View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Inflation, Unemployment, and Money: Comparing
the Evidence from Two Simple Models
KEITH M. CARLSON
T W O years ago, Professors Barra and Fischer intro­
duced their survey of monetary theory with the fol­
lowing statement:
Perhaps the most striking contrast between current
views of money and those of thirty years ago is the
rediscovery of the endogeneity of the price level and
inflation and their relation to the behavior of money.1
This assessment contrasts sharply with that of the
Council of Economic Advisers in their 1978 Annual
R eport .2 In a forty-one page chapter on inflation and
unemployment, there are only two oblique references
to monetary policy as a contributing factor to the
inflationary process.
The theory of inflation that underlies the Council’s
discussion is conventional — inflation is usually initi­
ated by excess demand, but once the momentum builds
up, “the rate of wage and price increase reacts very
slowly to idle resources and excess supply.”3 The
Council believes there is a trade-off between inflation
and unemployment, but rejects the terms of the
trade-off as too costly. They argue that
. . . it would take at least 6 years of the current de­
gree of economic slack (an unemployment rate near
6 V percent) to cut the inflation rate from 6 to 3
2
percent.4
Consequently, the Council’s recommended strategy for
inflation control is one of “voluntarism”, jawboning,
and structural improvements. Implicit in this strategy
is a stabilization policy stimulative enough to propel
the economy to high employment and full utilization
of capacity.5
The Council’s strategy for economic policy rests on
a belief in the inflation-unemployment trade-off and a
neglect of money. In particular, the apparent current
policy strategy is reminiscent of that applied in Au­
gust 1971 when the price-wage freeze was introduced.
At that time the same thinking prevailed — hold
prices down directly and reduce unemployment via
expansionary monetary and fiscal policy.6
1Robert J. Barro and Stanley Fischer, “Recent Developments
in Monetary Theory,” Journal of Monetary Economics (April
1976), p. 133.
2Economic Report of the President, 1978, pp. 138-78.
3Ibid., p. 150.
4Ibid.
®Ibid., pp. 73-75, 152-56.
6Economic Report of the President, 1972, pp. 22-27.

Page 2


The purpose of this article is to demonstrate that
the apparent trade-off between inflation and unem­
ployment is in fact the result of variable monetary
growth. The approach draws heavily on recent work
by Professor Stein of Brown University.7 The appear­
ance of a trade-off results from differences in the tim­
ing of the response of inflation and unemployment to
changes in monetary growth. However, the trade-off is
an illusion. Unemployment responds to monetary
growth in the short run, but tends towards a steadystate value in the long run. Effects of monetary growth
on inflation are just the opposite; there is little effect
in the short run, with the full and permanent effect
coming in the long run. These processes have implica­
tions that are strongly at variance with those advo­
cated by the Council of Economic Advisers.

The Relation Between Inflation and
Unemployment: T he Conventional
View and Modifications
The relation between inflation and unemployment
is usually depicted by the Phillips curve.8 According to
this relationship, high rates of inflation are associated
with low rates of unemployment; likewise, low infla­
tion rates are associated with high unemployment
rates. Within recent years, however, experience in the
United States and other countries has run counter to
the prediction of the original relation. In particular,
there have been times that inflation and unemploy­
ment have moved in the same direction, a phenom­
enon that has been labeled “stagflation.” Economists
have reacted to this experience by augmenting Phil­
lips curve theory with consideration of the effects of
inflationary expectations.9
Lately, the Phillips curve discussion has taken yet
another twist. Some economists have suggested that
accelerations and decelerations of inflation are related
’ Jerome L. Stein, “Inflation and Stagflation,” forthcoming in
Journal of Banking and Finance, ana “Inflation, Employment
and Stagflation,” Journal of Monetary Economics (April
1978), pp. 193-228.
8The original analysis is found in A. W. Phillips, “The Relation
Between Unemployment and the Rate of Change of Money
Wage Rates in the United Kingdom, 1861-1957,” Economica
(November 1958), pp. 283-99.
9For a survey of the Phillips curve literature, see Robert J.
Gordon, “Recent Developments in the Theory of Inflation and
Unemployment,” Journal of Monetary Economics (April
1976), pp. 185-220.

SEPTEM BER

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

to the level of unemployment. For example, according
to Modigliani and Papademos,
. . . historical experience clearly supports the propo­
sition that there exists some critical rate of unemploy­
ment such that, as long as unemployment does not fall
below it, inflation can be expected to decline. . . .10
They go on to refer to this critical unemployment rate
as the noninflationary rate of unemployment (N IRU ).
In this case, “noninflationary” is defined to mean that
the rate of inflation, at whatever level, is not
increasing.
The value of NIRU can be derived from an esti­
mate of the following simple relation:11
pt — p t-i = O o + O l U t-l
C
C

The symbol p is the year-to-year percent change in
the GNP deflator and U is the unemployment rate.
Using annual data from 1952 to 1976, this equation
is estimated as
(1 )

P t- •p t-i = 2.463 (1 .9 7 0 )

.453 U t-i
(- 1 .8 9 4 )

R2 = .15
SE = 1.41
D W = 1.79

Since the dependent variable is a second difference,
there is considerable variation in it. The unemploy­
ment rate explains only a small portion of this varia­
tion, although both the coefficients in the equation
are significant at the ten percent level (t statistics are
shown in parentheses).
Since NIRU is defined as that rate of unemployment
which is consistent with nonaccelerating inflation, its
value can be found by setting p t — pt-i = 0 in equa­
tion (1 ) and solving for U. The value of NIRU for
this estimated equation is 5.44 percent.
This estimate of equation (1 ) is consistent with the
Council of Economic Advisers’ assessment of the
terms of the inflation-unemployment trade-off in their
1978 Annual Report. A 6.5 percent unemployment
rate was used as an example of sufficient slack in the
economy such that a deceleration of inflation of 0.5
percent per year would be generated. Substituting
6.5 into equation (1 ) yields a decline in p of 0.48
percentage points per year.
By way of comment, it should be noted that this
model does not indicate how a particular rate of
inflation or rate of unemployment is attained. Rather,
the equation simply shows how inflation will change,
10Franco Modigliani and Lucas Papademos, “Monetary Policy
for the Coming Quarters: The Conflicting Views,” New
England Economic Review (March/April 1976), p. 4.
u The Modigliani-Papademos approach to estimating NIRU is
much more convoluted. For a critique of the ModiglianiPapademos results, see Stein, “Inflation, Employment and
Stagflation.”



1978

given the degree of slack in the economy as measured
by the unemployment rate. To complete the model,
an equation specifying the determination of the un­
employment rate would have to be added. In this way
the effect of monetary and fiscal policy could be cap­
tured via the effect on the unemployment rate.

T he Relation Between Inflation and
Unemployment: A Monetary View
An alternate view of the relation between inflation
and unemployment is that both variables are respond­
ing to the movements of a third variable. To the ex­
tent that there appears to be a relationship between
movements in prices and unemployment, it is in fact
a reflection of differential time responses to changes in
the third variable. This is the monetary view, which
stresses the long-run relation between money and
prices, but also takes into account transitory effects of
money on real product growth and unemployment.
According to the monetary view, shifts in the shortrun Phillips curve are associated with changes in the
growth rate of money. The hypothesis is that the
fundam ental determinant of the inflation rate is the
rate of monetary expansion. Regardless of the initial
conditions, the inflation rate will tend to converge to
the rate of monetary growth, and the unemployment
rate will tend toward its steady state value. This
steady state value is not, however, related to the
NIRU concept mentioned above. In fact, the mone­
tary interpretation denies the validity of the NIRU
argument.
In an attempt to keep the analysis simple, another
single equation is specified as representative of the
monetary view. Like Equation (1), the focus is on
accelerations and decelerations of inflation. Accord­
ing to the monetary view, inflation will accelerate if
money growth exceeds the ongoing inflation rate for
an extended period of time (approximately one year).
This simple representation of the monetary view ap­
pears as follows:
P t — p t - i = |3o + |3i( m t - i — p t - i )

Symbols are as defined above, with the addition of
m, the year-to-year percent change in the narrowly
defined money stock (M l).
Estimating this equation for the period 1954 to
1976, using annual data, yields the following:12
(2 )

pt — pt-i =

.449 (m t-i — p t-i)
(4.106)

R2 = .43
SE = 1.13
DW = 1.93

12| o was not significant, so the equation was reestimated with­
3
out the constant.
Page 3

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

These results indicate that inflation will accelerate
by 0.45 of a percentage point in each year following
that in which money growth exceeds the inflation rate
by one percentage point. Based on the specification
of this equation, inflation will not accelerate or decel­
erate if the money growth rate equals the inflation
rate. Comparing the monetary equation with the con­
ventional equation indicates that the monetary equa­
tion explains a larger proportion of the variation in
pt — pt-i, and the standard error of the equation is
reduced by 20 percent.
Suppose now that both views have merit. Can the
rate of monetary expansion and the unemployment
rate be used to explain accelerations and decelera­
tions of inflation? To investigate this possibility, the
following version was estimated:
pt — p t-i = yo + y i (U t-i — UE ) + y 2(m t- i — p t-i)

The value of the critical unemployment rate, as calcu­
lated from Equation (1 ), was inserted into the equa­
tion as UE. Again, using annual data from 1952
through 1976, the following results were obtained:
( 3 ) pt — p t-i =

.001 —
.177 (U t-i — 5 .4 4 )
(.0 0 4 )
( - .8 2 6 )

+ .406 (m t-i — p t-i)
(3 .3 0 1 )

R2 = .45
SE = 1.16
D W = 1.99

For this specification of the equation, supposedly
allowing for both conventional and monetary effects,
neither the unemployment rate nor the constant are
significant. However, the monetary variable remains
significant, although the value of the coefficient is
slightly less than in (2). The R2 and standard error
are only slightly changed from (2).
The implication of these results is that accelerations
and decelerations of inflation are not systematically
related to the degree of resource utilization as meas­
ured by the unemployment rate. Restricting the analy­
sis to very simple models, changes in the rate of infla­
tion are much more closely associated with monetary
growth, with no independent effect coming from the
unemployment rate.
What does the monetary view say about the deter­
mination of the unemployment rate? The monetary
view recognizes short-run impacts of money on out­
put and employment. This relationship can be speci­
fied as:
Ut — U t-i = 6o + SiUt-i + 5 2 (mt-i — p t-i)

This equation is simply a distributed lag response
of the unemployment rate, U, to monetary growth.
There is a transitory effect of money on unemploy­
ment when monetary growth is greater or less than
Digitized forPage 4
FRASER


1978

the inflation rate. Over the long run, however, steady
monetary growth has no effect on unemployment be­
cause, according to equation (2 ), monetary growth
and inflation are the same in equilibrium. As a result,
the mt-i — pt-i term goes to zero and the equilibrium
unemployment rate is determined by 6 0 and &!.
The estimated unemployment equation for the
period 1954-76 is as follows:
(4 )

Ut — U t-i =

3 .9 5 8 —
.721 U t-i
(5 .0 7 9 ) ( - 4 .8 6 2 )
—
.380 (m t-i — p t-i)
( - 4 .4 6 6 )

R2 = .61
SE = .80
D W = 1.57

The implied steady state value for the unemployment
rate, found by setting U t = Ut_i, is 5.49 (that is,
3.958 -i- .721), essentially the same result as in equation
(1 ). The interpretation of this equation is that money
growth in excess of the inflation rate has a temporary
effect on unemployment, but this effect disappears as
the inflation rate converges to the growth rate of
money in the long run. The steady state unemploy­
ment rate for the monetary view differs from NIRU
in that inflation can accelerate even if the unemploy­
ment rate is in excess of its critical value.

Policy Implications
The policy implications of these two views of the
relation between inflation and unemployment differ
substantially. The concept of NIRU suggests that pol­
icymakers need not consider inflation a problem until
unemployment approaches this critical value. On the
other hand, the monetary view stresses the effect of
excessive monetary growth on inflation, independent
of the prevailing unemployment rate. The lesson of
the monetary view is that, in the long run, a steady
growth of money will eventually result in a rate of
inflation equal to that of monetary growth, and a
rate of unemployment that will go to its steady state
value. Only by constantly accelerating monetary
growth is it possible to use monetary actions to re­
duce unemployment.
Consider the two views in terms of conditions as
they exist in 1978. The inflation rate for 1977 over
1976 was 5.9 percent, and the unemployment rate in
1977 averaged 7 percent. The conventional view ar­
gues that inflation will not accelerate as long as unem­
ployment stays above 5.44 percent. Consequently, it
appears that output can be stimulated until the critical
unemployment rate is reached. Then the stimulus can
be reduced to a rate commensurate with long-term
growth. The monetary view, on the other hand, indi­
cates that money growth in excess of the ongoing in­
flation rate can lead to an acceleration in inflation

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

T able I

1978

T able II

HYPOTHETICAL C A SE A: C O N V E N T IO N A L V IE W
Attempted G radual Return to Full Employment
(1 )

pt —

p t -l =

2.4 6 3 —

(2 )

Ut —

U t -i =

—

X

£

HYPOTHETICAL C A S E B: C O N V E N T IO N A L V IE W
Attempted Rapid Return to Full Employment

.453 U t - i

.412 (xt —
P.

(1 )

y

Pt —

p t-i =

2.4 6 3 —

(2 )

3.5)

Ut —

U t -i =

- - .412 (xt —

m

.453 U t -i
3.5)

X

£

P.

y

m

7 .2 %

1 9 7 7 (Act.)

4 .9 %

7 .0 %

5 .9 %

1 1 .0 %

7 .2 %

1 9 7 7 (Act.)

