation could con tribute to reducing BIF losses by constraining the extrem e behavior of a small m inority of failed banks. The data are examined for such extrem e cases in two ways. The first approach involves determining w hether BIF loss ratios with capital ratios below 10 percent one year prior to failure. Eliminating the banks in existence less than three years when they failed has a similar effect on the average BIF loss ratio of banks with ratios of the alternative capital measure to total assets in excess of 10 percent one year prior to failure. 13 Table 5 Characteristics of Banks with Relatively High BIF Loss Ratios Characteristics Banks with BIF loss ratios above 50 percent All banks in the sample 44 854 Mean percentage change in total assets in their last year -9.13% -9.79% Percentage that paid dividends in their last year 20.45 21.08 Percentage with equity capital ratio below 5 percent for five or more consecutive quarters before failure 54.55 43.79 Percentage in the West South Central region 75.00 56.21 Percentage supervised by the Office of the Comptroller of the Currency 56.82 37.70 Number of banks w ere relatively high among banks that engaged in extrem e behavior. These banks would have the following characteristics: equity capital ratio below 5 percent fo r five or m ore consecutive quarters b efore failure, and asset grow th and dividend payments in their last year. No banks in the sample had this com bination of ch arac teristics. The second approach involves examining the characteristics of banks with relatively high BIF loss ratios, to determ ine w h eth er they exhibited extrem e behavior that will be constrained under PCA. Table 5 presents some of the ch aracteris tics of 44 banks with BIF loss ratios that exceed 50 percent. Their m ean asset grow th and the proportion paying dividends in their last year are almost identical to those fo r the entire sam ple. The banks with relatively high BIF loss ra tios do have a somewhat higher percentage with equity capital ratios below 5 percent for relatively long periods b efore failure. It is possi ble, how ever, to find other ways in which these banks are even m ore distinct from the entire sample. T h eir relatively high loss ratios may reflect regional effects: three-fourths w ere locat ed in the W est South Central region of the na tion, com pared with about 56 percent for the entire sample.18 A relatively high proportion w ere supervised by the Comptroller of the Cur rency. Thus, an examination of extrem e cases does not provide clear evidence of the effective ness of PCA in reducing BIF losses. R EG R ESSIO N ANALYSIS Loss ratios vary substantially within each of the groups of banks in tables 2 and 4; standard deviations are about half as large as their me ans. Perhaps an inverse relationship betw een capital ratios before failure and BIF loss ratios is evident only if oth er factors are held constant in regression analysis. A Description o f Banks in the Regression Analysis The 854 banks in the sample failed in the years 1985-90 (table 6). Most banks w ere rela tively small: about 60 percent had total assets 18States in this region are Arkansas, Louisiana, Oklahoma and Texas. JULY/AUGUST 1992 14 Table 6 Characteristics of Failed Banks in Regression Analysis Year of failure Number of banks Percentage 13.1% 15.5 20.5 17.1 17.4 16.4 1985 1986 1987 1988 1989 1990 112 132 175 146 149 140 Total 854 100.0 508 209 90 47 59.5 24.5 10.5 5.5 Asset size on failure date (millions of dollars) Assets < $25 $25 < Assets < $50 $50 < Assets < $100 $100 < Assets 100.0 Region New England (NE) Middle Atlantic (MA) South Atlantic (SA) East South Central (ESC) West South Central (WSC) East North Central (ENC) West North Central (WNC) Pacific Northwest (PNW) Pacific Southwest (PSW) 5 9 19 17 480 16 174 34 100 0.6 1.1 2.2 2.0 56.2 1.9 20.4 4.0 11.7 100.1 Federal supervisor occ Federal Reserve FDIC 322 68 464 37.7 8.0 54.3 100.0 Method of resolving failure Purchase and assumption Transfer of insured deposits Liquidation 667 115 72 78.1 13.5 8.4 100.0 NOTE: States in census regions: New England: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont Middle Atlantic: New Jersey, New York and Pennsylvania South Atlantic: Delaware, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia East South Central: Alabama, Kentucky, Mississippi and Tennessee West South Central: Arkansas, Louisiana, Oklahoma and Texas East North Central: Illinois, Indiana, Ohio, Michigan and Wisconsin West North Central: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota Pacific Northwest: Alaska, Idaho, Montana, Oregon, Washingon and Wyoming Pacific Southwest: Arizona, California, Colorado, Hawaii, Nevada, New Mexico and Utah http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 15 less than $25 million, and about 95 percent had total assets less than $100 million. The failed banks w ere heavily concentrated in certain regions. About 56 percent w ere in the W est South Central region. About 78 percent of the cases w ere resolved w hen oth er banks bought some of the assets of the failed banks and as sumed their liabilities. In another 14 percent of the cases, the FDIC tran sferred the insured deposits of failed banks to other banks. In these cases, the FDIC liquidated the failed banks’ as sets and made partial payments to uninsured depositors, based on the proceeds of liquidated assets. Failed banks w ere liquidated in the r e maining cases. Identifying the Variables The dependent variable is the ratio of BIF loss to total assets as of failure date.19 Independent variables are described in table 7. Capital Ratios — The case for applying PCA legislation to the supervisors of com m ercial banks implies negative, significant coefficients on the capital ratios lagged one year, EC 4 and AC 4. Asset Growth — The coefficient on GROWTH is assumed to have a negative sign: an increase (decrease) in assets in the last year is assumed to increase (decrease) the denom inator of the BIF loss ratio, while having little, if any, effect on the size of the BIF loss. Dividends — Arguments for legislating PCA imply a positive sign for the coefficient on DIV: dividends in the last year, divided by total assets as of failure date. The coefficient on DIV may be positive for two reasons. First, dividends are pay m ents of capital to shareholders, leaving less capital to absorb reductions in the value of as sets. Second, dividends may be a signal that the 19Avery, Hanweck and Kwast (1985) report the results of regressions with the same dependent variable. It is difficult to compare the results in this paper to those, since their objective was to predict FDIC losses from bank failures, not to test hypotheses about coefficients on independent variables. They do not attempt to adjust the specification of equations for possible collinearity. In Bovenzi and Murton (1988) and James (1991), the dependent variable is the loss on assets of failed banks, a concept that is related to BIF loss. Some of the independent variables in Bovenzi and Murton and in James are included, with slight modifica tions, in this study; the major difference involves measures of asset quality derived from examination reports, which are not included in this study. Barth, Bartholomew and Labich (1989) and Barth, Bartholomew and Bradley (1990) estimate the coefficients of equations designed to explain the cost to the Federal Savings and Loan Insurance Cor shareholders saw little reason to attempt to p re vent failure. Instead, they may have paid out capital in anticipation of failure. These reasons, however, do not account for possible influences of supervisors over w hich banks paid dividends or the size of their dividend payments. Quality o f Bank Loans — One m easure of loan quality is the value of loans that are past due or nonaccrual. A second m easure is the value of interest accrued on loans that was not collected. W hen borrow ers fall behind on their scheduled payments, banks continue to accrue the interest due from them as income until their loans are classified as nonaccrual.20 These m easures of loan quality may help ex plain the BIF losses from the failure of individu al banks. The following two m easures of asset quality are included as independent variables: 1. NPL — the ratio of nonperform ing loans to total assets. 2. ACCRUED — interest accrued on loans that was not collected, divided by total assets. The coefficients on these variables will have positive signs under the following assumptions: First, these m easures accurately reflect loan quality. Second, the allowance for loan losses is not large enough to cover the gap betw een the book value of these loans and their value to the FDIC as the receiv er of failed banks.21 Market Value o f Securities — Securities (various types of bonds) are reported on bank balance sheets at book values (purchase prices plus any am ortized changes in value), not at their cu rren t m arket values. Thus, the book value of equity reflects the book value of securi ties. Banks also rep ort inform ation on the m ar ket value of their securities on the report of condition. The following independent variable is a m easure of the gap betw een the book and poration of resolving cases of failed savings and loan as sociations. Results in Barth, Bartholomew and Bradley are not comparable to those in this study, since they include observations for failed and surviving associations and use a different statistical technique (Tobit regression analysis). 20Accrued interest that was not collected may not reflect default by borrowers on scheduled loan payments. In some loan contracts, such as construction loans, the original loan contract specifies a delayed schedule of interest payments. 21See the appendix for a discussion of accounting principles which features the role of the allowance for loan losses. JULY/AUGUST 1992 16 Table 7 •'ji O UJ Ratio of equity capital to total assets four quarters before failure. > 0 1 A Identification of Independent Variables Ratio of the alternative capital measure to total assets four quarters before failure. GROWTH Change in total assets of failed bank in its last year, divided by total assets as of failure date. DIV Dividends on common stock paid in the year ending in failure, divided by total assets as of failure date. NPL Loans and leases past due 90 days or more, plus nonaccrual loans, divided by total assets as of failure date. ACCRUED Interest on loans that was accrued but not received on the last report of condi tion, divided by total assets as of failure date. MARKET Book value of securities in the investment account as of the last report of condi tion, minus the market value of the securities, divided by total assets as of failure date. IDR Last observation available on deposits in accounts up to $100,000 each, divided by total assets as of failure date. P&A Dummy variable with a value of unity if a failed bank case was resolved through purchase and assumption, zero otherwise. TID Dummy variable with a value of unity if a failed bank case was resolved through transfer of insured deposits to another bank, zero otherwise. OCC Dummy variable with a value of unity if the bank was a national bank, super vised by the Office of the Comptroller of the Currency, zero otherwise. FR Dummy variable with a value of unity if a bank was supervised by the Federal Reserve, zero otherwise. InA Natural log of total assets as of failure date. 1985-1989 Dummy variables for the years in which the banks failed. NE, MA, SA, ESC, ENC, WNC, PNW, PSW Dummy variables for the regions in which failed banks were located. m arket value of securities: MARKET — the book value minus the m arket value of securities, divided by total assets. The expected sign of the coefficient on MAR KET depends on the conditions under w hich su pervisors close banks. Suppose they close banks w hen the book value of equity is zero or nega tive, without adjustm ents to the book value of equity for the m arket value of assets. Under this assumption, the expected sign on MARKET is positive: BIF losses would be related positively http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis to the gap betw een the book value and the m ar ket value of securities. Methods o f Resolving Failed Banks — W hen a bank fails, the FDIC becom es the receiver. As receiver, the FDIC must dispose of the failed bank’s assets and make payments to its creditors. The options chosen to resolve each case may affect the BIF's losses. Those choices, in turn, may reflect additional inform ation about failed banks not captured by the other indepen dent variables, such as characteristics of the 17 custom ers of failed banks that m ake them valu able to oth er banks.22 One method of resolving failed bank cases is liquidation. Failed banks are closed and deposi tors are paid off up to the insurance limit per account. The FDIC liquidates the assets and makes payments to uninsured depositors and oth er creditors of the failed bank. Shareholders generally get nothing. Resolution methods other than liquidation may be less expensive to BIF. In many cases, a solvent bank purchases some of the assets of a failed bank and assumes its liabilities. The FDIC provides cash to cover the gap betw een assets purchased and liabilities assumed. This is called a p u rch a se an d assu m ption (P&A) transaction. The FDIC solicits bids from solvent banks for the assets and liabilities. Banks bid by offering premiums; the cash payment by the FDIC to the bank with the winning bid is net of the prem i um. The FDIC generally disposes of failed banks through P&A transactions if its staff estim ates that the losses would be lower than under liqui dation.23 As a result, the variable P&A (dummy variable for banks resolved through P&A tran s actions) is expected to have a negative coefficient. In some cases, the FDIC liquidates the assets o f failed banks but solicits bids from other banks to assume their insured deposits. Bidders may anticipate long-term profits on the accounts of custom ers who choose to keep their deposits with the winning bidder. This method of dispos ing of failed banks is called tra n sfer o f in su red d ep o sits (TID). The independent variable TID (dummy variable for bank failure cases resolved through TID) is expected to have a negative coefficient. Share o f Deposits Fully Insured — Jam es (1991) found a positive association betw een the premiums paid by the winning bidders in P&A cases and the shares of deposits of failed banks that w ere fully insured (accounts in denomina tions of $100,000 or less). The sm aller accounts tend to be more profitable to banks because banks pay less than m arket interest rates on them .24 22The appendix examines in more detail how resolution methods affect BIF losses. 23For a discussion of the conditions for disposing of failed banks through P&A transactions, see Federal Deposit In surance Corporation (1984), pp. 81-108, Bovenzi and Muldoon (1990) and Department of the Treasury (1991), pp. I-30 through 1-51. The variable IDR (fully insured deposits divid ed by total assets) is included to reflect the com position of deposits. It is expected to have a negative coefficient because premiums paid to the FDIC by winning bidders are assumed to be positively related to IDR. An increase in the premium reduces the loss to BIF. Federal Supervisory Agency — The prim ary supervisor of nationally chartered banks is the Office of the Comptroller of the Currency (OCC). For state-chartered banks that are m em bers of the Federal Reserve System, the Federal Reserve is the prim ary federal supervi sory agency, while, for other state banks, it is the FDIC. D ifferences in supervisory practices among these agencies may affect BIF losses. Dummy variables (OCC and FR) are used to cap ture such effects. Bank Siz.e — BIF loss ratios may be higher for smaller banks for two reasons. First, Jam es (1991) finds that FDIC administrative costs are higher, per dollar of assets, for smaller failed banks.25 Second, smaller banks may be subject to less frequent exam ination and less thorough surveillance betw een examinations than larger banks. W hen supervisors discover that relatively small banks are bankrupt, the percentage losses on assets may be larger than w hen larger banks fail. The bank size variable is the natural log of total assets as of failure date. Location and Year o f Failure — The re maining independent variables are dummy vari ables for the regions of failed banks and the years in w hich they failed, since BIF loss ratios may vary systematically by region and year of failure. Regression Results Table 8 presents the regression results. The equations use different m easures of capital in the lagged capital ratio. Lagged Capital Ratios — The coefficients on capital ratios four quarters before failure are not statistically significant. O ther m easures yield the same result. In oth er regressions not rep ort ed here, the coefficients on dummy variables 24See Brunner, Duca and McLaughlin (1991) for information on the rates banks pay on various types of deposit ac counts. “ James (1991), pp. 1234-36. JULY/AUGUST 1992 18 for banks with capital ratios below 5 percent for various lengths of time b efore failure also are not statistically significant.26 The coefficients on the variables designed to reflect capital ratios before failure may be bi ased toward zero by including independent vari ables that reflect the quality and m arket value of bank assets. To illustrate, suppose the banks with persistently low capital ratios shifted their assets to high-risk categories as they approached failure, resulting in high ratios of nonperform ing loans to total assets on their last reports of condition. In addition, suppose these banks sold securities w ith capital gains and kept securities with capital losses to boost the book value of equity as they approached failure. This selective pattern of securities sales would make values of the variable MARKET relatively high at the banks with persistently low capital ratios. The effects of low capital ratios b efore failure on BIF loss ratios would be captured to some ex tent in the coefficients on NPL, ACCRUED and MARKET. To test for this bias, equations 1 and 2 of table 8 w ere estim ated w ithout the varia bles NPL, ACCRUED and MARKET. In results not reported here, the coefficients on capital ra tios b efore failure w ere not statistically sig nificant. O t h e r I n d e p e n d e n t V a r i a b l e s — The coeffi cient on GROWTH is negative, as hypothesized. The coefficient on DIV is negative and insignifi cant; advocates of PCA legislation implied it would have been positive. 26The most comparable results for S&Ls are in Barth, Bar tholomew and Labich (1989). In a regression equation with costs of resolving failed S&Ls as the dependent variable, tangible net worth on the last quarter reported is a highly significant variable. The coefficient is negative unity (a $1 increase in capital reduces resolution costs by $1), with a t-statistic of 13.9. Another significant variable is the number of months an association was insolvent before failure, which has a positive coefficient. The contrast of the results in this paper to those in Barth, Bartholomew and Labich is con sistent with the view that the supervisors of commercial banks were more effective in limiting the risk assumed by poorly capitalized institutions than the supervisors of S&Ls. 27Bovenzi and Murton (1988) find that, without holding other factors constant, BIF loss ratios were about 7 percentage points lower in P&A cases than in liquidation cases in 1985-86. The coefficient on P&A in table 8 indicates about the same effect. 28Gilbert (1991) found differences in the behavior of banks in Texas with national charters and those with state charters that could be interpreted as evidence of differences in practices among the federal supervisory agencies. National banks were allowed to operate with capital ratios below the minimum capital requirement for longer periods than state-chartered banks, and national banks accounted for http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis The coefficients on NPL and ACCRUED are significant with the positive signs, as hypothe sized. The coefficient on MARKET is significant but the sign is opposite of that hypothesized: a wider gap betw een the book value and m arket value of securities is associated with a low er BIF loss. The negative, significant coefficient on IDR in dicates that failed banks with higher ratios of fully insured deposits to total assets are m ore valuable to potential bidders, thus tending to reduce BIF loss ratios. T he coefficient on P&A indicates that BIF loss ratios are low er in P&A cases than in liquidation cases, holding other variables constant.27 BIF loss ratios are not sig nificantly low er in TID cases. The coefficient on OCC is positive and statistically significant. Hold ing constant the influences of the other in dependent variables, BIF loss ratios are about 2 percentage points higher for failed banks with national ch arte rs.28 T he coefficient on FR indi cates that, among state-chartered banks, th ere is no significant effect of Federal Reserve m em ber ship on loss ratios, holding constant the oth er independent variables. The coefficient on the natural log of assets is not statistically significant. In oth er regressions not reported here, dummy variables for banks in various size ranges also w ere not significant. The results do not support the hypothesis that BIF loss ratios are larger for sm aller banks, holding constant other determ inants of BIF loss ratios. almost all of the Texas banks that operated at least a year with negative equity. The undercapitalized banks in Texas with rapid assets growth and those with higher insider loans while undercapitalized tended to be national banks. Most of these differences between national and statechartered banks were not statistically significant outside Texas. These contrasts might indicate that the positive, signifi cant coefficients on OCC in table 8 reflect differences be tween national and state-chartered banks in the Southwest. To test for such a regional effect, the regressions in table 8 were estimated separately for banks in the states covered by the Dallas office of the OCC (Arkansas, Louisiana, New Mexico, Oklahoma and Texas) and for banks in other states. In each regression, the coefficient on OCC was positive but not significant at the 5 percent level. The coefficient on OCC was larger, however, in the regressions for banks in states outside the Southwest and significant at the 10 percent level. Thus, the effect on BIF loss ratios of supervision by the OCC is not restricted to the Southwest. 19 Table 8 Determinants of Bank Insurance Fund Losses Due to Individual Bank Failures Dependent variable: Bank Insurance Fund loss divided by total assets as of failure date Regression Number Independent variables Intercept EC-4 ac 1 0.3539 * (5.69) Regression Number 1 2 1985 -0.0207 (1.18) -0.0200 (1.16) 1986 -0.0028 (0.18) -0.0034 (0.22) -0.0021 (0.02) 1987 0.0054 (0.38) 0.0048 (0.33) 2 0.3495 * (5.69) -0.0324 (0.22) _4 Independent variables GROWTH -0.0442 * (2.64) -0.0451 * (2.73) 1988 0.0214 (1.53) 0.0211 (1.50) DIV -1.4038 (1.34) -1.42 (1.37) 1989 0.0255 (1.87) 0.0255 (1.87) NPL 0.3554 * (4.74) 0.3533 * (4.69) NE -0.0544 (1.04) -0.0550 (1.05) ACCRUED 3.2125 * (6.22) 3.2210 * (6.24) MA -0.0732 (1.86) -0.0732 (1.86) MARKET -1.3307 * (2.31) -1.2988 * (2.25) SA -0.0693 * (2.53) -0.0689 * (2.51) IDR -0.0855 * (3.50) -0.0848 * (3.46) ESC -0.0883 * (304) -0.0877 * (3.02) P&A -0.0656 * (4.40) -0.0651 * (4.35) ENC -0.1069 * (3.60) -0.1066 * (3.60) TID -0.0024 (0.13) -0.0021 (0.12) WNC -0.0904 * (7.29) -0.0904 * (7.30) OCC 0.0218 * (2.39) 0.0222 * (2.45) PNW -0.0497 * (2.36) -0.0498 * (2.37) FR 0.0179 (1.13) 0.0178 (1.12) PSW -0.0659 * (4.99) -0.0662 * (5.03) InA -0.0014 (0.29) -0.0012 (0.25) 0.2290 0.2291 R2 N 854 854 Statistically significant at the 5 percent level. NOTE: t-sta tistics are in parentheses under regression coefficients. The coefficients on dummy variables for in dividual years are not statistically significant. Coefficients on several regional dummy varia bles are negative and significant. The excluded region is the W est South Central region. The negative coefficients on some o f the regional dummy variables indicate that, holding constant other independent variables, loss ratios are sig nificantly low er for banks in several regions than for banks in the W est South Central region. CONCLUSIONS The main reason for legislating prompt co r rective action (PCA) is to reduce losses to deposit insurance funds. The case for such legislation rests on the following assumptions: JULY/AUGUST 1992 20 First, depository institutions have an incentive to assume greater risk as their capital ratios decline. Second, the longer an institution oper ates with a low capital ratio, the greater its op portunity to act on incentives to assume risk. Third, supervisors have been ineffective in limit ing the risk assumed by poorly capitalized insti tutions. Fourth, the insurance fund losses due to the failure of individual institutions reflect, to some extent, the risk assumed by these institu tions after they becam e poorly capitalized. And fifth, the actions mandated for supervisors in the legislation will constrain the risk assumed by poorly capitalized institutions, thereby limit ing insurance fund losses if they fail. This paper considers the likely effects of PCA legislation on BIF losses resulting from the failure of com m ercial banks. The method in volves exam ining w hether the evidence about com m ercial bank behavior and BIF losses sup port the assumptions that underlie the case for PCA legislation. The assumptions imply that the longer a bank operates with a low capital ratio b efore failure, the larger the BIF loss. The evidence does not support this hypothe sis. The evidence, instead, is consistent with the hypothesis that, in recen t years, supervisors have been effective in constraining the risk as sumed by poorly capitalized banks. These results raise doubts about w hether PCA legisla tion will reduce BIF losses. REFEREN C ES Avery, Robert B., Gerald A. Hanweck, and Myron L. Kwast. “An Analysis of Risk-Based Deposit Insurance for Commer cial Banks,” Proceedings of a Conference on Bank Struc ture and Competition, May 1-3, 1985 (Federal Reserve Bank of Chicago), pp. 217-50. Barth, James R., Philip F. Bartholomew, and Carol J. Labich. “ Moral Hazard and the Thrift Crisis: An Analysis of 1988 Resolutions,” Proceedings of a Conference on Bank Struc ture and Competition, May 3-5, 1989 (Federal Reserve Bank of Chicago), pp. 344-84. Barth, James R., Philip F. Bartholomew, and Michael G. Bradley. “ Determinants of Thrift Institution Resolution Costs,” Journal of Finance (July 1990), pp. 731-54. Berger, Allen N., Kathleen Kuester King, and James M. O'Brien. “ The Limitations of Market Value Accounting and a More Realistic Alternative,” Journal of Banking and Finance (September 1991), pp. 753-83. Bernanke, Ben S., and Cara S. Lown. “ The Credit Crunch,” Brookings Papers on Economic Activity, (2:1991), pp. 205-39. Bovenzi, John F., and Maureen E. Muldoon. “ FailureResolution Methods and Policy Considerations,” FDIC Banking Review (Fall 1990), pp. 1-11. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Bovenzi, John F., and Arthur J. Murton. “ Resolution Costs of Bank Failures,” FDIC Banking Review (Fall 1988), pp. 1-13. Brookings Institution. Blueprint for Restructuring America’s Financial Institutions—Report of a Task Force (1989). Brunner, Allan D., John V. Duca, and Mary M. McLaughlin. “ Recent Developments Affecting the Profitability and Prac tices of Commercial Banks,” Federal Reserve Bulletin (July 1991), pp. 505-27. Buser, Stephen A., Andrew H. Chen, and Edward J. Kane. “ Federal Deposit Insurance, Regulatory Policy, and Optimal Bank Capital,” Journal of Finance (March 1981), pp. 51-60. Chirinko, Robert S., and Gene D. Guill. "A Framework for As sessing Credit Risk in Depository Institutions: Toward Regulatory Reform,” Journal of Banking and Finance (Sep tember 1991), pp. 785-804. Department of the Treasury. Modernizing the Financial System (February 1991). Federal Deposit Insurance Corporation. The First Fifty Years: A History of the FDIC, 1933-83 (1984). French, George E. “ Early Corrective Action for Troubled Banks,” FDIC Banking Review (Fall 1991), pp. 1-12. Garcia, Gillian. “ The FSLIC is ‘Broke’ in More Ways Than One,” Cato Journal (Winter 1988), pp. 727-41. Gilbert, R. Alton. “ Supervision of Undercapitalized Banks: Is There a Case for Change?” this Review (May/June 1991), pp. 16-30. Horne, David K. “ Bank Dividend Patterns,” FDIC Banking Review (Fall 1991), pp. 13-24. James, Christopher. “ The Losses Realized in Bank Failures,” Journal of Finance (September 1991), pp. 1223-42. Keeley, Michael C., and Frederick T. Furlong. "A Reexamina tion of Mean-Variance Analysis of Bank Capital Regula tion,” Journal of Banking and Finance (March 1990), pp. 69-84. Keeton, William R. “ The New Risk-Based Capital Plan for Commercial Banks,” Federal Reserve Bank of Kansas City Economic Review (December 1989), pp. 40-60. Levonian, Mark E. “ What Happens if Banks are Closed ‘Early’?” Proceedings of a Conference on Bank Structure and Competition, May 1-3, 1991 (Federal Reserve Bank of Chicago), pp. 273-321. Mondschean, Thomas. “ Market Value Accounting for Com mercial Banks,” Federal Reserve Bank of Chicago Econom ic Perspectives (January/February 1992), pp. 16-31. Peek, Joe, and Eric Rosengren. “ The Capital Crunch: Neither a Borrower Nor a Lender Be,” Proceedings of a Conference on Bank Structure and Competition, May 6-8, 1992a (Federal Reserve Bank of Chicago). ________“ The Capital Crunch in New England,” Federal Reserve Bank of Boston, New England Economic Review (May/June 1992b), pp. 21-31. Shadow Financial Regulatory Committee. “An Outline of a Program for Deposit Insurance and Regulatory Reform,” Statement No. 41, February 13, 1989, in George G. Kauf man, ed., Restructuring the American Financial System (Kluwer Academic Publishers, 1990), pp. 163-68. Spong, Kenneth. Banking Regulation: Its Purposes, Implemen tation and Effects, 3rd. ed. (Federal Reserve Bank of Kan sas City, 1990). U.S. General Accounting Office. Deposit Insurance: A Strategy for Reform (March 1991). Walter, John R. “ Loan Loss Reserves,” Federal Reserve Bank of Richmond Economic Review (July/August 1991), pp. 20-30. 21 A p p en d ix An In tro d u c tio n to B a n k A c c o u n tin g a n d th e FDIC’s P r a c ti c e s in R e s o lv in g F a ile d B a n k s The text assumes a basic understanding of bank accounting principles and the methods used by the FDIC in resolving failed banks. This ap pendix provides an introduction to these topics. the study up to two years b efore their failure. The bank could absorb loan losses up to $2 without reducing equity. The ratio of equity to total assets is above 5 percent. The accounting principles can be illustrated by referrin g to the balance sheets of a hypo thetical bank. Items in table A1 reflect book rath er than m arket values. For instance, the book value of loans is the sum of the outstand ing balances that borrow ers owe the bank, other than the loans that have been declared losses. Values of m arketable securities are book values, not cu rren t m arket values. The financial condition of the bank would look w orse if securities w ere marked to their m arket value of $35. Net w orth actually would be zero. One of the key balance sheet items for our purposes is the allowance for loan and lease losses, w hich represents an accum ulation of past earnings set aside to absorb anticipated fu ture losses on loans that becom e uncollectable. In accounting statem ents filed with bank super visors, the allowance fo r loan losses is reported on the asset side of the balance sheet as a deduc tion from loans. Thus, net loans are net of an ticipated losses, as reflected in the allowance. W hen a bank cannot collect from a borrow er, accounting principles indicate that management is to declare the loan a loss and charge the loss against the allowance fo r loan losses. The ac counting entries involve reductions in both loans and the allow ance.1 Increases in the allowance for loan losses com e out of current earnings. The relevant item in the incom e statem ent is called the "provision for loan losses,” which is included among bank expenses. If a bank must make a large provision fo r loan losses in a given period, because of ac tual or anticipated loan losses, cu rren t earnings may be negative. W hen cu rren t earnings are negative, equity is reduced. The top half of table A1 presents the balance sheet of a solvent bank, based on book value ac counting. Securities are recorded at their book value of $40. The allowance for loan losses is one-third of nonperform ing loans, which the text indicates is about average for the banks in 1See Walter (1991) for a thorough discussion of the al lowance for loan losses. 2 When the FDIC liquidates a bank, it becomes a creditor of the failed bank for the amount of its payment to the in The bottom half of table A1 is the balance sheet of the same bank after it recognizes some loan losses. All $6 of the nonperform ing loans turn out to be uncollectable, and an additional $1 of other loans is charged o ff as a loss. These losses reduce the allowance and equity to zero. At this point, the bank is closed and the FDIC becom es the receiver. The duties of a receiver of a bankrupt firm are to dispose of its assets and make payments to its creditors from the proceeds. The FDIC’s loss depends on the method used to resolve this case. Under the liquidation method, the FDIC would pay the fully insured depositors $70 and liquidate the assets, sharing the proceeds of the assets with the uninsured depositors.2 Equation A1 indicates the deter minants of the loss to BIF under the liquidation method. (Al) BIF loss = $70 (payment to fully insured depositors) - (70/(70 + 19)) [$5 (cash) + $35 (m arket value of securities) + $33 (liquidation value of loans)] = $12.58. The present value of payments to the uninsured depositors, on deposits of $19, would be (A2) (19/89)[$73] = $15.58. Another method of resolving failed banks is called p u rch a se a n d assu m ption . The FDIC solicits bids from oth er banks to purchase some of the assets of the failed bank and to assume its liabilities. In this illustration, the bank with the winning bid purchases the $5 of cash and pays $35 for the securities. W hether this bid would result in a low er loss to BIF than under sured depositors. The claim of the FDIC against the assets of the failed bank has equal priority to the claims of the uninsured depositors. JULY/AUGUST 1992 22 Table A1 Balance Sheet of a Hypothetical Bank PRIOR TO CHARGE-OFF OF LOAN LOSSES Assets Cash Securities Loans Nonperforming Other Allowance for loan losses Liabilities $ 5 40 Insured deposits Uninsured deposits $70 19 6 45 2 49 Net worth 5 $94 $94 Memo: Market value of securities is $35 AFTER CHARGE-OFF OF LOAN LOSSES Assets Cash Securities Loans Nonperforming Other Allowance Liabilities $ 5 40 0 44 0 44 Insured deposits Uninsured deposits Net worth $89 $70 19 0 $89 Memo: Market value of securities is $35. The present value of loans in liquidation, net of liquidation costs, is $33. liquidation depends on the size of the premium paid by the winning bidder, as indicated in the followed equation: (A3) BIF loss = $49 (payment by the FDIC to cover the gap betw een $40 of assets purchased and $89 of liabilities assumed - $33 (liquidation value of loans) - premium. The premium would have to exceed $3.42 to make the purchase and assumption transaction less costly to the FDIC than liquidation. A third resolution method is called tra n sfer o f in su red d ep osits. The FDIC solicits bids from other banks to assum e the insured deposit liabil http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis ities of the failed bank, but the FDIC liquidates the assets. The FDIC shares with the uninsured depositors the premium paid by the bank that assumes the insured deposit liabilities of the failed bank. Equation A4 presents the loss to BIF: (A4) BIF loss = $70 (cash to the bank that assumes the insured deposit liabilities) - (70/89) [$73 (liquidation value of assets) + premium], A com parison of equations A1 and A4 indicates that the BIF loss is sm aller under the tran sfer of insured deposits than under liquidation for any positive premium. 23 Daniel L. Thornton Daniel L. Thornton is an assistant vice president at the Federal Reserve Bank of St. Louis. Kevin White provided research assistance. Targeting M3: The Issue of Monetary Control I d e a l l y , AN INTERMEDIATE m onetary policy target should be both reliably associated with the goals of m onetary policy and readily controlled.1 In the 1970s and early 1980s, M l was the Federal Reserve's principal interm ediate m onetary aggregate target because of its close and stable relationship w ith nominal GDP. The principal issue then was how well M l could be controlled. As an outgrow th of the controversy over M l control, Congress passed the M onetary Control Act of 1980 (MCA) and the Federal Reserve replaced lagged reserve accounting (LRA) with contem poraneous reserve accounting (CRA). A principal objective of both the Act and the retu rn to CRA was to enhance M l control.2 The breakdow n of the relationship betw een M l and nominal GDP in the 1980s, however, caused the Federal Reserve to shift its emphasis away from M l. In 1986, the Fed dropped M l from its list of interm ediate policy targets and M2 becam e the Fed’s principal m onetary aggre gate. As with M l, the decision to focus on M2 was made on the basis of the long-run stability o f its relationship with nom inal GDP.3 The issue of M2’s controllability, how ever, has received scant attention. 1For modern survey of this literature, see Friedman (1990). 2The MCA extended Federal Reserve requirements to all depository institutions, removed differential reserve requirements by type of bank (Reserve City or Country) W hile the Federal Open M arket Committee currently sets target grow th rate ranges fo r M2, it is not the only aggregate that the Committee targets. M oreover, its grow th is but one of many factors that the Committee considers in form ulat ing and implementing m onetary policy. Neverthe less, M2 does receive considerable attention both in the Comm ittee’s deliberations and in the press. Consequently, this article analyzes the issue of M2 control. Under the existing system of reserve requ ire ments, the Fed can successfully target and control M2 only by implicitly targeting and controlling M l. At times, M2 control may require relatively large open m arket operations. O ther things the same, such large operations are potentially destabilizing fo r financial m arkets. M oreover, if M l or total reserves grow very rapidly while M2 grows slowly, the m arket may have difficulty in interpreting the thrust of monetary policy or the Fed’s intentions. To mitigate these problem s requires some changes in the existing stru ctu re of reserve requirem ents that, evidence suggests, would enhance significantly the Fed’s ability to control M2. These changes should have a minimal effect and removed reserve requirements from a large category of non-transaction deposits, not included in M1. 