Full text of Review (Federal Reserve Bank of St. Louis) : October 1982

View original document

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

CO
13
O

o
cu
QQ

October 1982
Vol. 64, No. 8

CO
CD

01
3 Is It Tim e To Give Up the Fight Against
Inflation?

“O
CD

7 The Mix of Monetary and Fiscal Policies:
Conventional Wisdom Vs. Em pirical Reality
22 Simple Analytics of the Money Supply Process
and Monetary Control

The Review is published 10 times per year by the Research and Public Information Department o f
the Federal Reserve Bank o f St. Louis. Single-copy subscriptions are available to the public fr e e o f
charge. Mail requests f o r subscriptions, back issues, or address changes to: Research and Public
Information Department, Federal Reserve Bank o f St. Louis, P.O. Box 442, St. Louis, Missouri
63166.
Articles herein may be reprinted provided the source is credited. Please provide the Bank’s
Research and Public Information Department with a copy o f reprinted material.

Is It Time To Give Up the Fight Against
Inflation?
Address by LA W REN C E K. ROOS, President, Federal Reserve Bank of St. Louis, before the Petroleum
and Chemical Industry Conference of the Institute of Electrical and Electronic Engineers, St. Louis,
Missouri, August 30, 1982

I AM pleased to have this opportunity to appear
before you at a time when critical decisions concerning
the future course of economic policy are being made.
For almost 15 years, from the mid-60s to 1980,
Americans endured accelerating inflation. During that
period when prices rose from an annual rate of less than
two percent to double-digit dimensions, our economy
became afflicted by problems which increasingly
weakened the industrial and commercial sinews of this
nation. By the end of the 1970s, it became apparent
that we were losing the war against inflation. Our
losses consisted of more than just higher prices; we
suffered significant declines in productivity, sizable
reductions in investment and savings growth, rapidly
rising interest rates, volatile financial markets, increas
ing unemployment and a host of other social problems.
Toward the end of 1979, the Federal Beserve took
decisive steps to halt the rising growth in money that
had spawned the inflation. Its goal became disinflation
— a gradual reduction in the rate of inflation through a
gradual reduction in the growth of the money supply.
Today, after nearly three years of the new policy, we
seem to be on the threshold of success. For the first
time in several years, inflation has dropped substan
tially. So far this year, on an annual basis, prices have
risen only about 5 percent, a significant improvement
over the double-digit figures of a few years ago.
Unfortunately, however, at the very moment of
meaningful progress, we are being besieged by a varie

ty of economic ills that are severely testing our resolve
to continue the fight. Among these are: increased un
employment, high interest rates, sharply reduced
housing and construction activity, falling commodity
prices, low rates of saving and investment, and what
some observers see as a declining competitive advan
tage for our products in foreign markets.
It is becoming increasingly popular in some circles
to attribute our current economic ills to our disinflation
efforts. Almost daily we are inundated by articles with
titles like “How Disinflation Hurts Us All,” “The Costs
of Disinflation, ” or, even more alarming, “The Curse
of Disinflation.” Congressmen have been quoted as
equating our current anti-inflation efforts to the PLO’s
efforts in West Beirut, hinting that, if continued, our
disinflation policies will reduce the nation’s economy
to a pile of rubble. When the Chairman of the Federal
Beserve System recently appeared before Congress,
he was urged to “consider a major change in monetary
policy before it is too late. ” Bills have been introduced
in Congress to force the Federal Beserve to abandon its
efforts to reduce monetary growth and to revert to
interest rate targeting — the very same procedure that
produced our past inflation.
In the face of this sort of prodding, it is not surprising
that more and more people are asking whether the
time has indeed come for us to move away from disin
flation and reinflate our way back to more “prosperous”
times. It is this issue that I would like to discuss with
you.
3

FEDERAL RESERVE BANK OF ST. LOUIS

Now, I am not going to tell you that disinflation is
without its costs. It is not costless. To bring down and
hold down inflation after 15 years of soaring prices
requires a major overhauling of our economic system.
And, as in any major overhaul, there will be a certain
amount of “downtime” — a temporary reduction in
output and a rise in unemployment — as we retool for a
different economic environment. However, I want to
stress that this downtime, for most sectors of the econ
omy, is only temporary. It marks a necessary transition
to a stable economic environment essential for longer
term growth and prosperity.
Also, it is important to recognize that not all indus
tries will do well even if inflation is halted. Some will
suffer permanent losses — especially those that, for a
variety of reasons, benefit from inflation. Among these
are speculative land and commodity firms, companies
that produce and sell “collectibles” of all kinds, and,
perhaps, even those publishing companies that market
books on “how to beat inflation.”
Reform is never without its costs, but in considering
whether the time has really arrived to reinflate, I
would like to pose two questions: First, is disinflation
solely responsible for our current economic problems?
Second, can we solve our problems by returning to our
old “inflation as usual” policies? I will argue that the
answer to both of these questions is “No. ”
First, let’s consider why disinflation, per se, is not
the sole cause of our current economic problems. Cer
tainly disinflation is responsible for some temporary
problems. However, some of our most serious prob
lems would have occurred even if inflation had con
tinued; they are the direct result of a new awareness on
the part of the public of the pernicious nature of infla
tion and people’s actions to attempt to protect them
selves from some of its effects.
In the early 1970s, when serious inflation was a new
experience for most Americans, people tended to view
rising prices as a temporary phenomenon that, like an
old soldier, would somehow fade away. While no one
in retrospect would question that the decade of the
1970s was one of rising inflation, many people at that
time behaved as if inflation either didn’t exist or, at
least, wouldn’t persist. Being unconditioned to infla
tion, they totally failed to anticipate either the extent
or the duration of the problem. As a result many people
— perhaps you and I included— continued to save and
lend money at interest rates that failed to compensate
for our loss of future purchasing power that came
about as a result of inflation. This was true even for

OCTOBER 1982

many sophisticated financiers whose activities in
fluence interest rates.
As a result, the real interest rate that savers and
lenders received — the actual return after adjusting for
the impact of inflation — was not only ridiculously low
during the inflationary 70s, it was negative for much of
the decade. Now, whenever the real rate of return on
financial assets — such as bonds, savings accounts and
the like — is unrealistically low — or even worse, when
it is negative — people who hold financial assets — the
savers — end up poorer and people who sell financial
assets to them — the borrowers — end up wealthier.
During the 1970s, people who borrowed funds to
purchase tangible assets — houses, cars, land, gold,
etc. — did well; those who lent funds did badly. Of
course, nobody planned it that way. It was, however, a
predictable result of the public’s failure to recognize
the true nature and extent of inflation.
To illustrate how this works, let’s take a simple ex
ample. Consider two individuals, each of whom ex
pects the rate of inflation to average 5 percent per year
over the next ten years. Let’s further assume that the
competitive real rate of interest at that time (i.e., the
actual return above the rate of inflation) is 3 percent
per year. If one borrows $100,000 from the other to
purchase a house, the borrower would be willing to pay
(and the lender would expect to receive) an 8 percent
interest rate on the loan — 5 percent to compensate for
expected inflation and 3 percent to provide a desired
real return.
Suppose, however, that the actual rate of inflation
over the ten year period turned out to be 9 percent per
year, instead of the 5 percent that was originally ex
pected and factored into the loan. If this happened, the
lender would be the loser. He would be receiving 8
percent each year on an investment that was eroding in
value, due to inflation, at 9 percent per year. Instead of
a positive real rate of return of 3 percent, he would be
losing one percent in terms of his annual real rate of
return.
On the other hand, the borrower would be doing

unexpectedly well. He would be paying 8 percent
annual interest on a house that was appreciating in
value at 9 percent per year — and getting a real return
from living in it as well. In times of unexpected infla
tion, the lender’s loss is the borrower’s gain. And many
people failed to compensate for this factor during the
inflationary 1970s.
However, as this disparity between individual rates
of return on real versus financial assets became in

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

creasingly noticeable over tim e, people becam e
attuned to the meaning of inflation and took steps to
protect themselves against it. They did this by pur
chasing tangible assets such as houses, land, commod
ities of all kinds — and by refraining from purchasing
financial assets that were depreciating in value.

process of disinflation itself, that accounts for current
relatively high interest rates and most of the disloca
tions that have occurred in the economy. And it is that
adjustment process that has important implications for
assessing whether reinflation would provide any relief
from our current problems.