4 .9 %

7 .0 %

5 .9 %

1 1 .0 %

1 97 8

4.7

6.5

5.2

10.1

6.6

1 97 8

7.4

5.4

5.2

13.0

9.5

1979

4 .7

6.0

4 .7

9.6

6.1

1 97 9

3.5

5.4

5.2

8.9

5.4

1 98 0

4.7

5.5

4.5

9.4

5.9

1980

3.5

5.4

5.2

8.9

5.4

1981

3 .7

5.4

4.4

8.2

4.7

1981

3.5

5.4

5.2

8.9

5.4

1 98 2

3.5

5.4

4.4

8.0

4.5

1 98 2

3.5

5.4

5.2

8.9

5.4

1 98 3

3.5

5.4

4.4

8.0

4.5

1 983

3.5

5.4

5.2

8.9

5.4

1984

3.5

5.4

4.4

8.0

4.5

1 98 4

3.5

5.4

5.2

8.9

5.4

1 98 5

3.5

5.4

4.4

8.0

4.5

1 98 5

3.5

5.4

5.2

8.9

5.4

N ote: A path for U was selected and then the path of p was calcu­
lated using equation ( 1 ) . The U path was used to derive the
implied x (the growth rate of output), assuming potential
output grows a t 3.5 percent per year. The x and p paths
were then used to derive y (the growth rate of nominal
G N P ), and then assuming velocity growth of 3.6 percent
per year, the path of m was derived.

even if the unemployment rate is above its critical
value.
Two different policy paths for the conventional
model are shown in Tables I and II. Table I sum­
marizes a policy directed toward a gradual return to
full employment (N IRU ) by 1980, and Table II shows
an attempt to reach full employment quickly — in
1978. These cases were constructed by selecting a
target path for unemployment and then calculating
the effect of unemployment on inflation using Equa­
tion (1). An Okun’s Law13 equation (See equation
(2), Table I) was added to the model to find the
growth of output consistent with the unemployment
path. By adding together the rates of increase in
output and the price level the implied growth of
nominal GNP was calculated as a step towards de­
riving the growth rate of money consistent with the
path of the other variables.14
According to Table I, based on a gradual return to
full employment, inflation and unemployment decline
simultaneously until 1980, and then stabilize. By add­
ing an assumption of constant velocity growth of 3.5
percent to the conventional model, steady state rates

N ote: See Table I.

of monetary growth and inflation are also derived.
These steady state rates appear little different than
those for the monetary model. However, the path to
this equilibrium differs substantially.
According to Table II, also based on the concept
of NIRU, there appears to be no obstacle to returning
to full employment quickly. The difference between
the results in Tables I and II is that a quick return
to full employment “locks” the model in at a higher
growth rate of money and inflation than does the
gradual approach. The reason for this disparity of
results for the conventional model is that the at­
tempted quick return to full employment allows the
effect of unemployment on inflation to operate for
only one year.
Table III

HYPOTHETICAL C A SE A: M O N E T A R Y V IE W
Attempted G rad ual Return to Full Employment
(1 )
(2 )

Pt —

Ut —

Ut--i =

Pt- 1 =

3.958 —

.449 (m t-i - — Pt-l!
.721 Ut -1 —

.3 8 0 (m t-i —

Pt-l)

where m :
v:
p:
x:

m+ v = p + x
rate of increase in money
rate of increase in velocity
rate of increase in the price level
rate of increase in output.




U

P

y

m

7 .0 %

5 .9 %

1 1 .0 %

7 .2 %

1978

3.4

5.4

6.5

10.1

6.6

1 97 9

2.9

5.4

6.5

9.6

6.1

1980

2.9

5.6

6.3

9.4

5.9

1981

13Arthur M. Okun, “Potential GNP: Its Measurement and Sig­
nificance,” The Political Economy of Prosperity ( Washington,
D.C.: The Brookings Institution, 1970), pp. 132-45.
14This is based on the following:

X

4 .9 %

2.0

5.7

6.1

8.2

4.7

1 9 7 7 (Act.)

1982

2.4

6.1

5.5

8.0

4.5

1983

2.8

6.0

5.0

8.0

4.5

1 98 4

3.1

5.8

4.8

8.0

4.5

1985

3.2

5.7

4.7

8.0

4.5

N ote: The path of m was taken from Table I and the path of p was
calculated from equation ( 1 ) . Given m and p, the U path
was then calculated from equation ( 2 ) . The y path from
Table I and the p path were used to calculate an implied x.

Page 5

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

Compare these results with those im­
plied by the monetary model of infla­
tion and unemployment. Using the
growth rates of money derived for the
conventional model in Tables I and II,
the paths for inflation and unemploy­
ment for the monetary model are traced
out in Tables III and IV. According to
Table III, attempting a gradual return
to full employment can be accomplished,
but in the early stages there is an accel­
eration of inflation rather than the de­
celeration predicted by the conventional
model. In 1980, inflation decelerates in
response to the slowing in the growth
rate of money. However, unemployment
also rises again before the steady state
is finally approached in 1984 and 1985.

SEPTEM BER

Table V

ALTERNATIVE CASES: M O N E T A R Y V IE W
Rapid Return to Steady M oney Growth
(1 )

U t-i =

3.958 —

.449

(m t-i —

.721 Ut - 1 —

4 % M oney
Growth

p t -l)

.3 8 0 (m t-i —

P t -l)

6 % M oney
Growth

8 % M oney
Growth

£

P.

U

P.

u

P

£

1 9 7 7 (Act.) 5 . 9 %

7 .0 %

5 .9 %

7 .0 %

5 .9 %

7 .0 %

5 .9 %

7 .0 %

197 8

6.5

5.4

6.5

5.4

6.5

5.4

6.5

5.4

1 97 9

4.5

7.2

5.4

6.4

5.9

5.7

7.2

4.9

1980

3.4

6.9

4.8

6.3

5.9

5.5

7.5

5.0

1981

2.8

6.4

4.4

6.0

5.9

5.5

7.7

5.2

£

198 2

2.4

6.1

4.2

5.8

6.0

5.4

7.9

5.3

198 3

2.2

5.8

4.1

5.6

6.0

5.5

7.9

5.4

1984

2.1

5.7

4.1

5.6

6.0

5.5

7.9

5.4

1 98 5

2.1

5.6

4.0

5.6

6.0

5.5

7.9

5.4

case, the steady-state rate of monetary growth is
begun in 1978. According to these simulations, there
appears to be little prospect for reducing the inflation
rate from its 1977 value without incurring a period of
rising unemployment during the interim. However,
the policies of inflation control (2 and 4 percent
money growth) show that once the period of rising
unemployment is weathered, both inflation and un­
employment decline from their 1977 values toward
their steady-state values.

Conclusions

Table IV

HYPOTHETICAL C A SE B: M O N E T A R Y V IE W
Attempted Rapid Return to Full Employment
Pt —
Ut —

Ut —

2 % M oney
Growth

To explore in greater depth the implications of the
monetary view, alternative simulations of steady
growth rates of money are shown in Table V. In each

(2 )

Pt-■1 =

Pt —

(2 )

Examination of the other case (Table IV ) — an
attempted quick return to full employment — indi­
cates severe oscillations in inflation and unemploy­
ment before the steady state is approached. The un­
employment target is overshot, and the rapid growth
in money in 1978 has its effect on the inflation rate for
several years.

(1 )

1978

U t-i =

.449 (m t-i —

pt-1 =

3.958 —

.721 U t -l —

p t-i)

.3 8 0 (mt-1 -— p t-i)
m

X

U

P

1 9 7 7 (Act.)

4 .9 %

7 .0 %

5 .9 %

1 1 .0 %

7 .2 %

1978

6.1

5.4

6.5

13.0

9.5

y

1 97 9

1.0

4.3

7.8

8.9

5.4

1980

2.1

6.1

6.7

8.9

5.4

1981

2.6

6.1

6.1

8.9

5.4

1 98 2

2.9

6.0

5.8

8.9

5.4

1983

3.1

5.8

5.6

8.9

5.4

1984

3.2

5.6

5.5

8.9

5.4

1985

3.2

5.6

5.5

8.9

5.4

N ote: See Table I I I except th at the paths for m and y are taken
from Table II.

Digitized forPage 6
FRASER


The Administration has taken an approach to con­
trolling inflation that is predicated on the assump­
tion that economic slack is a factor in determining
the inflation rate. In particular, the direct approach
to inflation control has been chosen by the Adminis­
tration because the terms of the trade-off between
inflation and unemployment are deemed unacceptable.
Policy based on this type of reasoning is potentially
disruptive. According to the simple monetary model
used here, attempts to stimulate output with expan­
sionary monetary policy will have accompanying effects
on inflation, despite apparent slack in the economy.
Even though there is a similarity in long-run targets,
substantially different paths to this equilibrium are
derived, depending on which model is used and how
fast the policymakers hope to achieve their targets.

Does the Stage of the Business Cycle
Affect the Inflation Rate?
JOHN A. TATOM

w

▼
▼ITH the U.S. economy well into its fourth year
of expansion and approaching high rates of resource
employment, renewed fears of accelerating inflation
have surfaced. One source of such concern is the
widely held view that the rate of inflation is a cycli­
cal phenomenon, falling during recessions and rising
as the economy approaches a cyclical peak. Accord­
ing to this explanation, inflation is influenced by the
degree of slack in markets for goods, services, and
resources. When there are ample supplies of unused
resources available, price pressures are presumed to
diminish. Similarly, the inflation rate is believed to
accelerate as high employment conditions arise. Dur­
ing such periods, resource availability becomes more
limited and firms, competing for scarce resources to
meet growing demand for their products, bid up
resource prices and consequently product prices.1
Such an explanation has considerable appeal since
it appears to be based upon standard supply-demand
considerations, but the analysis is incomplete and its
use for explaining inflation is limited. The explana­
tion obscures the nature of the inflationary process,
fostering confusion about the cause of inflation and,
more important, confusion over appropriate Govern­
ment policies.
An alternative view contends that inflation results
from a sustained rate of growth in the money stock
which exceeds the growth rate of the quantity of
money demanded by the nation’s wealth owners.
While the focus of this view is on the economy’s
rate of monetary expansion, it leaves open the possi­
bility that in the short run, slack, or its absence, can
1Such an explanation is part of the rationale for an impending
acceleration in inflation predicted by Irwin L. Kellner, Busi­
ness Report, Manufacturers Hanover Trust, Spring 1978. The
view that slack, or the lack of it, influences the outlook for
the inflation rate has also been expressed recently by Lindley
H. Clark, Jr. “How Much Slacks, The Wall Street Journal,
May 2, 1978 and in “The Labor Market May Be Breeding
Inflation,” Business Week, July 31, 1978, pp. 93-94. Such
analysis also plays a crucial role in the Council of Economic
Advisers’ recent discussion of the inflation problem. See Eco­
nomic Report of the President, 1978, pp. 149-50, 168-72.



exert an additional independent influence on the rate
of inflation. The analysis below indicates that such
an independent causal link between slack and the
inflation rate is not supported by recent experience,
once the rate of monetary expansion is taken into
account. An apparent relationship between the rate
of inflation and the extent of slack in resource and
goods markets can easily arise, however. When mone­
tary growth is procyclical, the timing of the impact
of changes in monetary growth on the extent of
capacity utilization and on the inflation rate can give
rise to such observations.
For example, a recession can be caused by a slow­
ing of the growth rate of the money stock. Such a
slowdown in money growth, if sufficiently sharp and
maintained, will reduce the growth rate of total
spending for final goods and services, expectations
of inflation, and the rate of inflation. As a conse­
quence, increased slack (temporarily) and reduced
inflation will both be observed, but each is solely
the result of the pattern of money growth. There
need be no independent causal relationship between
economic slack and the inflation rate.
Similarly, an acceleration in monetary growth can
cause a temporary reduction in slack in the economy
while fueling the longer-term trend rate of growth of
the money stock and, consequently, the inflation rate.
Such a sequence of events would ensure that an ac­
celeration in the rate of inflation, as well as tighter
markets for resources, goods, and services, would be
observed at about the same time, but again solely
as a consequence of the pattern of money growth.

THE EVIDENCE FROM POSTWAR
RECESSIONS
Slack is not a well defined economic concept, but
it refers to conditions in which existing resources
are not utilized to the extent associated with “full
employment.” Such underutilization is typically repre­
sented by a higher unemployment rate or a lower
Page 7

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

Table 1

Inflation Rate and Postwar Recessions
(Com pound A n n u al Rales of C han ge of the G N P Deflator)
IV / 4 8 - IV / 4 9

111/53-11/54

111/57-11/58

11/60-1/61

1 .9 %

IV / 6 9 -IV / 7 0

IV /73-1/75

Inflation Rate During:
Year Prior to Peak
Recession {Peak to Trough)
Year Following Recession

4.1 %

3 .4 %

1 .6 %

-2 .0

1.5

.9

.6

5.1

11.0

5.2

1.7

2.5

1.9

4.7

5.8

utilization rate of manufacturing plant and equip­
ment than the rates achieved during economic booms.
Even without a precise definition, however, most
observers would agree that during a recession the
economy has sufficient resources available at existing
resource prices to produce goods and services at a
faster rate than is observed. During the early periods
of economic recovery following a recession, more
slack would be present than during the peak period
before the recession. Thus, some evidence on the
relationship between the degree of economic slack
and inflation can be obtained by looking at the in­
flation rate which existed before, during, and after
the six postwar U.S. recessions. If inflation is a cycli­
cal phenomenon, rising and falling with the pace
of economic activity, then the inflation rate should
be greatest during the period immediately prior to a
cyclical peak, and lower during the recession and
subsequent recovery period.

5 .3 %

7 .5 %

pelling, it does, in the majority of cases, appear to be
consistent with the cyclical view of inflation.