3The empirical basis for focusing on M2 is established by Hallman, Porter and Small (1991). JULY/AUGUST 1992 24 on the operation of the reserve m arket and can be accom plished without extending reserve requirem ents to non-depository institutions or increasing the so-called "reserve tax” on depository institutions. T H E M O N ETA RY C O N TRO L P R O B L E M : AN O V E R V IE W O F T H E C EN T R A L ISSU ES Issues in m onetary control are often fram ed in term s o f target variables, targets and instru ments. For purposes of this analysis the target variable is taken to be M2, and the target is taken to be a specific level or grow th rate for it.4 The instrum ent is the tool the policym akers use to guide the target variable to the target. The degree of m onetary control is defined by the strength o f the relationship betw een the target variable and the policy instrum ent: the stronger this relationship, the m ore precise the control. Tw o possibilities exist. First, th ere could be a d irect relationship betw een the instrum ent and the target variable, in which changes in the instrum ent directly affect the target variable. Second, th ere could be an indirect link betw een the instrum ent and the target variable. In this case, changes in the instrum ent affect the target variable by affecting other variables, for example, the interest rate. M onetary control is m ore precise the smaller the role of factors oth er than the policy in stru m ent in determ ining the target variable. Indeed, control is best w hen th ere are no such "leak ages.” If the relationship betw een the target variable and instrum ent is indirect, precise control tends to be m ore difficult; factors other than the policy instrum ent affect not only the target variable, but also the relationship betw een the instrum ent and the target variable. Such leak ages exist w hen the relationships betw een the in 4Currently, the Federal Open Market Committee (FOMC) sets long-run target ranges for the growth rate of M2 from the fourth quarter of one year to the fourth quarter of the next. These growth rate ranges imply target ranges for the levels of the variables over the planning period. The FOMC also sets short-run growth rate ranges for M2 for the period between meetings, that is, the “ intermeeting period.” The growth rate ranges, in turn, imply targets for the level of M2. Hence, there is a one-to-one correspondence between targets for the growth rate and targets for the level of M2. 5For example, the uncharacteristically slow growth recently of the non-M1 components of M2 was unanticipated and is, http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis strum ent and the oth er variables or betw een the oth er variables and the target variable are neither strong n or precise. In any event, m ore and larger leakages imply less control. Fu rtherm ore, w hen control is indirect, the relationship betw een the policy instrum ent and the target may be unreliable and may change from time to time, in response to such things as financial innovation and regulatory change. Hence, the ability to control m onetary aggregates through such indirect channels may vary in ways that are both difficult to explain and impossible to predict.5 Implementing a m onetary control procedure is com plicated by oth er factors, such as the avail ability of inform ation, the time horizon over which the policym aker wishes to affect control, possible "feedback” effects betw een other varia bles and the instrum ent and the ability to predict factors that affect the aggregate that cannot be controlled either directly or indirectly. Since the purpose of this paper is simply to point out the fundam ental issues in controlling M2, the question of how best to implem ent a practical control procedure fo r M2 is not considered. Controlling M2 M2 consists of M l plus an array of savingstype deposits that are called the non-M l com ponents of M2 (NM1M2).6 The Fed’s ability to control M2 depends on its ability to control both M l and NM1M2. If th ere w ere a direct link betw een both of these M2 com ponents and both could be controlled equally well, there would be no difference betw een the Fed’s ability to control M l and its ability to control M2. But this is not the case. Historically, the Fed has established direct control over the non-currency com ponents of the m onetary aggregates through a system of as yet, not understood. See Bullard (1992) and Carlson (1992) for a discussion of this issue. 6The non-M1 components of M2 consist of savings deposits (including money market deposit accounts), small denomi nation time deposits, general purpose broker/dealer money market mutual funds, overnight RPs issued by all commer cial banks and overnight Eurodollars issued to U.S. residents by foreign branches of U.S. banks. See Hafer (1980) for a more detailed discussion of each component of M2 except money market deposit accounts. 25 reserve requirem ents.7 In 1959, NM1M2 con sisted prim arily of time and savings deposits, most of w hich w ere subject to the Federal Reserve’s reserve requirem ents. The MCA, how ever, eliminated reserve requirem ents on a broad class of NM1M2 and the rem ainder w ere eliminated in D ecem ber 1990. Consequently, currently th ere is no direct relationship betw een the Fed’s actions and NM1M2. In contrast, the MCA enhanced significantly the relationship betw een the Fed’s instrum ent and M l.8 Essentially, M2 now consists of one component, M l, which the Fed can influence directly, and another com ponent, NM1M2, over which the Fed has no direct influence. A detailed model of M2 control is presented in the appendix to this article; th ree conclusions em erge from it. First, the Fed's ability to control M2 is b etter the stronger the direct relationship betw een its policy instrum ent and M l and the stronger the indirect effects of policy actions on NM1M2. Second, because th ere is a strong direct link betw een policy actions and M l, other things the same, M2 control is b etter the larger the proportion of M l in M2. Finally, M2 control will be b etter the larger the indirect effects of policy actions on NM1M2 and, in particular, the larger such effects are relative to the total effect of policy actions on M l. To see why this last point applies, suppose policy actions have no effect on NM1M2, either direct or indirect. In this case, M2 can be controlled only by manipulating M l to com pletely offsett undesired m ovements in NM1M2. Since NM1M2 is large relative to M l, the re 7Some analysts point out that depository institutions would maintain vault cash to service deposit inflows and outflows from such deposits so that the money supply could be con trolled even in the absence of official reserve requirements. In effect, such institutions would be maintaining reserves equal to some fraction of these deposit balances, so effec tively they would be imposing reserve requirement on them selves. Indeed, currently a significant number of depository institutions hold vault cash in excess of their reserve re quirement. While it is no doubt true that depository institu tions would hold cash for some purposes, there is no guarantee nor evidence that this “ implicit reserve ratio” would be stable or systematically related to the level of deposits. Under the present system of reserve requirements, depository institutions attempt to economize on their hold ings of excess reserves. This is what makes reserve require ments an effective tool of monetary control. quired manipulation of M l could be quite large. If the indirect effect of policy actions on NM1M2 w ere large and positive, that is, if an open m arket purchase results in an increase in NM1M2, the required manipulation of M l would be m uch smaller. If, however, the indirect effects of policy actions on NM1M2 w ere negative, so that an open m arket purchase results in a decrease in NM1M2, open m arket operations would have to be pursued even m ore aggressively. In oth er words, the required change in M l would have to be larger to offset the decline in NM1M2.9 T H E R EC E N T B E H A V IO R O F M 2 The em pirical analysis of the basic issues raised above begins with a simple analysis of the behavior of M2 relative to that of M l. Figure 1 shows the share of M l in M2 during the period of the official published series on the m onetary aggregates, January 1959 to M arch 1992. The proportion o f M l in M2 declined through the late 1970s, decreasing from nearly 50 percent in 1959 to about 25 percent in 1977. Since then, the ratio has changed relatively little on average but has been somewhat variable. M oreover, the proportion of M2 grow th accounted for by NM1M2 grow th increased significantly betw een 1959 and 1977. This is illustrated in figure 2, which shows the growth rates of M2 and NM1M2 since 1959. Before the late 1970s, the grow th rate of NM1M2 was consistently higher than the grow th rate of M2. Since then, how ever, the grow th rates of M2 and NM1M2 have been very similar. under the Fed’s direct control. This constituted a potential source of leakage of monetary control for both M1 and M2. In addition, the reserve requirements on different deposits were different, hence, the relationship between the policy in strument and a particular monetary aggregate would change with shifts in the public’s preference for certain types of deposits or financial innovations. See Garfinkel and Thorn ton (1989, 1991a). 9ln the extreme indirect effects the sum of the actions on M1, and very unlikely case in which the negative of policy actions on NM1M2 were larger than positive direct and indirect effects of policy the process would be dynamically unstable. 8The MCA required other changes that enhanced control over M1. Prior to the MCA, Federal Reserve reserve requirements applied only to member banks. Hence, some components of both M1 and NM1M2 were not directly linked to the Fed’s policy actions and, therefore, were not JULY/AUGUST 1992 26 Figure 1 The Ratio of M1 to M2 Percent 1959 61 63 65 67 69 71 Monthly Data 73 75 77 79 81 83 85 87 891991 The Link Between Policy Actions and M l and NM1M2 Estim ates of the direct and indirect effects of policy actions on M l and NM1M2 can be obtained by regressing these variables on total 10Note that the regression analysis here takes the view that total reserves are exogenous. If that is not the case, then the correlation between reserves and say NM1M2 could be due to the effect of shifts in NM1M2 on reserves, rather than the other way around. For example, if NM1M2 declined the Fed might offset some of the effect of the decline in M2 by increasing total reserves and, consequent ly, M1. Note, however, that this would result in a negative relationship between NM1M2 and TR. "T h e question of stationarity naturally arises when monetary and reserve aggregates are used. Such variables tend to grow over time at widely variable growth rates. Therefore, the null hypothesis of non-stationarity is frequently not rejected when applied to such univariate time series. The null hypothesis of non-stationarity may not be rejected even when first differences of such variables are used. Of course, reserve requirements establish a link between re serves and checkable deposits. This is certainly the case for reserves and total checkable deposits since the elimi FEDERAL RESERVE BANK OF ST. LOUIS Percent reserves. Total reserves (TR) is taken as the policy instrum ent because cu rren cy is supplied on demand and because changes in total reserves are closely related to open m arket operations.10 The equations are estim ated with all variables in first-differences (A).11 Table 1 nation of reserve requirements on nontransaction accounts. As a result, these variables should be cointegrated. This does not necessarily imply that there is a stationary linear relationship between reserves and the other monetary aggregates like M1 (currency is non-stationary) and NM1M2 or M2. Furthermore, the first-difference of variables that are growing over time is not necessarily stationary. For exam ple, if a variable grows at a constant 5 percent rate, then first-differences of the variable will get larger and larger over time. In short samples like the one used here, how ever, such non-stationarity is not very important. Indeed, the null hypothesis of a unit root in the first differences of total reserves is rejected at the 5 percent significance level. Because of this and because the coefficients are more diffi cult to interpret when growth rates are used, all the equa tions are estimated using first-differences of the levels of the variables. See Dickey, Jansen and Thornton (1991) for a discussion of stationarity and cointegration. 27 Figure 2 The Growth of M2 and NM1M2 Percent 60- -60 50 50 40 40 30 30 •1Q 1959 61 --10 63 65 67 69 71 73 75 77 79 81 shows the results o f regressing first-differences of the various m onetary aggregates on ATR.12 The regression of AMI on ATR shows that there is a strong relationship betw een AMI and ATR, with ATR explaining about 80 percent of the variation in AMI. M oreover, the estimated 12The period begins with the effective implementation of the MCA in March 1984; see Garfinkel and Thornton (1989). Following the removal of reserve requirements on non transaction accounts, excess reserves rose significantly above their pre-December 1990 level for about three months, then declined to about their previous level as depository institutions were surprised by this action. Consequently, dummy variables are included for January, February and March of 1991. 13This is not precisely correct because some reserves are held in the form of excess reserves and because reserve requirements on government and certain foreign deposits are not included in either other checkable deposits (OCD) or M1. Hence, the multiplier is smaller than 8.33. The amount of excess reserves or reserves needed to support these other deposits, however, is not large relative to total reserves, so the difference between the effective multiplier and 8.33 is quite small. Also, this result simply could be due to the fact that the reserve series has been adjusted for reserve requirement Percent Monthly Data 83 85 87 89 1991 coefficient on ATR is not statistically different from 8.33, that is, 1/.12, w here .12 is the marginal reserve requirem ent on transaction deposits.13 This suggests both that total ch eck able deposits (TCD) and cu rren cy are u ncor related and that th ere are no indirect effects of changes so the coefficient is biased toward 8.33, the reciprocal of the marginal reserve requirement, .12. How ever, the Board of Governors uses the average rather than the marginal reserve requirement to adjust its series for re serve requirement changes. See Garfinkel and Thornton (1991b) and Meulendyke (1990). Nevertheless, an equation involving TCD and total reserves not adjusted for reserve requirement changes was estimated. These data are avail able only on a not seasonally adjusted basis. When the seasonal dummy variables were excluded, the adjusted R-square was .89 and the estimated coefficient was 8.80— not significantly different from 8.33 at the 5 percent sig nificance level. When monthly seasonal dummy variables were included, the adjusted R-square was .96 and the esti mated coefficient was 7.44. In this case the hypothesis that the coefficient was equal to 8.33 is rejected at the 5 percent significance level—the t-statistic is 2.52. As a practical matter, however, this qualification does not appear to be particularly important as the degree of the bias is not large. JULY/AUGUST 1992 28 Table 1 Estimates of the Effect of Policy Actions on Various Monetary Aggregates, Monthly Data, March 1984 - March 1992________ AM1 Constant AM2 ANM1M2 ATCD 1.863* (7.85) 10.342* (12.38) 8.469* (10.41) 0.619* (2.76) ATR 8.293* (19.37) 8.690* (5.77) 0.397 (0.27) 8.250* (20.37) D.W. 1.784 0.727 0.612 1.853 .802 .265 .000 .820 Adj. R2 ’ indicates statistical significance at the 5 percent level. policy actions on M l. This conclusion is rein forced by the fact that the adjusted R-square for the regression of the ATCD on ATR is nearly identical to that of the AMI regression, and the fact that the coefficients on ATR are nearly identical in the two equations. Consequently, all o f the effect of ATR on AMI comes through the direct relationship betw een TR and TCD that results from the Federal Reserve’s system of reserve requ irem ents.15 The results for ANM1M2 show that the indirect effect of policy actions on this com po nent of M2 are nil.16 The adjusted R-square is zero and the coefficient on ATR, which captures both the direct and indirect effects of policy 14This coefficient measures both the direct and indirect effects of policy actions on M1. See the appendix for details. Because the total effect is not significantly different from the direct effect, the indirect effect must be insignifi cantly different from zero. See Garfinkel and Thornton (1991 a) for an analysis of the relationship between currency and TCD. 15The lack of any significant serial correlation in the residuals of the estimated equation suggests that the remaining error is simply “ control error” and seasonals. 16Note that the D.W. statistic indicates significant first-order serial correlation in all but the equation involving TCD. This is to be expected because, in these cases, a simple regres sion of the changes in these variables on the change in to tal reserves does not adequately reflect the process generating these variables. See the appendix to http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis actions, is statistically insignificant. The lack of an effect on NM1M2 is reflected in the coefficient of ATR in the M2 equation. This coefficient too is not statistically different from 8.33, suggesting that the m arginal effect of policy actions on M2 comes solely through their effect on M l. It could be argued that the indirect effects of policy actions on M2, say, through interest rates, take time to w ork so that the potential for indirect control of M2 is not adequately reflected in the m onthly data. This issue is investigated first by using low er frequency (quarterly) data and second by including a sixmonth distributed lag of ATR. The results using quarterly data, presented in table 2, are similar Garfinkel and Thornton (1991a) for an illustration of this point using M1. This merely confirms the fact that the behavior of NM1M2 and, hence, M2 is not adequately explained by Fed policy actions. The presence of positive, first-order serial correlation does tend to bias the estimates of the standard errors downward. Hence, the reported t-stastistics may overstate the statistical significance of the change in total reserves in these equations. 29 Table 2 Estimates of the Effect of Policy Actions on Various Monetary Aggregates, Quarterly Data, 1984.2 - 1992.1 AM1 Constant ANM1M2 AM2 ATCD 4.790* (7.21) 32.270* (8.48) 27.481 * (6.99) ATR 9.683* (18.27) 8.353* (2.75) -1 .3 3 0 (0.42) 9.657' (19.85) D.W. 1.394 1.065 0.917 1.684 .917 .162 .000 .930 Adj. R2 0.988 (1.62) 'indicates statistical significance at the 5 percent level. Table 3 Long-Run Effects of Policy Actions on Various Monetary Aggregates, Monthly Data, March 1984 - March 1992 AM1 Constant AM2 ANM1M2 ATCD 1.471* (5.27) 11.352* (10.95) 9.881* (10.07) 0.170 (0.66) P 7.976* (16.23) 10.728* (5.87) 2.752 (1.59) 7.874* (17.35) e 9.878* (13.32) 3.567 (1.29) -6 .3 1 1 * (2.42) 9.958* (14.53) 1.902* (2.32) -7 .1 6 0 * (2.35) -9 .0 6 2 * (3.15) 2.084* (2.76) 1.622 0.750 0.09 1.733 .823 .275 D.W. Adj. R2 .043 .847 ‘ indicates statistical significance at the 5 percent level. to those using monthly data. Again, policy actions have no effect—direct or indirect—on NM1M2; their effect on M2 com es only through their effect on M l.17 17One difference is that the coefficient of the change in total reserves in both the M1 and TCD equations is larger than 8.33, and the difference is statistically significant. This result is puzzling. It appears, however, that it is due to the fact that the Board of Governors uses the average rather than the marginal reserve requirement when adjusting reserves for changes in reserve requirements. The average reserve requirement is significantly smaller than the margi nal. This means that the coefficient of a regression of the change in TCD on a change in total reserves so adjusted The estim ates including a six-month distrib uted lag of total reserves, presented in table 3, give a broadly similar picture. The coefficient /? m easures the contem poraneous relationship will be substantially larger than 8.33. At high frequencies, however, the change in total reserves so adjusted is likely to reflect the actual change in reserves so that the coeffi cient is approximately equal to the reciprocal of the marginal reserve requirement. At lower frequencies or in distributed lag specifications that capture the long-run effect of a change in constructed total reserve series, the estimated coefficient better reflects the reserve requirement used in the constructed series. JULY/AUGUST 1992 30 betw een the changes in the dependent variable and ATR; 0 m easures the total effect of cu rren t and past changes in total reserves on changes in the dependent variable; and \x m easures the sum of the lagged effects of ATR.18 T h ere is a significant association betw een changes in NM1M2 and changes in total reserves, as the adjusted R-square is statistically different from zero. The R-square is very small however, and all of the statistical significance is associated with the subsequent negative effect of total reserves on NM1M2. The contem poraneous effect of a change in total reserves on M2 is larger in this specifica tion than in table 1; note, how ever, that this is simply the sum of statistically significant and statistically insignificant effects (the coefficient for M l, 7.976, plus the coefficient for NM1M2, 2.752). For both AMI and ATCD, the results are similar to those using quarterly data.19 For M2 and NM1M2, the subsequent effect of policy actions largely offsets the initial effect. That the subsequent effect is negative and statistically significant is somewhat surprising. If this result w ere robust and not m erely the artifact of the particular sample period, it would create a potentially difficult problem fo r M2 control.20 To see this, assume that M2 is cu r rently below its target level and the Fed increases reserves to nudge M2 upward. This action would set in motion changes that would eventually lead to a reduction in NM1M2, creat ing a need fo r additional policy action. Antici pating this, policym akers would have to be m ore aggressive in increasing M l to hit their M2 target. 18Note that the estimated coefficients satisfy the restriction, p = 0-M - The coefficients were estimated from a simple reparameterizaton of the change in the appropriate monetary aggregate on a constant term and the contem poraneous and six lags of the actual change in total reserves. 19lt could be argued that the results are sensitive to the choice of the policy instrument. To investigate this possibili ty, two other policy instruments were considered, the ad justed monetary base and non-borrowed reserves. The evidence of monetary policy actions on short-term interest rates generally is strongest if non-borrowed reserves is used as the policy instrument [see Thornton (1988) and Christiano and Eichenbaum (forthcoming)]. Moreover, it is generally argued that the Fed controls M2 through its in fluence on short-term interest rates and the connection be tween these rates and the demand for M2. Consequently, non-borrowed reserves is a particularly important alternative policy instrument to consider. The results, however, indi cate that the general conclusions drawn above are insensi tive to the variable chosen as the policy instrument. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis The Recent Behavior o f NM 1M2 and M onetary P olicy The above analysis suggests that, if the Fed has been targeting M2, there should be more instability in the behavior of the policy instrument, and there should be an inverse relationship betw een the policy instrum ent and NM1M2. Data from the latter part of the 1980s is broadly consistent with M2 targeting. Figure 3 shows a 12-month moving average of the grow th rate of total reserves and M l since January 1959. Tw o things are evident from the figure: the relationship between M l and total reserves improves dram atically following the effective im plem entation of the MCA, and the volatility of the grow th rate of total reserves increases pretty dramatically in the 1980s. Figure 4 shows the 12-month moving averages of total reserve grow th and NM1M2 grow th fo r the same period. The growth rates of total reserves and NM1M2 are not negatively correlated as strict M2 targeting would suggest they should be in the latter part of the 1980s. W hile there are periods since the m id-1980s w hen sharp accelerations in reserve grow th are associated with significant decelerations in the grow th rate of NM1M2, a pattern of com pensating variations in the grow th rates of these variables does not em erge.21 Hence, these data do not appear to support the idea that the large, persistent swings in total reserve grow th are associated directly w ith targeting M2.22 Nevertheless, as table 4 shows, reserve grow th was m uch faster on average since the mid-1980s, and this faster reserve grow th is associated with a significant slowing in NM1M2 grow th. 20One explanation for this result stems from the fact that the first difference of NM1M2 has a statistically significant, negative linear time trend during the period. It appears that the negative lagged effect of the change in total reserves on the change in NM1M2 in table 3 merely reflects the negative trend in the latter variable over the sample period. The trend coefficient is -.1 0 2 with a t-statistic of -4 .5 9 . 21For example, as M2 growth slipped to the bottom of the Fed’s target range during the latter half of 1991, total reserve growth accelerated sharply. 22The fact that the estimate of in table 3 is negative, however, could be evidence of this behavior. See footnote 11 for a discussion of this point. 31 Figure 3 12-Month Moving Average of the Growth of M1 and Total Reserves Percent Percent 28 .2 8 24 - 20 - 8 | -24 20 i . - i /1 L 4195961 63 65 67 69 71 73 75 77 79 81 83 85 87 891991 Figure 4 12-Month Moving Average of the Growth of the Non-M1 Components of M2 and Total Reserves Percent 28 Percent ....... 28 -4-f 195961 63 65 67 69 71 73 75 77 79 81 83 85 87 891991 JULY/AUGUST 1992 32 the need fo r large swings in the policy actions of the Fed. Table 4 Average Growth Rates of Various Monetary and Reserve Aggregates Aggregate TR NM1M2 M2 1959.1-1984.1 1984.2-1992.1 3.53% 10.29 8.57 8.37% 5.26 5.78 Consistent with the Fed’s objective fo r M2 grow th during the period, M2 grow th has slowed significantly since the m id-1980s.23 Hence, while the evidence suggests that the Fed has not been attem pting to target M2 closely over periods of up to a year, it is consistent with the Fed's targeting of M2 over a somewhat longer time horizon. Indeed, the experience on average over the latter half of the 1980s is broadly consistent with the Fed’s paying increased attention to M2 and with the Fed’s objectives for M2 growth. ENHANCING M 2 CO N TRO L The analytical and em pirical analyses above suggest that M2 can be controlled only by pursuing m onetary policy actions to offset movements in NM1M2 over w hich the Fed has little or no control. W hile such actions are not necessarily destablizing, they could be, espe cially w hen actions are required to offset large, undesired m ovem ents in NM1M2. M oreover, such large changes in policy actions could be m isinterpreted. If the Fed wishes to target M2, changes in the stru ctu re of reserve requirem ents could be made that would significantly enhance its controllability. Such changes would eliminate 23The Federal Open Market Committee’s target range for M2 decreased in a series of steps from 6 to 9 percent in 1984 to 2.5 to 6.5 percent by 1992. 24See Garfinkel and Thornton (1989, 1991a). 25See Thornton (1983) for a discussion of the timing issue as it applied to LRA and CRA. 26Of course, depository institutions can also hold reserves in the form of non-interest-bearing vault cash. Since many institutions are currently holding vault cash in excess of their required reserves, it may not be correct to suggest that such holdings impose a tax on these institutions. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis The em pirical results h ere and elsew here suggest that reserve requirem ents, like those imposed on the checkable deposits in M l, can be an effective way to establish a direct link over the deposit com ponents of the m onetary aggregates.24 In oth er words, M2 control could be enhanced substantially by extending reserve requirem ents to the financial assets that make up NM1M2. Most effective m onetary control would be obtained if the percentage reserve requirem ent w ere the same fo r all assets that m ake up the aggregate. This would prevent shifts in the ag gregate that are simply due to shifts in the pub lic’s p referen ce betw een deposits with "high” marginal reserve requirem ents and those w ith “low” marginal reserve requirem ents. Control would also be best if the timing of reserve re quirem ents on all categories of deposits w ere the same. As long as the timing is the same, this issue is o f little consequence, especially if the objective is to control the m onetary aggregate over a period of a qu arter or m ore.25 The P ro b le m o f the R eserve Tax Reserve requirem ents are often thought of as a “reserve tax” because they fo rce depository institutions to hold a portion of th eir assets in the form of non-interest-bearing deposits at the Federal Reserve and because the marginal interest incom e from these funds, w hich the Fed invests in interest-bearing U.S. governm ent securities, is rebated to the U.S. T reasury.26 Imposing the cu rren t reserve requirem ent on transaction accounts to the non-transaction com ponents of M2 would significantly increase the reserve tax on depository institutions.27 This would put them at a competitive disadvantage and, undoubtedly, give rise to tax avoidance schem es and increased competition from other 27The reserve tax is only part of the net tax on depository institutions resulting from government supervision and regulation, and it may not be large relative to the other tax es and subsidies. For example, currently over three-fourths of the depository institutions satisfy their reserve require ments with vault cash, which they would probably hold in the absence of reserve requirements. Second, depository institutions are insured by the government at a fraction of the cost. On net, institutions probably receive a net subsidy from the government. 33 financial interm ediaries. T he adverse effect of extending reserve requirem ents to NM1M2 could be offset, how ever, by paying in terest on required reserve balances held with Federal Re serve banks.28 as long as changes in the quantity of deposits that are exem pt from reserve requirem ents are infrequent and relatively small. Another problem would rem ain: requiring depository institutions to hold a significant portion of their assets as reserves might alter significantly the composition of their assets away from loans. This would fu rth er reduce the role o f depository institutions in supplying credit to the econom y.29 Because of this, it would seem desirable to set the percentage reserve requirem ent on the com ponents o f M2 at a level that would leave the am ount of total reserves held at their cu rren t level. U nfor tunately, part of NM1M2—general purpose b rok er and dealer money m arket mutual funds—are not held at depository institutions. Hence, either these deposits would have to be exem pt from reserve requirem ents or reserve requirem ents would have to be extended to non-depository institutions. The form er option seems the most desirable for at least two reasons. First, extending reserve requirem ents to non-depository institutions would set a p rece dent and would raise oth er issues, such as w hether deposit insurance should be extended to such institutions or w h eth er they would be perm itted to borrow at the Federal Reserve’s discount window. Second, because such deposits account for only about 10 p ercen t of M2, they constitute a relatively m inor source of leakage fo r M2 control. The Effect on Bank Lending Rates o f Funds Obtained by Managed Liabilities Exempting money m arket mutual funds from reserve requirem ents and imposing uniform requirem ents on the rem aining non-currency com ponents of M2 would require an average reserve requirem ent of about 2 p ercen t.30 M onetary control would be best if the marginal and average reserve requirem ents w ere the same, that is, if no deposits are exem pt from reserve requirem ents. Logic suggests and the empirical evidence above supports the notion, however, that this is not a m ajor consideration 280 f course, it would require an act of Congress for the Federal Reserve to pay interest on reserves. 29See Kaufman (1991). 30The exact estimate of 1.76 percent is based on notseasonally-adjusted data and total reserves not adjusted for reserve requirement changes for April 1992. 31lt should be noted, however, that insurance premiums paid by depository institutions have increased significantly. It has been increasingly the case that deposi tory institutions have relied on "managed liabilities” to m eet changes in loan demand. During periods w hen loan demand is strong, institutions are m ore aggressive in setting higher rates on large and small tim e deposits and money m arket deposit accounts (MMDAs) to attract additional funds. Bank loan rates are equal to the rate paid on these deposits plus a spread that is determ ined by the competitive conditions in the m arket. If such funds w ere subjected to a 2 percent reserve tax, it would raise the m arginal cost of funds obtained from managed liabilities by about 2 percent (1/.98). W hether this would harm the competitive posi tion of depository institutions fu rth er, given that the total tax would be unchanged, is unclear. In any event, depository institutions have a com petitive advantage because their deposit liabilities are federally insured, while their com petitors’ are not.31 Nevertheless, any adverse effects of extending reserve req u ire m ents to most of NM1M2 could be mitigated by paying interest on required reserve balances with the Fed. The interest rate paid on these balances could be tied to m arket rates and set close enough to such rates to reduce the reserve tax to the point at which it plays an insignificant role in allocating credit.32 If these changes w ere made, the evidence sug gests that M2 could be controlled without large swings in the use of the Fed’s policy instrum ent. M oreover, increased M2 control could be achieved w ithout increasing the reserve tax and with little or none of the oth er adverse effects commonly associated w ith reserve requirem ents. 32For example, it could be paid in arrears and at a rate that is one-quarter of a percent below the rates depository institutions paid on their managed liabilities in M2 over the maintenance period. This would all but eliminate the reserve tax. If this were done on the basis of the average rate paid on such deposits, such a scheme would result in a slight subsidy to institutions that pay below average rates and a net effective cost to those paying above average rates. This might have the effect of tempering slightly the incentive of some institutions to bid aggressively for such funds. JULY/AUGUST 1992 34 SUM M ARY AND CONCLUSIONS Among oth er variables, the Fed currently sets target ranges for the M2 m onetary aggregate. W ithout considering its desirability, this paper analyzes the controllability of M2 under existing institutional arrangem ents. Both the analysis and the data suggest that, currently, M2 can be con trolled only through the Fed’s control of M l. The evidence also suggests that M2 control is difficult and that hitting an M2 target may, at times, require very large changes in open m arket operations. To cou nteract these problem s, the paper suggests several ways in w hich the Fed could enhance M2 controllability while virtually eliminating the large changes in policy actions that can be required under the cu rren t system of reserve requirem ents. Enhanced M2 control could be achieved without increasing the reserve tax on depository institutions and without fo rc ing depository institutions to shift their asset portfolios away from loans. R EFER EN C ES Bullard, James B. “ The FOMC in 1991: An Elusive Recovery,” this Review (March/April 1992), pp. 41-61. Carlson, John B. “ Recent Behavior of Velocity: Alternative Measures of Money,” Economic Trends, Federal Reserve Bank of Cleveland (April 1992), pp. 2-10. Christiano and Eichenbaum, “ Identification and the Liquidity Effect of a Monetary Policy Shock,” in A. Cuikerman, L.Z. Hercawitz and L. Leiderman, eds., Business Cycles, Growth and Political Economy (MIT Press, forthcoming). Dickey, David A., Dennis W. Jansen and Daniel L. Thornton. “ A Primer on Cointegration with an Application to Money and Income,” this Review (March/April 1991), pp. 58-78. Friedman, Benjamin. “ Targets and Instruments of Monetary Policy,” in Benjamin Friedman and Frank Hahn, eds. Handbook of Monetary Economics, Vol. 2, (Amsterdam: North-Holland, 1990), pp. 1185-230. Garfinkel, Michelle R., and Daniel L. Thornton. “ The Link Between M1 and the Monetary Base in the 1980s,” this Review (September/October 1989), pp. 35-52. ________“ The Multiplier Approach to the Money Supply Process: A Precautionary Note,” this Review (July/August 1991a), pp. 47-64. ________“ Alternative Measures of the Monetary Base: What Are the Differences and Are They Important?” this Review (November/December 1991b), pp. 19-35. Hafer, R. W. “ The New Monetary Aggregates,” this Review (February 1980), pp. 25-32. Hallman, Jeffrey J., Richard D. Porter and David H. Small. “ Is the Price Level Tied to the M2 Monetary Aggregate in the Long Run?,” American Economic Review (September 1991), pp. 841-58. Kaufman, George G. “ The Diminishing Role of Commercial Banking in the U.S. Economy,” Federal Reserve Bank of Chicago Working Paper 91-11, (May 1991). Meulendyke, Ann-Marie. “ Possible Roles for the Monetary Base,” in Intermediate Targets and Indicators for Monetary Policy (Federal Reserve Bank of New York, July 1990), pp. 20-66. Thornton, Daniel L. “ Lagged and Contemporaneous Reserve Accounting: An Alternative View,” this Review (November 1983), pp. 26-33. _______ . “ The Effect of Monetary Policy on Short-Term Interest Rates,” this Review (May/June 1988), pp. 53-72. A p p en d ix A S im p le M odel of M2 C o n tro l This appendix presents a simple model of M2 control. In the following analysis, the policy instrum ent is taken to be the change in total reserves, TR. T he general results, how ever, do not depend on the use of total reserves. O ther policy instrum ents such as the m onetary base or non-borrow ed reserves would yield similar results. M2 consists of M l and some savings-type deposits called the non-M l com ponents of M2, NM1M2. That is, http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis (1) M2 = M l +NM1M2. Thus, changes in M2 per unit of tim e can be w ritten as (2) M2 = M1 + NM1M2. M l consists of currency, C, and total ch eck able deposits, TCD. Consequently, by definition M l can be w ritten as (3) M l = (1 + k)TCD, 35 w here k is the ratio of cu rren cy to TCD. For the purpose of this illustration, k is assumed to be constant.1 w here g is obtained in a m anner analogous to that used to obtain f, and v denotes the stochastic part of NM1M2 that is unrelated to policy actions. The quantity of TCD is directly related to the Fed’s policy instrum ent through the Fed’s system of reserve requirem ents. That is, The control problem for M2 can be illustrated most easily by considering the general condition that the effects o f policy actions on NM1M2 are some proportion of their total effect on M l. That is, (4) TCD = (l/r)TR, w here r is the proportion of additional TCD that m ust be held in the form of reserves (vault cash and deposit balances at the Federal Reserve). Combining (3) and (4), yields (5) M l = ((1 + k)/r)fR , w hich establishes a direct link betw een M l and TR. (11) g = A[(l + k)/r+n. W hile th ere are no constraints on the value of A, the fact that policy actions have no direct effect on NM1M2 makes it likely that |A| < 1 . Combining equations 1 and 9-11 yields the following equation fo r M2: It may be that policy actions also affect M l indirectly, through their effect on other varia bles, X. That is, (12) M2 = [1 + A] [(1 + k )/r + f']T R + u + v. (6) M l = h(X), First, M2 control is generally better the smaller the control erro r and the stronger the indirect effects of policy actions on NM1M2, that is, the smaller are u and v. and (7) X = j(TR).2 Together, they imply that (8) M l = f' f R.3 Allowing for the possibility of both direct and indirect effects of policy actions on M l and the possibility of an additive stochastic control erro r, u, that is independent of both the direct and indirect effects, the total effect of policy actions on M l can be sum m arized as (9) M l = [(1 + k) It ) + f']TR + u. Since, by construction, policy actions have no direct effect on NM1M2, the effect of such actions on NM1M2 can be expressed as (10) NM1M2 = g f R + v, 'This assumption is not critical to the major findings of the analysis. See Garfinkel and Thornton (1991a) for a recent criticism of this common assumption. 