It was simply a situation of people awakening to what
was happening and adjusting their investments to com
pensate for some of the effects of inflation. As a con
sequence, land values increased dramatically. Hous
ing prices, on average, rose 2.5 percent per year faster
than the rate of inflation. Inventories became one of
the best ways to hold corporate wealth. Commodity
prices soared — the price of gold alone rose nearly 21
percent per year above the rate of inflation. Even if
OPEC had not existed, petroleum products would
have been a boom industry. On the other hand, finan
cial markets foundered. Investors in bonds and money
market instruments earned negative real rates of re
turn ranging from -0.2 to -0.5 percent per year.

The issue, therefore, is what will we gain if we
abandon our disinflation efforts? Can we reinflate our
way back to more prosperous times?

The consequences were inevitable: savers got poor
er and the supply of credit declined; borrowers got
wealthier and the demand for credit rose. As a result,
real interest rates began to rise. Moreover, the public
got smarter; after 15 years of living with inflation,
people finally came to realize the extent and persist
ence of the problem and adjusted their economic deci
sions so as to reduce, as much as possible, the disloca
tions that inflation produces. Both savers and lenders
learned to insist on a return that would preserve their
purchasing power as well as compensate them for the
greater risk of lending during a period of uncertain
inflation. And all of this, of course, put even further
upward pressure on real interest rates.
By the beginning of the 1980s, inflation awareness,
not inflation naivete, dictated the direction of financial
dealings. Real interest rates became positive and the
benefits people had previously enjoyed from borrow
ing to acquire real assets evaporated. As a result, cer
tain sectors of the economy such as housing, land spec
ulation, commodity purchases, etc., that had previous
ly been subsidized by invalid and erroneous infla
tionary anticipations began to experience a decline.
Similarly, gold prices fell, while bonds and money
market instruments finally began yielding positive real
rates of return.
What happened, in a nutshell, was that the public
learned to recognize and anticipate the effects of
changes in the rate of inflation and, as a result, is now
better prepared to adjust its economic decisions ac
cordingly. It is that adjustment process, and not the

To some extent, advocates of reinflation are simply
victims of misplaced nostalgia or selective amnesia.
Critics of today’s disinflation policies seem to have
forgotten that the 1970s were not good years for this
nation. During the inflationary 1970s real economic
growth fell about 25 percent below what it had been
during the 1960s; the unemployment rate, which had
averaged less than 4 percent in the last half of the 60s,
averaged 7 percent from 1975 to 1979. Long-term
interest rates, which had averaged less than six percent
during the late 60s, reached double-digit levels by the
end of the 1970s. It was precisely these problems that
led us to embark on the struggle against inflation. It is
precisely these long-run consequences of higher infla
tion that we can expect to return to if we do opt to
reinflate.
O f course, there are some who feel that a return to
reinflation is necessary to bring relief to certain indus
tries that have severely suffered during the disinflation
effort, such as the housing and construction industries.
That reinflation would accomplish this is wishful think
ing! As I’ve explained, the gains that inflation produced
for housing and similarly affected sectors during the
1970s arose from the fact that interest rates did not
correctly reflect the full impact of inflation.
Today, however, the public has developed an infla
tion awareness that would prevent this from happening
again. The public is now fully aware that it can protect
the value of its assets only by anticipating changes in
the future rate of inflation. Consequently, if the public
were to believe that reinflation was imminent, interest
rates would increase to protect investors’ real return.
Any anticipated reinflationary gains to the housing in
dustry, or to any other interest-sensitive sectors of the
economy, would be doomed by upward adjustments in
interest rates resulting from heightened inflationary
expectations.
This is why I find the current clamor for reinflation
so disturbing. It is not just that there is no evidence
that excessive monetary expansion would bring the
results its advocates seek; it is rather that pressure for
5

FEDERAL RESERVE BANK OF ST. LOUIS

reinflation raises doubts in people’s minds as to
whether inflation will continue to decline. In view of
the rising sentiment for reinflation, it is not surprising
that people remain skeptical as to the extent of our
current commitment to disinflation. Many of the sav
ers and lenders who were so badly burned in the 1970s
understandably are sensitive to what they read and
hear that implies a potential weakening of our resolve
to continue the struggle against inflation.
Just recently, the American Bankers Association re
leased a long-range inflation forecast predicting an
average inflation rate of more than 9 percent per year
over the next ten years. The latest Harris poll of busi
ness executives shows that many of those polled expect
inflation to begin rising within one year. If bankers and
business executives aren’t convinced that disinflation
will continue, we can hardly expect to convince lend
ers and savers that this will happen.

OCTOBER 1982

The solution to this problem, and the challenge
facing us for the future, is to make our anti-inflation
policies credible to a disbelieving public. There are
several things that would contribute mightily to this
effort. First, our current monetary policy stance must
be maintained; there must be no, even temporary,
reinflation “relapse. ” Second, federal deficits must be
reduced significantly. I say this not because I believe
that deficits by themselves cause inflation — they do
not. I say it because smaller federal deficits would
reduce the “temptation” facing monetary authorities to
monetize a portion of the deficit. Finally, the American
public must be prepared to resist pressures for policy
makers to revert to reinflation in order to alleviate the
temporary pain of the process of disinflation. Only
when people become convinced that our anti-inflation
fight is “for real” and will be pursued over a long period
of time, will inflation finally be eliminated and stabil
ity, so necessary for economic growth, be restored.

The Mix of Monetary and Fiscal
Policies: Conventional Wisdom Vs.
Empirical Reality
K EIT H M. CARLSON

T HE current economic situation of high interest
..Irates, high unemployment and large federal deficits
has prompted a call for a change in the mix of stabiliza
tion policies. The current mix seems to be one of “easy”
fiscal policy and “tight” monetary policy. Many
analysts feel the mix should be shifted toward “tighter”
fiscal policy and “easier” monetary policy, ostensibly
for purposes of putting the economy on the path to
recovery.
James Tobin, for example, recently stated that:
. . . the mix o f policies is unhealthy. To achieve a solid
reco v ery , such as th e adm inistration p rojects, and to
achieve it w ithout astronom ical in terest rates and se
rious crow ding out, w e n eed an easier m onetary policy
com b in ed w ith a tig h ter fiscal p o licy .1

Economists at the Brookings Institution have ex
pressed a similar view:
Beyond 198 2 , th e key to an im p roved econ om ic situ
ation m ust lie in a realign m en t of econ om ic policy—
a shift in th e mix o f fiscal and m o n etary policy, by
m atch in g red u ctio n s o f fu tu re b u d get deficits with
an easier m onetary policy. As p resen tly con stitu ted ,
fiscal and m o n etary policies ap pear to b e on a collision
cou rse . . .2

The Congressional Budget Office talks of the clash
between monetary and fiscal policy:
S tatem en ts from th e F e d e ra l R eserv e suggest that
m o n e ta ry p o licy will co n tin u e its an ti-in flatio n ary

^ames Tobin, “The Wrong Mix for Recovery,” Challenge (MayJune 1982), p. 25.
2Joseph A. Pechman and Barry P. Bosworth, “The Budget and the
Economy,” in Joseph A. Pechman, ed., Setting National Priorities:
The 1983 Budget (The Brookings Institution, 1982), p. 43.

stance in the com ing years . . . By con trast, th e budget
m easures en acted last su m m er will provide con sid er
able stim ulus to econ om ic activity ov er th e n ext few
years. This suggests th e possibility o f a clash b etw een
m onetary and fiscal policy unless the C ongress en acts
fu rth er sp en d in g cu ts and tax in creases to red u ce
federal borrow ing o r th e F e d e ra l R eserv e adopts a less
restrictive m o n etary policy. If th e clash m aterializes, it
will be reflected in high real in terest rates that crow d
out p rivate in v estm en t.3

The notion of policy mix is well-known and has been
a part of the macroeconomics literature for a number of
years, but seldom has it generated controversy as it has
now. Despite its recognition, however, little is known
about the exact terms of the mix or what indicators of
monetary and fiscal policy are most appropriate to use
in defining it. For example, neither Tobin, the Brook
ings economists, nor the Congressional Budget Office
state by how much fiscal policy should be tightened
and to what extent monetary policy should be eased.
The gravity of the current economic situation re
quires that the notion of “policy mix” be given a more
precise interpretation. Is current policy what it seems?
How does one measure the ease and tightness of
monetary and fiscal policies? What measures of eco
nomic performance are relevant to the mix argu
ment— interest rates, GNP, the ratio of investment to
GNP? What horizon is pertinent— short-run, longrun, a specific number of years? Can policies really be
traded off to achieve a specific economic objective?
These are the types of questions that are given short
shrift when the mix of policies is discussed.
Congressional Budget Office, The Prospects fo r Economic Recov
ery (U.S. Government Printing Office, 1982), p. 27.