THE CYCLICAL VIEW OF INFLATION
The cyclical view is embodied in what is referred
to as the “Phillips Curve,” which indicates a trade-off
relationship between the rate of inflation and the
unemployment rate such as that shown in Figure I.2
The existence of such a trade-off means that policy­
makers can only choose among the available com­
binations of unemployment and inflation in setting
policies. In the cyclical context, it is clear from Fig­
ure I that the lower unemployment rate associated
with a cyclical peak requires a higher inflation rate,
Figure I
The P h il l ip s C u r v e

In fact, the evidence from the four postwar reces­
sions prior to the 1970s appears to be fairly consistent
with the cyclical view (Table I). In each case, infla­
tion was not a significant problem during the reces­
sion, averaging no more than a 1.5 percent annual
rate. In the first recession (1948-49), prices actually
fell, on average. Also in each case, the rate of infla­
tion was lower than in the year prior to the recession.
A comparison of the inflation rate in the year follow­
ing a recession with that prevailing prior to the
recession yields mixed results. In two of the first
four recessions, prices rose slower after the recession
than they did before the recession, and in two cases
they rose faster following the recession.
In contrast, the rate of inflation was relatively
high during the two most recent recessions. Nonethe­
less, even these experiences appear to offer some
evidence supporting the cyclical view. In the 1969-70
recession, the inflation rate slowed slightly during
the recession and in each of the two cases, the in­
flation rate was lower in the year following the
recession than it had been in the year preceding the
recession. While the postwar evidence is not com­

Page 8


2See A. W. Phillips, “The Relation Between Unemployment and
the Rate of Change of Money Wage Rates in the United King­
dom, 1861-1957,’ Economica (November 1958), pp. 283-99,
and Richard G. Lipsey, “The Relation Between Unemploy­
ment and the Rate of Change of Money Wage Rates in the
United Kingdom, 1862-1957: A Further Analysis,” Economica
(February 1960), pp. 1-31. The original analysis was stated
in terms of a wage inflation-unemployment rate trade-off. This
was quickly translated into a price inflation-unemployment
rate trade-off by assuming that prices of goods and services
are a constant mark-up over wage costs.

F E D E R A L R E S E R V E B A N K O F ST. LO U IS

while a lower inflation rate occurs only when the
unemployment rate is higher, such as during a
recession.
During the decade of the 1960s, such a trade-off
was believed by many observers to be stable and to
arise from the application of supply and demand
analysis to the market for labor services. In the past
decade, however, it has become increasingly uncer­
tain that such a stable empirical relationship exists.
Chart I shows the combinations of the annual infla­
tion rate and unemployment rate for the United
States from 1954 to 1977. From 1954 to 1969 the hy­
pothesized relationship appears to exist and to be
fairly stable. The gyrations of inflation and unemploy­
ment in the seventies, however, eliminate confidence
in the existence and stability of the Phillips Curve.
Moreover, a considerable literature has developed
since the late 1960s on the factors which influence
the Phillips Curve, and views concerning it have
changed markedly.3 The most notable developments
concern the “natural rate hypothesis” and the “expectations-adjusted Phillips Curve.” The natural rate
hypothesis indicates that in the long run the Phillips
Curve is vertical at a “natural” unemployment rate
where unemployment is associated with only fric­
tional and structural characteristics of labor markets.
The expectations adjustment allows the short-run
Phillips Curve to shift in response to changing expec­
tations concerning the inflation rate.4
These two developments are represented in Figure
II where the long-run Phillips Curve is indicated at
the natural unemployment rate Un, and the short-run
Phillips Curve is drawn as before, but its position
can shift in response to changes in the expected rate
of inflation, u #. Unlike the earlier trade-off view, the
dynamics of inflation and unemployment adjustment
3See Thomas M. Humphrey, “Some Recent Developments in
Phillips Curve Analysis,” Federal Reserve Bank of Richmond
Economic Review ( January/February 1978), pp. 15-23, for a
discussion of the evolution and, according to some theorists,
the demise of the Phillips Curve analysis. Also, see Anthony
M. Santomero and John J. Seater, “The Inflation-Unemployment Trade-off; A Critique of the Literature,” Journal of
Economic Literature (June 1978), pp. 499-544. One of the
original empirical studies of these developments is Leonall C.
Andersen and Keith M. Carlson, “An Econometric Analysis of
the Relation of Monetary Variables to the Behavior of Prices
and Unemployment,” The Econometrics of Price Determina­
tion, Proceedings of a Conference Sponsored by Board of Gov­
ernors of the Federal Reserve System and Social Science Re­
search Council, Washington, D.C., October 30-31, 1970, pp.
166-83.
4These developments, as well as the importance of two different
views of the Phillips Curve for policy analysis, are discussed
in the accompanying Review article by Keith M. Carlson,
“Inflation, Unemployment, and Money: Comparing the Evi­
dence from Two Simple Models.”



SEPTEM BER

1978

C h art I

Inflation a n d U n e m p lo y m e n t
I n f l a t i o n Rate 11

1954-1977

Percent
U n e m p lo y m e n t Rate 12
S o u r c e s : U .S . D e p a r t m e n t o f C o m m e r c e a n d U.S. D e p a r t m e n t

of Labor

1_! P e r c e n t a g e C h a n g e in th e G N P Im p lic it P r ic e D e f la t o r.
[2_ P e r c e n t o f C i v i l i a n L a b o r Fo rce .
L a t e st d a t a p lo t te d : 1 9 7 7

introduced by these hypotheses does not require that
a unique relationship between slack and the inflation
rate exists. While an inverse relationship exists for
movement along a short-run Phillips Curve, shifts in
the curve can lead to the observation of both rising
inflation and rising unemployment ( such as from 1973
to 1974), or both could fall (such as from 1975 to
1976).
For example, suppose the economy is at point A
in Figure II and the monetary authorities increase
the rate of money stock growth in an attempt to
reduce unemployment. According to the old Phillips
Curve view, output demand and employment de­
mand would expand, reducing unemployment. More­
over, as employers bid up wages in their attempt to
expand employment, the inflation rate would rise to
Hj. The economy would move to point B. The change
introduced by the newer view is that such a policy
would lead to revised expectations of inflation, shift­
ing the short-run curve upward. The natural rate
Page 9

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

Figure II
The " N e w ” V i e w

of the P h i l l i p s C u r v e

SEPTEM BER

1978

While the extent of slack in the economy alone does
not appear ( Chart I ) to account for the inflation rate,
it may be that slack exerts som e independent influ­
ence. Does the state of economic activity, as indicated
by measures of “slack,” influence the rate o f inflation
independently of the course of monetary growth? To
examine this question, a standard formulation of the
money-price link is developed. Then the question of
the independent influence of slack on the inflation
rate, given m onetary growth, is addressed.

Money Growth and Inflation

would tend to be restored so that actual inflation
would equal the anticipated rate at point C. The
movement from point A to C could follow the path
shown by the upward arrow, where initially inflation
and slack would move in opposite directions, and
later in the same direction. Similarly, a policy action
to lower inflation to it0 could lead to the reverse path
indicated by the downward arrow from C to A.
Thus, even if there is a short-run Phillips Curve
which exhibits an inverse relationship between the
inflation rate and the unemployment rate, there is
little reason to expect that observed changes in the
two measures will represent movements along the
curve. Instead, expectational factors, according to the
new Phillips Curve view, can give rise to both direct
and inverse relationships between slack and the in­
flation rate. While the observations in Chart I may
or may not be compatible with the new view of the
Phillips Curve, such a view offers little support for
the existence of a unique trade-off and even less sup­
port for its relevance, in the short run. In the long
run, it denies the existence of the inverse
relationship.

AN ALTERNATIVE VIEW OF INFLATION
One of the oldest and most tested postulates of
economic theory is that inflation is a monetary phe­
nomenon. A sustained change in the rate of growth of
the stock of money inevitably causes a similar change
in the inflation rate. When inflation is viewed in this
context, the question addressed here must be restated.

Page 10


The linkage between the growth rate of money
and inflation is usually thought to arise from a stable
relationship between the stock of money (and its
growth rate) and the dollar value of the nation’s
income or total spending on final goods and services
(and its growth rate). An increase in the rate of
growth of the money stock, according to this view,
results in a proportionate rise in the growth rate of
total spending. This growth in total spending can be
divided into the rate of increase of the dollar prices
of the nation’s output (inflation) and the rate of
growth of the volume of output. In the long run, the
output-growth component of spending demand tends
to be constant.5 Thus, a change in the growth rate
of the money stock ultimately tends to be reflected
fully in the rate of change in prices or the inflation
rate.
As an empirical matter, the response of spending
to a change in the growth rate of the money stock
does not occur instantaneously but, instead, appears
to be completed over a four-quarter period. More
importantly, the initial spending response to a change
in the growth rate of the money stock is primarily
reflected in the real output growth component instead
of in the inflation rate. The proportional impact of
money stock growth on the inflation rate occurs only
after a fairly long period of adjustment.6
The hypothesis that the rate of inflation depends
upon the long-term rate of money growth has been
formalized in an equation which relates the rate of
inflation in a quarter directly to the rate of monetary
5The rationale for this steady growth in output is that the
growth of the nation’s ability to produce output is determined
by growth in the supplies of resources such as labor and capi­
tal and the growth rate of resource productivity and these
growth rates tend to be fairly constant.
6One study which provides a more detailed statement of the
theory and evidence supporting these conclusions is Leonall
C. Andersen and Keith M. Carlson, “A Monetarist Model for
Economic Stabilization,” this Review (April 1970), pp. 7-25.

SEPTEM BER

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

M o n e y G r o w th a n d Inflation

Percent

1
i §
1
y i

'0 ,
12
§
%
1

§p 1
n

n

It

1
i

1

%

1954

1955

1956

1957

1958

1959

1960

/

^5.9

n iii

1s
'k Money1
1

lilil

m

1961

1962

1963

1964

1965

1966

1967

1968

1969

1970

\
\

iiili

Uli

1

I

/,

. X'

i

J

J

12

i!if

m
9/
fH 1
W /1
i§

1rp
4i

i

lit

§§ i
m 1

1i

P e r t t nt

'W M . /
<
A lii ;
2
§\infl ilion 1
'/fX'//,
\

ii§
m

1978

1971

1972

1973

1974

1975

1976

1977

1978

S h a d e d a r e a s re p re se n t p e r i o d s o f b u s in e s s re c e s s io n s .
[X T w e n t y - q u a r t e r a n n u a l g r o w t h ra te o f M |.
[2 F o u r -q u a r t e r a n n u a l g r o w t h ra te o f th e G N P D e fla t o r, e n d i n g f o u r q u a r t e r s in th e future.
L a t e s t d a t a p lo t t e d : ln f la t io n - 2 n d q u a rt e r 1 9 7 7 ; M o n e y - 2 n d q u a rt e r 1 9 7 8

growth which prevailed in the past. One such equa­
tion, which can serve as the point of departure here,
n

(1 )

A in P = do + a i Z wi A in M-i

o

where prices, P, are measured by the GNP deflator,
and the money supply, M, is measured by M l. The
rate of increase of prices and money are measured by
changes in their logarithms (A in ).7 An estimate of
the equation for the period I/1954-I/1978 which con­
siders money growth in the current and prior twenty
quarters is:
( 1 ')

20

400* A In P = - .460 + 1.140 Z W 400-A ln M-i
i
( - 1 .2 2 3 ) (12.409) o
R2 = .64

d = 1.07

S E = 1.54

The equation has the characteristics typically hypoth­
esized — the constant term is not significantly different
7This equation is taken from Denis S. Kamosky, “The Link
Between Money and Prices — 1971-76,” this Review (June
1976), pp. 17-23. He argues that the sources of an auto­
regression problem indicated by the Durban-Watson statistic
for the estimate in (1 ') are the presence of wage and price
controls and their removal in the early 1970s as well as the
price level surge associated with the large increase in the
relative price of energy resources in 1973-75. Such an auto­
regression problem was not present up to mid-1971 and the
properties of the equation were the same. It must be noted
that the equation is intended as a short-hand description of
the fundamental inflation process, and excludes other poten­
tially important exogenous variables. The equation is esti­
mated using a third degree polynominal with a zero tail
constraint.

•



from zero ( t-statistics are indicated in parentheses)
and the sum of the past money growth coefficients is
not significantly different from unity.
A simplification of this result is that the rate of
inflation equals the trend rate of money stock growth.
Temporary developments, such as unusually adverse
weather or strikes, may temporarily influence the in­
flation rate from quarter to quarter, but the funda­
mental, permanent component of the inflation rate is
determined by the trend growth of the money stock.
Before examining the independent influence of slack,
given this view of inflation, it is useful to look at the
explanatory power of the trend rate of money stock
growth for the period since 1954 (Chart II).
The errors produced with this simple relationship
are most notable over the 1971-75 period. In 1971-73
the pattern of errors reflects the existence of wage
and price controls and their removal. Initially the
inflation rate was held below the rate indicated for
money growth but later, as controls ended, the infla­
tion rate exceeded the rate of money growth. The
pattern of errors cancels out over a period long
enough to allow prices to “catch up” to their normal
relation to the stock of money. From late 1973
through early 1975, the relationship understates the
rate of increase in prices by a sizeable amount. Dur­
ing this period, there was a substantial increase in
the relative price of energy resources which reduced
the economic capacity of the nation’s productive
facilities. This change caused a sharp, but brief,
once-and-for-all rise in the level of prices in 1974, and
Page 11

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

the inflation rate quickly fell back in line with the
rate indicated by past money growth.8

Does the State of Economic Activity
Affect the Money •Price Link?
If the presence or absence of underutilized re­
sources has an independent influence on the rate of
inflation, then the discrepancy between the inflation
rate and the trend growth of the money stock should
be systematically related to measures of the extent
of “slack” in the economy. For example, using the
cyclical view, one would expect that when there is
evidence of slack in the economy, the rate of inflation
would tend to be smaller than monetary growth alone
would indicate. Moreover, the extent of the reduction
in the inflation rate would presumably be related to
the extent of resource unemployment. Conversely,
when the nation’s resources are fully employed, one
would expect, again using the cyclical approach, that
the rate of inflation would exceed the rate indicated
by monetary factors alone.9
To examine this hypothesis, three measures of
slack are used. The first is the “GNP gap” which
measures the percent by which the economy’s ability
to produce goods and services, given its resources
and their productivity, exceeds its actual output of
goods and services as measured by real GNP.10 A
given percentage gap indicates the percentage by
which the nation’s output of goods and services