2ln the case of M1, one could think of it as a situation in which M1 was equal to the money multiplier (mm) times to tal reserves, where the multiplier is some function of X. That is, M1=mm(X)TR. Several aspects of equation 12 are worthy of note. Second, control will be b etter the larger the proportion of M l in M2. This is not the case if u > v , but that appears to be extrem ely unlikely. This con jectu re is supported by the empirical analysis in the paper. Third, control will be b etter the larger the in direct effect of policy actions on NM1M2 relative to their total effect on M l, that is, the larger the value of A. This is so because the proportion of M2 related to TR is larger in proportion to u and v the larger the value of A. Indeed, if A = 0 (which implies that g = 0), then the only direct control over M2 would com e through the Fed’s control over M l. Control of M2 could be obtained only by offsetting shifts in v by manipulating M l. Since NM1M2 are large relative to M l, this could require relatively large changes in M l. If A w ere negative, M2 control would requ ire even m ore aggressive M l policies. ship between these deposits and total reserves. Hence, there is nothing equivalent to a money multiplier for NM1M2. 3The function f is equal to h(j(TR)) which implies that TR = p(t), where t denotes time. These functions are written in their general form, however, in the empirical section of the paper, it will be assumed that they are linear. In this case, fill = mm(X)TR + [3m m /9X)(9X /9TR )TR ]. In the case of NM1M2, however, there is no direct relation JULY/AUGUST 1992 36 Steven Russell Steven Russell is an economist at the Federal Reserve Bank of St. Louis. Lynn Dietrich provided research assistance. Understanding the Term Structure of Interest Rates: The Expectations Theory I III'. INTEREST RATES on loans and securities provide basic summary m easures of their attrac tiveness to lenders. The role played by interest rates in allo catin g fun d s a cro ss fin a n cia l m a rk ets is very similar to the role played by prices in allocating resources in m arkets for goods and services. Ju st as a relatively high price o f a par ticular good tends to draw physical resources into its production, a relatively high interest rate on a particular type of security tends to draw funds into the activities that type of secu rity is issued to finance. And just as identifying the factors that help determ ine prices is a key area of inquiry among econom ists who study goods m arkets, identifying the factors that help determ ine interest rates is a key area of inquiry for those who study financial m arkets. Econom ic theory suggests that one im portant factor explaining the differences in the interest rates on d ifferent securities may be differences in their term s—that is, in the lengths of time before they m ature. The relationship betw een the term s of securities and their m arket rates of in terest is know n as the term stru ctu re of interest rates. To display the term stru ctu re of interest rates on securities of a particular type at a p ar ticular point in time, econom ists use a diagram called a y ield curve. As a result, term stru ctu re theory is often described as the theory of the yield curve. Econom ists are interested in term stru ctu re theory fo r a num ber of reasons. One reason is that since the actual term stru ctu re of interest rates is easy to observe, the accuracy of the predictions of different term stru ctu re theories is relatively easy to evaluate. These theories are usually based on assumptions and principles that have applications in other branch es of econom ic theory. If such principles prove useful in explaining the term structure, they might also prove useful in contexts in which their relevance is less easy to evaluate. One theory of the term stru ctu re that will be described here, fo r example, suggests that a behavioral trait called risk aversion may play a m ajor role in determ ining the shape of the yield curve. If sub sequent research lends credence to this theory, econom ists may give m ore emphasis to risk aversion in constructing theories of other aspects of financial m arket operation.1 A second reason why economists are interested in term structure theories is that they help explain the ways in w hich changes in short-term interest 'Examples include the role of financial intermediaries and the pricing of claims to physical assets (such as stocks). JULY/AUGUST 1992 37 rates—rates on securities with relatively short term s—affect the levels of long-term interest rates. Econom ic theory suggests that m onetary policy may have a direct effect on short-term interest rates, but little, if any, direct effect on long-term rates. It also suggests that long-term rates play a critical role in a num ber of im por tant econom ic decisions, such as firm s’ decisions about investment, and households’ decisions about purchases of homes and oth er durable goods. Theories of the term stru ctu re may help explain the m echanism by w hich m onetary policy affects these decisions.2 A third reason econom ists are interested in the term stru ctu re is that it may provide inform a tion about the expectation s of participants in financial m arkets. These expectations are of considerable interest to forecasters and policy m akers. M arket participants’ beliefs about what may happen in the future influence their cu r ren t decisions; these decisions, in turn, help determ ine what actually happens in the future. Thus, knowledge of participants’ expectations is critical to forecasting future events or determ in ing the effects of different policies. Many econom ists believe that the people best able to forecast events in a m arket are in fact the participants in that m arket. If this is true, interest rate forecasting and inferring the nature of financial m arket participants’ expecta tions amount to the same thing. The term stru c ture theory that will be described in this article, w hich is called the ex p ectation s theory, suggests that the observed term stru ctu re can indeed be used to in fer m arket participants’ expectations about future interest rates—and through them, what actual future rates might be, and how events that tend to influence these rates may unfold. These events could include changes in the rate of econom ic grow th o r changes in m onetary policy, fo r example. The goal of this article is to provide a simple but thorough description of the expectations theory. The first section of the article lays the groundw ork by explaining the basic concept and principles of interest rates and securities pricing. The presentation emphasizes issues that are particularly relevant to understanding how Term structure theories are traditionally stated in terms of nominal or money interest rates. Economic theory predicts, however, that it is primarily real interest rates—interest rates net of expected inflation—that influence the decisions of households and firms. It is possible to formulate versions of most term-structure theories, including the theory described in this article, that apply specifically to real interest rates. Since we cannot observe inflation expectations, however, http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis the financial m arket goes about assigning differ ent interest rates to securities with different term s. The second part of the article presents the expectations theory itself. The presentation is oriented around two widely noted observa tions about the term stru ctu re: (1) that yield curves are usually upward-sloping, and (2) that the steepness and/or direction of their slopes tends to change systematically as interest rates rise and fall. BUILDING B LO C K S O F T H E T ER M ST R U C T U R E Prices, Interest Rates and Time Since the expectations theory tries to explain certain aspects of the way interest rates are determ ined, it is impossible to understand the theory w ithout a thorough understanding of the nature and role of interest rates. A good starting point is the analogy we drew earlier betw een the prices of goods and services and the interest rates on securities. In our economy, purchasers of goods or services almost always pay with money, so the “p rice” of a given quantity of goods is simply the num ber of dollars paid for it. In m arkets w here the goods are readily divisible and m ore or less uniform in quality, such as m arkets fo r agricultural commodities, the price is usually thought of as a num ber of dollars p e r unit of goods. This way of thinking about prices reflects what econom ists call the Law of One Price: w hen inform ation is readily available and the num ber of buyers and sellers is large, each transaction involving a particular good tends to take place at the same unit price, regardless of the quantity of the good exchanged. Discount and Return Ratios—In the securi ties m arket, one can think of lenders as buyers, and of future paym ents as the items they pur chase. People lend to the federal governm ent, for instance, by buying U.S. Treasury securities, which are governm ent promises to repay the loans by making one or m ore future payments. T he direct securities m arket counterpart of a price in a goods m arket would be the num ber we cannot measure real interest rates directly. This makes it difficult to describe real-interest-rate versions of the theories in terms non-economists are likely to understand. 38 of dollars lent (paid) today per dollar repaid in the future (future dollar purchased).3 A security that cost $10,000 and returned $12,500 at a later date, fo r example, would have a unit price of 0.80. This price might be called a discount ratio.4 Econom ists usually conform to financial m ar ket practice by thinking about securities in term s of retu rn rath er than discount ratios— that is, ratios of am ounts repaid to the amounts lent, ra th er than the reverse. W e can define the return ratio on a single-payment security as the ratio of its m aturity paym ent to its price (that is, the amount lent). The retu rn ratio on the security ju st described would be 1.25—the reciprocal of its discount ratio. Accounting f o r the Time Dim ension— The retu rn ratio, it turns out, is not a very good analogue to the m arket price: it suffers from a serious problem that is directly connected to the topic of this article. In a competitive m arket, we think of the unit price as capturing all the price inform ation a prospective buyer needs to allow him to decide w h eth er to buy a particular good. Stated differently, a buyer should be in different betw een tw o purchases that take place at the same price.5 This raises the question of w hether a lender will actually be indifferent b etw een making tw o loans (purchasing two securities) that have the same retu rn ratio. Suppose, for instance, that a lender has a choice b etw een making a $10,000 loan that repays $12,500 at the end of two years, and a $10,000 loan that repays $12,500 at the end of five years. Each of these loans has the same retu rn ratio. W hich is he likely to choose? It seems fairly obvious that our hypothetical lender will p refer the form er of these loans to the latter: the form er loan repays the same amount at an earlier date. The fact that the two loans have identical retu rn ratios is not enough to m ake this lender indifferent betw een them . 3For the moment, we will make the (inaccurate) assumption that all loans/securities return a single payment at a tixed maturity date. T he retu rn ratio is flawed because it neglects an im portant aspect of securities transactions that is absent from most goods transactions. This aspect is the tim e dim ension. A securities transaction is an exchange that takes place over an interval of time, and the length of the in ter val is im portant to the parties in the tran sac tion. Lenders are likely to be less interested in the total am ount to be repaid than in the amount to be repaid per unit o f time. How can we adjust the retu rn ratio to take the time dimension into account? If all loans had the same term , no adjustm ent would be needed. Fortunately, any loan with a term of m ore than one period can be expressed as a sequence of one-period loans with identical one-period retu rn ratios. A five-year loan, for example, can be expressed as a sequence of five one-year loans with a com mon annual retu rn ratio. W e can use these annual-equivalent retu rn ratios to com pare the retu rns on loans with different term s. In order to be m ore con crete about this state ment, we need to define some notation. Let’s call the cu rren t date “date 0 ” and the m aturity date of a given security "date N,” so that the term of the security is N periods. From now on we will think of the periods as years; this is convenient, but not essential. Let V0 represent the am ount lent and VNthe am ount repaid. The retu rn ratio on the loan is thus VN/V0, and the p er-p erio d (usually annual) return ratio is:6 W e can com pute this ratio fo r any single paym ent loan, as long as we know the amount lent, the amount repaid and the term . It pro vides us w ith exactly what we are looking for: a num erical yardstick that can be used to 6The symbol “ = ” should be read “ is equal, by definition, to.” 4Since prospective lenders always have the option of storing their money, the discount ratio should always be less than one. (No lender with this option will make a loan that returns less money than he lent.) 5We must assume that the goods do not differ in quality, and that price information is freely available. We must also assume that the goods are readily divisible, so that any quantity can be purchased at the given unit price. These are standard assumptions in the theory of competitive markets. JULY/AUGUST 1992 39 com pare the retu rns on any two loans, regard less of their term s.7 To conform to financial m arket practice, we must modify the annual retu rn ratio a little fu rth er. M arket participants like to divide the repaym ent on loans into two com ponents: one equal to the am ount lent, w hich is called the principal, and another representing the rem ainder, w hich is called the in terest.8 They m easure the retu rn on loans as ratios of the interest to the principal. In our notation, m arket participants think of these retu rns in term s of net return ratios V - V , r — N ° _ VN V„ v„ Unfortunately, the net retu rn ratio suffers from the same problem s of term com parison as the retu rn ratio. However, we can define a net p er-p erio d (again, usually annual) in terest rate by r= a person who com es to the m arket offering to m ake a fixed repaym ent, at a fixed date in the future, will be able to borrow . If we let VN rep resen t the repaym ent a b orrow er promises to make exactly N years in the future, then he will be able to borrow (sell his security for) an amount V0, w here V„ V0 = (l + r*) This is the basic form ula fo r “pricing” (or dis counting) securities. So far, we have assumed that all loans/securities retu rn a single paym ent at a fixed m aturity date. W e know that in practice, how ever, most secu rities retu rn multiple payments at multiple future dates. As long as the am ounts and dates of these payments are known, we can simply price them separately and sum them to obtain the security’s total price, or p resen t value 1=R-1, , , which is a per-period version of r. The annual interest rate serves as the financial m arket’s basic m easure of the attractiveness of the retu rns on securities. Very often it is converted into a percentage by multiplying it by 100. If the annual interest rate truly serves as the analogue of the m arket price fo r securities, we can expect that in a com petitive m arket it will be determ ined by the interaction of supply and demand. Financial m arket participants will face a m a rk et interest rate r * , w hich they will view as beyond their pow er to influence, and will make their borrow ing and lending decisions accordingly.9 Pricing Securities The annual interest rate form ula can be used to determ ine the price of a security: the amount 7Suppose we construct a sequence of one-period loans represents the {(V0, v,), (Vr v2)....... (VN_,, VN)>, where amount lent at date j, and Vj + 1 the amount repaid one period later. This sequence has the properties that (1) the amount lent at date 0 is V0, (2) the amount repaid at date N is VN and (3) the amount repaid on the tth loan in the sequence, at any intermediate date t + 1, is identical to the amount lent on the t + 1st loan, which is extended at the same date. (Thus, the loans are “ rolled over” from date to date.) Properities (1) through (3) guarantee that, from the lender’s point of view, this sequence of one-period loans is identical to the multiperiod loan. It turns out that only one sequence of loans satisfies these three properties and is consistent with our requirement that the return ratios on each loan be http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis V1 V2 VN A V V t l + r* (l + r*)2 (l + r*) ti (l+r*) v„--------+ -------- - + ... + ------- - - 2_j-------- ,■ The present value of a sequence of futu re pay m ents is the cu rren t m arket value o f those pay ments, w h ere the m arket value is determ ined by discounting the futu re payments back to the present at the m arket interest rate. Here, V represents the paym ent at the end of any date t (if th ere is no paym ent at a particular date t, we say that V; = 0) and 1/(1 + r * ) 1 represents the discount factor applied to that payment. S e c o n d a r y M a r k e t P r i c i n g —W e are now ready to confront a pair of questions that are crucial in understanding the term structure. First, suppose the ow ner of a security wants to sell it before it com es due—that is, in the seco n d ary m arket. How m uch can he expect to receive fo r it? identical. This is the sequence produced when each succes sive one-period loan is extended at a return ratio of R, as defined above. 8Part of the reason for this is that, as was noted above, any one contemplating making a loan has the option of “ lending to himself” by simply storing the money. As a result, people are unlikely to make loans unless the dollar repayment exceeds the dollar principal—that is, unless they receive interest. 9Hereafter, the “ * ” superscript signifies that this particular value of the annual interest rate r is the one selected by the market. 40 T he key to answ ering this question is to recognize that from a lender’s point of view, a security purchased in the secondary m arket is essentially identical to (is a p erfect substitute for) a security he might purchase in the p rim ary or new issue m arket. The prim ary-m arket substi tute would have a term equal to the rem aining term on the secondary security—the num ber of years the security has left to run. It would retu rn payments in the same amounts, and at the same dates, as the rem aining paym ents on the secondary security—those that have y et to be made and would consequently be collected by the security’s purchaser. W e can use this substitution principle, along with w hat we have just learned about primarym arket pricing, to price a security sold in the secondary m arket. W e will call the date at which the security is sold date T, and the price of the security at that date VT. The rem aining term of the security is then N-T, and its rem ain ing payments are due at dates T + l , T + 2, ... , N - l , N?° The paym ents are consequently due 1, 2, ..., N - T - l , N - T periods in the future, relative to date T. (W e’ll assume that the pay m ent due at date T has already been made.) Continuing our notational convention that sub scripts rep resen t dates, w e’ll let r* denote the m arket in terest rate at date T. W e can then write V„ V 1 + r* , VT (1 + r *)2 V., (l + r*)N_T r ti (1 + r*)' It is im portant to note that r*, the m arket rate on the date w hen the security is sold, may be different from the m arket rate w hen the security was issued (which we will call r^ . If r* is relatively low then the secondary m arket price VT will be relatively high, and vice versa. This dependence o f cu rren t secondary m arket prices on cu rren t in terest rates (and of future secon dary m arket prices on future interest rates) will 10Some of these payments may be zero. In the case of a single-payment security, for example, there is only one remaining payment; it is received at date N. "T h e fact that this equation is not linear rules out standard algebraic solution methods. If the security in question has only two payments remaining (if N - T = 2), the equation can be transformed into a second-order polynomial equa tion and solved using the quadratic formula. play a key role in our ultimate explanation fo r the slope of the yield curve. Interest Rates and Yields—The securities pricing form ula just presented can be used to help us tackle a second im portant question. Suppose we have a multiple paym ent security that is selling in the m arket at a know n price. This could be eith er a newly issued security or a security sold in the secondary m arket. W hat is the annual interest rate on the security? Since this security returns multiple payments, we cannot apply the annual interest rate formula that was presented on page 39. W e can, however, exploit the fact that the annual interest rate on this security must be the rate that gives it its c u r ren t m arket price—that is, the rate that makes the present value of its stream of futu re pay ments equal to its m arket price. Consequently, the m arket interest rate r* must solve the equation VT + l VT= — — + l + rT ( l + r T) v„ (1 + r T) Here, VT is the price of the security—which we are now assuming that we know—and VT+1, VT+2, ... , VN are the rem aining payments on the security. Since this equation has only the single un known rT, we might expect to be able to solve it to obtain r *.11 This is usually accomplished using num erical methods. These methods pro ceed by starting with a guess fo r r*, computing the associated present value, and adjusting the guess according to the sign and size of the dif feren ce betw een this value and the actual m ar ket price. An annual interest rate computed in this m anner—that is, as the solution to a present value equation—is called a y ield .11 W e have now—finally!—learned enough to begin investigating the term stru ctu re o f in ter est rates. One way to start is by constructing a y ield curve: a diagram w hich, as noted above, displays the relationship betw een the rem aining term s of, and the yields on, different securities. 12For most of the rest of this paper the terms “ interest rate” and “ yield” will be used interchangeably. Unfortunately, participants in financial markets compute what they call interest rates on securities in a variety of ways, and some of them are significantly different from yields. These differ ences can be particularly important for securities with terms of less than a year. For details, see Mishkin (1989), pp. 82-92. JULY/AUGUST 1992 41 A problem we m ust confront in doing this is that many factors other than d ifferent rem ain ing term s can cause d ifferences in the yields on securities. These include d ifferences in credit risk (that is, in the likelihood of default by the borrow er) and in tax treatm ent. To isolate yield differences that are due solely to term differences, we need to com pare the yields on securities that do not differ in these other characteristics. One simple way to do this is to com pare the yields on securities issued by the U.S. Treasury. T reas ury securities are issued with a wide variety of term s and are traded in a large and active secondary m arket—a fact that m akes it possible to obtain a secondary m arket yield quotation for virtually any term . Treasury securities can also be thought of as essentially riskless, since the federal governm ent is the only organization in the United States that can legally print money to cover its debts. Finally, the interest on all these securities is taxed on the same basis. T H E E X P E C T A T IO N S T H E O R Y W hat does econom ic theory have to say about the term structure? As with most questions in econom ics, th ere are a num ber of differing views. The theory described below, how ever, is accepted, at least in part, by most econom ists interested in m onetary and financial issues. It is called the expectations theory.13 A basic challenge fo r term stru ctu re theory is to explain two em pirical regularities, or “stylized facts,” of the interest rate term structure. These regularities can be described as facts about the slope or steepness of the yield curve at d iffer ent points in time. One of them involves the direction the yield curve usually slopes: most of the time, the yield curve is gently upwardsloping. A nother involves circum stances that seem to produce curves with unusual slopes: when short-term interest rates are relatively high, the yield curve is often downward-sloping; w hen short-term rates are relatively low, the curve is often steeply upward-sloping. 13Early statements of the expectations theory include various works of Irving Fisher [see the citations listed by Wood (1964), p. 457, footnote 1]. The theory was elaborated by Keynes (1930), Lutz (1940) and H icks (1946); these authors proposed a variant of the expectations theory that has become known as the liquidity premium theory. This variant will be described at some length below. The most promi nent alternative to the expectations theory is the market segmentation theory of Culbertson (1957). Another variant of the expectations theory, which combines elements of both the liquidity premium and market segmentation http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Linking Short-Term and LongTerm Interest Rates A point o f departure fo r the expectations theory is the role of secondary m arkets in tran sform ing the effective term s o f securities. Suppose, fo r example, that a lender owns a five-year Treasury bond w hich he purchased in the pri m ary m arket. The bond is maturing, but the lender now wishes he had lent fo r 10 years. If he takes the maturity payment on his fiveyear bond and uses it to purchase a second five-year bond, he will, in effect, have lent fo r 10 years. The only difference betw een this and the single 10-year loan is that the rate of retu rn the lender receives over the coming five years will be determ ined by cu rren t m arket condi tions, rath er than conditions five years ago. Suppose, conversely, that this lender owns a 10-year Treasury bond w hich he purchased five years ago. He has now decided that he needs his money and would have p referred to have lent for five years. If th ere w ere no secondary m arket, he would be stuck: he would not be repaid by the T reasury until the bond m atured five years in the future. The secondary m arket allows him to receive early repayment indirectly, by selling his bond to another lender. If he chooses to sell the bond, he will, in effect, have lent for five rath er than 10 years. The only difference betw een this and a true five-year loan is that the amount of the repaym ent (the sale price of the bond) will depend on cu rren t m arket condi tions, rath er than conditions five years ago. Now suppose (rather unrealistically) that there is no uncertainty about future interest rates, so that lenders today know exactly what m arket yields on securities w ith different term s will be five years in the future. Suppose fu rth er that they know that the futu re five-year Treasury yield will be identical to the cu rren t five-year yield—say, 7Vz percent. How will this affect the cu rren t yield on 10-year Treasury securities? theories, is the preferred habitat theory of Modigliani and Sutch (1966). 42 W e can answ er this question by process of elimination, ruling out possibilities that are clearly w rong until we are left with a single one that must be right. Suppose first that the cu rren t yield on 10-year Treasury bonds is higher than 7V.2 percent. W e have seen that if a lender sells such a bond after five years, the yield to m aturity its buyer will receive must be exactly the same as the yield on a newly issued five-year bond he might purchase instead. This future five-year yield, we have assumed, will be exactly 7Vi percent. Consequently, the (five-year) yield the original lender will receive w hen he sells the 10-year bond, after holding it fo r five years, must be higher than 7V2 percent: oth er wise, the bond's 10-year yield, which is the average of its yields fo r the first and second five years, could not exceed that figure. But if it is possible to obtain a five-year yield of m ore than 7% percent by purchasing a 10-year bond and selling it after five years, why would any cu rren t lender buy a newly issued five-year bond, or a secondary bond with five years left to ru n—each of which, according to our assump tions, will yield exactly 7Vz percent? Clearly, if five-year bonds are to survive in the cu rren t market, the cu rren t yield on 10-year bonds must not in fact be higher than 7V.2 percent. Now suppose that the cu rren t yield on 10-year bonds is low er than 7 V2 percent. Then if a lender buys a five-year bond today, he will receive a yield over five years that is higher than the 10-year yield. If he wants to lend for 10 years, he can use the m aturity paym ent on the first five-year bond to purchase a second five-year bond. Since we have assumed that the yield on this second bond will be exactly 7xh percent, the average yield he receives over the 10-year period will also be exactly 7 V2 percent. This average yield is higher than the 10-year bond yield, how ever; consequently, no cu rren t lender will buy a 10-year bond. If 10-year bonds are to survive in the cu rren t m arket, their yields must not in fact be low er than 7 V2 percent. W e have just seen that if five- and 10-year bonds are to coexist in the m arket, the 10-year bond yield can be neither higher nor lower than the five-year bond yield. This means, of course, that it m ust be equal to the five-year yield. An argum ent of the same sort could be applied, with equal ease, to any long term , and any pair of short term s that sum to it. Thus, under these assumptions, i f lenders know that short-term rates will rem ain constant in the f u ture, cu rren t long-term rates m ust b e equal to cu rrent short-term rates, so that the yield curve will be perfectly flat. Now suppose that instead of knowing the fivey ear rate will rem ain constant fo r the next 10 years, we know it will rem ain constant (at 7Vi percent) fo r five years, and then rise to 10 p er cent. W hat m ust the cu rren t rate on a cu rren t 10-year security be? Notice that if a lender p u r chases a five-year bond yielding 7V.2 percent to day, and rolls it over fo r a second five-year bond yielding 10 percent, he will receive an average annual rate of 8% p ercen t over the 10-year period. Under the circum stances, he would be foolish to lend for ten years at any rate low er than 8% percent. Conversely, sup pose that the U.S. Treasury w ishes to borrow fo r a period of 10 years. If it borrow s by issu ing a five-year bond and then rolls the loan over for a second five years, it pays an average annual rate of 8% percent. Clearly, it would be foolish to o ffer m ore than 8% p ercen t on its 10-year bonds. Extending this argument to different long term s and different com binations of short term s that sum to them leads to the following prediction: i f there is no uncertainty about f u t u r e interest rates, cu rrent long-term rates m ust be an appropriately w eighted average o f cu rrent and fu t u r e short-term rates. Notice that, fo r the purposes of this predic tion, a “long” term does not have to be long by conventional standards. A two-year rate, for instance, m ust be a weighted average of cu rren t and future one-year rates, while a six-month rate must be a weighted average of cu rren t and futu re three-m onth rates, etc. Clearly, it would be helpful to have a baseline "very short-term ” rate to organize these sorts of predictions around. A natural candidate would be the rate on a riskless security with a term o f zero. W hat kind of security has a zero term ? One example would be a security on w hich you can get your money back at any time. W e have securities like this in the form of dem an d deposits or checking accounts. W hile these deposits are not issued by the U.S. Treasury, the fact that they are insured by the federal governm ent makes them virtually as safe as Treasury secu rities.14 W e can consequently 14Strictly speaking, this is true only for personal deposits, and only up to a maximum of $100,000 per deposit. JULY/AUGUST 1992 43 define the baseline interest rate, r°, as the rate on a perfectly safe zero-term security and iden tify it in practice with the m arket rate on fed er ally insured bank deposits.15 A yield curve drawn under the assumption that lenders know that the base rate will fall in the near future (that K is not very large, and that r" < r°) is displayed in figure l . 17 W e can now state a m athem atical rule for determ ining the rate of interest on a security w ith a term of N as a function of the base rate r°. (We must continue to assume that financial m arket participants know the levels of future rates.) Let r° rep resent the cu rren t rate of in ter est on a federally insured demand deposit. Let r" rep resen t the value of this same rate beginning at date K, when th ere will be a one time, perm anent change in the rate. Let rep resent the cu rren t rate on a security with a term of N. (We will re fe r to r j as a termadjusted rate; the rationale fo r this usage will becom e clear later in the paper. Notice that we are letting subscripts rep resen t dates, and superscripts term s to maturity.) Then The assumption that lenders have complete and p erfect knowledge about future interest rates is not very realistic. A m ore reasonable assumption might be that th ere is some u n cer tainty about future rates, but that lenders know their ex p ected values—that is, their best fo re casts, given the inform ation available. If lenders base their decisions entirely on these best fo re casts, then form ula (*) is still a valid description of the expectations theory provided that the rate r°K is interpreted as the expected value of the term -zero rate at date K. People who b e have like this—those who base their decisions entirely on the forecast provided by the expect ed value—are said to be risk neutral. = r “, 0 < N < K, and (*) r°K + r°(N -K ) r? = ----------- ----------, N > K. N The coefficients K of the cu rren t base rate r “, and N - K of the futu re base rate r°, are the appropriate weights referred to in the italicized prediction on page 42. Here, K is the num ber of years at w hich the base rate will stay at its original level r°, and N - K is the num ber of years at which it will stay at its new level r”. W hile this form ula has been stated for the case in w hich m arket rates will change only once, it is easy to generalize to cover the case of m ulti ple base rate changes.16 15A complication arises because demand deposit accounts do not pay interest, while functionally equivalent checkable accounts [negotiated order of withdrawal (NOW) accounts and money market deposit accounts (MMDAs), for example] are interest-bearing. Most economists believe that demand deposits pay interest indirectly, since banks that issue them typically do not charge fees that cover the costs of main taining the accounts and providing funds transfer (check ing, etc.) services. These issues are discussed and the implicit interest rates on demand deposits estimated by Klein (1974) and Dotsey (1983), among others. We will in terpret r° as this implicit demand deposit rate, or, equiva lently, as the explicit interest rate on NOW accounts or MMDAs issued by institutions that do charge cost-covering fees. Under this interpretation, r° will be a positive number. '^Suppose we know that the base rate will change at future dates K 1, K2....... KJt and that the new base rates at these dates will be r° , r ° ........r° . For notational convenience, call the current date (heretofore date 0) date KQ. Then the http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis S y s t e m a t i c s l o p e c h a n g e s —W e can now ex plain one of the two em pirical regularities iden tified in the introduction: the fact that yield curves tend to be steeply upward-sloping w hen w hen short-term interest rates are low and often slope downward w hen short-term rates are high. Before we can do this, however, we need to consider w hat we m ean w hen we say that interest rates are “low” or "high.” Is a 20 percent short-term rate high, fo r example? In the United States, the answ er to this question is almost certainly "yes.” In Israel, or Argentina, how ever, the answ er to the same question would almost certainly be "n o.” This is because in recen t U.S. history in terest rates have rarely risen as high as 20 p ercen t and, w hen they have done so, have quickly returned to low er levels. In recen t Israeli or Argentinian history, current term-adjusted rate on a security with a term of N (N > Kj) will be given by ~ J-1 Both formulas (*) and (**) are approximations of the exact formulas. For details, see the shaded insert on the following page. 17Along the horizontal axis in figure 1, N" represents a partic ular term longer than K, and r[* the term-adjusted rate on a security with that term. Since NT is fairly close to K, the weighted average that determines is strongly influenced by the K years at which the base rate will remain at its original, high level r°. As the term lengthens, the influence of this period wanes and the term-adjusted rate gets closer and closer to the new, lower base rate r£. 