FEDERAL RESERVE BANK OF ST. LOUIS

The purpose of this article is to provide some spec
ificity to the policy mix question. As a point of de
parture, some indicators of monetary and fiscal policy
are examined, and a historical classification of policy
mix is developed. A conventional macroeconomic
treatment of policy mix is then presented, providing
the basis for development of testable hypotheses.
These hypotheses are tested, and some policy implica
tions are derived.

D EFIN IN G EASE AND TIGHTNESS
OF POLICY
Many summary measures of monetary and fiscal
policy have been developed over the years. The main
reason for seeking such a measure is to provide a quick
interpretation of current policy stance. The criteria for
selection are that the measure primarily reflect move
ments in the instruments of policy and that it not be
influenced greatly by the pace of economic activity.4 In
other words, the indicator should reflect the thrust of
the policy on the economy rather than the reverse.

Measurement of Monetary Actions
Measuring the stance of monetary policy revives
many controversial issues, one of the oldest of which
centers on the choice between interest rates and
monetary aggregates. Although the level of interest
rates is often alluded to as an indicator of monetary
policy, most analysts have found it to be unreliable as a
policy measure, since a great many forces influence
interest rates besides monetary actions. The monetary
aggregates, on the other hand, tend to reflect more
accurately changes in the policy instruments— open
market operations, reserve requirements and the dis
count rate— without being influenced unduly by out
side forces.5
Chief among the candidates for a monetary policy
indicator are the money stock (M l) and the monetary
base. The monetary base has appeal because it reflects
more accurately changes in the instruments of policy
than does M l. The M l measure, however, tends to be
more closely related to GNP.6
4See, for example, Albert E. Burger, “The Implementation Prob
lem of Monetary Policy,” this Review (March 1971), pp. 20-30.
5See R. W. Hafer, “Selecting a Monetary Indicator: A Test of the
New Monetary Aggregates,” this Review (February 1981), pp.
12-18.
6For references to the literature, along with a contrasting inter
pretation, see William E. Cullison, “Money, the Monetary Base,
and Nominal GNP,” Federal Reserve Bank of Richmond Economic
Review (May/Iune 1982), pp. 3-13.

OCTOBER 1982

Labeling monetary policy as easy or tight is, of
course, quite arbitrary. The procedure followed here is
to examine the historical record of M l and develop a
classification of relatively easy and relatively tight
policy on the basis of this record. To make this classifi
cation meaningful, one must account for changes in the
trend of monetary growth, particularly if one intends to
focus on the impact of monetary policy on real vari
ables. Consumers and investors come to expect certain
growth rates of the monetary aggregates, basing such
expectations on past experience. It is the deviation of
the monetary aggregate around this expected growth
rate that affiects real economic activity.
Table 1 summarizes monetary policy since 1956.
The first column shows the four-quarter rate of change
of M l minus its trend (20-quarter rate of change) for
the period ending in the fourth quarter of each year
(except for 1982).7 The second column, which classifies
monetary policy as easy, tight or neutral, follows from a
three-part division of the observations in the first col
umn. The mean plus or minus Vi standard deviation
serve as points of demarcation. The classification is
relative and also approximate. Rigidly adhering to the
four-quarter rate of change can mask changes in policy
that occur within the year. A more exhaustive study
would not be tied to periods of fixed length. Nonethe
less, the classification seems to accord with common
interpretation of economic experience; for example, all
of the observations labeled as “tight” occurred near
recession periods.

Measurement o f Fiscal Actions
The measurement of fiscal actions also has been
researched extensively over the years.8 The chief con
clusion from this research is that recorded surpluses or
deficits do not provide an accurate measure of fiscal
actions. The reason is that a considerable amount of the
movement of receipts and expenditures reflects an
automatic response to the pace of economic activity
rather than policy actions. Consequently, only fiscal
m easures on a high-em ploym ent basis are con
sidered.9 These high-employment budget measures
justification for a 20-quarter rate of change as a measure of trend is
found in Denis S. Karnosky, “The Link Between Money and
Prices,” this Review (June 1976), pp. 17-23. See also Keith M.
Carlson, “The Lag from Money to Prices,” this Review (October
1980), pp. 3-10.
8Alan S. Blinder and Robert M. Solow, “Analytical Foundations of
Fiscal Policy,” in Alan S. Blinder and Robert M. Solow, eds., The
Economics o f Public Finance (The Brookings Institution, 1974), pp.
3-115.
9Frank de L eeuw and Thomas M. Holloway, “ The HighEmployment Budget: Revised Estimates and Automatic Inflation
Effects,” Survey o f C urrent Business (April 1982), pp. 21-33.

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

Table 1
Classification of Ease and Tightness of Monetary
and Fiscal Policy
Monetary
Measure1

Fiscal Measure1
A(S/D) H
-4

M1 _4 - M1

-2 0

EH< - FH2 0
c Value

yh
_4

R e EH4

Year2

Value

Class

1956

-1 .1 9

5.84

2.15

0.52

1957

-1 .6 7

6.83

-4 .6 3

-0 .6 9

Class

value

Class

Value

Class

1958

1.49

10.12

-1 0 .0 5

-1 .7 3

1959

0.33

-6 .7 1

9.46

1.72

1960

-0 .7 9

-2 .8 9

5.47

1.07

1961

1.19

2.68

-6 .3 4

-1 .0 2

1962

-0 .3 0

1.75

-2 .0 2

-0 .3 2

1963

1.75

-1 .2 2

1.51

0.33

1964

1.71

-3 .8 2

-1 .9 6

-0 .3 8

1965

0.91

4.35

-8 .3 4

-1 .4 8

1966

-0 .6 9

8.46

-2.01

-0 .4 8

1967

1.98

2.49

-3 .4 6

-0 .7 9

1968

2.31

-0 .0 8

7.89

1.31

1969

-1 .0 8

-6 .3 3

5.69

1.14

1970

-0 .1 6

-1 .9 8

-6 .5 9

-1 .3 2

1971

0.87

-1 .2 8

-1 .2 9

-0 .2 9

1972

2.26

6.83

-5 .3 4

-1 .1 5

1973

-0 .1 7

-2 .7 3

7.60

1.35

T
T

1974

-1 .3 5

6.32

3.43

0.58

1975

-1 .1 6

2.93

-8 .1 9

-1 .6 7

1976

0.16

-4 .0 2

2.09

0.30

1977

2.21

0.22

-1 .6 2

-0 .4 5

1978

1.77

-1 .5 8

4.82

0.81

1979

0.41

0.88

0.21

-0 .0 2

1980

-0 .1 4

5.56

0.62

0.05

1981

-2 .2 0

0.31

-1 .3 4

-0 .3 3

1982

-2 .1 1

-3 .1 4

-4 .8 2

-1 .0 0

Mean
Mean - Vkr
Mean + '/2a

0.23
-0 .4 5
0.92

1.27
-1 .0 6
3.54

-0 .4 7
-3 .1 4
2.20

-0 .1 1
-0 .6 0
0.38

E: Easy
N: Neutral
T: Tight
1For definitions of variables, see text. Ease is associated with relatively large positive values except for
the right-hand pair of fiscal measures. For these measures, ease is associated with relatively large
negative values.
2AII values are for year ending fourth quarter, except for 1982 which is for year ending second quarter.

can be assembled in different ways, however. Three
high-employment measures are examined: the rate of
change of expenditures minus its trend, the rate of
change of receipts minus the rate of change of expendi

tures, and the change of the surplus or deficit scaled by
the size of the economy as measured by potential GNP.
The latter two measures were tested for the presence
of trend, but none was apparent.
9

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Table 2
The Mix of Policy, 1956-82
Monetary

Fiscal
Tight

Tight

1956, 1960, 1969, 1974

Neutral
1966, 1981

Easy
1957,1975,
1982

Neutral

1973

1959,1962,1971,

1965,1970

1976, 1979, 1980
Easy

1968, 1978

1963, 1964, 1977

1958,1961,
1967,1972

NOTE: Monetary policy is measured by four-quarter rate of change of M1 minus 20-quarter rate of
change. Fiscal policy is measured by change in high-employment surplus/deficit over four
quarters divided by high-employment GNP at the beginning of the period.