8For a discussion of the experience in 1971-76 and the support
it provides for the monetary explanation of inflation see Karnosky, “The Link Between Money and Prices.” The price level
impact of the capacity loss, the mechanism linking prices to the
loss of economic capacity, in 1973-74 is explained in more de­
tail in Robert H. Rasche and John A. Tatom, “The Effects of
the New Energy Regime on Economic Capacity, Production,
and Prices,” this Review (May 1977), pp. 2-12.
9This statement of the hypothesis may be considered to be a
version of the new Phillips Curve view of inflation outlined
above when an additional assumption is added to that view.
The required assumption is that the expected rate of infla­
tion, which shifts the Phillips Curve, is the rate indicated
by the rate of monetary expansion, i.e. equation ( 1 ’ ). Then
departures of the actual inflation rate from the expected in­
flation rate should be systematically related to the extent
of slack, if the short-run Phillips Curve is negatively sloped.
Viewed in this light, the evidence presented below is a test
of the existence of a negatively sloped short-run Phillips Curve,
given the expectations assumption.
10The potential output series used here is a modified series
based upon the methods discussed in Robert H. Rasche and
John A. Tatom, “Energy Resources and Potential GNP,” this
Review (June 1977), pp. 10-24. The series is prepared by
this Bank and is available, together with a description of the
method, from the author.
Digitized forPage 12
FRASER


SEPTEM BER

1978

could be expanded by fully utilizing the capital,
labor, and energy resources available.
A second measure looks only at the extent of util­
ization of labor services. This measure is the differ­
ence between the actual unemployment rate of the
civilian labor force and the rate which would prevail
if labor were fully employed.11 The third measure
reflects slack in the utilization of capital resources.
It is calculated by subtracting the Federal Reserve
Board capacity utilization rate from 87.5 percent. The
latter figure is used here as full utilization of capac­
ity, since it is the rate generally achieved at postwar
cyclical peaks. Chart III shows the three measures
of slack.
According to the cyclical view, the relationship be­
tween the discrepancies of the inflation rate from
trend money growth in Chart II and the slack meas­
ures in Chart III would be expected to be signifi­
cantly negative. In fact, the simple correlation coeffi­
cients of the inflation rate residuals are -.07, -.24, and
-.27 for the GNP gap, excess unemployment rate, and
excess capacity measures, respectively. While the
correlation coefficients all show the correct sign to
support the cyclical view, they are much closer to zero
(indicating no relationship) than they are to minus
one (indicating perfect correlation). Moreover, regres­
sion analysis of the inflation rate discrepancy-slack
relationship indicates no significant relationship be­
tween inflation and slack, once trend money growth
is taken into account. The average error between the
inflation rate and trend money growth is not signifi­
cantly different from zero in such regressions.12
n The full-employment unemployment rate used here is that
prepared and described by Peter K. Clark, “Potential GNP
in die United States, 1948-1980,” U.S. Productive Capacity:
Estimating the Utilization Gap (Center for the Study of
American Business, Washington University, St. Louis, 1977),
pp. 21-66.
12For example, a regression equation for the hypothesis that
actual inflation during the year less trend money growth
through the current quarter depends upon the current gap
yields the estimated equation:
100 ( InPt + 4 — In P t) — 20 ( In Mt - l n M t - 20 ) =
.19 - .03 Gt
where Pt is the price index in quarter t, Mt is the stock of
money in quarter t and Gt is the existing gap in period t.
The standard error of the equation is 1.20, and the t-statistics
for the constant and G coefficient are 1.23 and —.66, respec­
tively. Thus, a significant negative relationship can be re­
jected at the 99 percent confidence level. In addition the
statistically insignificant constant indicates a zero mean dif­
ference between trend money growth and the subsequent
inflation rate. The adjusted R2 is zero to two decimal places.
The equations referred to in the text are estimated with a
Cochrane-Orcutt adjustment and in no case does the constant
have a t-statistic larger than .96, or a slack coefficient have
a t-statistic larger than 1.31, in absolute value.

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

C h o rt III

Measures of Slack
Pe r c e nt

Excess Unem ploym ent

Pe r c e nt

4

4

3

3

2

2

1

1

0

0

-1

-1

-2

-2

Percent

Percent

9

9

-6
1966

1968

1970

1972

1974

1976

1978

Sources: U.S. Departm ent of Labor a n d U.S. Departm ent of Commerce
S h a d e d a re a s represent p e rio d s of b u sin e ss recessions.
Latest d a ta plotted: 1st qu arte r

A more detailed test of the hypothesis that slack
influences the rate of inflation is to use equation (1')
directly. Then the cyclical view may be regarded as
an argument that “initial conditions” matter or that
the predetermined stage of the business cycle is a
significant omitted variable in equation (1) and (1').
This hypothesis can be tested by adding the lagged
value of a measure of slack to equation (1'). Table II
shows the results obtained for three alternative meas­
ures of slack: the GNP gap, the excess of the unem­



ployment rate over the full-employment unemploy­
ment rate, and the Federal Reserve Board’s capacity
utilization rate.13 It should be noted that the capacity
13When a Cochrane-Orcutt procedure is used to adjust for the
significant autoregression in the equations, none of the re­
sults are altered except that the d-statistics become satisfac­
tory. The results reported in Table II were also obtained for
the shorter sample period I/1954-II/1971. An alternative
hypothesis is that slack slows inflation, but only when it is
large and, otherwise, inflation is determined only by money
growth. Using a dummy variable of one for quarters in which
Page 13

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

zero, verifying the hypothesized absence of long-run
real effects of monetary growth.16

Table II

Estimates of the Effect of Initial Conditions

(1/1954 — 1 1978)
/

t-statistic

Sum of
Money
Growth
Coeffi­
cients

Gap

.07

.90

1.13

12.26

Excess Unemployment

.03

.16

1.14

12.29

Capacity Utilization Rate

.03

.70

1.14

12.38

Lagged Variable A dd ed
to Equation ( l ' l *

Coefficient

t-statistic

D urban-W atson
d - statistic
.64

1.10

.64

1.08

.64

1.08

*Ali equations are estimated using a third degree Almon polynominal with a zero tail constraint. The money growth coefficients
include a current and 20 lagged growth rate effects.

utilization rate measures the inverse of slack here
since the rate is not subtracted from the arbitrary
high-employment benchmark of 87.5 percent as in
Chart III. The effect of initial slack on the rate of
inflation has the wrong sign (positive when it is
hypothesized to be negative) in the first two cases and
is not significantly different from zero for any slack
measure. The inclusion of a slack variable has no
noticeable effect on the monetary growth coefficient
or on the quality of the fit of the equation.14
The monetarist view also suggests that the extent
of slack is influenced, in the short run, by changes in
the rate of monetary growth, but that in the long run
such changes have no effect on real economic activ­
ity. Table III presents estimates of the three slack
variables as determined by the current and past
money growth rates contained in the equations of
Table II.15 In each case, the sum of the money
growth coefficients is not significantly different from
slack is greater than one standard deviation above its sample
mean, and zero otherwise again does not alter the equation
( 1 ') and the dummy variable is statistically insignificant in
each case.
14Note that the d-statistic is not affected either which indicates
that slack is apparently not one of the important missing
variables. An alternative hypothesis might be that it is the
rate at which slack is reduced, rather than the level of
slack which affects the inflation rate. While it is difficult to
argue that such a rate is an exogenous variable, including
quarterly changes in the slack variables in the reduced form
above yields the same results as for the level of slack, that
is, they exert no independent influence on the inflation rate.
15The high adjusted R2 results from the use of a CochraneOrcutt adjustment. Without such an adjustment, the adjusted
R2 is smaller than .4 for each equation, indicating that multicollinearity is not likely to be the source of the lack of
significance of the slack variables in Table II.

Page 14


It is useful to examine the pattern of response of a
slack variable, such as the GNP gap, to a change in
the rate of money growth, since it sheds more light
on the cyclical variability introduced by a change in
the rate of money growth. Chart IV shows the pattern
of response of the GNP gap to a one percent increase
in the rate of growth of the money stock obtained
from the first equation in Table III.17 For the first
eight quarters of such an increase in money growth,
the GNP gap is reduced until it is about .77 percent­
age points smaller. In the subsequent three years,
however, such money growth leads to an increase in
the GNP gap so that, in the long run, there has been
no significant change in the size of the gap. Thus, an
increase in the rate of money growth has real effects
in the short run as the GNP gap is reduced. There
are no long-run real effects of a permanent change
in the rate of money growth; only the inflation rate
is affected.18
While the analysis above shows that the stage of
the business cycle does not exert an independent in­
fluence on the rate of inflation, the sometimes con­
trasting evidence from recession experiences may be
disconcerting. The apparent conflict is easily resolved
by the modem view of the Phillips Curve, which sug­
gests such dynamic changes may sometimes occur,
and by the monetary explanation of inflation and
short-run fluctuations in economic activity. Recessions
16The results tend to support another major proposition con­
cerning the functioning of the economy — the economy is
inherently stable. Neither the gap equation nor the excess
unemployment equation, yield a constant term which is sig­
nificantly different from zero, indicating that while changes
in the rate of money growth affect the output gap and ex­
cessive unemployment in the short run, both tend to zero in
the long run — independently of a constant rate of money
growth. Similarly, the significant constant term in the capac­
ity utilization rate equation of 82 percent may be considered
the steady-state capacity utilization rate which is also inde­
pendent of any given sustained rate of money growth. A
similar conclusion using annual data and a model of the
capacity utilization rate may be found in John A. Tatom,
“The Measurement and Meaning of Potential Output — A
Comment on the Perloff and Wachter Results,” CamegieRochester Conference Series on Public Policy, Journal of
Monetary Economics, forthcoming January 1979. Not only is
the natural rate the same as that derived here, the timing
of the adjustment of capacity utilization to changes in the
money growth rate is also the same as that used here.
17The t-statistics of the individual coefficients of money growth
exceed 2.9 in every quarter except at the turning point
where the change in the gap is quite small in quarters
6 to 8.
18Similar qualitative results have recently been found by Robert
J. Barro, “Unanticipated Money, Output, and the Price Level
in the United States,” Journal of Political Economy (August
1978), pp. 549-80. Using a rational expectations model and
annual data, he finds the timing of the impact of money

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

Table III

M onetary Growth and The Cycle

(1/1954 — 1 1978)
/
Constant

Dependent Variable *

t-statistic

Sum of M o n e y
Growth Coefficients

t-statistic

R2

_d_

P_

G ap

2.21

.93

.09

.18

.90

1.47

.93

Excess Unemployment

2.24

1.80

-.2 7

-1 .0 6

.92

1.02

.95

.06

.07

.85

1.31

.89

82.01

Capacity Utilization Rate

2 2 .9 5 **

♦All equations are estimated using a Cochrane-Orcutt adjustment and a third degree Almon polynominal with a zero tail constraint. The
money growth coefficients include a current and 19 lagged growth rate effects.
♦♦Significant a t one percent level.

usually develop as a result of a sharp drop in the
rate of growth of the stock of money.19 Since such a
drop in money growth lowers the trend rate of
growth of money, it is not surprising that the basic
inflation rate often slows. Such a reduction in money
growth slows the rate of growth of total spending for
the economy’s goods and services, temporarily lead­
ing to a recession. Thus, the cyclical variations in the
inflation rate arise because of the influence of mon­
etary policy on both the presence of the cycle and on
the rate of inflation.
Chart IV

C u m u la tiv e C h a n g e in the G N P G a p
D u e to a Pe rm a n e n t O n e Percent Increase
in M o n e y Stock G ro w t h
C u m u la t iv e C h a n g e i> G N P Gap
P e rc u t

CONCLUSION
Recent concern that the inflation rate will acceler­
ate in the 1978-79 period due to increasing limitations
on resource availability is unwarranted. The stage of
growth on inflation to occur over a longer period than indi­
cated here while the temporary effects on real output growth
occur over a similar period.
19This conclusion has a considerable history. See for example,
William Poole, “The Relationship of Monetary Decelerations
to Business Cycle Peaks: Another Look at the Evidence,”
Journal of Finance (June 1975), pp. 697-712.



the business cycle does not affect the inflation rate.20
The future course of the inflation rate will depend
instead on how monetary policy affects the growth of
the stock of money. Other factors, such as the impact
of unusual winter weather in 1977 and 1978 or the
recent coal strike, may have temporary impacts on the
rate of inflation, but the inflation rate is fundamentally
a monetary phenomenon. While some evidence exists
which appears to provide casual support for the view
that the inflation rate is influenced by the stage of
the business cycle, this evidence is misleading. The
confusion arises from a failure to account for both
the short-run influence of monetary growth on eco­
nomic activity and the long-run influence of monetary
growth on the inflation rate. To the extent that the
nation’s cyclical experience has been caused by pro­
cyclical variations in the money stock, one would ex­
pect procyclical movements in the inflation rate. The
evidence presented here provides support for this
view as well as indicating the unimportance of the
stage of the cycle as an independent determinant of
the inflation rate.
20Phillip Cagan argues in “The Reduction of Inflation by Slack
Demand,” in William Fellner, Project Director, Contempo­
rary Economic Problems 1978, (Washington, D. C.: Ameri­
can Enterprise Institute for Public Policy Research, 1978),
pp. 13-45, that slack does slow the inflation rate. He uses
two different models to study the relationship between six
measures of inflation and three slack measures like those
used here, for the periods 1953-69 and 1953-77. The alterna­
tive inflation series are the GNP deflator; consumer price
index; wholesale price index, crude materials; wholesale
rice index, intermediate materials; wholesale price index,
nished goods; and average hourly earnings. Actually, the
statistical evidence ( t-statistics) are only supportive of the
hypothesis in the cases of the consumer price index and
wholesale finished goods in his first model, and to some
extent, only average hourly earnings in the second model.
When the experiment reported in Table II is performed
using measures of inflation based upon other price indexes
(consumer price index, all items; consumer price index for
all urban consumers; producer price index, all commodities;
producer price index, industrial commodities; and hourly
compensation, private business sector) none of the three
measures of slack is significant. The data series used and
sample period are not identical. Nonetheless, it appears that
once the impact of money growth on inflation is accounted
for, slack does not affect the inflation rate even in the limited
number of cases supported by Cagan’s results.
Page 15