44 T h e E x a c t F o r m u la L in k in g S h o rt- an d L o n g -T e rm R a te s Both form ula (*) and the generalized version presented in footnote 16 are linearized approxi mations of the exact formula. The exact version of form ula (*) states that, if r° is the cu rren t base rate, and r “ is the base rate at date K, then the cu rren t N-period term -adjusted rate rN 0 satisfies the relationship (l + rN)N= (l + r “)K (l + r«)N- K, w hich implies r^ = ^ (l + r “)K (1 + r “)N_K - 1 . If we know the base rate will change at futu re dates K,, K2, ... , K,, and that the new base rates at these dates will be r ° , r “ , ... , r° [again, fo r notational convenience, calling the cu rren t date (date 0) date K0, and the term inal date (date N) date KJ+1], then the cu rren t term -adjusted rate on a security with a term of N (N > Kj) satisfies the relationship Fortunately, the approximations given by the linearized formulas are adequate fo r most purposes. In the case described on pp. 42 of the text, fo r instance, the yield given by the exact form ula is 8.743 percent, com pared to the linearized figure of 8.750 percent. The expectations theory can also be shown to imply that, if r* is the cu rren t N-period term -adjusted rate, and r* is the cu rren t K-period rate, then r£ _K, the term -adjusted rate on (N -K )-period securities that is expected to prevail at date K, satisfies the relationship (1 + r£ -K)N_K= (l + r£)N/(l + r*)K, w hich implies that rK N- * = N~ y ( l+ r ^ ) N/(l + r«)K - 1 . J+ l (1+r^)N= TT (1+rt )K-K— 0 [h e r e ] i = 1 V . is the multiplicative analogue of This implies by contrast, rates have rarely fallen as low as 20 percent and, w hen they have done so, have quickly returned to higher levels. W hen we say that interest rates are high or low, w hat w e usually m ean is that they are high or low relative to recen t historical experi ence, and that we feel this experience gives us a good deal o f guidance about the level of inter est rates in the future. Thus, w hen we say in terest rates are high we usually expect them to fall in the future, and vice versa. As w e have just seen, the expectations theory predicts that when we expect rates to fall the yield curve I. T he rate r£*-K is often referred to as the "K-period forw ard rate” on a security with a term of N -K . The expectations theory is often described as a theory that identifies the forw ard rate with the expected future spot rate. will slope downward, and that w hen we expect them to rise the curve will slope upward. The simple expectations theory has the virtue of great flexibility: if you are willing to make sufficiently artful assumptions about lenders’ ex pectations about the pattern of future interest rates, you can use this theory to explain the shape of virtually any yield curve. The theory provides an explanation fo r one basic empirical regularity about yield curves that is rath er diffi cult to believe, how ever. The regularity in ques tion is that most of the time, during the last century at least, yield curves have been distinctly JULY/AUGUST 1992 45 Figure 1 Term-adjusted Yield When the Base Rate Will Fall in the Future Yield upward-sloping.18 The simple expectations the ory could explain this only by assuming that lenders usually expect rates to rise persistently over time. This assumption does not seem plau sible; unless you believe that lenders w ere ex trem ely poor forecasters. W hile interest rates have varied considerably during the past century, th ere is little evidence that they have increased on average, or that m arket participants had any reason to expect them to do so. Indeed, the evidence suggests that people usually expect future short-term interest rates to rem ain near cu rren t levels.19 W hat we need, then, is a modified version of the theory that will predict 18See Malkiel (1970), pp. 5-6, 12; Kessel (1965), pp. 17-19; and Shiller (1990), p. 629. It is sometimes asserted that yield curves were usually downward sloping during the late nineteenth and early twentieth centuries: see Meiselman (1962), appendix C, and Homer and Sylla (1991), pp. 317-22, 403-09 for descriptions and explanations of this phenomenon. 19The simple statiscal models of interest rate behavior that explain the data best are based on the assumption that http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis an upward-sloping yield curve under this assumption. Interest Risk, Term Premia and the Slope o f the Yield Curve Any alternative explanation for the fact that yield curves are norm ally upward-sloping must be based on something about long-term securi ties that makes them system atically less attrac tive to lenders than short-term securities. As we have just seen, the expectations theory predicts that, if lenders know fo r certain that short-term interest rates will rem ain constant, they should rates have a long-run average or mean level and tend to return toward that level, rather slowly, after departing from it. These models imply that, if short-term interest rates are currently near their mean level (where they should be most of the time), they should be expected to stay near the cur rent level in both the short and the long run, and that, even if they are far from the mean level, they should be expected to stay near the current level in the short run. 46 be indifferent betw een lending by purchasing short-term securities and lending by purchasing long-term ones. Long- and short-term interest rates should consequently be equal, and the yield curve should be flat. This prediction implies that any alternative explanation fo r the upward slope of the curve must be based on the effects of u ncertainty about future interest rates. V = ------- -------T (1 + r * ) N_T If r* = r*, so that in terest rates have not changed since this security was issued, its price will be V = ------- ------ ■ T I n t e r e s t R i s k a n t i C a p it a l L o s s e s —One reason why uncertainty about interest rates may influence the behavior of lenders is that unan ticipated changes in interest rates affect the val ue of th eir securities in the secondary m arket. Suppose, to retu rn to an earlier example, that a lender buys a 10-year security that retu rns a yield of 7V2 percent and sells it in the secondary m arket after five years. If interest rates have rem ained unchanged in the interim, the secon dary m arket price of his security will give him a five-year yield of 7V2 percent. If they have ris en, the price will be lower, and he will receive a low er yield. As we have already noted, the reason for these price and yield changes is that a security sold in the secondary m arket must com pete with prim ary m arket securities with the same term as its rem aining term . If the m arket in ter est rate on prim ary securities has risen, the yield on secondary securities must rise to the same level; since the rem aining paym ents on these securities are fixed, this rise can be arranged only through a decline in the secu ri ties’ m arket price. A form al way to see this is by inspecting the secondary m arket pricing form ula for a single-payment security: 20The expectations theory offers no explanation for the reasons market participants might expect short-term rates to change. It is a theory that attempts to explain the levels of long-term interest rates relative to the current levels of short-term rates, not one that attempts to explain their absolute levels. Stated differently, the expectations theory is not a true theory of the determination of interest rates. Market participants may expect short-term interest rates to change because they expect changes in any of the innumerable factors economic theory predicts might influence them. Economic theory suggests that interest rates of the sort discussed in this article (money or nominal interest rates) are sums of real interest rates (rates expressed in terms of the purchasing power of the dollar amounts lent and repaid) and expected rates of inflation. This is the so-called Fisher equation. As a result, the question of interest rate determination is sometimes thought of as two questions: what determines real interest rates, and what determines inflation expectations. Most economists believe that nomi nal factors (such as changes in the levels or growth (l+ror - T It is easy to check, by applying the annual in ter est rate form ula, that both the T-year ex post yield on this security (the yield from date 0, when it was issued, to date T, w hen it is sold) and the N-T year ex ante yield (the yield from date T, w hen it is sold, to date N, w hen it will mature) are equal to the initial rate r* W e will call VT the anticipated price of this security. If the actual price VT exceeds the anticipated price VT, we say the original lender has experienced a capital gain. The amount of the gain is simply VT- V T. If the anticipated price falls short of the actual price, the lender has experienced a capital loss in the amount VT- V T. It is clear from our pricing form ula that capital gains occu r if r* falls short of r* (if m ar ket interest rates have fallen), and vice versa. This m eans that lenders' expectations about future capital gains and losses must be tied to their expectations about future interest rates. W hat should we assume about expectations regarding futu re in terest rates? As we noted to ward the end of the previous section, it seem s reasonable to assum e that m arket participants recognize that interest rates may change, but expect them to rem ain constant on average.20 rates of monetary aggregates) play the principal role in driving inflation expectations, while real factors (such as technological changes, changes in the perceived attractive ness of investment opportunities, changes in demographic structure or changes in the nature of financial regulation) play the principal role in real interest rate determination. There is, however, considerable disagreement about the degree of interaction between nominal and real factors, and especially about whether changes in nominal factors can have persistent effects on real interest rates. JULY/AUGUST 1992 47 Under this assumption, the expected capital gains on future secondary m arket sales of secu rities are approxim ately zero .21 It seems conceivable that this situation might not bother lenders. Economists usually assume, how ever, that the satisfaction a person derives from an extra dollar’s w orth of expenditures declines as the total value of his expenditures increases. If this is so, he will find the gain in satisfaction provided by the extra goods he can purchase if his retu rns exceed his expectations to be sm aller than the loss in satisfaction from the goods he will have to refrain from purchas ing if his retu rns fall short of his expectations. This should cause actuarially fair (zero expected loss) u ncertainty about the future retu rns on his securities to upset him. A person who behaves like this is said to be risk averse. Since buying term securities exposes lenders to actuarially fair retu rn uncertainty, while buy ing securities with zero term s (such as demand deposits) does not, risk averse lenders will be reluctant to buy term securities. They will insist on higher expected yields on term securities than on demand deposits to com pensate them selves for the uncertainty. The notion that financial decisionm akers are risk averse is wide ly accepted by econom ists, and we will adopt it without fu rth er discussion. I n t e r e s t R i s k a n d t h e T e r m S t r u c t u r e —W e have just explained why term securities tend to have higher yields than demand deposits w hen both are default-free: term securities carry in terest risk, but demand deposits do not. W e have not yet explained why securities with longer term s tend to have higher yields than those with sh orter ones. Our discussion certainly suggests a possible explanation, however: longer-term securities may carry m ore interest risk than shorter-term ones. But why should this be the case? W e will begin our investigation of this ques tion by posing another question that is closely related. Suppose w e have two single-payment securities with different term s, but the same original (date 0) prices and yields. If m arket interest rates rem ain unchanged, their cu rren t (date T) prices will also be identical, even 21Since the secondary market price is computed by dividing the maturity payment by the gross interest rate 1 + r, an increase in the rate by a given percentage causes a fall in the price that is slightly smaller than the rise in the price caused by an equal percentage decrease in the rate. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis though their m aturity payments will not be. But suppose that the m arket interest rate— specifically, the m arket “base rate” r°—rises by a fixed amount from date 0 to date T (so that r “ = r° + Ar, with Ar > 0). W hich security will fall furthest in price? Notice that the rem aining term of the short term security will be sm aller than that of the long-term security; if we call the short term Ns, and the long term N,, then the remaining term s of these securities are N .- T and N ,-T , resp ec tively. Since m arket yields have risen, the sh ort term secondary security must generate extra interest to com pete with newly-issued sh ort term securities. The amount of extra interest will be approximately ArVT(Ns- T ) ; this is the rate increase Ar, applied to the (common) secon dary m arket price VT, fo r each year of the re maining term (Ns- T ) . T he long-term security must also generate extra interest; in this case, the amount is ArVx(N ,-T ). This is the same rate increase, applied to the same base price, but continued fo r N ,-N s additional years. Of course, neither security can really produce “extra in terest” in the conventional sense. The interest is paid indirectly, as part of the maturity payment, and the tim e and date of that payment are fixed. Instead, the price of each security m ust decline far enough so that it can increase at the new (and higher) annual rate r°, while still reaching the fixed m aturity paym ent VN at the m aturity date N. Since the price of the long term security will have to increase at this rate for a much longer time, it will have to fall m uch fu rth er than the price of the short-term security. The relative sizes of the two price declines will be approxim ately equal to the relative sizes of the securities’ rem aining term s. A security with fou r years left to run will su ffer a price decline approximately double that of a security with the same secondary m arket price but only two years left to run, and so on. T h e T e r m P r e m i u m —If the risk o f capital loss on securities tends to increase in proportion to their rem aining term s, lenders who demand interest com pensation for bearing this risk will demand m ore com pensation on long-term As a result, the expected price change is slightly positive. Although this effect is never very strong, it becomes more pronounced as the remaining term of the secondary security increases. 48 securities than on short-term secu rites.22 This will tend to make the yields on longer-term securities higher than those on securities with sh orter term s—that is, it will tend to make the yield curve upward-sloping.23 W e can define the term p rem iu m on Treasury securities o f a given term as the difference betw een the yield on those securities and the yield on federally insured demand deposits. That is, TN = r N- r°, or equivalently r N= r° + T N, w here r N rep resents the yield on N-term T reas ury securities, and tn represents th eir term premium. W e now have a theory that predicts that the term premium should increase syste matically with the rem aining term , and, m ore specifically, that it should increase in proportion to the rem aining term . W e can form alize this by w riting tn= t (N) =m N , w here m is a positive constant of proportionality. A plot of the sort of yield curve consistent with this prediction is displayed in figure 2. W e might refer to the num ber m as the coefficient o f risk aversion. D ifferent values of m can be thought of as indicating different degrees of lenders’ risk aversion. If m is rela tively high, a small increase in the term and, thus, in the risk of capital loss, will cause lend ers to demand a good deal of com pensation in the form of a large increase in the term prem i um. This is the kind of behavior we would ex pect from very risk-averse lenders. If m is low, on the oth er hand, it will take a large increase in the term , and, thus, the risk to cause lenders to demand m uch additional compensation. 22ln reality, the increase in risk is slightly less than propor tional to the term, but the deviation from exact proportional ity is very small. We are implicitly assuming that the change in the base rate, if any, will occur at a known future date, and that the rate, having changed, will remain at its new level permanently. We are also assuming that T, the date of sale, is fixed and known. 23Early statements of the liquidity premium theory include Keynes (1930), Hicks (1946) and Meiselman (1962). The term “ liquidity premium” is based on the notion that liquidity—the ability to sell an asset rapidly and without loss— is valuable to lenders, and lenders will charge interest premia on assets that are relatively illiquid. Since the risk of capital loss is the risk that an asset may ultimately be saleable only at a loss, the premium for capital loss risk is in a sense a liquidity premium. This is the kind of behavior we would expect from lenders who are not very risk-averse.24 It was pointed out earlier that lenders may not always expect the level of short-term interest rates—in particular, the level of the term -zero rate—to stay constant on average. W hen they do not, the base rates to w hich the term premia must be added will also depend on the term . These term -dependent base rates have been referred to as termadjusted rates, and their cu rren t values have been denoted rN 0. The actual yield should be the sum of the term -adjusted rate and the term premium: A b n o r m a l Y ie ld C u r v e s —This latest addition to the expectations theory allows us to consider the role of the term prem ium in determ ining the shape of abnorm al yield curves—the sort that appear w hen lenders expect in terest rates to change in the future. In this case, the actual yield should be given by the sum o f the termadjusted rate (that is, the weighted-average base rate) and the appropriate term premium. This can produce curves that slope in one direction along one part of their range, but in the oppo site direction along another part. If lenders expect interest rates to rem ain constant fo r a short period, and then fall sharply, fo r example, the yield curve will appear humped, sloping upward at very short term s, peaking n ear the term corresponding to the date at w hich rates are expected to decline, and sloping downward fo r a range of term s th e rea fter (see figure 3).25 Curves with this shape are frequently observed shortly before econom ic recessions begin, presum ably because interest rates tend to fall sharply during recessions. 24lf m = 0, lenders do not require any compensation for the risk of capital loss. As noted earlier, lenders who behave in this manner are said to be risk-neutral. 25Note that if a normal yield curve (a hypothetical curve observed when interest rates are expected to remain con stant, on average) is upward-sloping, the expectations theory does not always interpret an upward-sloping yield curve as an indication that the market expects interest rates to rise. To obtain the right directional signal, the slope of the observed yield curve must be compared to the slope of a normal curve. The theory now interprets an ob served yield curve that is upward-sloping, but flatter than normal, as a signal that the market expects interest rates to fall slightly. JULY/AUGUST 1992 49 Figure 2 Yield Curve When the Base Rate Is Constant Yield Figure 3 Yield Curve When the Base Rate Will Fall in the Future Yield http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 50 CONCLUDING R EM A R K S This article presents a basic description of the concepts and issues involved in the study of the term stru ctu re of interest rates. It has also presented a simple version of the expectations theory of the term structure. This theory p re dicts that the shape of the yield curve is deter mined by the expectations of financial m arket participants about the level of futu re interest rates and by their u ncertainty about the accuracy of their expectations. The analysis presented h ere suggests that the expectations theory can help explain two im por tant "stylized facts” about yield curves: the fact that the steepness and direction of their slopes tend to vary system atically with the level of short-term interest rates, and the fact that they are usually upward-sloping. The explanation for the form er fact is that forward-looking lenders will refu se to purchase term securities unless long-term in terest rates are averages o f the short-term interest rates that the lenders expect at various points in the future. The explanation fo r the latter fact is that the interest risk on securities tends to increase with their term s; this causes risk-averse lenders to demand amounts of in terest com pensation that also increase with the term s. R EFEREN C ES Culbertson, John M. “ The Term Structure of Interest Rates,” Quarterly Journal o f Economics (November 1957), pp. 485-517. Dotsey, Michael. “ An Examination of Implicit Interest Rates on Demand Deposits,” Federal Reserve Bank of Richmond Economic Review (September/October 1983), pp. 3-11. Hicks, John R. Value and Capital, 2d ed. (Oxford University Press, 1946). Homer, Sidney, and Richard Sylla. A History of Interest Rates, 3d ed. (Rutgers University Press, 1991). Kessel, Reuben A. “ The Cyclical Behavior of the Term Structure of Interest Rates,” National Bureau of Economic Research Occasional Paper #91 (1965). Keynes, John M. A Treatise on Money, Vol. 1&2 (Harcourt, Brace and Company, 1930). Klein, Benjamin. “ Competitive Interest Payments on Bank Deposits and the Long-Run Demand for Money,” American Economic Review (December 1974), pp. 931-49. Lutz, Friedrich A. “ The Structure of Interest Rates,” Quarterly Journal of Economics (November 1940), pp. 36-63. Malkiel, Burton G. “ The Term Structure of Interest Rates: Theory, Empirical Evidence, and Applications,” Monograph (General Learning Corporation, Morristown, New Jersey), 1970. Meiselman, David. The Term Structure of Interest Rates (Prentice Hall, Inc., 1962). Mishkin, Frederic S. The Economics of Money, Banking, and Financial Markets, 2d ed. (Scott, Foresman and Company, 1989). Modigliani, Franco, and Richard Sutch. “ Innovations in Interest Rate Policy,” American Economic Review (May 1966), pp. 178-97. Russell, Steven H. “ Reference Notes on Interest Rates, Securities Pricing, and Related Topics,” Unpublished monograph, 1991. Shiller, Robert J. “ The Term Structure of Interest Rates,” (with an appendix by J. Huston McCulloch) in Benjamin M. Friedman and Frank H. Hahn, eds., Handbook of Monetary Economics, Vol. 1 (North Holland: Elsevier Science Publishing Company, Inc., 1990). Wood, John H. “ The Expectations Hypothesis, the Yield Curve, and Monetary Policy,” Quarterly Journal of Economics (August 1964), pp. 457-70. JULY/AUGUST 1992 51 Mark. D. Flood Mark D. Flood is an economist at the Federal Reserve Bank of St. Louis. James P. Kelley provided research assistance. The author would like to thank Bob Eisenbeis, Mark Flannery, Carter Golembe, and George Kaufman for many helpful com ments. All remaining errors are the author’s. The Great Deposit Insurance Debate In the stress o f the recent banking crisis ... there was a very definite appeal f r o m bankers f o r the United States G overnm ent itself to insure all bank deposits so that no depositor anyw here in the country n eed have any f e a r as to the loss o f his account. Such a guarantee as that would indeed have put a p rem iu m on bad banking. Such a guarantee as that would have m ade the G overnm ent pay substantially all losses which had been accum ulated, w hether by m isfortune, by unwise ju d gm en t, o r by s h e e r recklessness, and it might well have brought an intolerable bu rd en upon the Federal Treasury. —Sen. Robert Bulkley (D-OH), Address to the U. S. Cham ber of Commerce, May 4, 1933.1 T he only dan ger is that having learned the lesson, we m ay fo rg e t it. H um an nature is such a fu n n y thing. W e learn som ething today, it is im p ressed upon us, and in a f e w short years we seem to fo rg e t all about it and go along and m ake the sam e m istakes over again. —Francis M. Law (1934), p. 41. T h e ONGOING PROLIFERATION of bank and th rift failures is the forem ost cu rren t issue for financial regulators. Failures of federally insured banks and thrifts num bered in the thousands during the 1980s. The problem is especially im portant for public policy, because of the poten tial liability of the federal taxpayer. For example, by 1989, the Federal Savings and Loan Insur ance Corporation (FSLIC) was so deeply overex tended—on the order of $200 billion—that only the U. S. Treasury could fund its shortfall. The significance of insurance is seen elsew here as well: econom ists are quick to point to flat-rate deposit insurance as a factor in causing the high failure rates. Flat-rate deposit insurance is said to create a m oral hazard: if no one charges 'Quoted by Sen. Murphy (D-IA) in Congressional Record (1933), p. 3008. JULY/AUGUST 1992 52 bankers a higher rate for assuming risk, then bankers will exploit the risk-return trade-off to invest in a riskier portfolio. Why, then, do we have taxpayer-backed, flatrate deposit insurance?2 A simple answ er would be that the legislators who adopted federal de posit insurance in 1933 did not understand the econom ic incentives involved. This simple answ er seems wrong, how ever. It has been pointed out that certain observers articulated the problem s with deposit insurance quite clearly in 1933. In this view, the fault lies with the policym akers of 1933, who failed to heed those warnings. This fails to answ er why policym akers would ignore these argum ents. M oreover, it does not explain why it should have taken almost 50 years for the flaws in deposit insurance to take effect. This paper exam ines the deposit insur ance debate of 1933, first to see precisely what the issues and argum ents w ere at the time and, secondarily, to see how those issues w ere treat ed in the legislation. Briefly, I conclude that the legislators of 1933 both understood the difficul ties w ith deposit insurance and incorporated in the legislation num erous provisions designed to mitigate those problem s. The Banking Act of 1933 separated commercial and investm ent banking, limited bank securities activities, expanded the branching privileges of Federal Reserve m em ber banks, authorized fed eral regulators to rem ove the officers and direc tors of m em ber banks, regulated the paym ent o f interest on deposits, and increased minimum 2The Federal Deposit Insurance Corporation (FDIC) has re cently announced a move toward risk-adjustment of its in surance premia. 3The Act is often called the Glass-Steagall Act. It is referred to here as the Banking Act of 1933 to avoid con fusion with the separate Glass-Steagall Act of 1932. Sig nificantly, it also has the longer official title: “An Act to provide for the safer and more effective use of the assets of banks, to regulate interbank control, to prevent the un due diversion of funds into speculative operations, and for other purposes.” The temporary plan was later extended, and the perma nent plan delayed, for one year (to July 1, 1935) by the Act of June 16, 1934. The Banking Act of 1935 substantial ly emended the permanent plan to resemble closely the temporary plan. See the shaded insert on page 72 for fur ther details of the various plans. 4The Federal Reserve did not adopt an official position, although there is some evidence of opposition: “ Deposit guaranty by mutual insurance is not part of the Presiden tial program, nor is it favored by Federal Reserve authori ties,” “ Permanent Bank Reform” (1933); see also Kennedy (1973), pp. 217-18. Comptroller O’Connor favored deposit insurance; former Comptroller Pole opposed it. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis capital requirem ents for new national banks, among num erous lesser provisions. It also estab lished a temporary deposit insurance plan lasting from Jan u ary 1 to July 1, 1934, and a perm a nent plan that was to have started on July 1, 1934.3 Although this paper focuses on deposit insurance, it is im portant to b ear in mind that both the deposit insurance provisions of the bill and the debate that surrounded them each had a larger context. The various provisions of the Banking Act of 1933 constituted an interdepen dent package. The deposit guaranty provisions of the bill w ere initially opposed by President Roosevelt, C arter Glass (Senate sponsor of the bill and Con gress’s elder statesm an on banking issues), T re a sury Secretary Woodin, the American Bankers Association (ABA), and the Association of Re serve City Bankers, among others.4 Despite this opposition, on Ju n e 13, 1933, the bill passed virtually unanimously in the Senate, with six dissents in the House, and was signed into law by the President on Ju n e 16.5 Not surprisingly then, the public debate preceding and surround ing the adoption of federal deposit insurance was active and far-reaching. This paper is organized around the m ajor them es of the debate: the actuarial questions concerning the effects of deposit insurance, the philosophical and practical questions of fairness to depositors and of depositor protection as an expedient means to financial stability, and the political and legal questions surrounding bank chartering and supervision. Much of the debate 5The Senate did not record a vote, although even Sen. Huey Long (D-LA), who had been a flamboyant detractor, rose to speak in favor of the bill. Cummings (1933) claims that the Senate vote was unanimous. The House dis senters included Reps. McFadden (R-PA), McGugin (RKS), Beck (R-PA) and Kvale (Farmer/Labor-MN). See “ Congress Passes and President Roosevelt Signs GlassSteagall Bank Bill as Agreed on in Conference” (1933), p. 4192. Rep. McGugin’s request for a division revealed 191 ayes and 6 noes; a quorum of 237 was reported present; Congressional Record (1933), p. 5898. 53 was motivated by econom ic and political selfinterest and was structured rhetorically in term s of m orality and justice. Considerable at tention is paid h ere to rhetorical detail.6 As m uch as possible, I have attem pted to report the debate in its own term s—liberal use is made of quotations and epigraphs—ra th er than risk m isconstruing the meaning through inaccurate paraphrase. BACKGROUND T O T H E D E B A T E The banking debate in 1933 covered not only deposit insurance and the separation of com m ercial and investm ent banking, but the full catalogue of financial m atters: the gold stan dard, inflation, m onetary policy and the con trac tion of bank credit, interstate branching, the relative m erits of federal and state charters, holding company regulation, etc. By 1933, nearly anything to do with banks or banking was an im portant political issue. The Great Contraction The people know that the Federal Reserve octopus loaned ... to the gamblers o f this Nation in 1928 some sixty billion dollars o f credit money—bank money—hot air ... and then when the crisis came in the last 3 months o f 1929, cut that credit money—bank money—hot air—down to thirteen billion. No nation, no industry, can survive such an expansion and contraction o f money and credit. Give to me the power to double the money at will, and then give me the power to cut it square in two at will, and I can keep you in bondage,7 It is reasonable to begin a recollection of the debate over deposit insurance w ith the price collapse on the New York Stock Exchange of 6Most of what remains of the debate is formalized oratory: prepared speeches, Congressional debate, letters to the editor, etc. Because the debate was a cacophony of voices, rather than an orderly dialogue, no attempt has been made to present the arguments in chronological ord er. A time line of the significant events of 1933 is provided in the shaded insert on page 55. In terms of the written record, academic economists en tered the debate late, for the most part after the Banking Act of 1933 had already been signed into law. See H. Preston (1933), Westerfield (1933), Willis (1934), Willis and Chapman (1934), Taggart and Jennings (1934), Fox (1936) and Jones (1938). Phillips (1992) reports that Frank Knight and several colleagues at the University of Chicago advo cated federal guaranty of deposits as part of comprehen sive bank reforms proposed during the banking crisis in March 1933. W illis had been an advisor to Carter Glass since the debate over the Federal Reserve Act in 1912. Guy Emerson, who published in the Quarterly Journal of Economics, was not an academician, but an officer at Bankers Trust Co. and the 1930 president of the Associa O ctober 29, 1929. The stock m arket crash was popularly recognized as the start of the Great Depression. The rem ainder of the Hoover ad m inistration’s tenure witnessed historic declines in national econom ic activity. By the beginning of 1933, industrial production and nominal GNP had both been cut in half; unemployment had topped 24 percent. Bank failure rates, which had already been high throughout the 1920s, had increased fourfold, while both money sup ply and velocity had plummeted. The price level fell accordingly. For contem porary econom ic com m entators, the stock m arket crash was m ore than a m ar ker betw een historical eras. For many, there was a causal relationship betw een the stock m arket’s collapse and subsequent real econom ic activity. In most cases, this causality was m ore elaborate than p o st h o c ergo p r o p te r hoc. A p re scient Paul W arburg, for example, w arned in M arch 1929: If orgies of unrestrained speculation are per mitted to spread too far, however, the ultimate collapse is certain not only to affect the specu lators themselves, but also to bring about a gen eral depression involving the entire country.8 The logic was that stock m arket speculation “ab sorbs so m uch of the nation’s credit supply that it threatens to cripple the country’s regular bus iness.”9 A m ore radical theory was advanced by the "liquidationists,” who held sway in influen tial circles of governm ent and the academ y.10 For them, the cyclical contraction was a good thing: it reflected the liquidation of unsuccess ful investments that crept in during the boom years, thus freeing econom ic resources for a m ore efficient redeploym ent elsew here. tion of Reserve City Bankers; Emerson (1934) is largely a paraphrase of Association of Reserve City Bankers (1933), which he co-authored. 7Rep. Lemke (R-ND), Congressional Record (1933), p. 3908. W arburg (1929), p. 569. 9lbid., p. 571. 10De Long (1990) provides a valuable review of the liquidationist perspective. The liquidationists included Secretary of the Treasury Andrew Mellon, as well as the economists Friedrich von Hayek, Lionel Robbins, Seymour Harris and Joseph Schumpeter. More recent economic analyses have discounted the role of the crash in causing the Depres sion, emphasizing instead other forces, both monetary and non-monetary; see Wheelock (1992a) and the references therein. JULY/AUGUST 1992 54 Crisis and Unlimited Possibility We are confused. We grasp, as at straws, fo r the significance o f events and o f proposed govern ment action. Never before in our lives have we had such great need fo r someone to interpret un derlying movements fo r our guidance.11 By 1933, the correlation betw een econom ic ac tivity and bank credit was lost on no one. Dur ing the interregnum betw een Hoover's electoral loss in November 1932 and Roosevelt’s inaugura tion in M arch 1933, what had been a debilitat ing banking malaise becam e a desperate crisis. Starting with Michigan, on Valentine’s Day, whole states began to declare official bank holi days; elsew here, individual banks in scores w ere suspending withdrawals. By inauguration day, M arch 4, m ost states had declared a holiday.12 Even much earlier, bank failures had left whole towns w ithout norm al payment services, rele gating them to b a rte r.13 Theories of the connection betw een bank failures, m onetary contraction and the m ore general m acroeconom ic torpidity w ere w ide spread and varied. Roosevelt, in his inaugural address, suggested that the set of people who correctly understood the nation’s econom ic probblems did not overlap with the set of people who had held the reins: Their efforts have been cast in the pattern of an outworn tradition. Faced by failure of credit they have proposed only the lending of more money. Stripped of the lure of profit by which to induce our people to follow their false lead ership, they have resorted to exhortations, pleading tearfully for restored confidence. They know only the rules of a generation of selfseekers. They have no vision, and when there is no vision the people perish.14 To some extent, such a suggestion was accurate; Treasury Secretary Mellon and the liquidationists had initially refused even to admit that there was a problem. Some proposed that complex intrigues w ere at work to sap the nation’s wealth. Rep. Lemke (see the quote referen ced in footnote 7), for ex ample, advanced a m onetarist thesis that both the boom of 1929 and the Depression w ere the intentional result of Federal Reserve policy. More conspiratorial still was Rep. M cFadden’s belief, advanced on the House floor, that "money Jew s” lay behind the banking crisis.15 Rep. W eideman, offering the m etaphor that "the most dangerous beasts in the jungle make the softest approach,” claimed that “international money lenders” had duped the Congress into creating a system for skimming bank gold reserves into a central pool "to feed the maw of international speculation.”16 Alarm generated by the crisis and frustration at the lack of a rem edy com bined to expand the political horizons. Radical solutions w ere sug gested. Inform ed by the political experim ents under way elsew here, relatively sober proposals w ere submitted to scrap the inefficient b u reau cracies of representative dem ocracy in favor of a fascist dictatorship or state socialism .17 More 11Love (1932), p. 25. 