The right-hand portion of table 1 summarizes these
three fiscal measures over the 1956-to-1982 period.
The rate of change of high-employment expenditures
is included because previous studies have used it as a
summary fiscal measure even though it does not re
flect changes in tax policy.10 The other two highemployment measures— the rate of change of receipts
minus the rate of change of expenditures and the
change in the surplus/deficit scaled by potential
GNP— reflect changes in both tax policy and expendi
t u r e policy. In general, these latter two measures yield
the same classification of easy, tight and neutral.

Historical Record of Policy Mix
The measures of monetary and fiscal policy can be
combined to give a classification of each year in terms
of the mix of those policies. Table 2 provides this
summary on the basis of detrended M l and the highemployment surplus/deficit measure of fiscal action.
The years shown in the corners of this matrix are most
revealing. Periods when both policies were tight clear
ly were associated with recessions; those when both
were easy were associated with economic expansion.
The periods of contrasting policies, though few in num
ber, are interesting nonetheless. Easy monetary policy
and tight fiscal policy occurred only in 1968 and 1978,
both expansion years before business cycle peaks. The
10See Leonall C. Andersen and Jerry L. Jordan, “Monetary and
Fiscal Actions: A Test of Their Relative Importance in Economic
Stabilization,” this Review (November 1968), pp. 11-24.

other extreme of tight monetary policy and easy fiscal
policy occurred in 1957, 1975 and 1982; these were
recession years, although all of 1975 is not classified
so according to the National Bureau of Economic
Research.
The current situation (note that 1982 refers to the
four quarters ending in second quarter 1982) clearly
falls into the classification of tight monetary policy and
easy fiscal policy. According to the detrended measure
of monetary policy as shown in table 1, 1981 and 1982
are the tightest years for monetary policy for the 195682 period. The measures of fiscal policy, however, do
not indicate unusual ease. The expenditure measure
does not indicate ease at all, and the other two,
although suggesting ease, do not indicate that the de
gree of ease is unusual. There have been six or seven
of the last 27 years (depending on which measure is
used) when fiscal policy has been easier than in 1982.
The current concern about the mix of policies,
however, is not focused entirely on the recent past
(note, in particular, the quotations by Brookings and
the Congressional Budget Office above). There is con
cern about the near future. In other words, given
current trends, analysts seem to be most concerned
about developing trends in the mix. Consequently, a
full assessment of policy mix requires an extrapolation
of trends to determine if the mix of policies appears to
be worsening, that is, that monetary policy is tighten
ing further or at least remaining tight, and that expen
diture and tax policies are leading to a further easing
of fiscal policy.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Figure 1

D e te rm in a tio n

of R eal

In c o m e ,

In te re s t R a te an d

In v e s tm e n t-S a v in g

Interest
rate

Interest
ra te

(i)

R a t io

(')

rve

Real
in c o m e

IS curve depicts equilibrium in goods and services
market.
LM curve depicts equilibrium in money market.

DEVELOPMENT OF THE MIX
HYPOTHESIS: A FRAM EW ORK
OF ANALYSIS
The question of policy mix is broad, encompassing a
large body of macroeconomic theory and empirical
support for the theory. Since a broad review of the
theory is thus prohibitive, a typical example from mac
roeconomic textbooks is summarized to represent the
policy-mix literature.11

The Basic IS-LM Model
The notion of policy mix usually is explained by
using the well-known Hicksian IS-LM framework.
According to this framework, the level of economic
activity (real income) and the level of interest rates are
determined by the conjunction of conditions in two
n See, for example, Robert J. Gordon, Macroeconomics, 2nd ed.
(Little, Brown and Company, 1981), pp. 140-42; or William H.
Branson and James M. Litvack, Macroeconomics, 2nd ed. (Har
per and Row Publishers, 1981), pp. 86-90.

(l/X , S / X )

In v e s tm e n t,
S a v i n g ratio
(l /X , S / X )

Investment and saving curves are drawn for given
level of real income (X).

aggregate markets: the market for goods and services
and the market for money.12 Fiscal policy, that is,
changes in federal expenditures and tax rates, in
fluence the economy through the market for goods and
services, while monetary policy works through the
money market.
The IS-LM model is summarized in figure 1. The IS
curve is the locus of combinations of interest rate and
real economic activity consistent with equilibrium in
the goods and services market. The curve is downward-sloping because lower interest rates induce high
er levels of investment, which increase real income
through the multiplier. Including the federal govern
ment in the analysis broadens the equilibrium condi
tion to investment plus government purchases equals
savings plus taxes. The right-hand panel shows explic
itly the situation in the goods and services market that
underlies the IS curve in the left-hand panel. Although
in reality both savings and investment depend on real
12In the simplest version of the IS-LM model, there is no distinction
made between nominal and real interest rates because the price
level and price expectations are held constant.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Fig u re 2

Policy Mix and the IS-LM Framework
In t e r e s t
rate

Real
incom e

(X)
© Tight monetary policy ( M ^ M ) and easy fiscal
policy (G, G).
© Easy monetary policy (M2>M ) and tight fiscal
policy (G2<G).

income, the curves are drawn here only with reference
to the equilibrium level of income.
The LM curve, as shown in figure 1, is the locus of
combinations of interest rate and real economic activ
ity consistent with equilibrium in the money market.
Money is defined as currency plus checkable deposits.
The curve is upward-sloping because the demand for
real balances is assumed to be negatively related to
interest rates and positively related to real income.
Consequently, in order for the demand for real bal
ances to be equal to a fixed supply, an increased de
mand for money balances as real income increases
must be offset by reduced demand for money balances
via higher interest rates.
Real income and interest rates are determined
simultaneously by the intersection of the IS and LM
curves. Only this combination of interest rate and real
income is consistent with equilibrium in both the
goods and services and money markets. The way that
this equilibrium combination changes in response to
monetary and fiscal actions is of interest here.
Fiscal actions affect equilibrium by shifting the IS
curve, while monetary actions shift the LM curve.
Consequently, a given level of real income can be
12

achieved with different combinations, or mixes, of
monetary and fiscal actions. For example, in figure 2,
the combination of IS j and LM X represents “easy”
fiscal policy and “tight” monetary policy, and a given
level of real income is achieved with a higher interest
rate than at the original equilibrium. Similarly, the
intersection of IS2 and LM2 reflects “tight” fiscal policy
and “easy monetary policy. “High” interest rates
imply a lower rate of private investment (see righthand panel) and thus signify slower economic growth
over the long term than a set of policies that yields
“low” interest rates.
Within the context of the current economic situa
tion, the implication seems to be that the U.S. econ
omy is operating at a point corresponding to the in
tersection of IS j and LM i. This interpretation, how
ever, is not obvious. Rather, Tobin and Brookings
economists seem to draw the conclusion that the eco
nomic recovery cannot be started or sustained unless
interest rates are reduced by changing the mix of poli
cies. This interpretation suggests that they view the
IS-LM framework in dynamic rather than in static
terms. In other words, the level of real income and
interest rates are being moved over time by a combina
tion of policies in such a way that interest rates are
rising— or at least being sustained at high levels.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

F ig u re 3

In fla tio n a r y E x p e c ta tio n s and the IS-LM Fram ew ork

(X
)
Inflationary expectations (P E> 0 ) push up nominal
interest rate (i), but by less than PE.