Effectiveness of State Reserve Requirements
R. ALTON GILBERT

important decision made by each commercial
bank is whether to be a member of the Federal Re­
serve System (F R S). National banks are required by
law to be members but can withdraw from member­
ship with little difficulty by obtaining state charters.
State banks may choose whether to be members with­
out affecting the status of their state charters.
In recent years many banks have withdrawn from
the FRS, and the primary reason given is that they
must hold a larger share of their assets in non-earning
form as members than if they were nonmembers.1 The
FRS has proposed to reduce member bank reserve re­
quirements as a means of making membership more
attractive. State bank regulators, however, might wish
to counter this action in order to keep the number of
banks under their supervision from declining, and
might seek to do so by lowering reserve requirements
for nonmember banks. For states to offset the effects
of a reduction in FRS reserve requirements on the
attractiveness of membership, state reserve require­
ments would have to be effective, in the sense of in­
fluencing the cash holdings of nonmember banks or
other aspects of nonmember bank behavior. Thus,
effectiveness of state reserve requirements is one of
the issues to consider in estimating the attractiveness
of proposals for reducing member bank reserve
requirements.
Effectiveness of state reserve requirements is ana­
lyzed from three approaches. The first approach ex­
amines how close nonmember banks keep their cash
reserves to required cash reserves. A second approach
examines the influence of state reserve requirements
on the way nonmember banks report their uncollected
funds. Most states do not count cash items in the
process of collection (C IPC) as cash reserves; how­
ever, nonmember banks in such states can use uncol­
lected funds to meet reserve requirements by reporting
them as demand balances due from correspondents,
instead of as CIPC. This second approach tests
whether nonmember banks in states which do not
1Peter Rose, “Exodus: Why Banks are Leaving the Fed,” The
Bankers Magazine (Winter 1976), pp. 43-49.
Digitized for Page 16
FRASER


count CIPC as cash reserves report more of their un­
collected funds as demand balances due from corre­
spondents than nonmember banks in other states. A
third approach tests the effects of state reserve re­
quirements on the percentages of banks which are
FRS members in various states. Details of state re­
serve requirements are reported in a previous issue of
this R eview .2

FIRST APPROACH
Nature of Data Available and
Appropriate Comparisons
Most state banking authorities compare average
cash assets to required cash reserves over one-week or
two-week periods to determine whether banks are
meeting their reserve requirements. However, data
are available in a common format across states only as
2R. Alton Gilbert and Jean M. Lovati, “Bank Reserve Require­
ments and Their Enforcement: A Comparison Across States,”
this Review (March 1978), pp. 22-32. Another approach that
has been used to test the effectiveness of state reserve require­
ments is to estimate the relation between cash assets held by
nonmember banks and required cash reserves. See Lawrence
G. Goldberg and John T. Rose, “Do State Reserve Require­
ments Matter?” Journal of Bank Research (Spring 1977), pp.
31-39. That approach is not used in this paper for the follow­
ing reasons. If state reserve requirements influence cash
holdings of nonmember banks, demand for correspondent
balances by nonmember banks would also be a function of
additional variables, which should be held constant in testing
the influence of state reserve requirements on cash holdings
of nonmember banks. Data on some other determinants of
demand for correspondent balances, such as daily variability
of deposit liabilities, are not available for nonmember banks.
For evidence on the significance of deposit variability for
demand for correspondent deposits, see William G. Dewald
and G. Richard Dreese, “Bank Behavior with Respect to
Deposit Variability,” Journal of Finance (September 1970),
pp. 869-79. Another reason concerns the interpretation if a
positive relation is found between cash holdings and required
cash reserves of nonmember banks. Such a relation might
indicate that banks which hold relatively large percentages
of their assets in cash do so because of relatively high reserve
requirements. On the other hand, such a relation might indi­
cate that state banking authorities keep reserve requirements
relatively high in states in which nonmember banks hold
relatively high percentages of their assets in cash voluntarily.
In such states there would be little pressure on banking au­
thorities from banks to lower reserve requirements. In other
states in which banks wish to hold lower cash ratios, banking
authorities would be under pressure to keep reserve require­
ments no higher than voluntary cash holdings.

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

of individual days. One source is the quarterly Report
of Condition for all Federally insured banks; the other
is balance sheets as of each Wednesday for nonmem­
ber weekly reporting banks. Since these observations
are for individual days, at quarterly or weekly inter­
vals, observed cash holdings may be less than the
required amounts without necessarily indicating that
banks are violating state reserve requirements. Al­
ternately, reserves could be above required levels as
of individual days without necessarily indicating that
nonmember banks voluntarily hold more reserves than
required.
Another complication in drawing conclusions from
ratios of reserves to required reserves for effectiveness
of state reserve requirements is that banks often
choose to hold excess reserves. Relatively small
m em ber banks hold substantial amounts of excess re­
serves, although most of them would tend to hold less
cash if their reserve requirements were reduced.3
These problems of interpretation are dealt with by
comparing the ratio of cash reserves to required cash
reserves for nonmember banks with the ratio of re­
serves to required reserves for member banks of com­
parable size, calculated for the same individual days.
The nonmember ratios are calculated using state re­
quirements, and member bank ratios using FRS re­
quirements. Member bank reserve requirements are
used here as a standard for effective reserve require­
ments. To indicate how this standard is applied, sup­
pose nonmember banks have ratios of cash reserves
to required cash reserves which are significantly higher
than such ratios for member banks of comparable size.
State reserve requirements would be considered not
effective, in the sense that cash holdings of nonmem­
ber banks apparently would not be determined by
state reserve requirements to the same extent that
reserves of member banks are determined by their
required reserves.

Empirical Results
One recent quarterly R eport of Condition is used to
calculate ratios of cash reserves to required cash re­
serves for member and nonmember banks of compa3There tend to be economies of scale in managing a bank’s
reserve position. For instance, Treasury bills have minimum
dollar denominations, and correspondent banks generally
have minimum dollar units in which they invest excess re­
serves of respondent banks in the Federal funds market. Also,
there are efficiencies due to specialization, since the persons
who manage the reserve positions of relatively small banks
generally have additional responsibilities. Excess reserve ratios
of relatively small member banks indicate that the transactions
and cash management costs which are necessary to reduce
excess reserves are larger than the potential increases in
income from investing them.



SEPTEM BER

1978

rable size. These calculations indicate that nonm em ber
banks in most states hold cash reserves which are sub­
stantially larger than their required cash reserves.
Cash reserves several times as large as required cash
reserves were most common among the smallest non­
member banks, with larger banks having smaller ratios.
In all but two of the 38 states for which such compari­
sons are made, the average ratios of cash reserves to
required cash reserves w ere significantly higher than
those ratios for m em ber banks of com parable size.
Thus, based upon this information, state reserve re­
quirements appear to b e less effective than FRS
reserve requirem ents * Details of calculations and sta­
tistical tests are presented in section I of the Appendix.
In 1976 there were 23 weekly reporting banks
which were nonmembers. Two of those banks were
located in states with no cash reserve requirements.
Of the remaining 21 banks, 12 had average ratios of
cash reserves to required cash reserves which were not
significantly different from such ratios for member
banks of comparable size.5 These 12 banks are located
in seven states. Thus, results for nonmember weekly re­
porting banks provide evidence o f effective state re­
serve requirem ents for som e of the relatively large
nonm em ber banks in several states.
There are only a few nonmember banks that are
as large as weekly reporting banks. Evidence from
this approach indicates that state reserve require­
ments are not effective for most nonmember banks
in all but a few states, since their cash holdings are
so much larger than their required cash reserves.

SECOND APPROACH
All but seven states have reserve requirements
which must be satisfied completely, or in part, with
cash reserves, which include vault cash and demand
deposits with other domestic commercial banks.
Among the 43 states with reserve requirements which
must be met with cash assets, 17 allow banks to count
at least some types of cash items in the process of
collection (C IPC) as cash reserves. CIPC represent
primarily the dollar value of checks deposited with
correspondent banks for which the correspondents
have not received payment.
4Note that this result does not imply a comparison of the
burden of reserve requirements of FRS members to the bur­
den of state reserve requirements for nonmembers. The issue
being considered is how close member and nonmember banks
keep their cash reserves to their respective required cash
reserves.
5Each of the nonmember weekly reporting banks in 1976 had
total deposits greater than $180 million.
Page 17

FED E R A L. R E S E R V E B A N K O F ST. L O U IS

Differences among states in treatment of CIPC as
reserves could have significant implications for the
effective levels of state reserve requirements if non­
member banks reported all of their uncollected funds
as CIPC, because uncollected funds constitute sub­
stantial proportions of required reserves for most
banks.6 However, many banks report part or all of
their uncollected funds as demand balances due from
correspondents. Some banks may follow such an ac­
counting practice to use uncollected funds for meeting
reserve requirements. To illustrate why a bank might
do this, consider a nonmember bank which desires to
hold an amount of vault cash plus collected demand
balances with correspondents which is less than its
required cash reserves. If this bank is in a state which
does not count CIPC as reserves, it could increase its
reserves for purposes of meeting state requirements
by classifying its uncollected deposits at correspond­
ents as demand balances due from banks, rather than
as CIPC.
Regression analysis is used to test the influence of
state reserve requirements on the methods non­
member banks use for classifying uncollected funds.
Two hypotheses are tested: in states that have cash
reserve requirements and do not count CIPC as
reserves, (1) nonmember banks report less of their
uncollected funds as CIPC than do other banks, and
(2) the percentage of banks reporting uncollected
funds as CIPC is smaller in states with higher reserve
requirements and in states which enforce reserve
requirements more rigorously.
The regression results support both of these hy­
potheses (see Appendix, section III). The percentage
of nonmember banks reporting CIPC as zero is sig­
nificantly higher in states that have cash reserve
requirements and do not count CIPC as cash reserves.
Another measure of how nonmember banks report
uncollected funds is the percentage of banks report­
ing CIPC less than 25 percent of their demand bal­
ances due from correspondents. With this second
^Uncollected funds as a proportion of cash assets can be meas­
ured most accurately for member banks which send most
of their checks to a Federal Reserve Bank for collection. Mem­
ber banks receive credit for deposits with Reserve Banks
according to a time schedule which approximates the time
required for the FRS to make collection. Uncollected funds
which represent deposits at Federal Reserve Banks for which
member banks have not yet received credit must be reported
as CIPC. For a group of 49 member banks which regularly
deposit checks with their Federal Reserve Bank, CIPC was
about 83 percent of their reserve balances with their Federal
Reserve Bank. See R. Alton Gilbert, “Utilization of Federal
Reserve Bank Services By Member Banks: Implications for
the Costs and Benefits of Membership,” this Review (August
1977), p. 3.

Page 18


SEPTEM BER

1978

measure as the dependent variable, significant inde­
pendent variables are those which reflect treatment
of CIPC as cash reserves, the level of state reserve
requirements, and methods of monitoring reserve
positions of nonmember banks.
These results have implications for the level of
state reserve requirements relative to cash assets
nonmember banks would desire to hold voluntarily.
Cash reserve requirem ents of several states tend to
b e large enough relative to voluntary holdings of
vault cash plus collected dem and balances due from
correspondents to induce behavior by nonm em ber
banks w hich minimizes the burden o f state reserve
requirements. Whether nonmember banks are able
to fully offset the burden of state reserve require­
ments by reporting uncollected funds as demand
balances due from correspondents cannot be deter­
mined from this analysis.

THIRD APPROACH
The major cost of Federal Reserve membership is
reserves required of members, relative to reserves
held by nonmembers. If state reserve requirements
are effective, differences in requirements among states
would tend to induce differences among states in the
percentages of banks that choose Federal Reserve
membership: the percentage of banks within a state
that are members of the Federal Reserve System
would tend to be higher in states with relatively high
state reserve requirements and rigorous enforcement
by state banking authorities.
This hypothesis is also tested using regression analy­
sis.7 Results of those tests indicate that the percent­
age of banks in the Federal Reserve is not signifi­
cantly higher in states with relatively high reserve
requirements. Thus, by just examining the levels of
state reserve requirements, such requirements do not
appear to influence the membership choice of banks.
Two aspects of the enforcem ent of state reserve re­
quirements, however, do significantly influence the
choices of banks concerning FRS membership. The
most important variable reflects differences among
states in methods of monitoring the reserve positions
of nonmember banks. The most rigorous method state
bank supervisors use to monitor the reserve positions
of nonmember banks is frequent reports from banks
on their reserve positions. The percentage o f banks in
7See the Appendix, section IV, for a description of the data
and statistical tests.

F E D E R A L R E S E R V E B A N K O F ST. LO U IS

the F ed is significantly higher in states w hich require
nonm em ber banks to file frequen t reports on their re­
serve positions than in states which use less rigorous
m ethods to monitor com pliance with reserve require­
ments. This result is consistent with the hypothesis that
banks are more likely to choose Fed membership in
states with more rigorous enforcement of reserve
requirements.
The other significant aspect of state reserve re­
quirements is enforcement of penalties on reserve
deficiencies. Several states have dollar penalties which
are relatively low or, according to the state banking
supervisors, are seldom enforced. The percentage of
banks in the Fed is significantly lower in such states
than in other states which have higher dollar penalties,
enforce dollar penalties on reserve deficiencies more
rigorously or have various types of nondollar penalties.
These results indicate that enforcem ent of state re­
serve requirements, not the level of requirements,
influences the choice of banks concerning Federal
Reserve membership. One possible explanation for
this finding is that the measures of enforcement —
requirements for reporting on reserve positions, the
level of dollar penalties, and degrees to which penal­
ties are imposed —■reflect differences among states in
the nature of bank supervision in general, not just
enforcement of reserve requirements. Additional re­
search would be necessary to determine whether
states with relatively more rigorous enforcement of
reserve requirements also have more rigorous enforce­
ment of other banking regulations.

CONCLUSIONS
Empirical tests presented in this paper provide con­
flicting evidence on the effectiveness of state reserve
requirements. Most nonmember banks in all but a
few states hold ratios of cash reserves to required
cash reserves which are significantly larger than
ratios of reserves to required reserves for member
banks of comparable size. These results are consistent
with the view that the cash holdings by most non­
member banks are not determined by state reserve
requirements, but by cash requirements for banking
transactions. Under this interpretation, most nonmem­
ber banks would not tend to hold less cash if their
cash reserve requirements were reduced. Thus, states
could not offset Federal Reserve System (F R S ) ac­
tions intended to increase the attractiveness of mem­
bership — such as lowering member bank reserve
requirements — by lowering reserve requirements for
nonmember banks in response.