14Roosevelt (1938), p. 12. 12Before deposit insurance, banks in financial trouble were generally treated like any other business. Closure might be declared by supervisors or the directors of the bank. One option was then to seek protection from depositors and other creditors by declaring bankruptcy and accepting a court-appointed receivership. In the case of a temporary liquidity problem, a bank might instead suspend withdraw als or close to the public until the problem could be resolved. In practice, the terms “ failure” and “ suspen sion” were often used interchangeably. In the period 1921-32, roughly 85 percent of failed banks— holding 76 percent of the deposits in failed banks—were state banks (including mutual savings banks and private banks). See Bremer (1935), especially footnote 1 and pp. 41-49. See Federal Reserve Board (1934a), Colt and Keith (1933) or Friedman and Schwartz (1963) for a chronology of the banking crisis and the bank holidays. In a sense, Roosevelt had stage-managed the crisis. By refusing to participate with the outgoing administration over the bank ing situation, he projected the image of making a clean break with the past. At the same time, however, the result ing uncertainty surrounding his policy toward banking and the gold standard helped to provoke the crisis. See Kennedy (1973), pp. 135-55, or Burns (1974), pp. 31-51. 15McFadden lost his House seat over the incident. Scandal ized by his comments, the Republican and Democratic parties, both of which had endorsed him in 1932, repudiat ed him in the 1934 elections. See Martin (1990), p. 249, and Rep. McFadden (R-PA), Congressional Record (1933), pp. 6225-27. 13See “ What’ll We Use for Money?” (1933). http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 16Rep. Weideman (D-MI), Congressional Record (1933), pp. 3921-22. Weideman, in a conspiracy theory shared by the radio priest, Fr. Coughlin [see Chernow (1990), pp. 381-82], also claimed that the Great War had been or chestrated by international financiers, noting: “ Six months after the Federal Reserve Act was passed the war began.” 17See, for example, Ogg (1932), Calverton (1933) and Schlesinger (1960). Indeed, for many, the New Deal was state socialism. One must bear in mind that 1933 predat ed most of the failures and atrocities of the various Euro pean dictatorships. Although the collectivization of Soviet agriculture was largely complete, Stalin’s great political purges did not begin until the mid-1930s. Mussolini was still widely respected as the man who had brought order and unity to Italy; the invasion of Abyssinia was not until 1935. In Germany, Hitler was only beginning to wrest con trol from the notoriously ineffectual Weimar republic; he became Chancellor in late January 1933, and the Nazis burned the Reichstag four weeks later. 55 A C h ro n o lo g y 1/10/33 Sen. Huey Long’s filibuster of the Glass legislation begins. 1/21/33 Senate filibuster ends. 1/30/33 Hitler becom es Chancellor of Germany. 2/20/33 Prohibition repealed. 3/4/33 Franklin D. Roosevelt is inaugurated. 72nd Congress 2nd session ends. Senate of the 73rd Congress convenes in special session. 3/6/33 President Roosevelt declares a nation-wide bank holiday, lasting nine days. 3/9/33 Congress convenes in extraordinary session (first session of the 73rd Congress). The Em ergency Banking Act is introduced, passed and signed into law. 5/15/33 C arter Glass introduces S. 1631. 5/17/33 Henry Steagall introduces H. R. 5661. 5/19/33 A rthur Vandenberg introduces an amendment to the Glass bill. 5/23/33 House passes Steagall bill. 5/26/33 Senate passes Glass-Vandenberg bill. 5/27/33 The Securities Act of 1933 signed. 6/12/33 W orld M onetary and Economic C onference opens in London. 6/13/33 C onference com m ittee submits a con feren ce report on the Banking Act to Congress. Banking Act of 1933 is approved by Congress. 6/16/33 President Roosevelt signs the Banking Act of 1933. First session of the 73rd Congress adjourns. 9/4/33 9/17/33 1/1/34 ABA Convention begins in Chicago (ends 9/7/33). ABA President Frank Sisson dies. Federal Deposit Insurance Corporation is chartered. Tem porary deposit insurance begins. popular was a flirtation with governm ent by "tech n ocracy,” a small panel or cabinet of ex perts to replace the congressional and executive branches. Relative to alternatives such as these, federal deposit insurance—w hich had failed in Congress m ore than 150 times in the preceding 50 years—was a rem arkably m oderate option.18 M oral Overtones to the Debate The money changers have fled fro m their high seats in the temple o f our civilization. We may now restore that temple to the ancient truths. 18See FDIC (1951) and Paton (1932); Paton also cites H. R. 7806, introduced by Rep. Cable (R-OH) on January 15, 1932, and later revised as H. R. 10201. H. R. 7806 is omit ted from the FDIC (1951) digest. The m easure o f the re s to ra tio n lies in the extent to w h ich we a p p ly social values m ore noble than m ere m o n e ta ry p r o f it . 19 Both proponents and opponents of the deposit guaranty features of the Banking Act took the rhetorical high ground in arguing their point. Indeed, recou rse to morality in public debate was widespread. The "noble experim ent” with the prohibition of liquor was still an issue in the 1932 election.20 O ratory was laden with biblical imagery. Sen. Vandenberg (R-MI) referred to “ Prohibition was widely recognized as having failed by this time; see Kent (1932), p. 261. The Eighteenth Amendment was repealed in 1933. 19Roosevelt (1938), p. 12. JULY/AUGUST 1992 56 “B. C. days—which is to say, Before the Crash. ...”21 A. C. Robinson saw fit to lecture sub scribers to the ABA Jo u rn a l on the “Moral Values of T h rift,” advising bankers of the need fo r “an unshakeable conviction of these ideals [truth and m orali ty] and th eir ultimate trium ph. 'If thou faint in the day of adversity, thy strength is sm all.’ ”22 For many, the Depression represented an atonem ent for the excesses of the bull m arket. By all accounts, 1929 was characterized by stock m arket speculation.23 As the extent of the ava rice becam e clear w ith hindsight, the notion of econom ic depression as punishm ent for econom ic transgression took hold: We are passing through chastening experiences, as severe for the banker as for anyone else, many of the illusions have disappeared and the trappings of a meretricious prosperity have been stripped from most persons.24 The notion of recession as a necessary purga tive unfortunately extended to policym akers as well. Mellon’s advice to Hoover exposes the pi ous foundations to the liquidationist view of the Depression: It will purge the rottenness out of the system. High costs of living and high living will come down. People will work harder, live a more moral life. Values will be adjusted, and enter prising people will pick up the wrecks from less competent people.25 This fluency with righteousness revealed itself on all sides of the deposit insurance debate. Both proponents and d etractors of the deposit guaranty provisions of the Banking Act argued that their position w as ultimately a m atter of simple justice, w hich dare not be denied. The bankers declared that well-managed banks should not be forced to subsidize poorly run 2’ Vandenberg (1933), p. 39. 22Robinson (1931), p. 209. 23“ Orgy of speculation” was the catch phrase that captured the popular sentiment. For example, “ Our Orgy of Specu lation” (1929), p. 907, quotes Chancellor of the Exchequer Philip Snowden: “ There has been a perfect orgy of specu lation in New York during the last twelve months.” 24Robinson (1931), p. 209. 25Hoover, quoted in De Long (1990), p. 5. Bankers Magazine offered it as a modern paradox, “ that depressions are sent by heaven for the chastening of mankind.” See “ Modern Paradoxes” (1933). The liquidationists drew a sardonic retort from Keynes, who identified it as sanctimony masquerading as economics: “ It would, they feel, be a victory for the mammon of unrighteousness if so much prosperity was not subsequently balanced by universal bankruptcy.” See Keynes (1973), p. 349. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis banks. Supporters of the legislation maintained that depositors should not have to b ear the loss es accruing to their bankers’ mistakes. Those who felt that deposit insurance was a ploy to destroy the dual banking system painted a pic ture of the unit bank as the pillar of the na tional economy, untainted by corruption. The rem ainder of the paper is organized around these th ree loosely defined constituencies. A C TU A R IA L D IF F IC U L T IE S Opposition to deposit insurance can be roughly organized into two classes: objections on technical actuarial grounds, and objections to its anticipated impact on bank structure. The core constituency in the form er category consisted of the moneycenter banks, with ABA President Francis Sisson, himself a Wall Street banker, taking the lead.26 The economic motivation for their opposition was the belief that insurance meant a net transfer from big banks, where the bulk of deposits lay, to state-chartered unit banks, where they expected the bulk of the losses. Insurance and Guaranties In the law as written the guaranty plan is referred to not as a guaranty o f bank deposits, but as an insurance plan. There is nothing in this plan that entitles it to be classed as insurance.27 I think you gentlemen are all wrong to call this a guarantee o f deposits. There is not a thing in the bill that talks about guarantee. It is an insurance o f deposits.28 The actuarial correctn ess of the term “deposit insurance” as a description of th e proposed legis lation was a point of contention. The alternative label, offered by opponents, was "deposit guaran ty.” One’s choice of term s usually revealed w here Mellon’s advice also offers an example of a common tendency to anthropomorphize the economy, in this case as a system to be purged. For a more extreme example, see Taussig (1932), who draws an elaborate analogy be tween physicians and economists. 26The ABA (1933a) dissected the failure of the various state insurance schemes. The Association of Reserve City Bankers (1933) published a monograph late in the debate outlining the actuarial objections to deposit insurance. 27Association of Reserve City Bankers (1933), p. 27. 28C. F. Dabelstein, in ABA (1933b), p. 58. For similar re marks, see Rep. Beedy (R-ME), Congressional Record (1933), p. 3911; Sen. Glass (D-VA), ibid., p. 3726-27; and Donald Despain, quoted by Sen. Schall (R-MN), ibid., p. 4632. 57 one stood on the issue, and the semantic con troversy becam e a m icrocosm of the actuarial issues involved.29 By labeling the various schem es as plans to “guaranty” deposits, opponents w ere able to associate the plans immediately w ith the infelicitous recen t experience w ith state deposit guaranty schem es (discussed in the next subsec tion). The natural response for supporters was to insist on a different label. Both proponents and opponents devoted en er gy to identifying the desirable "insurance princi ple,” w hich then either accurately described or failed to describe the proposed legislation.30 Like blind men describing an elephant, however, few agreed on a definition for the insurance princi ple. This was so, despite Rep. Steagall's claim that the principle of insurance was "the most universally accepted principle known to the business life of the w orld.”31 Deposit insurance was clearly similar in many respects to other types of insurance, which had been in widespread use in the United States for decades. Even the m ost ardent detractor recog nized some resem blance: The general argument employed to promote the guaranty plan began with the premises that property can be insured and bank deposits are property. It travelled to the broad assumptions that the principle of the distribution of risk through insurance could be applied to bank deposits.32 The salient principles here, espoused repeatedly by supporters of the legislation, w ere the diver sification o f risk and the diffusion of losses. In 29The FDIC (1951), p. 69, provides a clear distinction be tween insurance and guaranty. By their definition, a guaranty is a promise from the U. S. government to pay off depositors in a failed bank; insurance is paid from an independent private fund. There was no agreed definition for insurance or guaranty in 1933, however, although the explicit acknowledgement that “ no clear distinction [be tween the terms ‘guaranty’ and ‘insurance’] has been made,” was rare; see Rep. Bacon (R-NY), Congressional Record (1933), p. 3959. W. B. Hughes also attempted to extricate the “ inexcusable mixture of the two terms ... Guarantee is where you make the good bank pay for the poor one. Insurance is where you make those who get the benefit pay for it.” See ABA (1933b) p. 59. I use the two terms interchangeably in this article. 30ln fact there were numerous conflicting legislative proposals afoot. That of Henry Steagall, who chaired the House Banking Committee, was taken most seriously; it eventually became law. See FDIC (1951) and Paton (1932). 31Congressional Record (1933), p. 3836. 32ABA (1933a), p. 7. 33Sen. Vandenberg, Congressional Record (1933), p. 4239. this respect, a national plan would differ from the state plans, w hich had "violated the prim ary insurance ten et that risks must be decentralized and sufficiently spread so as to avoid con cen trated losses.”33 For others, the distinction betw een govern m ent and private backing defined the difference betw een insurance and guaranty. Both Sen. Glass and Rep. Steagall w ere adamant that cov erage be provided privately, not by the government: This is not a Government guaranty of deposits. ... The Government is only involved in an initial subscription to the capital of a corporation that we think will pay a dividend to the Gov ernment on its investment. It is not a Govern ment guaranty.34 I do not mean to be understood as favoring Government guaranty of bank deposits. I do not. I have never favored such a plan. ... Bankers should insure their own deposits.35 The argum ent against governm ent backing was outlined by Sen. Bulkley.36 An insurance feature included in both the Steagall and Glass bills and in Sen. V andenberg’s tem porary insurance amendment to the Glass bill was a provision fo r depositor co-insurance.37 The Glass and Steagall bills called for a progres sive depositor copaym ent schedule: the first $10,000 would be covered in full, the next $40,000 would be covered at 75 percent, and only 50 percent of amounts over $50,000 would be covered; the Vandenberg amendm ent set a single coverage ceiling at $2,500. Some propo- 34Sen. Glass, Congressional Record (1933), p. 3729. See also footnote 28. 35Rep. Steagall, Congressional Record (1933), p. 3838. 36See the quote referenced by footnote 1. Similar concerns were voiced by Jamison (1933), p. 451: “ The great urgen cy for balancing the national budget precludes even the thought of piling another subsidy on the shoulders of the already overburdened taxpayers.” These sentiments are especially noteworthy in light of recent attempts to paint the insurance schemes as having taxpayer backing from the start. For example, Title IX of the Competitive Equality Banking Act of 1987 states that Congress “ should reaffirm that deposits up to the statutori ly prescribed amount in federally insured depository insti tutions are backed by the full faith and credit of the United States;” (emphasis added). 37Co-insurance is the insurance practice of involving the in sured party in some portion of the risk. Common tech niques of co-insurance are coverage ceilings, deductibles and copayment percentages. The aim of such provisions is to mitigate the problem of moral hazard or the tendency of people to behave more riskily when insured. JULY/AUGUST 1992 58 nents saw no need fo r such mitigating features. Rep. Dingell (D-MI), fo r example, offered bankers no quarter; his idea was "to guarantee every dollar put in by the depositor from now on and to make the banker and the borrow er pay the cost.”38 For Sen. Vandenberg, on the other hand, co-insurance was crucial; he complained angrily w hen Treasury Secretary W oodin proposed "not a limited insurance such as is included in the amendm ent which the Senate adopted, but a complete 100% guarantee.”39 Opponents in the banking industry w ere un impressed by such argum ents. Although all of the proposals achieved a spreading of losses and many had other fam iliar features of insurance, such as co-insurance or provision for a large reserve fund, they still w ere not “insu rance.”40 Francis Sisson was obstinate: “Detailed and tech nical d ifferences in this bill as com pared with form er guaranty schem es do not differentiate it in essential principle from them .”41 For all their trouble, crafters of the legislation had failed to m eet the b ank ers’ standard for insurance, the principle of selected risks: Insurance involves an old and tried principle. The essence of insurance is the payment by the insured of premiums in actuarial relation to the risk involved. Under the terms of the perma nent plan, however, the costs or premiums are not charged according to the risk.42 Roosevelt made a similar connection. In his first presidential press conference, he asserted: 3aCongressional Record (1933), p. 489. More thoughtful com mentators realized that the incidence of the cost could not be contained. Rep. Kloeb (D-OH), ibid., p. 489, challenged Rep. Dingell immediately: “Assuming that an assessment is made upon the bankers, how are we going to prevent that from sifting down to the depositors?” Similarly, Jami son (1933), p. 454, explained that, “ while the banks would remit the premiums,” they would also adjust their interest rates, so that, “ in the end the banks’ customers would pay the premiums.” ssQuoted in “ Congress Passes and President Roosevelt Signs Glass-Steagall Bank Bill as Agreed on in Confer ence” (1933), p. 4193. The proposal itself is surprising, given Woodin’s strong objections to deposit insurance. Many others shared Vandenberg’s view; see, for example, Sen. Glass, Congressional Record (1933), p. 3728; Sen. Bulkley (quoted by Sen. Murphy), ibid., p. 3007. 40There was disagreement about the reserve fund even af ter the legislation had been signed. The Association of Reserve City Bankers (1933), p. 28, asserted baldly that “ no provision is made for building up a reserve fund as would be the case under a true insurance plan,” while Sen. Vandenberg (1933), p. 39, contended that the plan was “ capitalized with truly prodigal reserves” (any irony in his use of the adjective “ prodigal” is doubtless unintend ed). The discrepancy lies in the fact that, unlike Van- http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis I can tell you as to guaranteeing bank deposits my own views, and I think those of the old Ad ministration. The general underlying thought behind the use of the word 'guarantee’ with respect to bank deposits is that you guarantee bad banks as well as good banks. The minute the Government starts to do that the Govern ment runs into a probable loss.43 Although he associates the "guaranty” term inol ogy with governm ent backing, its defining ch ar acteristic is clearly the absence of selected risks. Despite the attention given to selected risks in the debate, no significant attem pt appears to have been made to include a risk-based prem i um in legislation. Em erson, for one, thought such an arrangem ent could w ork.44 The ABA, on the oth er hand, thought it impossible: The apparently unsurmountable actuarial difficulty in the guaranty plan appears to be the impossibility of placing it on the basis of selected risks; the risks involved w ere “wholly unpredictable,” and banks w ere subject to “internal deteriora tion” w hen their deposits w ere guaranteed.43 History and Geography As to the history o f the guaranty plan, a wave o f guaranty o f state bank deposits laws swept over the seven contiguous western states o f Oklahoma, Kansas, Texas, Nebraska, Mississippi, South Dakota and North Dakota and the Pacific Coast state o f Washington in the period 1908-17. ... The laws establishing it were repealed or allowed denberg, the Association of Reserve City Bankers did not treat the FDIC’s capital as an insurance reserve fund. 41Sisson (1933b), p. 31. 42Association of Reserve City Bankers (1933), p. 27, (empha sis in the original). “ Selecting risks” refers to the practice of differentiating insured parties according to risk and charging insurance premia according to those risk class es. For example, 17-year-old men on average pose a great er risk to auto insurers than do 30-year-old men; therefore, 17-year-olds usually pay higher auto insurance premia. 43Roosevelt (1938), p. 37. 44Emerson (1934), p. 244, states, “ To put such a provision [assessments levied according to risk] into effect would re quire the classification of the banks of the country accord ing to various standards: geographical location, size, type, and character of banking policy. The last would present administrative difficulties, but these would not be insuper able.” Bankers Magazine had also thought it feasible: “ Presumably, an insurance company could be formed ..., which by carefully selecting its risks, might operate suc cessfully.” See “ Protecting Bank Depositors” (1931), p. 435. 45ABA (1933a), pp. 42-43. Similarly, Jamison (1933), p. 454, argued that selection of risks in this context would present “ complications that can not be easily overcome.” 59 to become in op erative as one a fte r a n o th e r o f the plans becam e fin a n c ia lly in solve nt an d was recog n iz e d as se rvin g to m ake ba n kin g m a tte rs w o rse.46 A s in the case o f b ra nch banking, N a tio n -w id e d iv e rs ific a tio n o f insurance ris k s w o u ld secure ba n kin g against a n y e ve n tu a lity except such a na tio n a l ca la m ity as w o u ld d e stro y the G overnm ent its e lf.47 The “guaranty” term inology connoted the defunct state deposit guaranty plans, a specter that terrorized the bankers. The m ere m ention of deposit guaranties could induce a banker to show "every sign of incipient apoplexy.”48 At the same time, the unvarying failure of the state plans provided a trove of evidence for foes of the federal schem e.49 Release of the ABA report coincided with the introduction of the Glass and Steagall bills in Congress. It found perverse delight in the failure of all eight of the state plans: Eight large scale tests, by practical working ex perience, of the guaranty of bank deposits plan as a means for strengthening banking condi tions and safeguarding the public interest are a matter of record. Each one of these attempts failed of its purpose. Taken separately, special circumstances such as technical defects in the plan or faulty ad ministration might be held accountable for the breakdown in any given instance, leaving it an open question as to whether the idea might not be successful under different circumstances. Taken as a composite whole, however, the failures of the various plans not only confirm one another in their defects, but each one also 46ABA (1933a), p. 7. The seven states listed are not, in fact, contiguous. 47Rep. Bacon, Congressional Record (1933), p. 3959. “^Stephenson (1934), p. 35. There is a hint of truth in Stephenson’s hyperbole. Francis Sisson died of heart failure within a fortnight of the ABA convention of Septem ber 1933 — which had included excoriating harangues [see Bell (1934)] delivered by Jesse Jones of the Recon struction Finance Corporation and soon-to-be FDIC board member J. F. T. O’Connor; see “ Death of Francis H. Sis son, Vice-President Guaranty Trust Co. of New York and Former President American Bankers Association” (1933), and O’Connor (1933). In a tribute at the next convention, Sisson’s ABA colleagues offered that his death was “ a tragic demonstration of devotion to duty even to the extent of exceeding the physical power of endurance ... He was a martyr to his work in your behalf.” Nahm (1934), p. 30. 49Several groups dissected the state plans in the course of the debate; see American Savings, Building and Loan In stitute (1933), ABA (1933a), Blocker (1929), Boeckel (1932), and the Association of Reserve City Bankers (1933). Refer ence was also made to an earlier essay by Robb (1921). supplies added special features that were tested and found wanting.50 This unbroken string of failures demanded an explanation from supporters of federal legisla tion. Proponents chose to distinguish clearly the new plan from the state schem es: "there is no logical relationship betw een these old State Guarantees and this new F ed era l Insurance; no analogy; no parallel; and no reason to confuse the m ortality of the form er with the vitality of the latter.”51 To make this case, supporters emphasized forem ost the m uch broader geographic—and th erefore industrial—diversification o f a federal insurance fund. "T h e fact that bank-depositguaranty projects have failed in local, restricted areas only proves one of the fundam ental prin ciples of insurance, that is, that th ere m ust exist wide and general distribution and diversifica tion.”52 In particular, the old plans w ere said to have suffered from a “one-crop” problem, that is, their application in states overwhelmingly de pendent upon agriculture: There is a vast difference between what can be accomplished by a small number of banks in one State dependent upon a single crop and what can be successfully accomplished by the banking system of this great Nation that holds the financial leadership of the world in its hands.53 On this point, at least, the bankers w ere forced to concede.54 The bankers revealed the geographic breadth of the federal plan to be a two-edged sword, There are also numerous retrospective accounts of the state guaranty plans, including Calomiris (1989 and 1990), Wheelock (1992b and 1992c), and Wheelock and Kumbhaker (1991); the most comprehensive, however, is Warburton (1959), parts of which appear in FDIC (1953 and 1957). The original legislation is collected in Federal Reserve Board (1925a and 1925b). 50A BA (1933a), p. 7. 51Vandenberg (1933), p. 39, (emphasis in the original). 52Donald Despain, quoted by Sen. Schall, Congressional Record (1933), pp. 4631-32. Virtually identical arguments are offered by Vandenberg (1933), p. 39, and Rep. Bacon, Congressional Record (1933), p. 3959. 53Rep. Steagall, Congressional Record (1933), p. 3838. 54For example, the Association of Reserve City Bankers (1933), pp. 31-32, acknowledged that, “ It is suggested ... that a single crop failure could shake the stability of all the banks in a State. On a national scale the plan would operate upon a broader base. This is true.” JULY/AUGUST 1992 60 Table 1 Estimated Assessments and Losses by Geographic Division Geographic division New England Middle Atlantic North Central Southern Mountain Southeastern Southwestern Western Grain Rocky Mountain Pacific Coast United States Percent of assessments in each division to total assessment 7.6% 44.0 18.6 3.5 2.8 4.3 8.0 1.8 9.4 3.7% 20.0 21.9 5.8 13.7 7.0 20.7 4.5 2.7 100.0% 100.0% how ever, and used it to fight back. They ex ploited the well-known fact that bank failures throughout the 1920s had occu rred dispropor tionately among small, rural banks (see table l) .55 This inform ation was used to argue that, with insurance premia assessed against deposits, the burden of funding federal deposit insu r a n c e -h a d it existed during the 1920s—would have been borne in large m easure by the money cen ter banks of the Northeast, w here m uch of the industry’s deposit base lay. The benefits of insurance, how ever—the payments to cover losses in failed banks—would have gone south and west. Subsidy and Discipline For it is to be remembered that the weak banks get the same insurance as the strong ones, and, unlike the situation in other kinds o f insurance, the bad risk pays no more fo r its insurance than the good one. This means competition among banks in slackness in the granting o f loans. The bank with the loose credit policy gets the busi ness and the bank with the careful, cautious credit policy loses it. The slack banker dances 55The table is reproduced from Association of Reserve City Bankers (1933), p. 26. See Bremer (1935) and Upham and Lamke (1934) for analyses of failures in the 1920s. 56E. W. Kemmerer of Princeton University, speaking to the Savings Bank Association of Massachusetts on September 14, 1933, and quoted in Association of Reserve City Bankers (1933), pp. 40-41. Kemmerer was the economic advisor to the comission that produced the latter. Similar thoughts were offered by Jamison (1933), p. 451: “ Govern http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS Percent of losses in each division during 1921-1931 to total losses and the conservative banker pays the fiddler. If the conservative banker protests, the slack one invites him to go to a warmer climate. Soon all are dancing and the fiddler, if paid at all, must collect fro m the depositors or fro m the taxpayers,56 For those who opposed deposit insurance on actuarial grounds, such technicalities w ere m erely m anifestations of a m ore fundamental is sue. As a m atter of principle, deposit insurance was held to be unjust. It involved the forced subsidization of poorly managed banks by well managed institutions; it subsidized the “bad” b anker at the expense of the “good.” This moral point provided substantial emotional force. Op ponents concluded that only good bank m anage m ent could ultimately assure safe and sound banking. Their argument, founded in actuarial theory and the experience of the state plans, proceeded in two steps. First, by protecting depositors against loss, a deposit guaranty would destroy discipline; insured depositors would take no in terest in the quality of their b an k’s m anage ment. Recalling the state plans, the guaranty had created “a sense of false security and lack of discrimination as betw een good and bad ment guaranty of bank deposits can be but one of two things — an outright subsidy ... or a plan of insurance.” Bradford (1933), p. 538, added: “ Such subsidization of weak banks by the Government, however, carried out on the basis of taxpayers’ money, is so monstrous as to be almost unthinkable.” 61 H $4 9 3 , o o o WORTH OF REGRETS ere rests T happened this way. He was the comptroller of a large corpora tion in New York City— a director of his local suburban bank — a fond father— he had the esteem of friends and business associates alike. T o day he is serving from three and a half to ten years for defrauding five banks and three brokerage houses of $493,000.00. I Wall Street proved his Waterloo. Naturally interested in market move ments, his interest led him gradually into heavy speculation. As the mar ket went down so did he — deeper and deeper. Finally, desperate, he forged stock certificates o f his own company which he used as collateral to bolster his personal brokerage accounts. Then, one day the axe fell. A check up revealed that he had defrauded five banks and three brokerage houses out of $493,000. With the money swallowed up in the greatest bear market of all times, the banks lost every penny. * * * * The stark reality o f these facts de mand eternal vigilance in granting every loan. Particularly in granting commercial loans, make sure that your borrowers are adequately cov ered by Fidelity Bonds. You always insist that your borrowers carry fire insurance to safeguard their physical assets. Ask for the same protection against the possible peculations of their employees. Insist that your loans be protected against the frail ties of hum an nature. For an em ployee, as well as a fire, can wreck a firm. F id e l it y & D e p o s it COM PAN Y O F M A RYLA N D H om e Office: Baltim ore Representatives Everywhere (i r — 11 K l 1 I WA U s**1 Fi del i ty and Surety Bonds banking.”57 In many minds, this dichotomy b e tw een good and bad bankers was the central is sue.58 B an kers M agazine editorialized that “the surest reliance of good banking is to be found in the men who manage the banks rath er than in the laws governing their operations.”59 In 1931, ABA President Rome Stephenson contended that, a large element in the internal conditions of the banks that failed was bad management 57ABA (1933a), p. 13. and that a predominant element in the internal conditions of the bank that remained sound in the face of the same external conditions was good management.60 W hat was needed was to teach "the conception of scientific banking.”61 The second step in the logic of opposition was an objection to the subsidy implicit in a guar anty. In the tones of a prudish parent, the ABA complained that the beneficiaries of state sys tem s had been the "bankers with easier stan dards,” w ho gained competitive advantages over those with “sounder but less attractive m eth ods.”62 The subsidy was especially problem atic among those banks “w hich have little chance of ultimate success.”63 A bank which does not earn a fair average rate of return over a period of years not only is un able to build up reserves against bad times, but, in order to improve profits, is under constant temptation to take risks which in the end are likely to lead to failure. The tendency of a guaranty plan will be to nurture these unprofitable units and keep them going temporarily in the knowledge that upon failure the losses can be shifted to other banks.64 Thus, the subsidy was seen to extend beyond the simple protection of unsound institutions from the competitive pressures of vigilant depos itors. Given their contention that, “no provision is made fo r building up a reserve fund,” losses charged to the insu rer by failing banks would have to be recouped after the fact from the sur vivors.65 Such a system would necessarily entail tran sfers of w ealth from surviving to failed banks. T h ere was no consensus in Congress on the im portance of discipline; some m em bers pointed out that life insurance was no incentive fo r sui cide.66 The fram ers of the Glass and Steagall bills, how ever, recognized the validity of the bankers’ objections and addressed the issue directly. Both bills, as well as the tem porary inmethods to be allowed to conduct banks, were able to command public trust and patronage and to attract large deposits to their institutions through high interest rates and trading on faith in the guaranty plan.” ABA (1933a), p. 17. 58The advertisement above depicts an insurer’s characteri zation of the bad banker. Coincidentally, President Roosevelt had been a vice-president for the Fidelity and Deposit Co. of Maryland after his unsuccessful VicePresidential bid in the 1920 election. “ Association of Reserve City Bankers (1933), p. 29. 59“ Federal Guaranty of Bank Deposits” (1932), p. 381. 64lbid., pp. 19-20. 60Stephenson (1931), p. 592. 65lbid., p. 28. 61Ibid., p. 592. 66See, for example, Rep. Luce (R-MA), Congressional Record (1933), p. 3918. Sens. King (D-UT) and Glass brie fly debated the role of immortality in the context of this analogy; Congressional Record (1933), p. 3728. 62ABA (1933a), p. 25. More specifically, “ greater numbers than ever of undercapitalized, ill-situated banks, as well as of persons wholly unfitted as to training, character or JULY/AUGUST 1992 62 surance am endm ent in the Senate, w ere careful to limit coverage. Sen. Vandenberg stated ex plicitly the rationale fo r coverage ceilings: the State Guarantees involved complete protec tion for a ll banking resources. ... Federal Insur ance, on the other hand, leaves the individual bank and banker so seriously responsible for such a preponderance of their resources that there is no appreciable immunity at all.67 Sen. Glass noted a second source of discipline inherent in the plan. Because the banks insured each other, deposit insurance would "lead to the severest espionage upon the rotten banks of this country that w e have ever had.”68 Under both the tem porary and perm anent plans, the small depositor was to be covered in full, in recognition of his inability to m onitor bank m anagem ent adequately: At present the depositor is at the mercy of his fellow depositors, over whom he has no con trol, and of the management of the bank, about which he is not usually in a position to be well informed. The depositor takes the risks, and the banks take the profits.69 A survey conducted by the Comptroller of the C urrency and the Federal Reserve in May 1933 revealed that the ceiling of $2,500 under the tem porary plan would fully cover 96.5 percent of depositors and 23.7 percent of total deposits in m em ber banks.70 PR O T E C T IN G D EPO SIT S W hile m ost industry opponents fought the deposit insurance plan on actuarial grounds, supporters argued that deposits p e r se required protection, to stabilize the medium of exchange and prom ote a renew ed expansion of bank credit. M ore significantly, proponents responded 67Vandenberg (1933), p. 39, (emphasis in the original). Congressional Record (1933), p. 3728. 69Rep. Bacon, Congressional Record (1933), p. 3959. 68Sen. Glass, 70See Federal Reserve Board (1933c), p. 414. The point to be made was that even the temporary plan succeeded in fully covering the vast majority of depositors. The survey, of course, took place before depositors had an incentive to split larger deposits into multiple accounts to achieve full deposit insurance coverage. 71Rep. Luce, Congressional Record (1933), p. 3914. 72Rep. Bacon, Congressional Record (1933), p. 3952. Comp troller Pole was instrumental in dichotomizing the industry into “ two definite types of banking, namely, that carried on by the small country bank and that of the large city bank.” See “ Comptroller Pole’s Views on Rural Unit Bank ing,” (1930), p. 468. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis with an argum ent of pow erful simplicity: the losses to innocent depositors in a bank failure w ere a plain injustice. Given the status of banks in the political clim ate of 1933, this was a charge that the bankers ultimately could not counter. The Agglomeration o f Deposits f o r Speculation The use o f ba n kin g fu n d s f o r speculation became a stench in the n o s trils o f the people.71 T h ere w as a strong sense that the banking in dustry in the 1920s had functioned as an elabo rate netw ork to collect savings at the local level and funnel them into lending on securities speculation: Another cause for many banking collapses was the domination of smaller banks by their large metropolitan correspondents, which drained funds from the country districts for speculative purposes and loaded up the small bank with worthless securities.72 Indeed, this was a prim ary motivation fo r those sections of the Banking Act requiring a separa tion of com m ercial and investm ent banking. Similar argum ents w ere brought against pro posals for nationwide branch, chain and group banking.73 A sensitivity to such a possibility was doubt less nurtured by the popularity of Ponzi schem es in the 1920s, including the infamous Florida land swindles.74 W ith such analogies in mind, banks cam e to be seen as merely fueling departments in enterprises run not by bankers concerned with operating banks but by promoters whose object was to exploit the credit resources of the bank. ... 