Expected Inflation and the IS-LM Model
The version of the IS-LM model discussed above is
based on a number of simplifying assumptions. One of
the most important simplifications is that inflation and
expectations of inflation can be ignored. It would seem
that any discussion of current economic developments
should be based on a framework that allows for the
effect of expected inflation.
The effect of introducing expectations into the ISLM model is shown in figure 3. The interest rate axis
now represents both nominal and real interest rates.
ISo represents equilibrium in the goods and services
market with expected inflation equal to zero (implying
nominal and real interest rates are the same). The
effect of introducing expected inflation of some posi
tive amount is to shift the IS curve upward and to the
right with respect to nominal rates, but leave it un
changed in terms of real rates. The reason for this is
that consumption and investm ent decisions are
assumed to be based on real rather than nominal rates
of interest. This means that IS Xis raised above IS0 by
the amount of expected inflation.
The effect on the LM curve of introducing expected
inflation is somewhat more complicated. If, for exam

ple, the demand for real balances depends on the
difference between rates of return on money and all
other assets, then expected inflation will affect all of
these rates of return equally, including the return on
real balances, and the LM curve will not be affected
when drawn in terms of the nominal interest rate. If
the assumption of equal effects of expected inflation on
all asset returns is relaxed, however, the results will
differ. An increase in expected inflation generally
can be expected to increase the nominal return on
capital relative to bonds and money. Wealth-holders
will attempt to rearrange their portfolios to hold more
capital and less money and bonds. Prices of capital will
rise as those on bonds fall; that is, interest rates will
rise. This means the LM curve will shift upward when
drawn with reference to the nominal rate of interest.
Because all wealth is not affected by a change in ex
pectations, however, the upward shift of the LM curve
will be less than the change in expected inflation.
The effect on policy mix of introducing expectations
into the IS-LM model depends on the response of
expectations to any change in policy. If expectations do
not change in response to a shift in policy, the mix can
be changed as in the basic model. A given level of
income can be achieved with different combinations of
policies and different interest rates. On the other
hand, if expectations are responsive to either monetary
or fiscal actions, a larger number of possibilities is
introduced, depending on the nature of the expecta
tions response. In general, the faster expectations
react to actual changes in monetary or fiscal policy, the
less the effect of policy changes on the real variables,
that is, real income, real interest rates and the invest
ment ratio.

TESTING THE M IX HYPOTHESIS
The IS-LM framework provides a rationale for
assigning the mix of policy an important role in plan
ning the course of economic activity. Though there are
many problems involved in moving from the classroom
blackboard to economic reality, the general framework
can still serve as a guide in formulating a test of the mix
hypothesis.

Testing Procedure
The question of policy mix focuses on the aggrega
tive effects of monetary and fiscal actions and thus
lends itself to a reduced-form approach to hypothesis
testing.13 The details of the transmission mechanism
13See Andersen and Jordan, “Monetary and Fiscal Actions.”

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Table 3
Variables Used in Regressions
Dependent variable

Monetary

Fiscal

Aaa: corporate Aaa
bond rate

M1: compounded annual
rate of change of
money stock narrowly
defined

EH: compounded annual
rate of change of highemployment federal
expenditures

Aaa - P 16: Aaa rate
minus 4-year rate
of change of GNP
deflator

X: compounded annual
rate of change of
real GNP

a (^):

change in the ratio

of fixed investment
to GNP

RH- E H: compounded
annual rate of change
of high-employment
federal receipts minus
the rate of change of
high-employment
expenditures

A(S/D)H .
H : change in the
Y _i
high-employment
federal surplus/
deficit divided by
high-employment GNP

NOTE: All variables are expressed in percent terms

from a policy action to economic activity are of second
ary interest, as is the process by which inflationary
expectations are formed. Rather, the chief concern is
whether certain values of key economic variables can
be achieved with different combinations of monetary
and fiscal actions.
The variables used in the analysis and summarized
in table 3 are restricted to those implicit in the IS-LM
framework. The dependent variables are real GNP,
real and nominal interest rates, and the ratio of fixed
investment to GNP. These variables were regressed
on current and lagged values of the monetary and fiscal
variables.14 The regressions involving interest rates
demonstrated substantial autocorrelation in the re
siduals, so these equations were adjusted using the
Cochrane-Oreutt procedure. All other equations were
estimated with ordinary least squares. Lag length was

14Although M l minus trend was used in table 1 in defining the
stance of monetary policy. M l without trend adjustment was used
in the regressions, mainly because of its simple interpretation.
Explaining the quarter-to-quarter variation of a variable is little
affected by this choice, except for the value of the constant term.

Digitized for14
FRASER

determined by varying the lag by multiples of four and
choosing from these regressions on the basis of max
imum adjusted R2. The Almon lag technique was used,
with the coefficients constrained to lie on a thirddegree polynominal with a tail constraint. These esti
mates were tested against the unrestricted leastsquares estimates to ensure that the smoothness
assumption could not be rejected by the data.

Regression Results
Tables 4 -7 summarize the statistical results. The
effects of the policy variables on the relevant depen
dent variable are summarized as one-year effect, twoyear effect and full effect. The final column, labeled
“mix effect, ” of tables 4 -7 is of particular interest here.
The mix effect is defined as the percentage change in
\11 that would be required to offset a tighter fiscal
policy and keep the dependent variable constant. A
tighter fiscal policy is defined as a decrease in federal
spending that would increase the change in the highemployment budget by $5 billion for 1980 and 1981.
An appendix provides the specifics of how this tighter
fiscal policy is defined.

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

Table 4
Effects of Monetary and Fiscal Actions on the Aaa Bond Rate
Sample Period; 11/1959—
IV/1981
Monetary Effect
Dependent
variable
Aaa

Aaa

Variable

Lags

3
7
20

Sum of
coefficients1

Fiscal Effect
Sum of
coefficients1

Variable

Lags

.114
.394*
1.395*
(4.49)

3
7
16

.010
.030
.098
(1.24)

3
7
24

.118*
.401*
1.750*
(4.92)

RH- EH

3
7
16

.017
.054
.101
(1.40)

3
7
24

.108
.363*
1.725*
(4.96)

3
7
16

.541
1.617
2.931
(1.80)

A(S/D)H
Y -i

Constant

1.539
(.51)

1.103
(.36)

1.178
(.40)

.17

.18

.20

1.74

1.77

Rho

Mix
effect2

.99

Undef.
0
0

.99

0
0
0

.99

Undef.
0
0

'Sums are cumulative for number of lags indicated. Absolute value of t-statistic in parentheses. * indicates sum is significant at 5 percent
level.
2Mix effect is the change in M1 (in percentage points) required to offset a tightening of fiscal policy in the form of an increase in A(S/D)H of
$5 billion. To offset means to keep the dependent variable unchanged. See appendix for details. “ Undef." means the mix effect is
undefined; neither the monetary or fiscal effect is significantly different from zero (5 percent level).

N om inal in terest rates and policy mix— Many
analysts consider the level of nominal interest rates to
be the primary cause for the current economic malaise.
Yet the IS-LM model depicts the interest rate as a
dependent variable. So, the first question investigated
here is whether the nominal interest rate is systemati
cally related to monetary-fiscal actions. The rela
tionship of the corporate Aaa bond rate to the selected
measures of monetary and fiscal policy is summarized
in table 4.
Three regressions were run, corresponding to the
three measures of fiscal action. The number of lags that
maximized the adjusted R2 of the equation ranged from
20 to 24 for M l and was 16 for the fiscal variable. The
results are summarized as one-year effects (current and
three lags), two-year effects (current and seven lags),
and the full effect, which takes into account all of the
lags. Relevant summary statistics also are shown.
According to the simple IS-LM interpretation as
shown in figure 2, a tighter fiscal policy should be
accompanied by a tighter monetary policy in order to
keep interest rates unchanged. Remember that the
model of figure 2 ignores the influence of price ex
pectations. A tightening of fiscal policy with no change

in monetary policy shifts the IS curve to the left, lead
ing to a fall in interest rates. To keep interest rates
unchanged, monetary policy would have to be tight
ened, shifting the LM curve to the left. Consequently,
the implication of the simple IS-LM model is that the
“mix effect” would be negative.
The results in table 4 are clear-cut. The mix effect is
either zero or undefined. The effect of E H on interest
rates is of the expected sign, but not significantly dif
ferent from zero. In the case of both RH — E H and
AS/DH ,
.
r i
—v j'i , the effect on interest rates is not of the exi-i
pected sign, but is also not significant. A tightening of
AS/Dh
fiscal policy, that is, an increase of R h — E Hor vH >
i )
is associated with an increase in interest rates. The
effect of monetary policy on interest rates in all cases is
positive, running contrary to the implication of the
simple IS-LM model.
In general, fiscal policy has no effect on the Aaa bond
rate; thus, there is no mix effect. The effect of monetary
expansion, on the other hand, builds up over time and
appears to be permanent. These results suggest that it
is necessary to augment the IS-LM model with price
15