SEPTEM BER

1978

However, other evidence presented above calls for
qualifications to this general conclusion. Several rela­
tively large nonmember banks (total deposits of $180
million and above) keep their cash reserves as closely
tied to their required cash reserves as do member
banks of comparable size. This evidence indicates that
state reserve requirements are effective for some of
the relatively large nonmember banks in several
states.
Other evidence which is not necessarily consistent
with the general conclusion on effectiveness of state
reserve requirements is that on reporting of uncol­
lected funds by nonmember banks. In states which
do not count cash items in the process of collection
(CIPC) as cash reserves, nonmember banks report
CIPC which is a smaller percentage of their demand
balances due from correspondents than do nonmem­
ber banks in other states. This evidence indicates that
nonmember banks tend to use their means of report­
ing uncollected funds to minimize the burden of state
reserve requirements.
Additional evidence which supports the conclusion
that state reserve requirements are not effective con­
cerns the influence of state reserve requirements on
the percentage of banks in various states which are
FRS members. The level of state reserve requirements
does not significantly influence the percentage of
banks which are FRS members. However, some dif­
ferences among states in methods of monitoring the
reserve positions of nonmember banks and enforcing
reserve requirements are significantly related to dif­
ferences in the percentage of banks that are FRS
members.
An overall assessment of results in this analysis
supports the view that in general state reserve re­
quirements are not effective. Evidence cited above
which is inconsistent with this general conclusion
calls for only limited qualifications, and may raise
more questions than it answers.
Only a small number of nonmember banks have
total deposits over $180 million. Thus, evidence on
effectiveness of state reserve requirements for several
nonmember banks in that size range applies to only a
small percentage of banks which would posssibly
be influenced by a reduction in FRS reserve
requirements.
Evidence that nonmember banks in some states
attempt to minimize the burden of state reserve re­
quirements by the way they report uncollected funds
does not indicate whether any burden remains after
banks take such actions. Nonmember banks in states
Page 19

F E D E R A L R E S E R V E B A N K O F ST. LO U IS

SEPTEM BER

1978

which do not count CIPC as cash reserves may be
able to avoid all burden of state reserve require­
ments by reporting their uncollected funds as demand
balances due from correspondents.

are FRS members may indicate as much about the
influence of differences among states in overall bank
regulation as it does about the influence of state re­
serve requirements on membership choice.

Differences among states in methods of monitoring
reserves of nonmember banks and enforcing reserve
requirements may be related to differences among
states in overall stringency of banking regulation.
Therefore, the evidence cited above concerning vari­
ables which influence the percentages of banks which

Thus, evidence developed in this paper indicates
that the FRS could increase the attractiveness of
membership by lowering member bank reserve re­
quirements. With only a few exceptions, states could
not offset the effects of such an action by lowering
reserve requirements for nonmember banks.

APPENDIX
Specification of Data and Empirical Results

RATIOS OF CASH RESERVES TO
REQUIRED CASH RESERVES FOR
MEMBER AND NONMEMBER BANKS
Ratios from the Report of Condition
Ratios of cash reserves to required reserves are calcu­
lated for all nonmember banks in states that have cash
reserve requirements, using data as of June 30, 1 9 7 6 .1 The
ratios are averaged for nonmember banks in each state
within the following size groups in terms of total deposits:
(a ) up to $10 million,
(b ) $10 million to $50 million,
(c ) $50 million to $ 1 0 0 million, and
(d ) $100 million to $ 5 0 0 million.2
Average ratios of reserves to required reserves are pre­
sented in Table A-I. E ach t-statistic (calculated for the
iSee Appendix, section II, for discussion of a possible bias in
the Report of Condition data.
-In most states there are few, if any, nonmember banks with
total deposits over $500 million. The influence of bank size
on the ratios of reserves to required reserves is held constant
by dividing nonmember banks in each state into these size
groups. A few banks which had extreme ratios were elimi­
nated from the analysis. The banks which were eliminated
from calculations in this section were also eliminated from
analysis in the following sections. Another study has drawn
inferences about the effectiveness of state reserve require­
ments based upon reserve ratios from the Report of Condi­
tion. One limitation of the study is that no criterion was de­
veloped from determining how large reserves can be in

Page 20


difference between the mean ratio for nonmember banks
and the mean ratio for members of comparable size) is
used to test the hpothesis that reserve requirements of a
state are effective.3 In each case in-which the difference
in mean ratios is not significantly different from zero, re­
serve requirements of that state are considered as signifi­
cant in determining the cash holdings of nonmember
banks as FRS reserve requirements are in determining the
reserves of member banks.
Results in Table A-I indicate that nonmember banks in
most states hold cash reserves which are substantially
above their required cash reserves. Nonmember banks in
all but two states, South Dakota and Wisconsin, had aver­
age ratios of reserves to required reserves which were
significantly higher than the average reserve ratios for
member banks of comparable size. Reserve requirements
are relatively high in both of these states. Thus, based
upon the criterion used in this section, state reserve re­
quirements are as effective as FRS reserve requirements
in only two of the 3 8 states examined.
relation to required reserves as of an individual day and yet
be consistent with effective state reserve requirements. See
Perry D. Quick, Appendix A, “Nonmember Bank Reserve
Requirements,” in “The Burden of Federal Reserve Member­
ship, NOW Accounts, and the Payment of Interest on Re­
serves,” prepared by the Staff of the Board of Governors of
the Federal Reserve System, June 1977, pp. 71-96.
3Mean ratios of reserves to required reserves, based upon the
Report of Condition for June 30, 1976, are calculated for the
combined group of member banks in those states which have
cash reserve requirements. This group of states includes most
member banks in the nation.

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

Table A-l

RATIO S OF C A S H RESERVES T O REQUIRED C A S H RESERVES O F N O N M E M B E R B A N K S 1

State

Sire G roup
(millions of
dollars)

Percent Reserves
are of
Required Reserves

t-statistic2

State
North Dakota

California

1 6 0 .3 1 %

6.568

137.11

up to $ 10

2 0 9 .0 7

13.958

170.41
25 4 .8 9

2 2 .2 4 6

O h io

193 .09

24.315

156 .44

1 5 .4 4 7

$ 1 0 0 to $ 5 0 0

138.52

16 .745

O regon

$ 1 0 to $ 5 0

2 4 9 .1 4

4 6 .2 9 0

Pennsylvania

3 7 3 .1 9
3 1 1 .5 4
13 5 .9 7

2.158

126 .74

4.5 1 5

$ 5 0 to $ 1 0 0

121 .86

up to $ 1 0

168 .42

198.01

9.821

$ 1 0 to $ 5 0

168.33

17.578

up to $ 1 0

3 9 3 .9 3

3 7 .6 0 2

$ 1 0 to $ 5 0

Kentucky

3 6 6 .1 7

3 3 .3 0 0

175.85

23.2 4 8

up to $ 1 0

2 0 5 .09

12.276

172 .73

19.001

$ 1 0 to $ 5 0

2 0 8 .0 0

32.5 3 8

6 6 .5 3 7

up to $ 1 0

2 7 8 .1 5

27.375

196.38

27.1 8 6

159.11

16.678

181.31

4 4 .8 5 9

up to $ 1 0

218.11

16.866

18 5 .0 9

2 4 .5 4 2

up to $1 0

133 .50

1.850

$ 1 0 to $ 5 0

9.5 1 3

up to $ 1 0

2 3 .989

21 8 .6 7

$ 1 0 to $ 5 0

6.6 62

144.05

265.28

$ 1 0 0 to $ 5 0 0
South Carolina

South Dakota

131.18

6.1 6 7

up to $ 1 0

171.08

7.971

$ 1 0 to $ 5 0

Tennessee
Kansas

up to $ 1 0

2.0 09

$ 1 0 to $ 5 0

Iow a

up to $1 0

6.7 1 4

$ 5 0 to $ 1 0 0

4 8 .7 6 0

$ 1 0 to $ 50

2.3 8 4

133.11

$ 1 0 to $ 5 0

23.858

$ 1 0 to $ 5 0
Indiana

up to $ 1 0

136 .63

$ 1 0 to $ 5 0

$ 1 0 to $ 5 0

G eorgia

up to $ 1 0

$ 5 0 to $ 1 0 0
O klaho m a

$ 5 0 to $ 1 0 0

Connecticut

t-statistic2

$ 1 0 to $ 5 0

18.672

up to $ 10

Percent Reserves
are of
Required Reserves

$ 1 0 to $ 5 0

8.0 0 7

$ 1 0 to $ 5 0

A rkansas

up to $ 10
$ 1 0 to $ 5 0

Alabam a

Size G roup
(m illions of
dollars)

151.91

12.648
19.659

$ 5 0 to $ 1 0 0

166 .33

up to $ 1 0

180.81

8.8 7 0

$ 1 0 to $ 5 0

Texas

166 .74

15.762
14.250

174.68

38.8 9 8

up to $ 1 0

257 .78

2 4 .392

up to $ 1 0

2 5 0 .8 6

2 0 .3 6 7

2 4 8 .2 0

4 4 .1 9 2

$ 1 0 to $ 5 0

157 .20

15 .127

up to $ 1 0

7 5 0 .5 7

8 2 .2 9 7

up to $ 1 0

190.05

11.944

6 9 4 .4 3

158 .27 5

$ 1 0 to $ 5 0

145.62

11.093

up to $ 1 0

2 2 3 .7 5

17.321

up to $ 1 0

193 .29

12.598

20 4 .3 3

29.772

$ 1 0 to $5 0

155.98

14.665

17 7 .7 0

23.772

up to $ 1 0

206 .05

13.773

$ 1 0 0 to $ 5 0 0

160.48

3 2 .3 4 9

$ 1 0 to $ 5 0

1 7 9 .97

22.361

up to $ 1 0

2 2 2 .9 7

12.884

$ 1 0 to $ 5 0
M ississipp i

155.93

$ 1 0 0 to $ 5 0 0

$ 5 0 to $ 1 0 0

Minnesota

$ 5 0 to $ 1 0 0

3 8 .815

$ 1 0 to $ 5 0

M ichigan

5 4 .9 2 0

2 1 7 .3 5

$ 1 0 to $ 5 0

Massachusetts

3 0 5 .88

$ 1 0 to $ 5 0

M a ryla n d

$ 1 0 to $ 5 0
$ 5 0 to $ 1 0 0

Louisiana

161.08

12.532

up to $ 1 0

17 5 .3 4
14 2 .8 6

9 .4 7 4

132.51

6.1 7 4

up to $ 1 0

153.83

5 .5 8 6

119.45

2.1 26

up to $ 1 0

17 4 .9 6

7.1 2 2

W ashington

W est Virginia

up to $ 1 0

134.81

1.898

$ 1 0 to $ 5 0

W isconsin

109 .74

$ 5 0 to $ 1 0 0

113 .53

-1 .1 6 1
-1 .2 3 3

$ 1 0 to $ 5 0

1 6 9 .97

19.538

7.8 2 9

$ 1 0 to $ 5 0
N ebraska

3 2 .7 7 9

Virginia

2 9 .688

2 1 7 .5 7

$ 5 0 to $ 1 0 0
M ontana

30 7 .8 2

$ 1 0 to $ 5 0

M issouri

up to $1 0
$ 1 0 to $ 5 0

Utah

$ 1 0 to $ 5 0

153 .16

13.2 5 7

New Hampshire

up to $ 1 0

22 7 .7 2

18.810

$ 1 0 to 5 0

148 .16

12.039

New Jersey

$ 1 0 to $ 5 0

148.78

12.083

$ 1 0 0 to $ 5 0 0

121 .99

6.053

New Mexico

$ 1 0 to $ 5 0

266.21

5 1 .0 5 3

New York

$ 1 0 to $ 5 0

134.28

7.198

North Carolina

up to $ 1 0

20 5 .9 0

14.929

$ 1 0 to $ 5 0

171 .30

19.862




W yom ing

•Observations of cash assets and deposit liabilities for member and
nonmember banks are derived from the R ep o r t o f C ondition, Ju ne
30, 1976. Average ratios of reserves to required reserves are calcu­
lated for nonmember banks in each of the four size groups in
c
each state with <ash reserve requirements. Ratios are reported for
size groups with ten or more nonmember banks. Ratios of reserves
held to required reserves are calculated for member banks m 43
states th at have cash reserve requirements for nonmember banks.
Inform ation on the ratios of reserves to required reserves for
member banks is provided below
Standard
Size G ro u D
Number
Mean P ercent that
Deviation
Reserves are of
(Total Deposits in
of
of Ratio
Required Reserves
Millions of Dollars)
Banks
up to $10
$10 to $50
$50 to $100
$100 to $500

1082
2608
512
408

123.18
113.29
116.79
113.25

48.710
28.942
33.183
33.658

2t-statistics are calculated fo r differences between mean ratios of
reserves to required reserves of nonmember banks and mean ratios
of reserves to required reserves of member banks of similar size.
W ith the exception of all three size categories for Wisconsin and
the "up to $10 m illion" category for South Dakota, all t-statistics
are significant at the 5 percent level.

Page 21

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

Ratios for Nonmember W eekly
Reporting Banks
Weekly reporting banks comprise a national sample of
relatively large commercial banks which report balance
sheet information as of each Wednesday. In 1976, 23
weekly reporting banks were nonmembers, each with total
deposits over $180 million. Two of those banks are located
in states with no cash reserve requirements. Ratios of
cash reserves to required cash reserves were calculated for
the remaining 21 banks as of each Wednesday in 1976
and averaged for each bank over the year (see Table A-II).
As a basis for comparison, average ratios of reserves
to required reserves, under reserve requirements of the
Federal Reserve, were calculated for 18 member banks
in the Eighth District, with total deposits of at least $180
million. For comparability with data for nonmembers, the
measure of cash reserves for each member bank is its re­
serve balance at the Federal Reserve at the close of each
Wednesday plus average daily vault cash during the
reserve settlement week ending two weeks earlier. Aver­
age daily required reserves are based upon deposit liabil­
ities two weeks earlier. Ratios of reserves to required
reserves are calculated for each member bank for each
Wednesday in the period from September 15, 1976
through January 12, 1977.
Mean ratios of reserves to required reserves of the 18
member banks are used to establish an acceptance region
for testing the hypothesis that the mean reserve ratio for
each nonmember bank was drawn from the same distri­
bution as that for member banks. This hypothesis is not
rejected, at the 5 percent level of significance, if the
mean ratio for a nonmember is in the range from 0.585
to 1.509.
Using this criterion, the hypothesis that reserve require­
ments are effective is not rejected for 12 of the 21 non­
member banks, located in California, Hawaii, Michigan,
New York, North Carolina, Ohio, and Pennsylvania. Thus,
results in Table A-II provide evidence of effective reserve
requirements in several states for some of the relatively
large nonmember banks.