73Group banking and chain banking are essentially variants of the modern bank holding company form of organiza tion. Group banking presumed some degree of standardi zation among the subsidiary banks in the holding company, while chain banks were operated as largely in dependent franchises within the holding company. 74A Ponzi scheme is a fraudulent investment plan, such as a chain letter, in which returns to existing investors are paid directly from the deposits of new investors, with the director of the scheme skimming the difference. Some of the Ponzi schemes had been run by Charles Ponzi him self. After several jail terms and a stint on the lam, Ponzi was finally deported to his native Italy in 1934. This was not his first one-way ticket. In 1903, his family had bought him a one-way ticket to Boston on the S. S. Vancouver in a successful bid to get rid of him. See Grodsky (1990). 63 The primary evil in our banks for many years has been the incessant efforts of promoters to get control of the funds which flow into the banks. The bank is the depository of the com munity’s funds and as such is the basis of the available credit of the community. The promoterbanker needs nothing so much as access to these credit pools.75 Such accusations w ere inevitably tinged with at least a hint of the conspiratorial.76 In keeping with this them e, the issues w ere fram ed for popular consum ption as a morality play in which the naive depositor is pitted against the sophisticated banker. The depositor tucks away the hard-earned wages of his honest labor, only to be systematically duped by the cunning intrigues of the banker. At the extrem e, some politicians played the religious card face up: "W e discovered that what we believed to be a bank system was in fact a respectable racket and so many connected with it only cheap, petty loan sharks and Shylocks.”77 In the end, a provi dential governm ent was seen to intercede on beh alf of the depositor, and deposit insurance was trum peted as "the shadow of a great rock in a w eary land.”78 The notion of the small depositor as an inno cent victim had im mense popular appeal. M cCutcheon’s 1931 political cartoon celebrating the blam elessness of the depositor in a failed bank w on the Pulitzer Prize (above right). Such popularity, of course, was plainly evident to politicians, who responded by introducing deposit insurance legislation in Congress. Rep. Steagall is reported to have told House Speaker Garner in April 1932, "You know, this fellow Hoover is going to w ake up one day soon and come in h ere with a message recom m ending guarantee of bank deposits, and as sure as he does, h e’ll be re-elected.”79 75Flynn (1934), pp. 394-96. 76Rep. Steagall, for example, avowed that a “ campaign was turned on urging bankers everywhere to ... employ their facilities in investment banking, in speculation, in stock gambling, and in aid of wild and reckless international high finance.” Congressional Record (1933), p. 3835. The Seventy-first Congress had formed a Senate Banking and Currency Subcommittee to investigate the extent to which the Federal Reserve and National Banking systems had been co-opted to “ finance the carrying of speculative securities.” Sen. Bulkley, quoted by Sen. Murphy, Con gressional Record (1933), p. 3006. See also footnotes 15 and 16 an the related text. 77Rep. Dingell, Congressional Record (1933), p. 3906. 78Rep. Hill (D-AL), Congressional Record (1933), p. 5899. Hill’s pronouncement was met with a round of applause in the House. Reprinted by permission: Tribune Media Services. For obvious reasons, bank failures con cen trat ed the attention of large num bers of voters, and Congressmen w ere anxious to associate them selves with the legislation. Sen. Vandenberg, up for re-election in 1934, was always careful to call his tem porary insurance amendment to the Banking Act of 1933 "The Vandenberg Amend m ent.” Rep. Dingell announced: "guaranty of bank deposits is my baby in M ichigan.”80 A peti tion circulated in the House in June 1933 to postpone adjournm ent indefinitely until a de posit insurance bill was made law.81 Figure 1 79Timmons (1948), p. 179. Garner responded, “ You’re right as rain, Henry, so get to work in a hurry. Report out a deposit insurance bill and we’ll shove it through.” The result was H. R. 11362, which passed the House on May 27, 1932. 80Rep. Dingell, Congressional Record (1933), p. 3906. It is noteworthy that both Sen. Vandenberg and Rep. Dingell were from Michigan, where, on February 14, 1933, William A. Comstock had become the first governor to declare a state banking holiday during the crisis; see Colt and Keith (1933), pp. 6-8. In light of the temporary insurance amend ment, any dispassionate observer would have to regard deposit insurance as Vandenberg’s baby in Michigan. 81See H. Preston (1933), p. 589, and Rep. McLeod (R-MI), Congressional Record (1933), p. 5825. JULY/AUGUST 1992 64 Figure 1 The Cause of Deposit Insurance Bills reveals that the num ber of guaranty bills in troduced in Congress correlated neatly with the num ber of bank failures. Theatrics aside, the central point for propo nents of the legislation rem ained, and it was difficult to refute: "T h e main point is always this—the d ep o s ito r ow n s th e m oney. If he puts it in for safe-keeping it should be safely kept.”82 Indeed, opponents conceded directly that depos itor losses in bank failure w ere unjust.83 In stead, they tried to red irect the debate to the question of “w h eth er the guaranty plan will in fact cure the defects in our banking system and give depositors the safety which they seek and 82Ford (1933), p. 9, (emphasis in the original). Similarly, Sen. Vandenberg stated: “ The savings of America must be made safe.” Congressional Record (1933), p. 4428. The question of legal title to deposited funds was somewhat more subtle than Ford’s quote suggests; see, for example, Amberg (1935), pp. 49-51. 83Amberg (1935), p. 51, felt that the struggle and fear of a bank run per se were bad, and that “ a great social pur pose would be served if the occasion of such fear could be removed.” http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Failures to which they are entitled.”84 On this latter question, the bankers rem ained obstinately negative; they favored “reform methods for banking that really strengthen banking,” and th erefore opposed deposit guaranties.85 The Stabilization o f the Medium o f Exchange We think o f the busy bee and the ant as tireless, but they are loafers com pared with the activity o f a busy dollar.1'6 We got the guarantee o f bank notes after having had wildcat banking in connection with State bank notes and after having had people injured who held notes o f the State banks. ... 84Association of Reserve City Bankers (1933), p. 2. 85Sisson (1933a), p. 563. He added: “ There can be no ques tion that the people of the United States should have a banking structure based on conditions rendering the banks immune from failure.” 86Donald Despain, quoted by Rep. Schall, Congressional Record (1933), p. 4631. 65 It is much m ore important in principle to guarantee bank deposits, because the real cir culating medium o f the country is bank deposits.87 Although, as a strictly political m atter, deposi tor protection was the central motivation responsible for the progress of deposit insur ance in Congress, other forces w ere at issue. Chief among these was the role of banking in the real economy. Regarding bank failures, it was recognized that causality ran two ways: just as the general drop in real incom es had caused loan defaults and thus widespread bank failures, bank failures and the concom itant re striction of bank services had caused real in com es to fall. The latter effect was seen to operate both directly and indirectly. Bank suspensions and failures could trap depositors’ w ealth for a period of m onths or even years until the bank either reopened or its bankruptcy was resolved. T he direct result was reduced consumption and investm ent spending by the affected depositors. In the extrem e case, w hen a tow n’s lone bank failed, even the sim plest form s of exchange could be hopelessly en cum bered: [The unacceptability of failure] would perhaps not be so if they were grocery stores or butch er shops, where failure would be disastrous to only a few people at most: but bank failures paralyze the economic life of whole communi ties, not only through the loss of money ac cumulations but by the destruction of the deposit currency which is the principal medium of exchange in all business activity.88 Under such circum stances, some affected re gions instituted scrip currencies, wooden coinage or system atic b arter arrangem ents, the most elaborate of which was the Em ergency Ex change Association in New York, headed by Leland Olds.89 A depositor's natural response to these possi bilities was to withdraw his funds b efore failure 87Fisher (1932), p. 143. 88Greer (1933b), p. 538. "S e e “ What’ll We Use for Money?” (1933). 90See Ives (1931) for colorful accounts of depositor runs and the various responses of bankers. Rep. Bacon, Congres sional Record (1933), p. 3959, estimated hoarding at $1.5 billion in January 1933. The extent of hoarding was also roughly gauged by tracking deposits in the U. S. Postal Savings system. Such deposits roughly quadrupled in the two years ending June 30, 1933 [see O’Connor (1933), p. 23], Friedman and Schwartz (1963), p. 173, state that such occurred. Both bank runs and the hoarding of cu rren cy received considerable attention.90 W ithdrawals for the purpose of safeguarding one's w ealth w ere deemed unpatriotic; legisla tion was even proposed to outlaw the practice. Banks had a natural response to the th reat of runs: "Credit was tightened in the desire to re main as liquid as possible to m eet the em ergen cies of ru ns.”91 Bankers maintained large cash reserves rath er than lend: It is estimated that banks now have available billions of dollars of collateral for use in extend ing loans, but the plain fact is that for more than 3 years bankers have given little thought to anything except to keep their banks in liquid condition. ... The fear that grips the minds and hearts of bankers, keeping ever before them the nightmare of bank runs, makes it impossi ble for them to extend the credits that are in dispensable to trade and commerce.92 This analysis is confirm ed by the facts. The ag gregate excess reserves of Federal Reserve m em b er banks, for example, had ballooned from $42 million in O ctober 1929 to a peak of $584 mil lion in Jan uary 1933, even though the num ber of m em ber banks had fallen from 8,616 to 6,816 over roughly the same period.93 Thus, bank failures w ere seen to have an indirect ef fect on output, as both depositors and bankers in solvent institutions prepared for the possibili ty of runs and failures. In the final analysis, depositor protection and stabilization of the medium of exchange w ere recognized as opposite sides of the same coin: We may talk about percentage of gold back of our currency, we may discuss technical provi sions of legislation ... The public does not un derstand these technical discussions, but from one end of this land to the other the people un derstand what we mean by guaranty of bank deposits; and they demand of you and me that we provide a banking system worthy of this great Nation and banks in which citizens may place the fruits of their toil and know that a deposits remained a “ minor factor” in spite of their growth. The system was established by the Postal Savings Bill of 1910 and was intended primarily for the savings of new immigrants. Deposits were guarantied in full. VicePresident-elect Garner reportedly told Roosevelt, “ You’ll have to have it [deposit insurance], Cap’n, or get more clerks in the Postal Savings banks.” See Timmons (1948), p. 179. 91Rep. Bacon, Congressional Record (1933), p. 3959. 92Rep. Steagall, Congressional Record (1933), p. 3840. 93Federal Reserve Board (1943), pp. 72-74, 371. JULY/AUGUST 1992 66 deposit slip in return for their hard earnings will be as safe as a Government bond. [Ap plause.1 They know that banks cannot serve the pub lic until confidence is restored, until the public is willing to take money now in hiding and return it to the banks as a basis for the expan sion of bank credit. This is indispensable to the support of business and the successful financ ing of the Treasury. It will bring increased earnings, higher incomes, and make it possible to balance the Government’s Budget without resort to vicious and vexatious methods of taxa tion.94 As such, they should be considered inseparable; it is clear that supporters of the legislation in tended it to achieve both ends. Attempts to rank the two issues according to their relative im portance are likely to be inconclusive.95 The Chastening o f Wall Street One banker in my state attempted to marry a white woman and they lynched him.91’ The opposition to federal deposit guaranties emanated largely from the nation’s bankers. This fact was a crushing liability to their cause in the political clim ate of 1933. The introduction of the Glass and Steagall bills cam e on the heels of the banking panic and, not entirely coin cidentally, amid the daily revelations of selfdealing and oth er cupidities from the Pecora hearings.97 The banker had becom e a pariah. 94Rep. Steagall, Congressional Record (1933), p. 3840. 95Golembe (1960) has argued that, among the motives for deposit insurance, depositor protection was secondary to protection of the circulating medium. Others have gone further, arguing that protection of depositors was a ration alization created after the fact. The issue raised by Golembe is certainly plausible; Rep. Bacon, for example, appears to have ranked them this way [Congressional Record (1933), p. 3959], On the other hand, it is notewor thy that Sen. Glass in 1933 abandoned his earlier plan for a liquidation fund, which would have prevented the freez ing of funds in suspended banks while still not protecting depositors from loss. The latter notion of depositor protec tion as an ex-post or revisionist justification is clearly false, however. 96This was a popular quip that made the rounds in 1933. In this instance, it is attributed to Carter Glass; see Kennedy (1973), p. 133; Bell (1934), pp. 262-63, also cites it. The joke is startling in its insensitivity. Examples of bankers of the day indulging in overtly racist humor are also availa ble; see, for example, Dyer (1933), pp. 91 and 94, and Amberg (1935), p. 49. 97The hearings were organized in January 1933 by the Senate Committee on Banking and Currency, and were run by the Committee’s counsel, Ferdinand Pecora; see Pecora (1939). The dust jacket relates that, in one in- http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Roosevelt fired the opening volley for his ad m inistration in his inaugural address; Plenty is at our doorstep, but a generous use of it languishes in the very sight of the supply. Primarily this is because the rulers of the ex change of mankind’s goods have failed, through their own stubbornness and their own in competence, have admitted their failure, and abdicated. Practices of the unscrupulous money changers stand indicted in the court of public opinion, rejected by the hearts and minds of men.98 He w ent on to demand safeguards against the “evils of the old ord er”; strict supervision of banking, an end to speculation with "oth er peo ple's m oney,” and provision for an adequate but sound cu rren cy.99 O thers w ere happy to follow this lead. It was comm onplace to hold the bankers, and particu larly their "speculative orgy” of 1929, responsi ble for the nation’s woes: You brought this country to the greatest panic in human history! ... There never was such an economic failure in the history of mankind as your outfit has brought upon us at this time, and it is due to this same speculation that you are defending here more than any other one thing.100 But these affiliates, I repeat, were the most un scrupulous contributors, next to the debauch of the New York Stock Exchange, to the financial catastrophe which visited this country and was stance, a journalist “ begged Mr. Pecora not to break so many front-page stories daily because it was physically impossible to cover them all.” See Benston (1990) for a thorough, revisionist view of the hearings. 98Roosevelt (1938), pp. 11-12. "R oosevelt (1938), p. 13. His reference to "other people’s money” was a nod to Justice Brandeis’s book of the same title, a reprint of his articles on the money trust that ap peared in Harper’s Weekly in 1913-14. Those who hold that all the great thoughts have long since been had will be pleased to learn that Kane’s (1991) reference to the “ Sor cerer’s Apprentice” segment of Walt Disney’s Fantasia as a metaphor for bank regulation was anticipated by Brandeis. Lacking Mickey Mouse’s rendition, however, Brandeis was forced to use the German original, Goethe’s Der Zauberlehrling; see Brandeis (1933), p. vii. 100Sen. Brookhart (R-IA) speaking to a New York Stock Ex change official at a Senate committee hearing in 1932; quoted by Danielian (1933), p. 496. 67 mainly responsible for the depression under which we have been suffering since.101 In the previous year, Huey Long had announced his intent to campaign for Roosevelt under the slogan: "Rid the country of the m illionaires."102 A popular ditty mocked: Mellon pulled the whistle, Hoover rang the bell, Wall Street gave the signal, And the country went to hell.103 In short, the bankers w ere vilified. Although some felt such indiscrim inate abuse was slanderous, they fought against the tid e.104 One of the casualties of the anti-banker senti m ent was the bankers’ battle against deposit in surance. Some in Congress announced that the b ankers’ opinions should be openly ignored: I believe that the myopic banker as an adviser should receive about as much consideration at the hands of the House as a braying jackass on the prairies of Missouri. They proved by their inability to maintain their own business that they have absolutely no right to advise the House as to what course we should follow.105 conservative way than to have ourselves run over in a stam pede.”107 Roosevelt held out until the very end, thus forcing Congress to concede in delaying implementation of the tem porary plan until Jan uary 1934. BAN K M A R K E T S T R U C T U R E The ram ifications o f deposit insurance w ere recognized as far-reaching. In many ways, the central and most contentious battle concerned neither actuarial feasibility nor the desirability of protecting deposits, but the regulatory issues of bank chartering and supervision. Because of the fundam ental legal issues involved, it was h ere that the econom ic and political aspects of the debate becam e most fully intertw ined. This was a fight with the weight of a long tradition behind it, and argum ents w ere often self consciously historical. Regulatory Competition and Lax Supervision Bank examinations to be effective must be made by experienced men, fr e e fro m political influence. ... We will never have proper banking supervi sion, national or state, until it is taken entirely away fro m political influence.108 The bankers, while they acknowledged the m erit of individual aspects of the deposit insur ance proposals, obstinately refused to coun tenance any of the schem es as a realistic reform . Even as the legislation was signed into law, Francis Sisson called a crusade, rallying ABA m em bers to fight "to the last ditch against the guaranty provisions” o f the bill.106 That the bankers’ concerns w ere not ignored entirely resulted largely from the presence in govern m ent of opponents of deposit guaranties who w ere m ore politically astute than the bankers them selves. Sen. Glass, for example, com pro mised his principles in a bid fo r some control over the legislation, explaining that it was "bet te r to deal with the problem in a cautious and a Much of the blam e fo r high rates of bank failure throughout the 1920s was placed upon com petition betw een state and federal authori ties. Because banks could choose the less costly of federal and state ch arters—and the associated regulations—state and federal regulators w ere forced into a "com petition in laxity” if they w ere to sustain the realm of their bureaucratic influence.109 For example, as a prelude to recom m ending broader powers for national banks, Comptroller Pole emphasized that: 101Sen. Glass, Congressional Record (1933), p. 3726. Glass is referring to the proposed separation of investment affili ates from Federal Reserve member banks. 106Sisson’s telegram is quoted in Pecora (1939), pp. 294-95. ' “ Kent (1932), p. 260. ’ ^Kennedy (1973), p. 26. 104See, for example, Bell (1934). Sisson (1933b), p. 30, offered that the treatment of bankers as “ demons of dark ness” and as an “ unseen mythical power for evil which spreads its baneful influence over [human beings]” merely satisfied an emotional need for a scapegoat. If Congress therefore would protect itself from the loss of its present banking instrumentality, it must make it to the advantage of capital to seek the national rather than a [state] trust company charter. ... 107Sen. Glass, Congressional Record (1933), p. 5862. 108Andrew (1934b), p. 93. 109Daiger (1933), p. 563, attributes coinage of the phrase “ competition in laxity” to Eugene Meyer in 1923 testimony to the House Banking and Currency Committee. The phrase attained some popularity; it was also used, for ex ample, by Wyatt (1933), p. 186, and Await (1933), p. 4. 105Rep. Dingell, Congressional Record (1933), p. 3906. JULY/AUGUST 1992 68 It is within the power of Congress to turn the advantage in favor of the national banks and thereby make it to the interest of all banks to operate under the national charter110 the inevitably slow and unsensational process of strengthening the banking system by strict regulation, vigilant public opinion and strict re quirements.116 In the eyes of opponents of deposit insurance, an especially im portant m anifestation of the com petition in laxity was the “promiscuous granting of bank ch a rters.”111 The immediate result of loose chartering was a condition called “over-banking,” or The Association of Reserve City Bankers went fu rth er, predicting that m anagers of the insur ance fund would be slow to close troubled insti tutions.117 In addition to regulatory competition, some saw political influence as a secondary force debilitating the supervisory process: a host of weak, unreliable banks that crowd one another out of existence by being too nu merously organized in places where there is no support for the multifarious institutions that have been established there.112 We never will have such supervision under po litical regulation and examination; we will never have any supervision worthy of the name that does not have real authority and heavy respon sibility tied to it.118 This “indiscreet indulgence of ch arter appli cants” was held responsible for the vast num b ers of bank failures throughout the previous decade:113 There are too many banks in the United States. The areas of greatest density of banks per capi ta coincide with the areas where failures are proportionately highest.114 The function of a deposit guaranty under such circum stances would be to exacerbate the problem by mitigating one source of public scrutiny: inspection by depositors. Opponents confirm ed their contention by referen ce to the ill-fated state guaranty schemes: In practice the guaranty of deposits plan gener ally tended to induce an unsound expansion in the number of banks ... This was clearly con nected with the indiscriminate popular confi dence created toward the banks under the guaranty.115 It is to be feared that the adoption of deposit guaranty laws may have somewhat retarded 110Pole (1929), p. 23. 111Association of Reserve City Bankers (1933), p. 30. 112H. Parker Willis, quoted in Lawrence (1930), p. 105. Only a few supporters of insurance addressed directly the plan’s implications for the regulatory process, w hich they presented as a cou n ter weight to incentives fo r bad banking under a guaranty. Rome Stephenson felt that the addi tional regulatory pow ers in the Banking Act differentiated the FDIC markedly from the state plans: Right there is the crux of the debate: Will banks under the federal plan be permitted the abuses which were tolerated in every one of the states where guaranty was tried? If so, then failure is inevitable. If not, success is practically certain. ... Let me assert unequivocally that the men who drew up the federal plan profited by the mistakes of the state guaranty failures and avoided them. ... None of the state laws had teeth in them. The federal law has teeth like a man-eating shark, and already has done some highly effective biting.119 C arter Glass, railing that “the C om ptroller’s office has not done its duty—its sworn duty— for unneeded banks or to unqualified promoters to open new institutions;” ibid., p. 22. The result was seen to be less over-banking and fewer failures relative to Oklahoma and Nebraska. 113Lawrence (1930), p. 104. Lawrence took this priggish tone one step further, admonishing that “A little birth control of banks on the part of the states which now suffer most from bank failures might have had a wholesome effect on the rate of mortality;” ibid., p. 84. 116A Saturday Evening Post editorial of August 9, 1924, quot ed in Association of Reserve City Bankers (1933), p. 42. 114Westerfield (1931), p. 17; the "m ultiplicity of banks” was first on his list of the six causes of bank failures since 1920. Andrew (1934b), p. 93, concurred that “ Everyone agrees that one of the main causes of our banking trouble was too many banks.” See also Bremer (1935). Await (1933), p. 4, attributes the boom in charters to “ lax State laws” and the 1900 reduction in the minimum capitaliza tion for national banks from $50,000 to $25,000. 118Donald Despain, quoted by Sen. Schall, Congressional Record (1933), p. 4632. 115ABA (1933a), p. 42. Mississippi was held up as the excep tion that proved the rule: “ The banking authorities in Mis sissippi had full discretion in the matter of granting new charters and used it liberally in refusing permission http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS 117See the quote referenced by footnote 64. 119Stephenson (1934), p. 46. In addition to authorizing the supervisory power of the FDIC, the Banking Act of 1933: increased the punitive authority of the Federal Reserve for member banks financing securities “ speculation,” prohibit ed insider lending for member banks, authorized federal regulators to remove the officers and directors of member banks for illegality or unsound banking practice, and re quired deposit-taking private banks to submit to supervi sion by the Comptroller’s office. 69 and has perm itted this great num ber of banks to engage in irregular and illicit practices,” a r gued that mutual responsibility inherent in the insurance plan implied mutual supervision: if the strong banker "know s that he has got to b ear a part of the burden of my irregular bank ing, he is going to rep ort me to the Comptroller of the C urrency and is going to insist that his exam iners com e th ere and do their duty.”120 The Dual Banking Question The fa ct is, o f course, that the deposit insurance schem e would not have been permitted by the conservative leaders in Congress if its organiza tion could not have been so shaped as to fu rther their idea o f a unified system o f banking in the country under the Reserve System. On the other hand, the m ore radical elements, in response to popular demand f o r som e sort o f protection f o r bank depositors, could not have built a nation wide guaranty system upon any other foundation than the Reserve organization.lzl Questions about the effect of insurance on the quality of chartering and supervision w ere side shows to the main event, however. At the heart o f the debate lay a decades-old controversy over the dual banking system. Given its far-reaching nature, the proposed legislation was universally regarded as a prim e opportunity for fundam en tal changes in banking policy. Com ptroller Pole had campaigned vigorously throughout his four-year tenure for some form of interstate branching fo r national banks. He drew a strong distinction betw een the small, state-chartered, rural unit bank—the “country” bank—and the large, nationally chartered insti tution. W hile he pretended to maintain great respect for the small unit bank as the “single type of institution which has contributed the most to ... the foundation of our national de velopm ent,” he was fighting to have them re placed by branch netw orks of national banks.122 He justified this split sentim ent by arguing that 120Sen. Glass, Congressional Record (1933), p. 3728. 121Anderson (1933c), p. 17. 122Pole (1929), p. 24. 123See Pole (1930a, 1931, 1932a and 1932b), “ The Need of a New Banking Policy” (1929) and “ Comptroller Pole’s Views on Rural Unit Banking” (1930). irreversible social changes—telephone, radio, and especially the automobile—had forever obvi ated the ru ral isolation that had made the unit bank competitively viable. Accompanied by a long parade of statistics, he emphasized the high failure rate of small, state-chartered banks during the 1920s.123 The country bank, he said, could not survive in com petition with large metropolitan institutions, w hich had m ore professional m anagem ent and w ere inevitably b etter diversified. Comptroller Pole was not alone in this cru sade. The McFadden Act had already broadened the branching pow ers of national banks; in 1930, the House Banking Committee arranged new hearings into the possibility of national or regional branch banking.124 The unsuccessful Glass bill of 1932 included limited provisions for statewide branching by national banks. Business W eek staked out the extrem e position, announc ing that "w hat we really need is just one big bank with 20,000 b ran ch es.”123 Supporters of bran ch banking took h eart in the Canadian ex perience: Canada has branch banking, and Canada has not had any bank failures during the depres sion. Is this a matter of cause and effect? ‘It is,’ declare the advocates of branch bank ing in the United States.126 Such highly concentrated branch netw orks w ere offered as an alternative to deposit insur ance as a means of geographic diffusion of loan losses and the diversification of credit risks.127 Comptroller Pole, o f course, felt branching to be the b etter option: Any attempt to maintain the present country bank system by force of legislation in the na ture of guaranty of deposits or the like, would be economically unsound and would not accom plish the purpose intended.128 Deposit guaranties had long been advocated as a way of diversifying risk for the unit bank w ith out a fundam ental change in the ownership 127For example, Rep. Bacon, Congressional Record (1933), p. 3961, noted that “deposit guaranty is undoubtedly a guaranty of reckless banking. ... Safety for the depositor can best be achieved by a unified branch banking sys tem.” 128Pole (1930b), p. 5. This same sentence appears in Pole (1930a), p. 4. 124U. S. Congress, House of Representatives, Committee on Banking and Currency (1930). 125“ The Ideal Bank” (1933), p. 16. 126Greer (1933a), p. 722. See also Lawrence (1930), and Rep. Bacon, Congressional Record (1933), pp. 3949-50. JULY/AUGUST 1992 70 stru ctu re of the banking industry.129 The vari ous histories of Populism, "Bryanism ,” the Panic of 1907 and the Pujo hearings all contained ele m ents of a deep popular m istrust of money cen ter banks. The publicity of the Pecora h ear ings in 1933 clearly did not assuage this mis trust. It w as not pure coincidence that the w estern agricultural states—the heart of the Grange and Populist m ovem ents—had been the ones to enact state deposit guaranties. In this context, then, it is ironic that, in 1933, federal deposit insurance should most often have been viewed as a lethal th reat to the country bank. That it was such a th reat testifies to the in fluence and legislative skill of C arter Glass. Sen. Glass, who had shepherded the Federal Reserve Act through the House in 1913, was protective of his handiwork: I took occasion to tell the Secretary of the Treasury the other day that if they pursue present policies much longer they will literally wreck the Federal Reserve System; that Woodrow Wilson in history will enjoy the dis tinction of having set up a banking system that fought the war for us and saved the Nation in the post-war period, and if they keep on mak ing a doormat of it this Congress will enjoy the distinction of having wrecked it.130 His prim ary con cern in the banking legislation o f 1933 was to bu ttress that system. Thus, the Glass bill required all FDIC m em ber banks to join the Federal Reserve System, ostensibly to give the Fed the legal right to exam ine FDIC m em b ers (the Fed was to be a prom inent share holder in the FDIC).131 Because an uninsured country bank facing insured com petitors was not considered viable, and because Fed m em ber ship would require at least $25,000 minimum capital, deposit insurance represented the end fo r the small, state non-m em ber banks.132 Deposit insurance would force a consolidation of banking within the Federal Reserve System. 129White (1982, 1983, 1984) reviews the historical connections between deposit insurance and bank chartering. 130Sen. Glass, Congressional Record (1933), p. 3728. 13'See the interchange between Sens. Glass and Couzens (R-MI), Congressional Record (1933), p. 3727. 132Section 17 of the Glass bill “ provides for the amount of capital of national banks depending upon the population of the places where they are to be located and also pro hibits the admission of a bank into the Federal Reserve System unless it possesses a paid-up unimpaired capital sufficient to entitle it to become a national bank.” See Glass (1933b), p. 16, (emphasis added). The population schedule for minimum capital was: $25,000 for areas un http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS it is instructive to note that Glass had aban doned an earlier schem e that would have forced the same consolidation within the Fed: unification of banking in the National Banking System. Comptroller Pole had sought to accom plish the same thing indirectly, by providing na tional banks with an undeniable competitive advantage in the form of interstate branching privileges. In 1932, Glass had requested of Gov. M eyer of the Federal Reserve a constitutional method of unifying banking: Meyer: "Do you want to bring about unified banking?” Glass: "Why, undoubtedly, yes.” Meyer: "I shall be glad to help you.” Glass: "I think the curse of the banking busi ness in this country is the dual system.” Meyer: "Then the Board is entirely in sympa thy with the Committee on the sub ject.’’133 The result was a legal opinion prepared by the General Counsel of the Federal Reserve Board on the constitutionality of such unification in the absence of a constitutional am endm ent.134 W hile Board Counsel confirm ed that such a con stitutional means existed, Sen. Gore introduced a constitutional am endm ent.135 Constitutionality was crucial, because champions of the ru ral unit bank w ere certain to raise the pow erful specter of states' rights in opposition: The fight regarding the American Dual System of Banking is a clear-cut issue between those who believe in the sovereignty of our states and home rule, and those who are in favor of a ‘unification of our banking system’ into one Washington bureau.136 Indeed, the political sensitivity of the states' rights issue was sufficient to fo rce Sen. Glass to abandon such a direct assault on the state banks b efore it could earnestly begin.137 der 3000 persons; $50,000 for 3000 to 6000 persons; $100,000 for 6000 to 50,000 persons; $200,000 for areas over 50,000 persons; see Steagall (1933a), pp. 18-19. 133Quoted by Anderson (1932b), p. 678. 134The opinion was published as Wyatt (1933). The Attorney General had felt it was not possible, and had told Glass that; see Anderson (1932b), p. 678. 135Joint resolution S. J. Res. 18 was introduced by Sen. Gore (D-OK), Congressional Record (1933), p. 249. 136Andrew (1934b), p. 95. 137See Bums (1974), pp. 11-12. 71 Arrayed against Sen. Glass in the battle for unification within the Fed was a coalition led by Henry Steagall in the House and Huey Long in the Senate.138 Sen. Long had crippled Glass's banking bill in the previous Congress with a ten-day filibuster; as champion of the common man, he had objected to an envisioned con cen tration of pow er implicit in the bill's branching provisions.139 This coalition indeed viewed de posit insurance as a means of survival for the small bank: If there is one purpose more than another which is inherent in the amendment which is now at stake in this conference, it is the pur pose to protect the smaller banking institutions, and to make the reopening of closed banks pos sible as speedily and as safely as it can be done.140 The final legislation was a two-stage com promise betw een Sen. Glass’s push for unifica tion and the Steagall-Long coalition’s desire to preserve the dual banking system. In the first stage, Glass agreed to support a deposit guaranty in exchange for provisions for significantly ex panded Federal Reserve authority: With these provisions, dependent upon them in fact, the Senate bill drafters were willing to ac cept the new Steagall bill for the insurance or guaranty of bank deposits in Federal Reserve member banks—but in member banks only.141 In the second stage, the dual banking support ers obtained several concessions, most notably: 138See Anderson (1933a), p. 17. They were joined by Sen. Vandenberg, whose temporary plan extended insurance to state non-member banks upon certification of soundness by the relevant state banking authority. 139There was little fondness connecting the two Southern Democrats. Smith and Beasley (1939), pp. 346-47, relate that, in the heat of the banking debate and in response to a series of Long's ad hominems, Glass unleashed a string of invective that literally chased the Kingfish — his hands clamped over his ears — off the Senate floor. This version of events is apocryphal, however. 140Sen. Vandenberg, referring to the temporary insurance amendment, Congressional Record (1933), p. 5256. See also Vandenberg (1933), p. 43. 141Anderson (1933a), p. 63. 