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

Table 5
Effects of Monetary and Fiscal Actions on a Proxy for the Real Interest Rate
Sample Period: 11/1959-IV/1981
Monetary Effect
Dependent
variable
Aaa - P .16

.114
.330*
.698*
(3.13)

3
7
20

.116*
.321*
.846*
(2.90)

3
7
20

.111
.299*
.798*
(2.78)

Sum of
Variable coefficients1
Lags

3
7
16

Variable

X
LU

A a a -P .16

Sum of
coefficients1

I
C
C

A a a - P 16

Lags

Fiscal Effect

A(S/D)H

3
7
16

.010
.022
.009
(.11)

3
7
16

.009
.016
-.0 1 5
(.21)

3
7
16

.202
.604
.385
(.24)

Constant

- .1 0 5
(.06)

-.5 5 9
(.27)

-.3 9 3
(.21)

.04

.06

1.79

1.82

1.81

Rho

Mix
effect2

.98

Undef.
0
0

.99

0
0
0

.98

Undef.
0
0

1Sums are cumulative for number of lags indicated. Absolute value of t-statistic in parentheses. * indicates sum is significant at 5 percent
level.
2Mix effect is the change in M1 (in percentage points) required to offset a tightening of fiscal policy in the form of an increase in A(S/D)H of
$5 billion. To offset means to keep the dependent variable unchanged. See appendix for details. “ Undef." means the mix effect is
undefined; neither the monetary or fiscal effect is significantly different from zero (5 percent level).

expectations, when those expectations seem to depend
on the rate of monetary expansion. The explanatory
power of each of the three equations is dominated by
the rate of change of money.

Real interest rates and policy mix— Recent concern
about the level of interest rates also has been couched
in terms of the real rate.15 Many consider the current
high level of real rates an obstacle to economic recov
ery. Since real rates are not observable, however,
proxies have to be developed. Although much work
has been done, the simple proxy of a nominal rate
minus the recent rate of inflation is still most common
ly used. For this analysis, the real rate is proxied by the
Aaa bond rate minus the rate of change of the GNP
deflator over the four previous years.
The results are summarized in table 5, following the
same format as before. The explanatory power of each
equation is very low, although the monetary effect is
significant in each case. The fiscal effect is not signifi

15See, in particular, the study by the Congressional Budget Office
cited in footnote 3.

Digitized for16
FRASER

cant for any of the three measures. As with nominal
rates, there is no mix effect applicable to the real rate.
In general, monetary and fiscal actions do little to
explain the movement of the real rate as measured by
the Aaa bond rate minus past inflation. To the extent
that the equation has explanatory power, it comes from
the monetary variable. Even that effect runs counter to
the conventional wisdom, as might be implied by an
expectations-augmented IS-LM model. More expan
sionary monetary policy is associated with increases in
the real rate. Such an effect should probably not be
taken too seriously, however, because of the problems
inherent in measuring the real rate.

Output growth and policy mix— Another interpreta
tion of the mix problem is that output growth is being
retarded by the particular combination of policies in
effect. The next set of regressions examines the rate of
output growth as a function of the monetary and fiscal
variables. These results are summarized in table 6.
The monetary and fiscal variables explain between
35 and 40 percent of the movement of output growth,
and no correction for serial correlation is necessary.
What is apparent from these regressions is the impor

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Table 6
Effects of Monetary and Fiscal Actions on the Rate of Output Growth
Sample Period: 11/1959-IV/1981___________________________________
Monetary Effect
Dependent
variable
X

Variable

Lags

3
7
20

Sum of
coefficients1

Fiscal Effect
Sum of
coefficients1

Variable

Lags

1.010*
.434
.001
(.01)

3
7
12

.030
-.0 9 2
-.3 0 7
(1.70)

3
7
20

1.008*
.520
-.3 0 1
(1.61)

RH- E H

3
7
12

-.0 8 2
-.0 8 2
.020
(.12)

3
7
20

1.076*
.670
-.2 7 8
(1.56)

3
7
12

-2 .7 7 6
-3 .1 2 9
-.9 4 6
(.24)

A(S/D)H
Y -,

Constant

6.055
(4.83)

4.506
(4.61)

4.279
(4.51)

.39

.37

.39

2.28

2.20

2.25

Rho

Mix
effect2

N.A.

0
Undef.
Undef.

N.A.

0
Undef.
Undef.

N.A.

0
Undef.
Undef.

1Sums are cumulative for number of lags indicated. Absolute value of t-statistic in parentheses. * indicates sum is significant at 5 percent
level.
2Mix effect is the change in M1 (in percentage points) required to offset a tightening of fiscal policy in the form of an increase in A(S/D)H of
$5 billion. To offset means to keep the dependent variable unchanged. See appendix for details. “ Undef.” means the mix effect is
undefined; neither the monetary or fiscal effect is significantly different from zero (5 percent level).

tance of the length of horizon. Money growth stimu
lates output growth in the short run, but this stimulus
fades after a year. Consequently, any potential mix
effect is applicable only in the short run, but none is
found because the fiscal effect on output is not signifi
cantly different from zero.
The conventional wisdom indicates that a tightening
of fiscal policy should be accompanied by an expansion
ary monetary policy in order to achieve a given rate of
output growth. The directions of the effects are gener
ally found, but, because the fiscal effect is not signifi
cantly different from zero, there is no mix effect. And,
for the long term, the mix effect is undefined because
neither the effects of monetary or fiscal actions are
significantly different from zero.

Investment ratio and policy mix—A final variable of
interest to those concerned with the current mix of
policies is the investment ratio. Easy fiscal policy is
thought to discourage private investment because the
federal government preempts the use of loanable funds
(see figure 2). To investigate this effect, the change in
the ratio of fixed investment to GNP is run against the
monetary and fiscal variables. Table 7 summarizes the
results.

The striking feature of these regressions is that the
explanatory power of these equations is quite high,
with about 55 percent of the movement in the invest
ment ratio explained by the monetary and fiscal vari
ables. Also, no correction for serial correlation is neces
sary. The effect of monetary policy on the investment
ratio is first positive, then negative (shown by a decline
in the sum of coefficients as the horizon is lengthened);
only in combination with E H is the effect significant
after 20 quarters. The fiscal effect, on the other hand,
builds up over time and is significant at the 95 per
cent level after 16 quarters for each of the three fiscal
variables.

The mix effect for the investment ratio, according to
conventional wisdom, should be negative. A tighten
ing of fiscal policy should be accompanied by a tighten
ing of monetary policy in order to keep the investment
ratio constant. That is, a tightening of fiscal policy is
supposed to encourage investment; if that is to be
offset, monetary policy also should be tightened. For
all cases in table 7, the conventional wisdom is upheld.
But, as the horizon is lengthened, there really is no mix
effect, because monetary actions do not have a perma
17

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Table 7
Effects of Monetary and Fiscal Actions on the Investment Ratio
Sample Period: 11/1959-IV/1981
Monetary Effect
Dependent
variable
A(l/Y)

A(l/Y)

Variable

Lags

3
7
20

Fiscal Effect

Sum of
coefficients1

Variable

Lags

.085*
.045
.048*
(2.82)

3
7
12

-.0 1 6 *
- .024*
- .032*
(2.91)

3
7
20

.096*
.050
- .0 0 0
(.02)

RH- E H

3
7
12

.009*
.020*
.034*
(3.40)

3
7
20

.099*
.051
.009
(.82)

3
7
12

.177
.412*
.699*
(2.88)

A(S/D)h
wH
Y -i

Constant

Rho

Mix
effect2
-0 .7
0
0

.080
(1.04)

-.0 1 7
(.30)

-.0 4 2
(.71)

.55

.56

.55

2.36

2.28

N.A.