IN

ANALYSIS OF POSSIBLE BIAS
REPO RT O F CO NDITIO N DATA

One possible problem with relying upon the Report of
Condition for information on cash holdings of nonmember
banks is that banks might increase their cash holdings on
the known dates for the Report of Condition and reduce
them immediately afterwards. Banks might behave that
way if they generally hold cash reserves which are less
than required reserves, since that report is disclosed to the
public and made available to state banking authorities.
Determining whether cash holdings of nonmember
banks from the Report of Condition are unusually high
requires information from other sources for comparison.
One source is the data for nonmember weekly reporting
banks discussed above.

Page 22


SEPTEM BER

1978

Table A -II

RATIO O F RESERVES T O REQUIRED RESERVES:
C O M P A R IS O N FOR LARGE MEM BER A N D
NONM EM BER BANKS

State
California

Bank
Num ber

1
2
3
4

Connecticut

1
2
3

A verage Ratio
for Each
W ednesday. 1 97 6

Interpretation:
Hypothesis
of Effective
State Reserve
Requirements*

1.166
1.484
1.091
1.020

Accept
Accept
Accept
Accept

1.810
2.329
2.241

Reject
Reject
Reject

Delaware

5.6 2 5

Reject

Haw aii

0 .9 8 6
0 .8 3 9

Accept
Accept

M a ryland

1.846

Reject

M ichigan

1.353

Accept

M issouri

2.004

Reject

New York

2.029

1.112

Reject
Accept

North Carolina

1.800
1.445

Reject
Accept

O h io

1.300

Accept

Pennsylvania

1.553
1.062
1.333

Reject
Accept
Accept

♦The hypothesis of effective state reserve requirements is not re­
jected if the average ratio of reserves to required reserves for a
nonmember bank is in the range from 0.585 to 1.509.

Although weekly reporting banks are larger than most
nonmembers used in the calculations from the Report of
Condition, they are probably part of the nonmember
group which would have the greatest incentives to hold
unusually high cash reserves on the Report of Condition
dates. Nonmember banks in the smaller size groups in
most states report cash reserves which are substantially
above required cash reserves. These banks would not have
incentives to hold cash reserves that much larger than
their required cash reserves for just the day of the report.
In contrast, the larger nonmember banks in most states
tend to have lower ratios of cash reserves to required
cash reserves than the small banks. Therefore, if any
nonmember banks increase their cash reserves on Report
of Condition dates to appear to be meeting reserve re­
quirements, the relatively large nonmember banks would
be most likely to do so.
One Wednesday in 1976 occurred on June 30, which
is a Report of Condition date. For each of the 21 nonmember weekly reporting banks in states with cash re­
serve requirements, cash reserves reported as of June 30
are compared to the average of their cash reserves as of
the four previous Wednesdays and the following four
Wednesdays.
Eight of the 21 banks had higher cash reserves on
June 30 than the average of both the previous and fol-

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

SEPTEM BER

1978

Table A -lll

IDENTIFICATION OF INDEPENDENT VARIABLES A N D SPECIFICATION O F HYPOTHESES
Direction of Influence on Dependent Variables:

Sym bol

Percentage of Banks Reporting CIPC
(a ) equal to zero, ( b) less than
10 percent of due from balances,
or (c) less than 25 percent of
due from balances

Description of Variable
of

member

banks

MPC 0

Percentage
to zero.

equal

+

M P C 10

Percentage of member banks reporting CIPC less than
10 percent of their dem and balances due from cor­
respondent banks.

+

M P C 25

Percentage of member banks reporting CIPC less than
25 percent of their dem and balances due from cor­
respondent banks.

+

EFF

Dummy variable with value of unity if a state has
cash reserve requirements an d C IPC are not counted
as reserves.

+

RR

M easure of state cash reserve requirements. For each
nonmember bank in a group, cash reserves required
by the state are subtracted from reserves that would
be required as a Federal Reserve member, and the
difference is divided by total deposits. These ratios are
averaged for nonmember banks in each group.

W EEKLY

Value of unity if the
weekly, zero otherwise.

settlement

period

is

+

+

BIW EEKLY

Value of unity if the reserve settlement
biweekly or semimonthly, zero otherwise.

period

is

?

?

REPORT

Value of unity if nonmember banks must file frequent
reports with the state ba n kin g authorities on deposit
liabilities and reserve positions, zero otherwise.

+

+

REC E X A M

Value of unity if nonmember banks do not have to
report on reserve positions regularly, but must keep
records on reserve positions to be inspected b y state
examiners during regular exam inations, zero otherwise.

?

?

REP DEF

Value of unity if nonmember banks must report re­
serve deficiencies to state ba nking authorities within
a short period of time after deficiencies occur, zero
otherwise.

+

+

LO PEN

Value of unity if there are small dollar penalties on
reserve deficiencies or if dollar penalties are infre­
quently enforced, zero otherwise.1

?

?

HI PEN

Value of unity if there are relatively large dollar
penalties for reserve deficiencies which are enforced
with relative frequency, zero otherwise.2

+

+

NE

Dummy variable with value of unity if a state has
no cash reserve requirements, zero otherwise.

—

Ratio 150

Dummy variable with value of unity if the average
ratio of cash reserves to required cash reserves for
a group of nonmember banks is less than 1.5, zero
otherwise.

+

reserve

reporting

CIPC

Percentage of Banks
in the Federal
Reserve System

lSt&tes in this category are Georgia, Iowa, Louisiana, Mississippi, New Mexico, North Dakota, South Dakota, and Wisconsin.
2States in this category are Alabama, Arkansas, California, Minnesota, Nebraska, New York, Oklahoma, Oregon, Texas, Washington, and
W est V irginia.

lowing four weeks. However, such results may reflect
largely the degree to which cash holdings of banks fluctu­
ate on a daily basis. To illustrate such an effect, seven
banks had cash reserves on June 30 smaller than their
average in the previous and the following four weeks.
Also, some of the eight banks that had higher cash
holdings on June 30 would have litde incentive for holding



unusually high cash reserves on the Report of Condition
date. Three of them are located in states that require
nonmember banks to file reports on daily reserve positions
shortly after each reserve settlement period. Of the re­
maining five banks with especially high cash reserves on
June 30, two had exceptionally high average ratios of
reserves to required reserves over the year 1976 (average
ratios of 1.85 and 2.33), indicating that they generally
Page 23

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

hold excess cash reserves. Thus, data for nonmember
weekly reporting banks provide little evidence that they
hold unusually high reserves on Report of Condition dates.

EFFECTS OF STATE RESERVE
REQUIREMENTS ON REPORTING OF
UNCOLLECTED FUNDS RY
NONMEMRER RANKS
Specification of Variables
Banks in each state are divided into the size groups
used in Table A-I. The following measures are used sepa­
rately as dependent variables.
(a ) percentage of nonmember banks which report
CIPC that is equal to zero on their June 30, 1976
Report of Condition,
(b ) percentage of nonmembers which report CIPC
that is less than ten percent of their demand bal­
ances due from correspondents, and
( c ) percentage of nonmember banks which report
CIPC that is less than 2 5 percent of their demand
balances due from correspondents.
Independent variables are described below. Their hy­
pothesized influences are summarized in Table A-III.
Influence of Bank Size — Means of classifying uncol­
lected funds appear to be related to bank size, the per­
centages specified above tending to be higher for smaller
banks. Influences of bank size are estimated by using
dummy variables (see Table A -III for specification of those
variables).
Classification of Uncollected Funds by M em ber Banks
— Ratios of CIPC to demand balances due from corre­
spondents for m em ber banks may be systematically related
to the same ratio for nonmember banks of similar size in
the same state. Independent variables reflecting the prac­
tices by member banks of reporting uncollected funds are
constructed in the same way as the dependent variables
specified above.
Geographic and transportation factors may influence
the speed with which checks are collected by both mem­
ber and nonmember banks in different states. Including
independent variables based upon the ratios for member
banks of CIPC to balances due from correspondents
would account for these common influences on uncollected
funds.
Another reason for including these measures for mem­
ber banks is the variation among correspondent banks in
methods of accounting for uncollected funds. Most of the
observations in this paper are for banks with total deposits
of less than $50 million. Many member banks in that size
range clear checks through correspondents instead of
through the FR S .4 F o r these member banks, the practice
of classifying uncollected funds as CIPC or balances due
4R. Alton Gilbert, “Utilization of Federal Reserve Bank Services
By Member Banks: Implications for the Costs and Benefits of
Membership,” this Review (August 1977), pp'. 2-15.

Page 24


SEPTEM BER

1978

from correspondents will be influenced by the accounting
practices of the correspondent banks through which they
and nonmember banks clear checks. Member banks have
no incentive to classify uncollected funds as due from
balances, since both CIPC and demand balances due from
correspondents are subtracted from gross demand deposits
to determine demand deposits subject to member bank
reserve requirements.
Use of these measures for member banks as independent
variables could bias the results. Correspondents might
adjust their methods of accounting for uncollected funds
to accommodate the desire of nonmembers to use un­
collected funds to m eet state reserve requirements. Meth­
ods of accounting for uncollected funds by member banks
would reflect, to some extent, the accommodation of cor­
respondents to nonmember bank wishes. In this case, in­
clusion of variables for classification of uncollected funds
by member banks in the regression analysis would bias
downward the estimated influence of state reserve require­
ments on the classification of uncollected funds by non­
members. To allow for such bias, variables for member
banks are removed in some regression equations.
Classification of CIPC in State R eserve Requirem ents
— 'A dummy variable is specified to reflect the incentives
of nonmember banks to classify uncollected funds as de­
mand balances due from correspondents: E F F has a value
of unity for states that have cash reserve requirements and
do not count CIPC as reserves, and has a value of zero
otherwise.5
M easurem ent of State Reserve R equirem ents — Levels
of state reserve requirements are difficult to compare.
Some apply to demand deposits only; others apply to all
deposits grouped together. Most states have different re­
serve requirements for demand and time deposits. Reserve
requirements are flat percentages in some states and grad­
uated in others. Thus, comparison of reserve requirements
among states depends upon the size of banks for which
comparisons are made and the composition of their de­
posit liabilities.
If a state allows nonmember banks to m eet all of their
reserve requirements with interest-earning assets, that
state is considered to have no cash reserve requirements.
Levels of reserve requirements are not calculated for those
states. F o r each nonmember bank in other states, the
relative level of state cash reserve requirements is meas­
ured by calculating cash reserves that would be required
as a Federal Reserve member, subtracting cash reserves
required as a nonmember, and dividing the difference by
total deposits. This ratio, denoted as RR, is averaged for
banks in each size group in the various states.
Monitoring and Enforcing State R eserve Requirem ents
—•There is substantial variation among states in proce­
dures for monitoring the reserve positions of nonmember
banks and for enforcing state reserve requirements.
Dummy variables are used to reflect differences in reserve
settlement periods, in methods of monitoring reserve posi5Values for the levels of state cash reserve requirements and
the indicators of monitoring and enforcement discussed below
are set equal to zero for states that count CIPC as reserves
and for those states with no cash reserve requirements.

FEDERAL

Table A -IV

EFFECTS O F STATE RESERVE REQUIREMENTS O N THE REPORTING O F UNCOLLECTED FU N D S BY N O N M E M B E R B A N K S

Equation
Num ber

Size 10

Size 5 0

Size 100

MPC 0

M P C I 25

EFF

EFF’ RR

E F F 'R R *
REC E X A M

EFF*RR*
REP DEF

E F F 'R R *
W EEKLY

Constant

R2

Degrees
Standard
of
Error
Freedom*

1

3 8 .6 5
(5 .0 9 4 )

8.64
(1 .1 6 9 )

-0 .9 9
( -0 .1 1 3 )

2

8 .47
(1 .4 2 9 )

-3 .1 9
(-0 .6 2 8 )

-3 .1 0
(-0 .5 3 3 )

0 .87
(9 .8 3 4 )

3

8.80
(1 . 5 0 4 )

-3 .1 7
(-0 .6 3 3 )

-3 .2 2
(-0 .5 6 1 )

0.85
(9 .5 8 3 )

4.01
(1 .7 3 4 )

4

7.91
(1 .3 3 1 )

-3 .6 7
(-0 .7 2 7 )

-3 .5 6
(-0 .6 1 9 )

0.85
(9 .6 1 0 )

4.75
(1 .9 2 9 )

5

3 7 .7 7
(5 .0 9 1 )

8.09
(1 .1 2 0 )

-1 .3 1
( -0 .1 5 5 )

7.39
(2 .1 8 0 )

6

37.6 5
(4 .9 8 4 )

8.01
(1 .0 9 7 )

-1 .3 7 1
( -0 .1 6 0 )

7.52
(2 .0 6 7 )

7

34.3 5
(5 .2 7 2 )

8.2 7 6
(1 .2 7 9 )

-3 . 7 1 8
(-0 .5 0 1 )

7.02
(2 .3 4 3 )

0 .4 9 5

15.582

77

15.92
(3 .4 1 2 )

0 .7 7 5

10.404

76

14.41
(3 .0 7 4 )

0.781

10.270

75

14.86
(3 .1 4 7 )

0 .7 8 0

10.284

74

16.49
(2 .3 7 7 )

0 .5 1 9

15.216

76

16.5 7
(2 .3 5 8 )

0.5 1 3

15.315

75

16.66
(2 .6 7 1 )

0 .5 0 7

14.829

93

80.50
(2 1 .0 7 3 )

0 .2 4 6

8.542

77

75.62
(1 8 .1 7 6 )

0 .2 9 6

8.254

76

-2 .4 4
(-1 .3 3 4 )