142Rep. Luce reported that bank structure issues predominat ed in the conference committee reconciling the Glass and Steagall bills: “ There were but two points of serious con troversy in the discussions of the conferees — those to which I have just referred, branch banking, the member ship requirement together with other details of insurance of bank deposits,” Congressional Record (1933), p. 5896. Much of the force of Glass’s requirements for Fed mem bership was lost when deposit insurance was revamped by the Banking Act of 1935; see, for example, Woosley (1936), pp. 24-26. See also the shaded insert on the fol- immediate insurance coverage fo r non-mem ber banks under the tem porary plan, and grand fathering of small state banks under the new minimum capital standards for Fed membership. Non-member banks would still have to apply for Federal Reserve m em bership by July 1, 1936, at the latest. W ith these changes, Sen. Long sup ported the bill, w hich then passed the Senate without objection.142 CONCLUSIONS Prophesying the future o f Federal Deposit Insur ance is at the sam e time both difficult and sim ple. It is difficult because the subject cannot be treated independently, that is, without relation to banking structure, banking practice, political and economic trends and human emotions. It is easy, on the other hand, because ... any man's guess is as good as that o f another.143 It is obvious from an examination of the record that the debate surrounding the adop tion of federal deposit insurance was both wideranging and well informed. T he banking crisis in M arch 1933, coming at the depths of the Great Depression and breaking on inauguration day, had focused attention with unique intensity on all aspects of public policy toward banks. While some contended that the urgency accom panying the crisis injected haste into the proceedings, it also ensured that all m ajor in terests w ere roused to o ffer their views and ar gue their cases. lowing page. The membership requirement was dropped entirely in 1939; see Golembe (1967), pp. 1098-1100. Opinions varied on the significance of the consolidation of bank regulation implicit in the final act. Bankers Maga zine editorialized that, “ while this development will bring the state banks under a considerable degree of Federal control, it will not — for a time at least — result in that unification of banking regarded by many as desirable. The state banks, by coming into the deposit-guaranty scheme have escaped with their lives.” “ State Banks Qualifying for Insurance of Deposits” (1933), p. 490. Anderson (1933c), p. 17, warned that, “ with all this variation, this glorification of the unit bank principle, however, comes the hard fact that these institutions, for the first time in their history, will be under one direct control whose authority is such as practically to set aside all the principle privileges for which state banks have fought so long.” 143Amberg (1935), p. 49. JULY/AUGUST 1992 72 The Four That Passed Law Banking Act of 1933 (temporary plan) Banking Act of 1933 (permanent plan) — Never operational — Act of 1934 Extending Temporary Deposit Insurance Banking Act of 1935 Period of operation From Jan. 1, 1934, to July 1, 1934, or earlier if the President so proclaims. From July 1, 1934 (or earlier if the President so proclaims). From July 1, 1934, to July 1, 1935 (extended to Aug. 31, 1935, in June of 1935). From August 23, 1935, onward. Coverage All deposits covered in full up to $2,500 100% coverage up to $10,000, 75% on the next $40,000, 50% of all over $50,000 All deposits covered in full up to $5,000. All deposits covered in full up to $5,000. Member ship All Fed member banks re quired to join. Non members allowed in with state certification and ap proval of the corporation. All Fed member banks re quired to join. Non members allowed in from 7/1/34 to 7/1/36 (with state and FDIC approval); Fed membership required by 7/1/36. All Fed member banks re quired to join. Non members allowed in until 7/1/37 (with state and FDIC approval); Fed mem bership required by 7/1/37. All Fed member banks re quired to join. Non members allowed in with FDIC approval. Non members with 1941 aver age deposits over $1 mil lion must join by 7/1/42. Assess ments on insured banks 0.5% of insured deposits, one half paid in cash, the other half subject to call. One more such assess ment as needed. Surplus as of 7/1/34 to be refunded. 0.5% of total deposits, half in cash, half subject to call. Extra assessments of 0.25% of total deposits, as needed and without upper limit. Same as under the tem porary plan of the Banking Act of 1933, except the surplus is to be measured and refunded as of 7/1/35. Annual assessment of 1/12 of 1% of average total deposits, payable in two installments. FDIC’s capital Provided according to the assessment schedule. $150 million on call from Treasury (to pay 6% div.) plus one-half the surplus of Federal Reserve banks (ca. $139 million) for $100-par, no-div., non voting stock plus 0.5% of deposits of FDIC banks ($150-200 million) for $100-par, 6% div., non voting stock. Provided according to the assessment schedule. Same as under the perma nent plan of 1933, except: all stock is no-par, no-div., non-voting; insured banks do not buy FDIC stock; and Federal Reserve bank surpluses are measured as of 1/1/35, rather than 1/1/33. Control Board of three: the Comp troller and two Presidential appointees. Same. Same. Same. FEDERAL RESERVE BANK OF ST. LOUIS 73 It has been suggested that the fram ers of the Banking Act of 1933 failed to consider the warnings about the potential dangers of governm ent-sponsored deposit insu rance.144 It is significant, then, that an exam ination of the historical record clearly shows that bill’s chief patrons w ere aware of the failure of the state schem es, the actuarial argum ents against deposit guaranties, and the various chartering issues involved. M oreover, they took these is sues into account w hen crafting the bill. In the end, even the Association of Reserve City Bankers was able to recom m end the tem porary insurance plan: It appears to this Commission that if guaranty is retained after July 1, 1934 [the date for im plementation of the permanent plan], this tem porary plan, in some modified form, would meet every emergency need, and eliminate many of the dangers in the permanent plan.145 Under the tem porary plan, coverage ceilings w ere conservative, the insurance corporation was em phatically segregated from the federal taxpayer, chartering standards for national banks w ere raised, and supervisory authority was broadly increased. These characteristics w ere retained under the perm anent plan of the Banking Act of 1935. As such, deposit insur ance, as construed in the Banking Acts of 1933 and 1935, succeeded in simultaneously protect ing the small depositor and leaving the banker answ erable to both supervisors and large depos itors for the quality of his management. At the same time, the deposit insurance provi sions of the Banking Act of 1933 w ere used as leverage to consolidate the industry within the Federal Reserve, although the Banking Act of 144Kaufman, for example, claims that the opinions of Emer son (1934) — and, by association, those of the banking community as a whole — regarding flaws in the actuarial basis for the plan were unheeded at the time. In particular, Kaufman (1990) states, pp. 1-2: “ Some of the problems are new, however many have been around for many years and were even clearly foreseen at the time they were forming or, worse yet, even earlier, at the time their underlying causes were put in place in the form of legislation or regulation. This is the case with the extant structure of federal deposit insurance. Among those fore casting the problems that this innovation would come to cause was Guy Emerson, a long-time economist for the Bankers Trust Company (New York). His warnings are evi dent in his article “ Guaranty of Deposits Under the Bank ing Act of 1933” published in the February 1934 Quarterly Journal of Economics and reprinted in this volume. Much of this book is necessitated because policy makers did not listen to Emerson and others more than half a century ago.” Related remarks appear on pp. xi-xii of the preface to the same volume. 1935 significantly w eakened the requirem ents for Fed m em bership of insured banks. A piecemeal dismantling of other provisions of the original legislation has also occu rred in the in tervening decades: coverage ceilings have risen steadily, even after accounting for inflation and b efore considering brokered deposits or too-bigto-fail policies; the full taxing authority of the U. S. Treasury has, d e fa c t o , been inserted behind the deposit insurance corporations; and deregu lation has subjected both banks and thrifts to increasingly harsh er com petition—and, in some cases, relaxed regulatory scrutiny—without simultaneously making bankers responsible to depositors fo r the riskiness of bank assets.146 It is perhaps with this m ore recen t negation of in dividual elem ents of a complex and interdepen dent package of bank reform s that we should seek the proxim ate cause of our recen t deposit insurance troubles, rath er than with policy flaws in the Banking Act of 1933 itself. R EFER EN C ES This list of references contains several sources that are rele vant to the debate, but which are not cited directly in the text. These additional references are included to provide others in terested in the topic with a more comprehensive listing of the primary source materials. “A Good Start,” Business Week (March 22, 1933), p. 32. Amberg, Harold V. “ The Future of Deposit Insurance,” Association of Reserve City Bankers: Proceedings, Twentyfourth Annual Convention (1935), pp. 49-61. American Bankers Association (ABA). “ The Guaranty of Bank Deposits,” (Economic Policy Commission: American Bankers Association, New York, 1933a). . “ Forum Discussion— Uniform Banking Law—Guar antee of Deposits,” Commercial and Financial Chronicle (American Bankers Convention Supplement, September 23, 1933b), pp. 58-59. 145Association of Reserve City Bankers (1933), p. 7. They were, however, at pains not to appear eager in their praise: “ What we are recommending, therefore, is co operation in an emergency measure of the sort that has been deemed necessary in almost all branches of our eco nomic life, but we are not, directly or indirectly, endorsing the principle of deposit guaranty" ibid., p. 7, (emphasis in the original). The permanent plan was never operational; it was in fact ultimately superseded by a modified form of the temporary plan. 146The technical legal question of the de jure liability of the United States government for deposit insurance is surpris ingly complex, and the answer is not entirely clear. As a practical matter, however, the question is neither complex nor unclear. See FDIC (1990), pp. 4438-39. JULY/AUGUST 1992 74 . "Report of Resolutions Committee— Insurance of Deposits Declared Unsound,” Commercial and Financial Chronicle (American Bankers Convention Supplement, Sep tember 23, 1933c), p. 59. American Savings, Building and Loan Institute. Guarantee of Bank Deposits and Building and Loan (American Savings, Building and Loan Institute, Chicago, 1933). Anderson, George E. “ The Glass Bill is a Medley,” American Bankers Association Journal (February 1932a), pp. 498, 532-35. ________"Washington Looks at the State Banks,” American Bankers Association Journal (May 1932b), pp. 677-78, 718. ________“ Bank Law Making,” American Bankers Associa tion Journal (May 1933a), pp. 17, 63. . “ The Price of Deposit Insurance,” American Bankers Association Journal (October 1933b), pp. 17-19, 51. ________“ Washington Epic: II. Prospectus— National Finan cial Control,” American Bankers Association Journal (November 1933c), pp. 16-17, 48. . “ Deposit Insurance, First Phase,” American Bankers Association Journal (January 1934a), pp. 20-21. Blocker, John G. “ The Guaranty of State Bank Deposits,” Bureau of Business Research of the University of Kansas, Kansas Studies in Business No. 11, (Department of Jour nalism Press, Lawrence, 1929). Boeckel, Richard M. The Guaranty of Bank Deposits (Editor ial Research Reports, Washington, D. C., 1932). Bogen, Jules I., and Marcus Nadler. The Banking Crisis (Dodd, Mead and Company, New York, 1933). Bradford, Frederick A. “ Futility of Deposit Guaranty Laws,” Bankers Magazine (June 1933), pp. 537-39. Brandeis, Louis D. Other People’s Money and How the Bankers Use It (National Home Library Foundation, Washington, 1933). Bremer, C. D. American Bank Failures (Columbia University Press, New York, 1935). Burns, Helen M. The American Banking Community and New Deal Banking Reforms 1933-1935 (Greenwood Press, West port, 1974). Calomiris, Charles W. “ Deposit insurance: Lessons from the record,” Federal Reserve Bank of Chicago Economic Per spectives (May/June 1989), pp. 10-30. ________“ Bank Owners,” Banking (November 1934b), pp. 11-12. ________“ Is Deposit Insurance Necessary? A Historical Per spective,” Journal of Economic History (June 1990), pp. 283-95. Andrew, L. A. “ Reconstruction: Individual Initiative,” Ameri can Bankers Association Journal (January 1934a), pp. 1516, 53, 69. Calverton, V. F. “ Is America Ripe for Fascism?,” Current History (September 1933), pp. 701-04. ________“ The Future of the Unit Bank,” Proceedings of the Forty-fourth Annual Convention of the Missouri Bankers As sociation (1934b), pp. 91-101. Carter, W. E. “Annual Address of the President,” Proceed ings of the Forty-fourth Annual Convention of the Missouri Bankers Association (1934), pp. 21-27. “Are State Banks to be Suppressed?,” Bankers Magazine (November 1929), pp. 663-64. Chernow, Ron. The House of Morgan: An American Banking Dynasty and the Rise of Modern Finance (Atlantic Monthly Press, New York, 1990). Association of Reserve City Bankers, Commission on Bank ing Law and Practice. “ The Guaranty of Bank Deposits,” Bulletin No. 3, (Chicago, November 1933). Collins, Charles W. Rural Banking Reform (MacMillan, New York, 1931). Await, F. G. Annual Report of the Comptroller of the Cur rency, 1932 (U. S. Government Printing Office, Washington, D.C., 1933). “ Bank Bill,” Business Week (June 10, 1933), p. 8. “ Bank Reform,” Business Week (January 4, 1933), pp. 3-4. “ Banking Issues in the Campaign,” American Bankers As sociation Journal (August 1932), pp. 22, 65. “ Banking Reform,” Business Week (March 8, 1933), p. 32. “ The Banks Reopen,” Business Week (March 22, 1933), pp. 3-4. Beebe, M. Plin. “A National View of State Banks,” American Bankers Association Journal (October 1931), pp. 217-18, 287. Bell, Elliott V. “ The Bankers Sign a Truce,” Current History (December 1934), pp. 257-63. ________“ Who Shall Rule the Money Market?,” Current History (July 1935), pp. 353-59. Bennett, Frank P., Jr. “A Word to National Banks,” Ameri can Bankers Association Journal (October 1931), pp. 23233, 272. Benson, Philip A. “ Government Guaranty of Bank Deposits," American Bankers Association Journal (December 1932), pp. 14-15, 69. Colt, Charles C., and N. S. Keith. 28 Days: A History of the Banking Crisis (Greenberg, New York, 1933). “ Comptroller Pole’s Views on Rural Unit Banking,” Bankers Magazine (April 1930), pp. 463-69. “ Congress Passes and President Roosevelt Signs GlassSteagall Bank Bill as Agreed on in Conference,” Commer cial and Financial Chronicle (June 17, 1933), pp. 4192-93. Congressional Record (daily edition), 73d Congress, 1st Ses sion, (1933). Crowley, Leo T. “ The Benefits of Deposit Insurance,” Proceedings of the Forty-fourth Annual Convention of the Missouri Bankers Association (1934), pp. 101-10. ________“ The Necessity of Cooperation Between State Su pervising Authorities and the F.D.I.C.” (with discussion), Proceedings of the Thirty-fourth Annual Convention of the National Association of Supen/isors of State Banks (Brandao Printing, New Orleans, 1935), pp. 65-68. Cummings, Walter J. “ The Federal Deposit Insurance Corpo ration,” Bankers Magazine (November 1933), pp. 577-78. Daiger, J. M. “ Toward Safer and Stronger Banks,” Current History (February 1933), pp. 558-564. Danielian, N. R. “ The Stock Market and the Public,” Atlantic Monthly (October 1933), pp. 496-508. Benston, George J. The Separation of Commercial and In vestment Banking: The Glass-Steagall Act Revisited and Reconsidered (Oxford University Press, New York, 1990). “ Death of Francis H. Sisson, Vice-President Guaranty Trust Co. of New York and Former President American Bankers Association,” Commercial and Financial Chronicle (Septem ber 23, 1933), p. 2195. Berle, A. A., Jr. “ Reconstruction: Central Control,” American Bankers Association Journal (January 1934), pp. 13-14, 52, 68-69. De Long, J. Bradford. “ ‘Liquidation’ Cycles: Old-Fashioned Real Business Cycle Theory and the Great Depression,” NBER Working Paper (October 1990). http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 75 “ Deposit Insurance,” Business Week (April 12, 1933a), p. 3. “ Deposit Insurance,” Business Week (May 31, 1933b), p. 20. “ Deposit Insurance Draws Near,” Review of Reviews and World’s Work (December 1933), pp. 50-51. “ Depositors Want Insurance,” Business Week (June 7, 1933), p. 12. Dyer, Gus W. “ Federal Control of Business,” Proceedings of the Forty-third Annual Convention of the Missouri Bankers Association (1933), pp. 91-98. Elliott, W. S. “ State Banks and the Future Thereof” (with discussion), Proceedings of the Thirty-fourth Annual Conven tion of the National Association of Supervisors of State Banks (Brandao Printing, New Orleans, 1935), pp. 80-85. Emerson, Guy. “ Guaranty of Deposits Under the Banking Act of 1933,” Quarterly Journal of Economics (February 1934), pp. 229-244. “ Extension of Branch Banking,” Bankers Magazine (Novem ber 1929), pp. 661-63. Federal Deposit Insurance Corporation (FDIC). “ History of Legislation for the Guaranty or Insurance of Bank Deposits,” Annual Report of the Federal Deposit Insurance Corporation, 1950 (FDIC, Washington, D.C., 1951), pp. 61-101. _______ . “ Insurance of Bank Obligations Prior to Federal Deposit Insurance,” Annual Report of the Federal Deposit Insurance Corporation, 1952 (FDIC, Washington, DC., 1953), pp. 57-72. ________“ State Deposit Insurance Systems, 1908-1930,” An nual Report of the Federal Deposit Insurance Corporation, 1956 (FDIC, Washington, D.C., 1957), pp. 45-73. ________Federal Deposit Insurance Corporation: The First Fifty Years (FDIC, Washington, D.C., 1984). ________“Are Deposits in Financial Institutions Guaranteed Directly by the Federal Government or by the FDIC and its Resources,” FDIC Advisory Opinion FDIC-90-6, in: Federal Deposit Insurance Corporation: Law, Regulations and Re lated Acts, Volume 1 (Prentice-Hall, 1990), pp. 4438-39. “ Federal Guaranty of Bank Deposits,” Bankers Magazine (April 1932), pp. 380-82. Federal Reserve Board. “ Guaranty of Bank Deposits,” Federal Reserve Bulletin, (September 1925a), pp. 626-40. ________“ State laws relating to guaranty of bank deposits,” Federal Reserve Bulletin, (September 1925b), pp. 641-68. ________“ Recent amendments to the Federal reserve act (Glass-Steagall bill),” Federal Reserve Bulletin, (March, 1932), pp. 180-81. ________“ Review of the Month,” Federal Reserve Bulletin, (March 1933a), pp. 113-33. ________“ Banking Act of 1933,” Federal Reserve Bulletin, (June 1933b), pp. 385-401. ________ “ Review of the Month,” Federal Reserve Bulletin, (July 1933c), pp. 413-18. _______ . Twentieth Annual Report of the Federal Reserve Board, Covering Operations for the Year 1933 (U. S. Govern ment Printing Office, Washington, 1934a). . “Act of June 16, 1934, extending for 1 year the tem porary plan for deposit insurance, etc.,” Federal Reserve Bulletin, (July 1934b), pp. 486-88. ________“ Banking Act of 1935,” Federal Reserve Bulletin, (September 1935), pp. 602-22. ________Banking and Monetary Statistics, 1914-1941 (U. S. Government Printing Office, Washington, 1943). “ Final Chapter in the History of Bank Deposit Guaranty,” Bankers Magazine (April 1930), p. 472. Fisher, Irving. “ Statement of Prof. Irving Fisher, Professor of Economics, Yale University, New Haven, Conn.,” in To Pro vide a Guaranty Fund for Depositors in Banks, U. S. Con gress, House of Representatives, Committee on Banking and Currency. Hearings, 72d Congress, 1st Session (U. S. Government Printing Office, Washington, D. C., 1932), pp. 143-54. Flynn, John T. “ The Wall Street Water Pump,” Harper’s Monthly Magazine (September 1933), pp. 404-13. ________“ Wanted: Real Banking Reform,” Current History (January 1934), pp. 394-401. “ For Better Banks,” Business Week (May 17, 1933), p. 5. Ford, Henry. “ Essentials of Sound Banking,” Rotarian (April 1933), pp. 6-9, 56-57. Fox, Mortimer J. “ Deposit Insurance as an Influence for Stabilizing the Banking Structure,” Journal of the American Statistical Association (March 1936), pp. 103-12. Friedman, Milton, and Anna Jacobson Schwartz. A Monetary History of the United States 1867-1960 (Princeton University Press, Princeton, 1963). Galbraith, John K. The Great Crash, 1929 (Houghton Mifflin, Boston, 1961). Gayer, A. D. “ The Banking Act of 1935,” Quarterly Journal of Economics (November 1935), pp. 97-116. Glass, Carter. “ Carter Glass urges the Need of Banking Re form,” Review of Reviews and World’s Work (January 1933a), pp. 20-22. ________“ Operation of the National and Federal Reserve Banking Systems: Report to accompany S. 1631,” 73d Con gress, 1st Session, Senate, Report No. 77 (May 15, 1933b). “ The Glass and Steagall Bills,” American Bankers Association Journal (June 1933), pp. 22-23. “ The Glass Bill,” Business Week (May 24, 1933), p. 32. Golembe, Carter. “ The Deposit Insurance Legislation of 1933: An Examination of its Antecedents and its Purposes,” Politi cal Science Quarterly (June 1960), pp. 181-200. ________“ Our Remarkable Banking System,” Virginia Law Review (June 1967), pp. 1091-1114. Greer, Guy. “ Why Canadian Banks Don’t Fail,” Harper’s Monthly Magazine (May 1933a), pp. 722-34. ________“ Wanted: Real Banking Reform,” Harper’s Monthly M agazine (O ctober 1933b), pp. 533-46. Grodsky, Marcia. “ Charles Ponzi,” in Larry Schweikart, ed., Encyclopedia of American Business History and Biography, Banking and Finance, 1913-1989 (Facts On File, New York, 1990), pp. 355-59. “ Guaranty of Deposits: How Can This Threat To Sound Bank ing Be Met?,” American Bankers Association Journal (Sep tember 1933), p. 96. Hapgood, Norman. “ Protect the Depositor,” Nation (March 15, 1933), p. 283. Harris, S. E. “ Banking and Currency Legislation, 1932,” Quarterly Journal of Economics (May 1932), pp. 546-57. Hecht, Rudolf S. “ The Banking Outlook” (with discussion), Proceedings of the Thirty-fourth Annual Convention of the National Association of Supervisors of State Banks (Bran dao Printing, New Orleans, 1935), pp. 42-49. “ The Ideal Bank,” Business Week (April 19, 1933), p. 16. Ives, Mitchell. “ How Banks Have Been Saved,” American Bankers Association Journal (August 1931), pp. 71-74, 112-13. JULY/AUGUST 1992 76 Jamison, C. L. “ Bank Deposit Guaranties: The Insurance Aspects of the Problem,” Bankers Magazine (May 1933), pp. 451-55. Jones, Homer. “ Insurance of Bank Deposits in the United States of America,” Economic Journal (December 1938), pp. 695-706. Kane, Edward J. "The S&L Insurance Mess,” Center for the Study of American Business, Contemporary Issues Series, No. 41 (February 1991). Kaufman, George G. “ Purpose and Operation of the Shadow Financial Regulatory Committee,” in George G. Kaufman, ed., Restructuring the American Financial System (Kluwer Academic Publishers, Boston, 1990). “ The Need of a New Banking Policy,” Bankers Magazine (November 1929), pp. 659-61. “ New Bank Bills,” Business Week (May 24, 1933), pp. 5-6. “ The New Banking Picture,” American Bankers Association Journal (April 1933), pp. 11-13, 37. “ New Banks — Model T,” Business Week (March 8, 1933), pp. 5-6. “ The ‘New Deal in Banking’,” Business Week (June 24, 1933), pp. 22-23. “ New Deal, New Money, New Banks,” Business Week (March 15, 1933), pp. 3-4. Kelly, Edward J., III. “ Legislative History of the Glass-Steagall Act,” in Ingo Walter, ed., Deregulating Wall Street (John Wiley, New York, 1985), pp. 41-65. O’Connor, James F. T. “ Defense of Banking Act of 1933 — Deposit Insurance Provision,” Commercial and Financial Chronicle (American Bankers Convention Supplement, Sep tember 23, 1933), pp. 23-27. Kennedy, Susan Estabrook. The Banking Crisis of 1933 (University Press of Kentucky, Lexington, 1973). ________“ Temporary Deposit Insurance,” American Bankers Association Journal (January 1934), p. 21. Kent, Frank R. “ The Next President,” Scribner’s Magazine (November 1932), pp. 257-261. ________“Address by J. F. O ’Connor, Comptroller of the Cur rency” (with discussion), Proceedings of the Thirty-fourth Annual Convention of the National Association of Supervi sors of State Banks (Brandao Printing, New Orleans, 1935), pp. 95-105. Keynes, John M., in Donald Moggridge, ed., The General The ory and After: Part 1 Preparation (Macmillan Press, 1973). Kiplinger, Willard M. “ The Reconstruction Workshop,” Ameri can Bankers Association Journal (April 1932), pp. 619, 634, 644. Landon, Alf. M. “ The Necessity of a Strong State Banking System Without a Deposit Guaranty,” Commercial and Financial Chronicle (American Bankers Convention Supple ment, September 23, 1933), pp. 50-53. Law, Francis M. “ Banking and the Country,” Proceedings of the Forty-fourth Annual Convention of the Missouri Bankers Association (1934), pp. 37-44. Lawrence, Joseph Stagg. Banking Concentration in the United States: A Critical Analysis (Bankers Publishing Company, New York, 1930). . “Are Big Banks More Profitable?” Bankers Maga zine (January 1931), pp. 9-13. ______ “ Burying the Banker,” Review of Reviews and World's Work (June 1934), pp. 19-23. Love, J. S. “ President’s Address,” Proceedings of the Thirtyfirst Annual Convention of the National Association of Super visors of State Banks (Brandao Printing, New Orleans, 1932), pp. 25-31. Martin, Tiarr. “ Louis Thomas McFadden,” in Larry Schweikart, ed., Encyclopedia of American Business History and Biography, Banking and Finance, 1913-1989 (Facts On File, New York, 1990), pp. 248-50. McLaughlin, George V. “ The Need for Revision of the GlassSteagall Act and a Sane Legislative Program for Banking," Commercial and Financial Chronicle (American Bankers Convention Supplement, September 23, 1933), pp. 16-19. McWhirter, Felix M. “ The Unit Bank Accepts the Challenge,” American Bankers Association Journal (July 1932), pp. 11-13, 61-62. “ Modern Paradoxes,” Bankers Magazine (August 1933), p. 119. Ogg, Frederic A. “ Does America Need a Dictator?” Current History (September 1932), pp. 641-48. “ Opinions on the New Banking Act, Charted from West to East,” American Bankers Association Journal (September 1933), pp. 22-23, 50, 52, 58. “ Other People’s Money,” Business Week (April 5, 1933), p. 9. “ Our Orgy of Speculation,” Bankers Magazine (December 1929), pp. 907-08. Paton, Thomas B. “ 16 Deposit Guaranty Bills in Congress,” American Bankers Association Journal (April 1932), pp. 621, 652. Pecora, Ferdinand. Wall Street Under Oath (Simon and Schuster, New York, 1939). “ Permanent Bank Reform,” Business Week (March 29, 1933), p. 3. Phillips, Ronnie J. “ The ‘Chicago Plan' and New Deal Bank ing Reform,” Jerome Levy Economics Institute, Working Paper No. 76 (June 1992). Platt, Edmund. “ Branch Banking: A Reply,” American Bankers Association Journal (August 1932), pp. 13-15, 62-65. Pole, John W. “ Plight of National Banking System,” Bankers Magazine (July 1929), pp. 23-26. ________Report of the Comptroller of the Currency, 1929 (U. S. Government Printing Office, Washington, D.C., 1930a). _______ . “ Statement of Hon. John W. Pole, Comptroller of the Currency,” in Branch, Chain, and Group Banking U. S. Congress, House of Representatives, Committee on Banking and Currency, Hearings, 71st Congress, 2d Ses sion, Volume 1 (U. S. Government Printing Office, Washington, D. C., 1930b), pp. 3-419. ________Report of the Comptroller of the Currency, 1930 (U. S. Government Printing Office, Washington, D.C., 1931). Nahm, Max B. “ Report of Resolutions Committee —Tributes in Memory of Francis H. Sisson, Peter W. Goebel and Mel vin A. Traylor,” Commercial and Financial Chronicle (Ameri can Bankers Convention Supplement, November 17, 1934), p. 30. ________ Report of the Comptroller of the Currency, 1931 (U. S. Government Printing Office, Washington, D.C., 1932a). National Industrial Conference Board. The Banking Situation in the United States (National Industrial Conference Board, New York, 1932). ________“ Statement of Hon. John W. Pole, Comptroller of the Currency,” in To Provide a Guaranty Fund for Depositors in Banks, U. S. Congress, House of Representatives, Com- http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 77 miltee on Banking and Currency. Hearings, 72d Congress, 1st Session (U. S. Government Printing Office, Washington, D. C., 1932b), pp. 6-56. Preston, Howard H. “ The Banking Act of 1933,” American Economic Review (December 1933), pp. 585-607. Preston, Thomas R. “As a Southern Banker Views the Situa tion,” American Bankers Association Journal (August 1932), p. 15. “ Protecting Bank Depositors,” Bankers Magazine (April 1931), pp. 433-35. Robb, Thomas B. The Guaranty of Bank Deposits (Houghton Mifflin, Boston, 1921). Robinson, A. C. “ The Moral Values of Thrift,” American Bankers Association Journal (October 1931), pp. 209-11, 285-86. Roosevelt, Franklin D. The Public Papers and Addresses of Franklin D. Roosevelt: Volume Two, The Year of Crisis, 1933 (Random House, 1938). Salter, Sir Arthur. “A New Economic Morality,” Harper’s Monthly Magazine (May 1933), pp. 641-49. Schlesinger, Arthur M. The Age of Roosevelt: The Politics of Upheaval (Houghton Mifflin, 1960). Sisson, Francis H. “ The Strength of Our Banking System,” Review of Reviews and World’s Work (December 1932), pp. 30-32, 71. ________“ How We May Have Safer Banks: The Solution Does Not Lie in the Government Guaranty of Deposits," Bankers Magazine (June 1933a), pp. 563-65. ________“Annual Address of the President,” Commercial and Financial Chronicle (American Bankers Convention Supple ment, September 23, 1933b), pp. 30-33. Smith, Rixey, and Norman Beasley. Carter Glass: A Biogra phy (Longmans, Green and Co., 1939). “ Start,” American Bankers Association Journal (January 1934), p. 80. “ State Banks Qualifying for Insurance of Deposits,” Bankers Magazine (November 1933), pp. 489-90. Steagall, Henry B. “ Banking Act of 1933: Report to accompa ny H.R. 5661,” 73d Congress, 1st Session, House of Representatives, Report No. 150 (May 19, 1933a). ________"Banking Act of 1933: Conference Report to ac company H.R. 5661,” 73d Congress, 1st Session, House of Representatives, Report No. 251 (June 12, 1933b). Stephenson, Rome C. “ Providing Safety for Future Banking,” Bankers Magazine (May 1931), pp. 591-94. . “ Making Banks Safe,” Rotarian (September 1934), pp. 34-35, 46-48. Taggart, J. H., and L. D. Jennings. “ The Insurance of Bank Deposits,” Journal of Political Economy (August, 1934), pp. 508-16. Vandenberg, Arthur H. “A Defense of the Bank Deposit Insur ance Law and an Answer to the American Bankers Associ ation” (with discussion), Proceedings of the Thirty-second Annual Convention of the National Association of Supervi sors of State Banks (Brandao Printing, 1933), pp. 38-67. Warburg, Paul M. “A Banking System Adrift at Sea,” Bankers Magazine (March 1929), pp. 569-73. Warburton, Clark. Deposit Insurance in Eight States During the Period 1908-1930, unpublished manuscript, (FDIC 1959). ________“ Origin, Development, and Problems of Deposit In surance,” Lectures in Monetary Economics, mimeo, Univer sity of California at Davis (1967). “ Washington and the Banks,” Business Week (March 8, 1933), pp. 3-4. “ Washington Reads the Signs,” Business Week (March 15, 1933), pp. 4-5. Watkins, Myron W. “ The Literature of the Crisis,” Quarterly Journal of Economics (May 1933), pp. 504-32. Westerfield, Ray B. “ Defects in American Banking,” Current History (April 1931), pp. 17-23. ________“ The Banking Act of 1933,” Journal of Political Economy (December 1933), pp. 721-49. “ What’ll We Use For Money?” Business Week (January 11, 1933), pp. 10-11. Wheelock, David C. “ Monetary Policy in the Great Depres sion: What the Fed Did, and Why,” this Review (March/April 1992a), pp. 3-28. . “ Deposit Insurance and Bank Failures: New Evi dence from the 1920s,” Economic Inquiry (July 1992b), pp. 530-43. ________“ Regulation and Bank Failures: New Evidence from the Agricultural Collapse of the 1920s,” Journal of Econom ic History (forthcoming 1992c). Wheelock, David C., and Subal C. Kumbhaker. “ Which Banks Choose Deposit Insurance? Evidence of Adverse Selection and Moral Hazard in a Voluntary Insurance System,” Fed eral Reserve Bank of St. Louis Working Paper 91-005A (September 1991). White, Eugene Nelson. “ The Political Economy of Banking Regulation, 1864-1933,” Journal of Economic History (March 1982), pp. 33-40. ________The Regulation and Reform of the American Bank ing System, 1900-1929 (Princeton University Press,1983). ________“A Reinterpretation of the Banking Crisis of 1930,” Journal of Economic History (March 1984), pp. 119-138. Willis, H. Parker. “ What Shall We Do with Our Banks?” Na tion (April 13, 1932), pp. 422-425. Taussig, F. W. “ Doctors, Economists, and the Depression,” Harper's Monthly Magazine (August 1932), pp. 355-65. ________"The Banking Act of 1933 — An Appraisal,” Ameri can Economic Association: Papers and Proceedings (March 1934), pp. 101-110. Timmons, Bascom N. Garner of Texas: A Personal History (Harper & Brothers, 1948). Willis, H. Parker, and John M. Chapman. The Banking Situa tion (Columbia University Press, New York, 1934). “ Unit Banking Not So Bad After All,” Bankers Magazine (April 1930), p. 470. Woollen, Evans. “ The Insurance of Bank Deposits,” Atlantic Monthly (November 1933), pp. 609-611. U. S. Congress, House of Representatives, Committee on Banking and Currency. Branch, Chain, and Group Banking, Hearings, 71st Congress, 2d Session (U. S. Government Printing Office, 1930). Woosley, John B. "The Permanent Plan for the Insurance of Bank Deposits,” Southern Economic Journal (April 1936), pp. 20-44. Upham, Cyril B., and Edwin Lamke. Closed and Distressed Banks: A Study in Public Administration (The Brookings In stitution, 1934). Wyatt, Walter. “ Constitutionality of Legislation Providing a Unified Commercial Banking System for the United States,” Federal Reserve Bulletin (March 1933), pp. 166-86. JULY/AUGUST 1992 78 Michael J. Dueker Michael J. Dueker is an economist at the Federal Reserve Bank of St. Louis. Richard I. Jako provided research assistance. The Response of Market Interest Rates to Discount Rate Changes I t IS WELL-ESTABLISHED that discount rate changes o f the same size can have markedly different effects on m arket interest rates. Studies of such effects, starting with Thornton (1982), have generally divided discount rate changes into two groups: "techn ical” changes, those made solely to keep the discount rate in line with m arket rates, and other "non-technical” changes.1 T he form er generally do not have a significant im pact on m arket rates, while the latter generally do. T he use of this technical/ non-technical dichotomy is predicated on the assumption that the m arket responds to a discount rate change based on the reasons for the change. Hakkio and Pearce (1992) find that the reasons generally fall into three categories: “(1) conditions in the m arket fo r bank reserves (2) m ovements in interm ediate targets such as the money supply and the foreign exchange value of the dollar; and (3) m ovements in ultimate targets such as inflation and econom ic grow th.” They observe that “changes in the rate because of type (1) factors are likely to be used to com plement open m arket operations, while changes because of type (2) or (3) factors are m ore likely to be used as signals of future Fed policy.’’2 Thus, technical changes result w hen the oppor tunity cost to banks of borrow ing reserv es—the federal funds rate less the discount rate—is too high or low to be consistent w ith attaining the Fed’s operating target. Since O ctober 1982 that target has been the level of borrow ed reserv es.3 Non-technical changes, on the oth er hand, encom pass all o f the oth er reasons the Fed might change the discount rate. Clearly a com bi nation of the factors identified by Hakkio and Pearce can be behind a given discount rate change, so the reaction of m arket interest rates to discount rate changes might be m ore h eter ogeneous than the technical/non-technical dichotomy would suggest. M oreover, as the efficient m arkets hypothesis implies, the response of m arket interest rates to a discount rate change should vary with the amount of new inform ation the discount change im parts regarding the Fed’s policy intentions or the state o f the econom y in general.4 'The technical/non-technical dichotomy has subsequently appeared in analyses of the effects of discount rate changes on interest rates [Roley and Troll (1984), Smirlock and Yawitz (1985), Thornton (1986, 1991), Cook and Hahn (1988)] and exchange rates [Batten and Thornton (1984)]. rate throughout the 1970s until October 1979 when the Fed began to target non-borrowed reserves. 2Hakkio and Pearce (1992), pp. 56-57. 3Thornton (1988) discusses under what conditions targeting borrowed reserves is equivalent to targeting the federal funds rate. The Fed’s operating target was the federal funds http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis This article presents results on the differential response of m arket in terest rates to discount 4lt is not surprising that the theoretical links between the discount rate and market interest rates have found empirical support in previous studies, given that from 1973 to 1989, for example, 6.2% of the variation in the T-bill rate took place on only 1.3% of the days, the 56 days when the discount rate changed. 79 rate changes using an econom etric fram ew ork that explains m ore heterogeneous responses in m arket interest rates than the technical/non technical dichotomy allows. The m ixture model employed here assumes that the m arket response is determ ined by eith er a “high-response" or "low-response” data-generating process. Inferences about w hich process governs a given period’s interest rate depend on the inform ation policy m akers cite w hen they change the discount rate. Thus, we can consider hypotheses like “the higher the unemploym ent rate, the larger the response of m arket interest rates to a discount rate change of a given size.” W ith the technical/non-technical dichotomy, in contrast, a discount rate change is described as non technical if the Fed m entions any num ber of things in its announcem ent, such as the inflation rate, unemployment rate, industrial production, money grow th rate, etc. T he technical/non technical dichotomy tells us little about the relative im portance of these individual factors. A principal aim of the m ixture model employed h ere is to study the influence these individual factors have on the m arket response. A M IX T U R E M OD EL O F T -B IL L R ES P O N S ES Given the limited num ber of discount rate changes (only 56 from 1973 to 1989), the model estim ates tw o levels o f response of 90-day Treasury bills to discount rate changes. The yield on T-bills is chosen because o f the im portant role it plays in calculating present values for stock dividends, bond coupons, annuities, housing rents, etc.5 W hile the statistical model assumes that one o f tw o mutually exclusive processes generates the change in the T-bill rate from any given discount rate change, the two response levels, "high” and "low ,” should be understood as upper and low er bounds w here all fitted responses are a probability-weighted com bina tion of the two boundary values.6 For example, if ATB is the change in the T-bill rate, ADR is the change in the discount rate and £ is a meanzero stochastic disturbance, then the m ixture model estim ates tw o data-generating processes, Process 1: ATB, = /?0 + /J,ADRt + £t Process 2: ATB( = /30 + /J2ADRt + £t This paper also includes some conjectural interpretations of the em pirical results. For example, if the m arket rates respond strongly to discount rate changes w hen the unemploym ent rate is high, one might conclude that the m arket believes that the Fed will consistently change m onetary policy in reaction to shifts in the unemploym ent rate. Objectively, how ever, the m ixture model’s fit and forecasts of the interest rate response serve as m easures of its perform ance relative to the standard technical/non technical dichotomy. T he second half of the paper addresses the im plication of the efficient m arkets hypothesis that a discount rate change must be “new s” fo r m arket rates to respond by testing w hether the timing of discount rate changes is sufficiently predictable to require that models of the m arket’s response to discount rate changes distinguish explicitly betw een anticipated and unanticipated changes. 