-2 .2

N.A.

-0 .4
0
0
0
0

N.A.

0
0
0
0
0

1Sums are cumulative for number of lags indicated. Absolute value of t-statistic in parentheses. * indicates sum is significant at 5 percent
level.
2Mix effect is the change in M1 (in percentage points) required to offset a tightening of fiscal policy in the form of an increase in A(S/D)H of
$5 billion. To offset means to keep the dependent variable unchanged. See appendix for details. °c means the mix effect is infinite, that is, a
nonzero effect is divided by zero; the fiscal effect is significant but the monetary effect is not (5 percent level).

nent effect on the investment ratio, while fiscal actions
do (this result is shown as °o).

IMPLICATIONS OF THE RESULTS
These regression results carry certain implications
for economic policy that are at variance with the con
ventional wisdom on policy mix. The general policy
implications of the statistical results require further
discussion. To give some indication of the magnitude of
effect, some different policy mixes are simulated for
the period from 1981 to 1985.

higher, not lower. The implication for real rates is not
clear; even though real rates were positively related to
money growth, the regressions were not significant. To
the extent that interest rates can be explained, money
growth, not fiscal actions, provides the explanation. An
easing of fiscal policy alone does not guarantee a fall in
interest rates. The effect of fiscal actions on interest
rates was not found to be significant.

These regression results can be pulled together into
a general conclusion. Consider Tobin’s recommenda
tion, cited at the beginning of this paper:

The idea of a solid recovery is that real output growth
could be stimulated by shifting the mix toward tighter
fiscal policy and easier monetary policy. The regres
sion results indicate that this desired effect could be
achieved, but only temporarily. Easier monetary pol
icy stimulates output growth initially, but this effect
dissipates after a year. Furthermore, the regression
results indicate that fiscal effects on output growth are
not significant, either in the short or long run.

To achieve a solid recovery, such as the administration
projects, and to achieve it without astronomical interest
rates and serious crowding out, we need an easier
monetary policy combined with a tighter fiscal policy.

Tobin’s recommendation has some validity in its
assertion that the mix should be changed to avoid
“serious crowding out. ” According to the regression
16

What light do these regression results shed on this
recommendation? First, if easier monetary policy
means faster money growth, following Tobin’s recom
mendation will yield nominal interest rates that are

16For a survey article on "crowding out, see Keith M. Carlson and
Roger W. Spencer, “Crowding Out and Its Critics,” this Review
(December 1975), pp. 2-17.

General Policy Implications

Digitized for18
FRASER

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

Table 8
Simulation of Alternative Policy Mixes
Easy Fiscal Policy
and
Tight Monetary Policy

Tight Fiscal Policy
and
Tight Monetary Policy

Tight Fiscal Policy
and
Easy Monetary Policy

Corporate Aaa Bond Rate
1981 Actual
1982
1983
1984
1985

14.17%
14.26
13.01
11.37
9.99

14.17%
14.28
13.26
12.00
10.87

14.17%
14.39
13.78
13.16
12.82

Growth Rate of Real GNP
1981 Actual
1982
1983
1984
1985

0.80%
-0 .1 3
2.16
3.10
3.67

0.80%
-0 .3 5
1.10
2.14
3.41

0.80%
1.36
2.84
2.54
2.83

15.36%
14.62
13.96
13.43
12.95

15.36%
14.64
14.19
14.16
14.52

15.36%
15.00
15.24
15.48
15.73

Investment Ratio
1981 Actual
1982
1983
1984
1985
NOTE: Easy
Tight
Easy
Tight

fiscal policy is a
fiscal policy is a
monetary policy
monetary policy

steady increase of the high-employment deficit to $150 billion in fiscal 1985.
steady movement toward a balanced high-employment budget in fiscal 1985.
is 6.5 percent growth rate of M1.
is 4.5 percent growth rate of M1.

results, tighter fiscal policy would indeed encourage a
rise in the investment ratio, but interest rates would be
minimally affected. This shift to tighter fiscal policy,
however, need not be accompanied by easier monetary
policy. There is a positive effect of monetary actions on
the investment ratio, but it appears to be temporary,
whereas the fiscal effect appears to be permanent.

Simulation of Alternative Policy Mixes
To provide some specific indication of the magni
tudes involved in the policy mix controversy, the re
gression results were used to simulate three different
mixes of policy for the 1982-85 period. Point estimates
of the coefficients were used even if they were not
significantly different from zero.1'
Two fiscal scenarios were chosen; one is based on
fiscal actions that lead to a $150 billion deficit in the
high-employment budget by 1985, the other a tighter

17Real interest rates were not simulated because none of the regres
sions was significant.

fiscal policy that yields a balanced high-employment
budget by 1985. The easy policy conforms roughly with
the prospective course of fiscal action as it appeared to
be developing in early 1982. The tight policy is consis
tent with a recommendation by the Brookings Institu
tion in their annual report on the federal budget.18
Two monetary policies were simulated: one is a
steady expansion of M l at a 4.5 percent rate, the other
is expansion at a 6.5 percent rate. The first policy is
labeled “tight” and would be within the 1982 target
range announced by the Federal Reserve. The 6.5
percent scenario for money is called “easy,” and would
be above the upper end of the 1982 target range.
Table 8 gives the results of these simulations. The
result of tightening fiscal policy and easing monetary
policy (compare the first and third columns) is to push
up nominal interest rates. The growth of output is
18See Charles L. Schultze, “Long-Term Budget Strategies,” in
Joseph A. Pechman, ed., Setting National Priorities: The 1983
Budget (The Brookings Institution, 1982), pp. 187-220. Their
recommendation was made before passage of the Tax Equity and
Fiscal Responsibility Act of 1982.

FEDERAL RESERVE BANK OF ST. LOUIS

actually worsened by the change in the mix, the reason
being that inflation accelerates with easier monetary
policy. The investment ratio, however, is increased by
changing the mix of policy. As the middle column of
table 8 shows, there is little to be gained by expanding
money more rapidly when fiscal policy is tightened.
Although easing monetary policy appears desirable
because of its beneficial effects on fixed investment,
recall that the long-run effects of money growth on the
investment ratio were not statistically different from
zero.

SUMMARY AND CONCLUSIONS
The notion of policy mix has been presented in
textbooks and discussed by eminent analysts almost as
if it were a self-evident truth. A recommended change
in policy mix seems to be based on the well-known
IS-LM model. When scrutinized more closely, the
question of the appropriate mix of monetary-fiscal poli
cies is not as clear-cut as the simple IS-LM model
implies. The lagged effects of policy actions must be
taken into account, as well as the empirical realities of
certain economic relationships.
This paper examined four dependent variables that
seem relevant in any discussion of policy mix— nomi
nal interest rates, real interest rates, the rate of output

OCTOBER 1982

growth, and the investment ratio. The conclusions are
as follows:
(1) M ovem ents o f nom inal in terest rates and, to a lesser
exten t, real rates are dom in ated by m o n etary actions.
The effect of fiscal actions on in terest rates is not statisti
cally significant.
(2) T h ere is a short-ru n effect o f m o n etary actions on
output grow th, b u t it is only tem p orary . O v er the long
run, m ovem en ts in ou tp ut grow th are unaffected by
eith er m onetary o r fiscal actions.
(3) T h e in vestm en t ratio is influenced tem p orarily by
m onetary actions, b u t th e effect appears to be p erm a
n ent for fiscal actions.

These conclusions imply that the IS-LM framework
must be carefully interpreted when used as a guide for
policy analysis, and that current recommendations for
a change in the mix are only partly valid. Fiscal policy
should indeed be tightened in order to stimulate an
increase in the investment ratio, if long-term economic
growth and/or housing investment is a national goal.
There is little evidence, however, to support the no
tion that interest rates would be affected greatly. There
is no basis for thinking that a tightening of fiscal policy
should be accompanied by an easing of monetary
policy. The effect of easier monetary policy would be
higher nominal interest rates and only a temporary
surge of output growth.