76.52
(1 8 .2 4 6 )

0.3 0 3

8.212

75

81.42
(2 0 .9 6 6 )

0.251

8.516

76

81.53
(2 0 .7 3 8 )

0.241

8.569

75

80.69
(2 2 .6 5 9 )

0.361

7.865

74

8 1 .3 7
(2 2 .3 8 8 )

0 .2 2 5

8.903

94

81.95
(2 3 .6 0 4 )

0 .3 0 4

8.436

91

13.34
(3 .2 0 7 )

8.14
(2 .0 1 0 )

9

5.73
(1 .1 4 5 )

2.59
(0 .5 7 9 )

-4 . 5 8
(-0 .9 8 3 )

0.15
(2 .5 4 5 )

10

5.90
(1 .1 8 4 )

2.69
(0 .6 0 3 )

-4 .5 0
( -0 .9 7 1 )

0 .16
(2 .5 9 7 )

11

13.61
(3 .2 7 7 )

8.32
(2 .0 5 7 )

-2 .6 4
( -0 .5 5 5 )

-2 .3 0
( -1 .2 1 1 )

12

13.45
(3 .1 8 1 )

8.22
(2 .0 1 1 )

-2 .7 2
(-0 .5 6 7 )

-2 .1 2
( -1 .0 4 1 )

13

14.20
(3.671 )

8.53
(2 .2 7 6 )

-0 .0 9
(-0 .0 2 0 )

14

11.08
(2 .8 3 3 )

7.24
(1 .8 6 4 )

-4 .3 4
(-0 .9 7 4 )

15

1 1.29
(3 .0 3 3 )

7.45
(2 .0 1 6 )

-2 .3 8
( -0 .5 5 9 )

of dem and balances at correspondents)

-2 .7 4
( -0 .5 7 5 )

-0 .2 9
( -0 .2 5 4 )
7 .50
(3 .7 2 0 )

6.01
(2 .994)

-8 .0 2
(-3 .5 5 9 Y

-7 .9 7
(-3 .4 9 5 )

-2 .1 2
( -1 .4 0 9 )

-1 .9 1
(-1 .1 8 4 )




1978

Page 25

““ ■ ™ 11™ 3 -are fo r s“ groups in states with 10 or more nonmembers and 10 or more members, except equations 7, 14, and 15, irhich are for groups with 10 or more noiimembers
’A
with no minimum number of members.

SE P TE M B ER

8

-0 .2 1
( -0 .1 0 4 )

S T . LOUIS

(Dependent variable: percent of nonmember banks reporting CIPC as less than 2 5 %

-1 .2 1
( -0 .8 9 0 )

OF

19.45
(2 .7 9 1 )

BANK

(Dependent variable: percent of nonmember banks reporting C IPC as zero)

RESERVE

IN D E PE N D E N T VA R IA BLE S
(•-statistics in parentheses under regression coefficients)

F E D E R A L R E S E R V E B A N K O F ST. L O U IS

tions of nonmember banks, and in penalties for reserve
deficiencies.
Interaction Terms — Variation in the level of state re­
serve requirements may have a stronger effect on the
classification of uncollected funds by nonmember banks
in those states with more strict monitoring and enforce­
ment of state reserve requirements. Interaction terms for
the level of reserve requirements and dummy variables
for enforcement are included as independent variables to
test this hypothesis.

Empirical Results
In the first seven equations in Table A-IV, the depend­
ent variable is the percentage of nonmember banks re­
porting CIPC equal to zero. The percentage of member
banks reporting CIPC as zero is positively related to that
percentage for nonmember banks [equations (2) - (4)].
Thus, the accounting practices of member and nonmem­
ber banks appear to reflect the common influences dis­
cussed above.
The one aspect of state reserve requirements which
influences the percentage of nonmember banks that report
CIPC as zero is the variable for states that have cash
reserve requirements and do not count CIPC as reserves
(E F F ), having a positive influence as hypothesized
[equations (5) - (7)]. Among states with cash reserve
requirements which do not count CIPC as reserves, the
level of reserve requirements (RR) does not add signifi­
cantly to the explanation of the dependent variable
[equation (6)].
However, when the percentage of member banks re­
porting CIPC as zero is included as an independent var­
iable, the variable that reflects the status of CIPC in state
reserve requirements (EFF) is not significant [equations
(3) and (4)]. This result is consistent with the view that
both member and nonmember banks base their methods
of accounting for uncollected funds upon the accounting
methods of correspondent banks, and that correspondent
banks adjust their accounting methods to serve the
interests of nonmember banks in meeting reserve
requirements.
Equations (8) - (15) of Table A-IV present regression
results with another dependent variable — the percentage
of nonmemebr banks reporting CIPC which is less than
25 percent of their demand balances due from correspond­
ents.6 Several measures of state reserve requirements are
significant, if the variable reflecting the reporting of un­
collected funds by member banks is eliminated from re­
gressions. The combination of measures of reserve require­
ments which yields the lowest standard error [reported in
equation (13)] includes levels of state reserve require­
ments (RR), dummy variables reflecting differences
among states in treatment of CIPC as reserves (E F F ),
and methods of monitoring reserve positions of nonmem­
ber banks (REC EXAM, REP DEF, and WEEKLY).7
^Effects of state reserve requirements were insignificant with
the percentage of banks reporting CIPC which is less than ten
percent of due from balances as the dependent variable.
7Value of the F-statistic for testing the combined influence of
these three variables, compared to the explanation due to bank
size variables alone, is 5.61. With 3 degrees of freedom in the
Page 26



SEPTEM BER

1978

EFFECTS OF STATE RESERVE
REQUIREMENTS ON CHOICE OF
MEMBERSHIP STATUS BY BANKS
Specification of Variables
Percentage of banks that are members (as of June
1976) is the dependent variable. Membership status is
strongly related to bank size; most very small banks are
nonmembers and most large banks are members (see
Table A-V). Effects of bank size are held constant by
calculating the percentage of banks that are members in
individual size groups in the various states, using the
same size categories as in the previous sections.8 Inde­
pendent variables and hypotheses concerning the direction
of influence of these variables on the percentage of banks
that are Federal Reserve members are presented in
Table A-III.

Empirical Results
Regression results are presented in Table A-VI. One test
involves two measures of reserve requirement levels as
Table A -V

PERCENT O F INSU RED C O M M E R C IA L B A N K S
IN EACH SIZE G R O U P THAT W ERE M EM BERS
O F THE FEDERAL RESERVE SYSTEM
A S O F DECEMBER 31, 1976
Asset Size
(in millions)
$

Percent

5 or less

1 8 .7 %

5 -

25.5

9.9

10-

24.9

38.0

25 -

49.9

48.9

50-

99.9

58.5

10 0 - 299.9

66.3

3 0 0 - 49 9 .9

78.6

5 0 0 or more

86.7

Source: Federal Deposit Insurance Corporation

numerator and 74 in the denominator, the F-statistic is sig­
nificant at the 0.5 percent level. Equations (8 ) and (1 3 ) were
reestimated with tne expanded sample used in equation ( 7 ) .
The combination of variables reflecting state reserve require­
ments in equation (1 5 ) is significant at the 5 percent level
(F-statistic of 4.564 with degrees of freedom of 3 and 9 1 ).
8The influence of state reserve requirements on choice of mem­
bership status by banks has been tested in other studies. See
Chris Joseph Prestopino, “Do Higher Reserve Requirements
Discourage Federal Reserve Membership?,” Journal of Finance
(December 1976), pp. 1471-80; John T. Rose, “Do Higher
Reserve Requirements Discourage Federal Reserve Member­
ship?: Comment,” Board of Governors of the Federal Reserve
System, mineo, June 1977. However, a major problem with
those studies is that percentages of banks in the Federal Re­
serve System are calculated for entire states. They include
measures to reflect the size distributions of banks in individual
states. The approach in this study probably deals with that
effect more directly. Another advantage of the approach in
this paper is that it increases the number of degrees of free­
dom for statistical tests.

F E D E R A L R E S E R V E B A N K O F ST. LO U IS

a

0O
1

a

£

O'
O

*
—
*
“

M EM B ERS

■*

K

N.
O'

K
O'

CO
CO
K

N00
"O
*”

K

K

CO
CN
CO
*“

•o

V)

■Sfi
O UJ

o
*
“

N.

CN
CN

oo
O'

CO

m
m

r~

*“

SEPTEM BER

independent variables: the measure of the
level of cash reserve requirements (RR)
and a dummy variable for states in which
there are no cash reserve requirements
(NE) [equation (2)]. The regression coeffi­
cients of those variables are not significant.

>
o

o
O'

O'

O'
00
CN

K
*0
o

K

K

CO

co
CO

CO
CO
CO

N
O'
CN

«o
•O
CO

o
o

00
S3
CO

o

o

o

o

o

o

o

o

•o
CN O
00 —
K CO
>o —

o o

«o «o

»o
— K
K CO

CO V)

«o CO
K —

o o

0*

CN

«o o

—

CO CO
O —

RESERVE

The regression coefficient for the level of
state reserve requirements might be insig­
nificant due to variation among states in the
amounts of cash assets that banks hold in
relation to their required reserves. In some
states nonmember banks hold cash assets
which, on average, are several times larger
than required reserves, while in other states
the ratios of reserves to required reserves
are close to unity. Influence of the level of
state reserve requirements on the percent­
age of banks that are Fed members might
be greatest in those states in which cash
holdings are relatively close to required
cash reserves.

00
CO

O

°I T
K CO
• —
o

K

•o
—

•o ^

o —

K
K
CN
O

o

K
r«-^
r-

r—

O CN

O j-

FEDERAL

O K

r-

O —

CN r -

ARE

OO

THAT

V ) CN
CO <N

BANKS

•
O

«0 cn
^C
M

<
>
zl 3
U M
Q

O'
a

.
«o 00
K K
> CN
—*

—

B IC

UJ

a

Z
—

g
o
Q.

T
3
2

a

H

8
s

: .2o>
c
: *; _
o«

o
co o

O ' CO

• r-I
O

CN O '

— o

o6
a

RESERVE
J)
■
O

CN
O

«

CN
CN 00
00 •O
o^
l
1

STATE
OF
EFFECTS

>
<

CN
»0
«o
I—

z S

REQUIREMENTS

ON

THE

PERCENT

OF

a
CO
O ' 00
O CO
co co
» w
—

o
•O

«o
O' o

o —#

O
N

CO ^
CN I

(/)

1 —

__

CN ^
<> »o
1 -L .

o

•O oo
co o
co ^
<N

^

O
'
-o
CN

p .

??
O'
O K
K CO
00 '
I 1
1 i

O'

00
CO
K CN
00 ~
1 w

■ 0
't 0
«0 0
0

—

CN CO

K

CN
O
K CO
00 r -'
1 1
1 1

o
>o «o

O' o

Cb ^
1 -L

CO
O CN
CN CN
00 r-I

»o
«o 00
00
K

1 -L

o

w

CO

o o

CO «o
CN

o

cn

CN CO

o

co

^

o

co

o

co

7

7

-

7

~

CO

|

1 —

-

__

7 ~

__

_

o
K
O
co
— CO
•O K
0 to o CN
0
r- lO
*
o«
o
o
«
o
oo *o
K « *o -o
0
« CO « ^ «
o « o CO Io
C I 7 J- ? ! o »o
O
I
7^
7I
I—
0<
u
• .£
j: )
S E

E. 3




22
60 C
O«
J fl
.2 §
«
S®
«E
V

This hypothesis is tested by adding an
independent variable calculated as the level
of reserve requirements (RR) multiplied by
a dummy variable with a value of unity if
cash reserves are less than 150 percent of
required cash reserves, and zero otherwise
(Ratio 150). This variable is insignificant.
Therefore, if state reserve requirements in­
fluence choice of membership status, the
effects will have to reflect aspects of those
requirements other than just the levels of
cash requirements.
The most significant aspect of state re­
serve requirements in influencing member­
ship choice is the procedure for reporting
reserve positions to state banking authori­
ties. The percentage of banks which are
members is significandy higher in states that
require nonmember banks to file periodic
reports on their reserve positions (REPORT)
[equation (5 )].9
Some states monitor the reserve positions
of banks by requiring them to report re­
serve deficiencies to the banking author­
ities shortly after incurring reserve defi­
ciencies. The regression coefficient of a
dummy variable which reflects this require­
ment (REP DEF) is not significantly dif­
ferent from zero [equation (7)]. With the

o o
*

«o o

K

«o

1978

9The dummy variable for states in which non­
member banks do not send reports to their
state banking authorities, but keep records of
reserve positions for inspection by examiners
( REC EXAM ), has a negative regression
coefficient which is approximately the same
in absolute value as the regression coefficient
for REPORT [equation (6 )]. This result indi­
cates that these two variables provide essen­
tially the same information. In all but 18 of
the 102 observations, either one or the other
has a value of unity. Of these two variables,
only REPORT is used in the other equations.
Page 27

variable for periodic reporting of reserve positions to
state banking authorities (REPORT) as an independent
variable, the requirement of reporting reserve deficiencies
(REP DEF) does not significantly influence the member­
ship choice of banks.

alties for reserve deficiencies are relatively low or seldom
imposed (LO PEN). The regression coefficient for this
variable is negative [equation (8) ], indicating that the per­
centage of members is relatively low in such states.10

Another variable which does significantly influence the
percentage of banks in the Federal Reserve is a dummy
variable for states which indicated that their dollar pen­

zero. Additional tests were conducted to determine whether
differences in levels of reserve requirements among states
have significant influences on the percentage of banks that
are members if significant features of state policies on moni­
toring and enforcement are held constant. Those tests involved
adding independent variables derived by multiplying the
measure for levels of reserve requirements ( RR) by each of
the significant dummy variables for enforcement (REPORT,
LO P EN ). In those equations (not reported in Table A-VI)
the regression coefficient for those additional independent
variables were insignificant.

10In regressions not reported in Table A-VI, the dummy variable
for states with more strict enforcement of dollar penalties for
reserve deficiencies ( HI P EN ) was substituted for LO PEN,
other variables the same as in equation ( 8 ) . The regression
coefficient of HI PEN was not significantly different from


Page 28