5This is because the T-bill rate serves as, or at least proxies, the “ risk-free” rate of return. Applications of the term struc ture theory of interest rates also treat the T-bill rate as an anchor, whose current and expected future values largely determine longer-term interest rates, which are relevant for investment decisions and the level of economic activity. Portfolio insurance, through the writing and buying of options, is another activity that must constantly refer to the T-bill rate; options must be priced such that riskless hedges, which create synthetic riskless assets, do not vio late arbitrage bounds relative to T-bill yields. w here p2 is greater than so that Process 2 governs the highest responses. A single equation can describe the m ixture model if we define a dummy variable, St, w hich equals one if Process 1 holds and zero if Process 2 holds. (1) ATB, = /30 + /?1S,ADRt + /}2( l - S t)ADRt + £t Equation 1 is a m ixture model because the dependent variable is assumed to be drawn from a m ixture of data-generating processes, in this case tw o.7 Because we do not observe St, only probabilistic inferences about its value are forthcom ing. Hence, the in ferred value of St can lie anyw here betw een zero and one, making the m ixture model m ore general than the technical/non-technical dichotomy, which restricts St to equal eith er zero or one. 6The assumption that there are only two response levels is not to be taken literally. It is a convenient way to estimate upper and lower bounds for the T-bill response and thus generate, through mixtures of the two levels, a continuum of response levels the model can explain, while estimating only a few parameters. Of course, some responses will lie outside these bounds: the difference is simply part of the residual and not explained by the econometric model. 7See Quandt and Ramsey (1978). JULY/AUGUST 1992 80 Table 1 Mixture Model Coefficients Parameter Description Value t-statistic Po Intercept Process 1 Response Process 2 Response Constant Magnitude and Sign of ADR Unemployment Rate St. Dev. Outside 1979-82 St. Dev. During 1979-82 When ADR * 0 .0018 .1449 .7743 7.141 -4 .3 7 4 - .4633 .096 .280 .726 1.12 4.57 10.40 3.11 2.77 1.99 Pi h eo ei 02 °o R2 Fu rtherm ore, since a prim ary objective is to use the m ixture model to create one-step-ahead forecasts of the T-bill response to discount rate changes, we pay special attention to the prior probabilities of Process 1 relative to Process 2. In particular we exam ine w hether the prior probabilities are constant or w hether they vary according to the magnitude of the discount change, previous discount rate changes, or various indicators of econom ic activity like infla tion, output, unemploym ent, etc. Such variables (denoted Zt) might indicate w hether financial m arkets believe that the Fed is actively changing policy in response to econom ic conditions. Because drawing inferences about the likelihood of Process 1 vs. Process 2 is analogous to draw ing inferences from a logit model, the logistic function provides a useful param eterization of the prior probability of Process l : 8 1 + exp(Z'0) w here all elem ents of Zt are known at tim e t - 1 , except the change in the discount rate. For policy makers, then, all of Zt is known b efore the Fed actually changes the discount rate, while for m arket w atchers, the Prob.(St= 1 |Zt) is useful fo r making in ferences conditional on the o ccu r ren ce of a given-sized discount rate change.9 8The parameters 0 represent the derivative of the log of the odds of Process 1 versus Process 2 with respect to Z. 9Many professional forecasters will present different forecasts for different “ scenarios,” where one scenario might include an easing in monetary policy accompanied by a discount rate change of 25 basis points. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Table 1 gives results from estimating the param eters in equations 1 and 2, and 6. Fu rth er details on the m ixture model and its estim ation are in the Appendix. Estimation Results f o r Mixture M odel The prior probabilities fo r Process 1 and Pro cess 2 are conditioned on the following variables in the results in table 1: a constant; the change in the discount rate multiplied by the sign of the previous change; and the unemploym ent rate. As an explanatory variable, the change in the discount rate multiplied by the sign of the previous change responds to the following observation: Generally, large absolute changes in the discount rate lead to relatively large responses in the T-bill rate; exceptions occur, however, w hen the discount rate change represents a change in the direction of the discount rate (increases to decreases and vice versa). For this explanatory variable, the relationship betw een the absolute magnitude of the discount rate change and the T-bill response will reverse itself w hen the direction changes. An alternative approach would be to estim ate a separate coeffi cient on a change-in-direction dummy variable, but, given that only eight changes in direction occu r in the sample, the additional coefficient cannot be 81 estimated precisely.10 The unemployment rate is included because it might summarize the effects of real shocks on the econom y.11 The hypothesis that /?, = fi2 is easily rejected, so that qualitative differences among discount rate changes of the same size do indeed cause them to differ in their effects on the T-bill rate. It is also useful to interpret the signs of the 0 param eters, all three of which are significantly different from zero. The positive constant implies that, other things equal, the lowresponse process is m ore likely to hold. The negative coefficient on the magnitude variable implies that increasing the size of the discount rate change leads to m ore than a proportionate increase in the T-bill response, provided that the change is in the same direction as the previ ous one. Thus, perhaps markets interpret 100 basis-point changes in the discount rate as especially convincing signals of a changing environment. The negative coefficient on the unemployment rate indicates that relatively large responses in the T-bill rate are m ore likely when the unemployment rate is high. One interpretation is that the m arket believes that the Fed reacts to high unemployment with active policy steps to stimulate the economy, so the market tends to key off discount rate changes and Process 2 is likely to hold. Table 2 Sum of Squared Residuals when ADR + 0 Sample period Full sample 1979-1982 Outside 1979-1982 Mixture model Technical/ Non technical 1.837 1.448 .390 3.076 1.936 1.141 In fitting the change in the T-bill rate on the days the discount rate changes, the mixture model attains an R" of .726 (on days when the discount rate does not change, the R“ is zero by construction).12 Estimation of the T-bill response, using the technical/non-technical classifications from Federal Reserve announce ments, results in a lower R of .459.13 Further more, as table 2 shows, the m ixture model provides a superior fit across both the October 1979-October 1982 period, when the Fed targeted non-borrowed reserves, and the rest of the sample. The generality of the mixture model, relative to the technical/non-technical dichotomy, is that the probability of Process 1 vs. Process 2 can lie anywhere between zero and one; table 3 shows that the probabilities of the high-response process lie between 10 and 90 percent for five responses. Table 3 also indicates that the differences between the m ixture model and the technical/ non-technical regression derive mainly from the fact that 33 of the 56 discount changes are non technical, yet the estimated probabilities of Pro cess 2 determining the T-bill responses in the mixture model sum only to 12.2, which indi cates that non-technical discount rate changes are considerably heterogeneous with respect to the market response. This concurs with Thornton (1991) who notes that the T-bill rate does not change significantly following some non-technical changes. Nevertheless, almost all high-response cases are non-technical, and on only three occasions did the probability of Process 2, the high-response case, exceed 0.9 outside of October 1979-October 1982, the period of nonborrow ed reserves targeting. It is not yet clear, then, w hether the large T-bill responses between 1979 and 1982 w ere due to the operating proce dure or the abnormally high unemployment rates. The next section shows that both the 10Such a version of the model was estimated with separate coefficients for the magnitude and the sign change. Not surprisingly, the coefficient on the sign-change variable suggests that changes in direction lead to small responses in the T-bill rate; with only eight occurrences, however, the standard error is large, making the point estimate unreliable. The coefficient on the magnitude of the discount rate change, which can use all 56 observations, is statis tically significant. Overall, both the version reported in the paper and the one described here give nearly identical estimates of the response levels and the number of highresponse cases. "O ther variables tried but found not to be significant were the most recent change in the inflation rate and the growth rate of industrial production. 12Note that a mixture model with constant prior probabilities fits almost as well as the one with time-varying prior proba bilities. Nevertheless, the prior probabilities do exhibit statistically significant variation, and by estimating their co-movements with other variables, we gain some insight as to what lies behind the T-bill responses. 13This regression follows Thornton (1982) who first documented that dividing discount rate changes into “ technical” and “ non-technical” changes leads to a regression of interest rate changes on discount rate changes where non-technical changes are significant and technical changes are insign ificant: ATB, = <J0 + A(L)ATB, , +d1DtADRt + (J2 (1- D () ADRt +£t, where D( is a dummy variable that equals one when the discount change is technical. The estimates of d1 and d2 are .036 and .540, respectively, for this data set. JULY/AUGUST 1992 82 Table 3 Specific Discount Rate Changes Date Change in discount rate! Change in T-bill rate Probability of highresponse process Non-technical = 1 1-15-73 2-26-73 4-23-73 5-11-73 6-11-73 7-02-73 8-14-73 4-25-74 12-09-74 1-06-75 2-05-75 3-10-75 5-16-75 1-19-76 11-22-76 8-30-77 10-26-77 1-09-78 5-11-78 7-03-78 8-21-78 9-22-78 10-16-78 11-01-78 7-20-79 8-17-79 9-19-79 10-09-79 2-15-80 5-29-80 6-13-80 7-28-80 9-26-80 11-17-80 12-05-80 5-05-81 11-02-81 12-04-81 7-20-82 8-02-82 8-16-82 8-27-82 10-12-82 11-22-82 12-14-82 4-09-84 11-23-84 12-24-84 5-20-85 3-07-86 4-21-86 7-11-86 8-21-86 9-04-87 8-09-88 2-24-89 0.50 0.50 0.25 0.25 0.50 0.50 0.50 0.50 -0.25 -0.50 -0.50 -0.50 -0.25 -0 .5 0 -0.25 0.50 0.25 0.50 0.50 0.25 0.50 0.25 0.50 1.00 0.50 0.50 0.50 1.00 1.00 -1.00 -1.00 -1.00 1.00 1.00 1.00 1.00 -1.00 -1.00 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 0.50 0.50 0.50 0.030 0.210 0.060 0.230 0.080 0.380 0.230 0.190 -0.180 -0.060 -0.150 0.060 0.010 -0.080 -0.060 0.020 -0.050 0.390 -0.070 -0.060 -0.040 0.110 0.060 0.100 0.160 0.060 -0.200 1.120 0.570 0.220 -0.020 0.160 0.460 0.800 0.980 0.600 -0.060 -0.580 -0.400 -0.810 -0.580 0.700 -0.370 -0.140 -0.320 -0.090 -0.100 -0.130 -0.140 -0.080 0.000 -0.100 -0.130 0.190 0.220 0.040 1.5028E-06 0.043218 0.0099814 0.13455 0.00061177 0.91483 0.073415 0.023996 0.023127 0.0015696 0.029039 3.9024E-05 0.021211 0.0027500 0.037609 2.8641 E-06 0.0038216 0.96901 7.5165E-06 0.0024606 1.9156E-05 0.035044 0.00048244 1.3174E-10 0.012015 0.00052925 1.0186E-07 0.99437 0.75012 2.0124E-06 0.076875 0.022486 0.00035858 0.96824 0.99060 0.86819 2.4887E-05 0.90221 0.56798 0.86346 0.72376 0.019068 0.99147 0.070041 0.96461 1.0819E-07 4.8352E-05 0.010614 0.014009 0.0019898 0.00013878 0.0034851 0.0088348 0.00044880 0.076418 0.00018991 0 1 0 0 1 1 0 1 1 1 0 1 0 0 0 0 0 1 0 0 1 1 1 1 1 1 0 1 1 0 0 0 1 1 1 1 0 0 1 1 1 0 0 1 1 0 1 1 1 1 0 0 1 1 1 1 FEDERAL RESERVE BANK OF ST. LOUIS 83 Table 4 Sum of Squared Forecast Errors Sample period Full sample 1979-1982 Outside 1979-1982 Forecast 1 Technical/ Non-technical Forecast 2 3.552 2.721 .831 3.076 1.936 1.141 2.404 1.711 .694 operating procedure and the unemployment rate m atter for forecasting. H o w G ood A re The One-Step-Ahead Forecasts? Substitution of the prior probabilities, Prob.(St= l|Zt), into equation 1 for St gives onestep-ahead forecasts for this model. Comparing the m ixture model’s sum of squared forecasts errors, found in table 4 under forecast 1, with the sum of squared residuals from the technical/ non-technical regression provides a relative m easure of forecast perform ance. The mixture model’s forecast 1 does not fare well from October 1979-October 1982, although it performs better than the technical/non-technical regression outside this period. One interpretation is that Federal Reserve announcements of discount rate changes, on which the technical/ non-technical classifications are based, take on special importance during periods when the Fed is targeting non-borrowed reserves. To learn about this, we add a dummy variable, which equals one when there is a non-technical change during the 1979-82 period, into Zt in the prior probabilities of equation 2 of the mixture model.14 The sum of squared forecast errors is reported in table 4 under forecast 2. Knowing w hether the discount change is technical greatly improves the forecasts between 1979-82. One possible explanation is that m arket w atchers can directly observe discrete shifts in Fed policy by watching the federal funds rate when it is the operating target. Under non-borrowed reserves targeting, however, the funds rate is market-determined, ' “Adding a second dummy variable for non-technical changes outside 1979-82 does not improve the estimates significantly. 15lt is easy to formulate in-sample forecasts that suggest, for example, that people in 1932 should have known that it so discrete shifts in Fed policy are more likely to be revealed through the discount rate, there by enhancing the informational value of discount rate changes, as it takes time for shifts in policy to translate into sustained changes in the rate of reserves growth. Out-of-Sample Forecasts f r o m 1990-92 Compared with in-sample forecasting, out-ofsample forecasting offers a stiffer and more economically meaningful test of an empirical model. Thus, it is useful to com pare forecasts from the mixture model and the technical/non technical regression for the seven discount rate changes beginning in December 1990, using the coefficients estimated over the 1972-89 period.13 Table 5 summarizes the results. The time-varying prior probabilities of Process 1 vs. Process 2 are clearly illustrated in table 5. As the unemployment rate increases, the prior probability of Process 2 increases, perhaps as markets expect active policy steps from the Fed to combat recession. Also, the change in Decem ber 1991 leads to a much higher prior probabili ty of the high-response process, because it was a change of 100 basis points and the sign of the discount rate change did not change from the previous one. The technical/non-technical regression, in contrast, consistently overpredicts the T-bill responses with its characterization that all non-technical discount rate changes of the same size should have the same effect on the T-bill rate. was a great time to buy stocks. When making real-world decisions, however, people have to forecast into the very uncertain future, a fact captured in out-of-sample forecasting. JULY/AUGUST 1992 84 Table 5 Out-of-Sample Forecasting Date Unemployment rate Technical change Prior Probability Process 1 Change in Discount Rate Change in T-bill Rate Technical forecasted ATB Mixture forecasted ATB 12-19-90 2-04-91 4-30-91 9-13-91 11-6-91 12-20-91 7-02-92 6.1 0 .999 -.5 0 -.11 -.273 -.0 2 7 6.5 0 .875 -.5 0 -.0 2 -.273 -.064 6.6 0 .869 -.5 0 -.0 8 -.273 -.066 6.8 0 .859 -.5 0 -.0 6 -.273 -.069 6.9 0 .853 -.5 0 -.1 3 -.273 -.070 7.1 0 .372 -1 .0 -.3 0 -.546 -.475 7.8 0 .205 -.5 0 -.31 -.273 -.27 4 Overall, the m ixture model with time-varying prior probabilities fits the changes in the T-bill rate better than the technical/non technical regression; it also provides better one-step-ahead forecasts, given that the prior probabilities use information about w hether the change is technical or non technical during periods when the operating target is non-borrowed reserves. Furtherm ore, the variables determining the prior probabilities of the two response levels may reveal something about the market's beliefs about discount rate policy. ARE DISCOUNT RATE CHANGES ANTICIPATED? Previous research has considered that w hether a discount rate change is anticipated or not is a potentially important factor in determining how strongly the T-bill rate responds.16 In other words, when m arket rates do not respond to a non technical change in the discount rate, it might be due to the fact that the m arket antici pated the change and market rates had already moved before the discount change. The relevance of this scenario hinges on w hether m arket p ar ticipants can predict with reasonable accuracy both the timing and magnitude of discount rate changes. The analysis here will follow the work of Hakkio and Pearce (1992) by lumping together 16Examples are Thornton (1986, 1991), Roley and Troll (1984), Smirlock and Yawitz (1985), and Hakkio and Pearce (1992). 17This restriction is simply due to a lack of a sufficient number of 25, 50, and 100 basis-point increases and decreases to allow for full separation of discount changes based on their sizes. Smirlock and Yawitz (1985), on the other hand, obtain an estimate of the expected change in the discount rate, not only the prior probability of a change. This comes at a cost, however, because their model does not consider the discrete nature of discount rate changes, i.e., their FEDERAL RESERVE BANK OF ST. LOUIS different-sized changes in the discount rate and concentrating on w hether the direction and tim ing of changes are predictable.17 The distinction will be that Hakkio and Pearce either estimate sub-samples of discount rate increases and decreases separately, or estimate a multinomial logit model, neither of which recognizes the ordering inherent in discount rate changes (decrease, no change, increase). The ordered probit model employed here takes into account that the probability of a decrease in the dis count rate, relative to the probability of no change, does not remain constant as the probability of an increase changes; the multi nomial logit requires this assumption.18 Maddala (1983) presents the basic ordered probit model, written here in term s of discount rate changes: (3) Prob. [d ecrease |Xt ,) = F(X"t ,/3) Prob.(no change\Xl = F(X* ,/? + c )- F(X’ fl) Prob. (increase |Xt ,) = l-F (X "t j/J + c) w here Xt l is a vector of information available at time t - 1 , F( ) is the cumulative normal density function and c is a positive constant. Furtherm ore, rather than view the anticipated/ unanticipated dichotomy as an alternative to model ignores the fact that the Fed always changes the discount rate by a minimum of 25 basis points, which effectively makes the likelihood of a discount rate change trivially small in many time periods. 18Applications of the multinomial logit model are often criticized for assuming an “ independence of irrelevant alternatives” when this property fails to hold for the choices being modeled. See Maddala (1983) for some examples. 85 Table 6 Response Coefficients for T-Bill Technical Increase Technical Decrease Non-technical Increase Non-technical Decrease «1 a5 a2 - “5 *3 a6 °4 ~ a6 Anticipated Unanticipated technical/non technical as Smirlock and Yawitz (1985) do, we can estimate the m arket’s responses to polychotomous discount rate changes: antici pated technical increases in the discount rate; anticipated non technical decreases; unanticipated technical changes; etc. In all there are eight different responses, as outlined in table 6. Hence, the hypothesis that anticipations of discount rate changes do not significantly move the T-Bill rate cannot be rejected if a 1= a 2 = a 3= a 4 = 0 cannot be rejected. The model imposes symmetrical responses for unanticipated increases and decreases in the discount rate simply due to sample-size constraints. With only 23 and 33 technical and non-technical changes, respectively, it is not possible to obtain good estimates of separate coefficients for either unanticipated technical increases and decreases or unanticipated non-technical increases and decreases. The sequential nature of the model means that we first use time t - 1 information to esti mate the respective probabilities of a decrease, no change or an increase in the discount rate at time t. Then, given the direction of the discount rate change, we use time t - 1 information to estimate the probabilities of technical and non technical changes in the discount rate. Together these prior probabilities give the prior proba bility of a technical discount rate increase: (4) Probability (tech. in crease |Xt_1) = Prob. (increase |Xt t) x P rob.(tech.chan ge\ increase, Xt l) 19We say “ anticipations of discount rate changes” and not “ anticipated discount rate changes,” because the model should include the effect on the T-bill rate of cases in which a discount change seemed likely, but none occurred. The estimates of Smirlock and Yawitz (1985) and Thornton (1991) do not fully account for unfulfilled anticipations of discount rate changes. The objective here is to regress changes in the T-bill rate on the prior probabilities of dis count rate changes, such as the one in equation 4, to see w hether market interest rates react to changing anticipations of discount rate changes.19 Results f r o m the Ordered P robit M od el Estimates from this model help determine which explanatory variables are useful in predicting discount rate changes and to what extent discount rate changes are predictable.20 The results from estimating equation 3 with weekly data (Friday-to-Friday) are in table 7 and Table 7 Ordered Probit Coefficients Variable Intercept Spread Industrial Production Unemployment Rate Constant (c) Coefficient t-statistic 3.88 -.21 7 -27.94 .265 4.34 8.94 4.09 3.88 5.00 24.11 indicate that discount rate changes are some what predictable in a qualitative sense; figures 1 and 2 show that the prior probability of a discount rate decrease or increase often peaks near the actual changes, but it never reaches one-half. Significant explanatory variables for the discount rate changes are the spread between the repurchase rate and the discount rate, 20The variables tried had been suggested in Hakkio and Pearce (1992). JULY/AUGUST 1992 86 Figure 1 Prior Probability of a Discount Rate Decrease Probability 1972 Probability 74 76 78 80 82 84 86 88 90 NOTE: Vertical lines represent dates of d iscount rate cuts. industrial production and the unemployment rate; money grow th is not significant. The signs of the ordered probit coefficients imply that, as the repurchase rate rises above the discount rate, the probability of a discount rate hike increases; low industrial production and high unemployment raise the probability of discount rate cuts, so all coefficients have the expected signs. The grow th rate of industrial production is not significant in determining the prior probabilities in the mixture model, but is significant in predicting discount rate changes, which means that industrial production helps indicate when discount rate changes will take place, but not how market rates will respond. The unemployment rate, on the other hand, is significant in both contexts. Figure 3 provides FEDERAL RESERVE BANK OF ST. LOUIS some interpretation by showing that early in reces sions sometimes m onetary policy easings bring discount rate decreases, yet other times the discount rate simply follows the cyclical path of m arket rates. Late in recessions, however, when the unemployment rate reaches its cyclical peak, m onetary policy easings usually take the discount rate substantially below its pre-recession level. Correlations between the probabilities of the high-response process and the unemployment rate and the growth rate of industrial production also support the idea that the market believes that the Fed shifts m onetary policy m ore often in response to unemployment than output. The correlation coefficient between the probability of the high-response process and the unemploy ment rate is .31; it is only .09 between the high- 87 Figure 2 Prior Probability of a Discount Rate Increase Probability 1972 Probability 74 76 78 80 82 84 86 88 90 N O TE : Vertical lines represent dates of discount rate increases. response probability and the grow th rate of industrial production. Probability That a Discount Rate Change Will Be Technical The ordered probit gives the first probability on the right-hand side of equation 4. The second, the Prob.(technical ch an ge]in crease, X( 1 ), comes from modeling the technical/non technical binary variable with an ordinary probit, using all the discount rate increases in the sample.21 The estimated probit coefficients and Xt_, can then be used to calculate Prob.{technical change]increase, Xt_1) for each observation. W hen using the Fed eral Reserve announcements to form the binary dependent variable (technical/non-technical), however, none of the explanatory variables is a statistically significant predictor of w hether dis count rate increases are likely to be technical or non-technical. Results for both probit models, one each for increases and decreases, appear in table 8. The probability of the discount change being technical is F(Xt JJ), w here F( ) is the cumulative 21The ordinary probit model is similar to the ordered probit of equation 3, except that the dependent variable is binomial, rather than trinomial. JULY/AUGUST 1992 88 Figure 3 The Discount Rate and the Unemployment Rate Percent density function for the normal distribution, so that a positive coefficient on a variable means that the probability that a change is technical increases with that variable. Despite the lack of statistical significance for all coefficients except that on the spread between the federal funds rate and the discount rate for the decreases, we nevertheless use fitted values generated with these coefficients in testing w hether anticipa tions of discount rate changes affect T-bill rates. This is because the limiting factor with respect to anticipating the timing and nature of discount rate changes is most likely an inability to predict the timing, given that in the ordered probit the prior probabilities of discount rate decreases and 22Because the anticipated discount rate change variables are generated regressors (they come from the sequential ordered probit model), the reported standard errors, as FEDERAL RESERVE BANK OF ST. LOUIS Percent increases never reach 50 percent and 25 per cent, respectively, as shown in figures 1 and 2. T-Bill Responses to Anticipated Discount Rate Changes The results of estimating table 6’s response coefficients appear in table 9. None of the four anticipated variables has a significant coeffi cient, although the F-test of joint significance gives an F4905 statistic of 3.115, which lies between the critical value F4,00 .„ = 3.32 at the 99 percent confidence level and the 95 percent critical value of 2 .3 7 .22 Thus when using the Pagan (1984) demonstrates, are valid for hypothesis testing only under the null hypothesis that their coefficients are zero. 89 Table 8 Probability of Technical vs. Non-technical Variable (Xt 1 ) Intercept Spread M1 Growth Rate Industrial Production Growth Unemployment Rate Discount Rate Increases Coefficient t-statistic -.626 .119 8.48 -38.24 -.079 technical/non-technical classifications, it might appear that anticipations of discount rate changes have an effect on the T-bill rate. It is unclear, however, w hether this result holds when we use the m ixture model’s classifications. Consequently, we repeat the exercise using a binary variable generated from the posterior probabilities from the mixture model, whereby a discount rate change is classified as coming from Process 2 if the Prob.(Process 2|A TB)>0.5. Only the dependent binary variable (Process 1/Process 2) changes from the previous analysis; the probabilities of discount rate changes from the ordered probit still apply. Table 10 contains new estimates of the T-bill response coefficients. With the mixture model classifications, the timing of discount rate changes does not appear to be sufficiently predictable to uncover evi dence that anticipations of discount changes lead to movements in the T-bill rate. In table 10, no coefficient on an anticipated variable is significant, and the F4 905 statistic for joint significance is only 1.77, which is less than the 95 percent critical value of 2.37. W e conclude that the timing of a discount rate change is difficult to predict, even at the weekly horizon, and anticipations of discount rate changes do not appear to be major determinants of move ments in the T-bill rate, especially when classi fying the discount rate changes as high- or low-response changes. CONCLUSIONS This paper presents a mixture model of two levels of T-bill responses to discount rate changes. All of the model’s results are compared with results obtained from classifying discount rate .323 .412 1.32 .654 .251 Discount Rate Decreases Coefficient t-statistic 1.71 -1.29 2.61 30.73 -.272 .794 2.273 .494 .823 1.01 changes as technical or non-technical, which is the standard approach in the literature. The mixture model yields superior results with the single exception of forecasting T-bill responses during the 1979-82 period of non borrowed reserves targeting. Conditioning the mixture model’s forecasts on whether the dis count change is technical or non-technical from 1979-82 remedies this shortcoming. Moreover the mixture model is well-suited to forecasting because it derives prior probabilities for each response level, which policymakers and market participants can use to analyze the likely impact of a discount rate change on market interest rates. Table 9 T-Bill Response Coefficients Variable Intercept Spread Unanticipated Non-technical Unanticipated Technical Anticipated Non-technical Decrease Anticipated Technical Decrease Anticipated Non-technical Increase Anticipated Technical Increase Coefficient t-statistic .030 -.049 .827 .230 -.201 -.692 3.123 -.616 1.50 2.45 5.95 1.10 .234 1.624 1.639 .834 Estimates of the m arket’s responses to discount rate changes are consistent with the idea that the market believes in several stylized facts. First, discount rate changes of larger absolute magnitudes appear to generate proportionately larger responses in the T-bill rate. Second, markets look for the Fed to respond actively when the JULY/AUGUST 1992 90 ipations of discount rate changes when we use the mixture model to classify the discount rate changes. Table 10 Alternative T-Bill Response Coefficients Variable Intercept Spread Unanticipated Process 2 Unanticipated Process 1 Anticipated Process 2 Decrease Anticipated Process 1 Decrease Anticipated Process 2 Increase Anticipated Process 1 Increase Coefficient t-statistic -.037 -.047 .826 .174 -.785 -.217 .516 -.273 1.85 2.47 6.03 .833 1.230 .547 .831 .168 unemployment rate is high. Third, discount rate policy apparently becomes an important source of information transmission during periods of non-borrowed reserves targeting. This is prob ably because discrete shifts in Fed policy are not revealed through the federal funds rate under non-borrowed reserves targeting, thereby boosting the status of Federal Reserve announce ments of discount rate changes as indicators of shifts in m onetary policy. The technical/nontechnical dichotomy is much less able to separate these individual influences behind the market response to discount rate changes. Furtherm ore, the m ixture model provides an econometric fram ew ork within which such stylized facts can be quantified to further our understanding of when and why some discount rate changes will have a significant impact on m arket interest rates. The second half of the paper uses a sequential ordered probit model, an econom etric model that is arguably m ore suited to estimating the extent to which discount rate changes can be anticipated than ones used previously in the literature. The estimates are consistent with Smirlock and Yawitz (1985) in that anticipations of discount rate changes might appear to affect the T-bill rate when the changes are classified as technical or non technical. The evidence, however, does not support such a role for antic FEDERAL RESERVE BANK OF ST. LOUIS REFERENCES Batten, Dallas S., and Daniel L. Thornton. “ Discount Rate Changes and the Foreign Exchange Market,” Journal of International Money and Finance (December 1984), pp. 279-92. Cook, Timothy, and Thomas Hahn. “ The Information Content of Discount Rate Announcements and Their Effect on Market Interest Rates,” Journal of Money, Credit and Banking (May 1988), pp. 167-80. Dempster, A. P., N. M. Laird, and D. B. Rubin. “ Maximum Likelihood from Incomplete Data via the EM Algorithm,” Journal of the Royal Statistical Society (Ser. B Vol. 39, No. 1, 1977), pp. 1-38. Hakkio, Craig, and Douglas Pearce. “ Discount Rate Policy Under Alternative Operating Regimes: An Empirical Investi gation,” International Review of Economics and Finance (Vol. 1, No. 1, 1992), pp. 55-72. Hamilton, James D. “ A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle,” Econometrica (March 1990), pp. 357-84. Maddala, G. S. Limited-Dependent and Qualitative Variables in Econometrics (Cambridge University Press, 1983). Pagan, Adrian. “ Econometric Issues in the Analysis of Regressions with Generated Regressors,” International Economic Review (February 1984), pp. 221-47. Quandt, Richard E., and James B. Ramsey. “ Estimating Mixtures of Normal Distributions and Switching Regressions” with Comment, Journal of the American Statistical Association (December 1978), pp. 730-41. Roley, V. Vance, and Rick Troll. “ The Impact of Discount Rate Changes on Market Interest Rates,” Federal Reserve Bank of Kansas City Economic Review (January 1984), pp. 27-39. Smirlock, Michael J., and Jess B. Yawitz. “ Asset Returns, Discount Rate Changes, and Market Efficiency,” Journal of Finance (September 1985), pp. 1141-58. Thornton, Daniel L. “ Discount Rates and Market Interest Rates: What’s the Connection?” this Review (June/July 1982), pp. 3-14. ________“ The Discount Rate and Market Interest Rates: Theory and Evidence,” this Review (August/September 1986), pp. 5-21. ________“ The Borrowed Reserves Operating Procedure: Theory and Evidence,” this Review (January/February 1988), pp. 30-54. _______“ The Market’s Reaction to Discount Rate Changes: What’s Behind the Announcement Effect?” unpublished manuscript, Federal Reserve Bank of St. Louis (September 1991). 91 A ppen dix Estimating the M ixture M odel An intuitive method of estimating mixture models with unknown sample separation across the different processes is the ExpectationMaximization (EM) algorithm of Dempster, Laird and Rubin (1977). Following the EM algorithm, we write the joint density of the change in the T-bill rate and the unobserved state, St, condi tional on Zt as (Al) f(ATBt, St = j|Zt) = #ATBt|St = j) Prob. (S, = j |Zt), j = 0,1. Taking logs and differentiating with respect to y = (/3, 0, o) in A l, we obtain scores of the loglikelihood under the assumption that the changes in the T-bill rate are normally distributed, so that when <f> denotes the normal density function, the probability-weighted scores to be set to zero are (A2) E ^ r , a ln f<ATB,' s =o|zt) LProb- (S, = 0|ATBt) -------------- r ------------- t= l + Finally, Bayes’ Law allows for calculation of Prob. (St = 0|ATBt): (A3) Prob. (St = 0|ATBt) = Prob. (S, = 0 1Zt)^(ATBt|S, = 0) / [prob. (St = 0 1Z)t)^(ATBt|St = 0) + Prob. (St = 1 1Zt)^(ATBt |St = 1)J The EM algorithm calls for the following steps to be taken in the estimation of (/J, 0, o): 1 d ln f(ATB, S, = l| z n Prob. (S, = l|ATBt) ------------- ^ ------------ J The variance ot is assumed to take on either of two values: ot = o, if t G (Oct. 1979-Oct. 1982) = o0 otherwise. Hence, the model allows for o, > o0, reflecting the greater volatility of interest rates experi enced under the Fed’s non-borrowed reserves operating procedure from October 1979 until October 1982. In the case w here changes in the T-bill rate are not normally distributed, the esti mates are still consistent, but not as efficient as they would be if the true density w ere known and maximized. Furtherm ore, Hamilton (1990) has shown that disturbances to real GNP growth appear more homoscedastic and normal when modelled with a non-linear, state-switching model than with a linear model. Step 1 Given starting values of the parameters, calcu late Prob.(St = 0|ATBt) using Bayes’ Law. Step 2 Find (/}, 0, o) which sets the probabilityweighted scores equal to zero. Step 3 With new estimates of (/?, 0, o), update the estimates of Prob.(St = 0|ATBt). Step 4 Iterate over 2 and 3 until convergence. JULY/AUGUST 1992 F e d e ra l R e s e rv e B a n k o f St. L o u is Post Office Box 442 St. Louis, Missouri 63166 The R e v ie w is published six times p er year b y the Research and Public Information Department o f the Federal R eserve Rank o f St. Louis. Single-copy subscriptions are available to the public f r e e o f charge. Mail requests f o r subscriptions, back issues, o r address changes to: Research and Public Information Department, Federal R eserve Rank o f St. Louis, P.O. R oy 442, St. Louis, Missouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal R eserve Rank o f St. Louis o r the Federal R eserve System. Articles herein may b e reprinted provided the source is credited. Please provide the Bank’s Research and Public Information Department with a cop y o f reprinted material.