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.
@FedFRASER