FEDERAL
RESERVE

Receipts
Level

Expenditures

Change

Level

Change

Tighter Fiscal Policy (High-Employment Values)

Surplus/Deficit

Receipts

Level

Change

Level

Expenditures

Change

Level

Change

$20.1

-$ 1 3 .3

-$ 5 .6

13.7

-1 1 .7

1.6

589.4

20.1

- 9 .2

2.5

610.3

20.9

3.2

12.4

626.4

16.1

23.7

20.5

622.8

-3 .6

40.6

16.9

645.4

22.6

39.3

- 1 .3

672.2

26.8

16.0

-2 3 .3

1980:1

$542.3

$14.5

$560.6

$25.1

-$ 1 8 .3

-$ 1 0 .6

$542.3

$14.5

$555.6

557.5

15.2

579.3

18.7

-2 1 .7

-3 .4

557.5

15.2

569.3

III

580.1

22.6

604.4

25.1

-2 4 .2

-2 .5

580.1

22.6

613.5

33.4

630.3

25.9

-1 6 .8

7.4

613.5

33.4

1981:1

650.2

36.7

651.4

21.1

-1 .3

15.5

650.2

36.7

663.3

13.1

652.8

1.4

10.6

11.9

663.3

13.1

III

684.7

21.4

680.4

27.6

4.3

-6 .3

684.7

21.4

688.3

3.6

712.2

31.8

-2 4 .0

-2 8 .3

688.3

3.6

no change

Change

Surplus/Deficit

O ST. LOUIS
F

___________ Actual Fiscal Policy (High-Employment Values)

BANK

Appendix
Change in Fiscal Policy Used in Calculating Mix Effect

$5 billion difference

For the fiscal variables used in the regressions, this change in policy translates as follows:

RH- E H

YH,

1st year

-3 .7 4

+ 3.74

+ .19

2nd year

-3 .2 8

+ 3.28

+ .18

Beyond 2nd year

-3 .2 8

+ 3.28

+ .18

OCTOBER
1982

A(S/D)H
EH

Simple Analytics of the Money Supply
Process and Monetary Control
D ANIEL L. TH O RN TO N

O
n O C TO BER 6, 1979, the Federal Reserve
adopted a new procedure for implementing monetary
policy that would place more emphasis on controlling
the money supply and less on controlling the level of
the federal funds rate. This procedure has been im
plemented by establishing an intermediate target for
nonborrowed reserves.1 While the Federal Reserve
has succeeded in slowing the rate of growth in the basic
monetary aggregate (M l) since adopting the new pro
cedure, it has failed to smooth the erratic short-run
movements in M l.

ernors is now committed to implementing the last of
these recommendations.3

Analysts have suggested a number of changes to the
Federal Reserve’s operating procedure to achieve
more stable short-run monetary growth. Among the
most frequently cited proposals are: adopting a mone
tary base target, tying the discount rate to a market
interest rate and adopting a system of contempora
neous reserve accounting (CRA).2 The Board of Gov

A M O D EL OF THE MONEY STOCK

'F o r a discussion of the new operating procedures, see Fred J.
Levin and Paul Meek, “Implementing the New Operating Proce
dures: The View from the Trading Desk,” New Monetary Control
Procedures, Volume I, Federal Reserve Staff Study (Board of
Governors of the Federal Reserve System, February 1981);
Stephen Axilrod and David E. Lindsey, “Federal Reserve System
Implementation of Monetary Policy: Analytical Foundations of the
New Approach,” American Economic Review (May 1981), pp.
2 46-52; and R. Alton Gilbert and Michael E . Trebing, “The
FOMC in 1980: A Year of Reserve Targeting,” this Review (August/
September 1981), pp. 2-22.
2There have been a number of suggestions for these reforms. For
example, Economic Report o f the President (1982); Anatol B. Balbach, “How Controllable is Money Growth?” this Review (April
1981), pp. 3-12; James M. Johannes and Robert H. Rasche, “Pre
dicting the Money Multiplier,” Journal o f Monetary Economics

The purpose of this paper is to analyze the effects of
adopting each of these proposals. Since each of these
recommendations, in some way, are linked to the
others, the analysis will proceed serially, beginning
with the effects of base targeting and ending with the
effects of CRA. These proposed reforms are analyzed
within the context of a simple linear stochastic model of
the money stock.4

The model used here is complete enough to provide
useful insights into the effects of each of these propos
als, yet simple enough to be readily understood. The
reader need not follow each step in the development of
the model in order to understand its implications. All
(July 1979), pp. 301-26, and “Can the Reserves Approach to
Monetary Control Really Work?” Journal o f Money, Credit and
Banking (August 1981), pp. 298-313; William Poole, “Federal
Reserve Operating Procedures: A Survey and Evaluation of the
Historical Record Since October 1979,” presented at the Confer
ence on Current Issues in the Conduct of Monetary Policy, Amer
ican Enterprise Institute (February 1982).
"The Federal Reserve will implement CRA on February 2, 1984.
‘‘The model is a more complicated version of linear stochastic mod
els recently considered by Poole and LeRoy. See Poole, “Federal
Reserve Operating Procedures: A Survey and Evaluation of the
Historical Record Since October 1979;” and Stephen F . LeRoy,
“Monetary Control Under Lagged Reserve Accounting, ” Southern
Economic Journal (October 1979), pp. 460-70.

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

Table 1
A Simple Linear Model of the Money Stock
Equations:

Definitions o f Variables:

Money Supply Equations
Mf = D, + C,

the supply of nominal money, composed of checkable
deposits and currency

checkable deposits of depository institutions

B, = RR, + Ct + ER,

the currency component of money

C, = kD, + pi, + Uc, k > 0, p < 0

the monetary base

TDt =

+ p,it + uTl, t > 0, p. ^ 0

required reserves of depository institutions

ER, = eD, + 5i, + uet, e > 0, 8 < 0

time and savings deposits of depository institutions

RR, * r(D, + TD,)

excess reserves of depository institutions

RR, + ER, = NBR, + BR,

NBR

nonborrowed reserves: total reserves of depository
institutions less depository institutions’ borrowing from
the Federal Reserve

depository institutions’ borrowing from the Federal
Reserve

the nominal market interest rate

the discount rate

nominal GNP

the demand for nominal money

the equilibrium nominal money stock

t D,

BR, = a(id, — i,) + Ub,, a < 0
Money Demand Equations
M,d = ($Y, + Xi, + umt, 0 > 0, X < 0
Market Equilibrium Condition
M f = M? =M ,

Uc, uT
,,
uet, ubt,

um
,

random variables with zero means and constant
variances.

Descriptions of Equations:
The first two equations are the definitions of money stock and
the monetary base, respectively. The next three equations are
parametric equations relating currency held by the nonbank pub
lic, time deposits and excess reserves to checkable deposits and
the market interest rate. These equations also contain random
components that represent both purely stochastic elements and
the effects of omitted variables. The sixth is an identity relating
required reserves to total deposits and the seventh is an equilib
rium condition requiring the demand for reserves to equal the

supply of reserves. The eighth equation makes depository institu
tions’ demand for borrowed reserves a function of the spread
between the Federal Reserve discount rate and the market in
terest rate, and the random error.1 The model is completed by
including a simple demand for money equation and a money
equilibrium condition. The demand for nominal money is assumed
to depend on nominal income and the market interest rate.2 Infla
tion and, hence, expectations of inflation are ignored.

1This borrowing equation differs from the usual one. The usual borrowing equation would relate borrowing to bank deposits para
metrically, say
BR, = bD,

a (id, — it).

This practice is not adopted here for the following reasons. First, there is little theoretical justification for parametrically relating borrowing to
the level of deposits. Second, this practice results in including the b-term in the multiplier. This gives the erroneous impression that the link
between money and the base will change with changes in b, even under base targeting, but this is not the case. This point will be made clear
later in the paper. It is true that there would be some frictional level of borrowing even if id, = i,. Thus, it might be appropriate to include a
constant term in equation 8.
Third, it is possible to obtain a nonborrowed reserve multiplier by completely ignoring equation 2 when the borrowing equation is written in
the above form. This encourages one to ignore the fact that nonborrowed reserves are linked to money only via their link to the base or total
reserves.
2That is, we follow common practice in assuming the absence of a “ money illusion” on the part of money holders.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

of the results presented are derived in the appendix.
Nevertheless, the complete model along with a de
scription of the equations and variables is presented in
table 1 for the reader’s convenience. Wherever possi
ble, the analysis is presented graphically.

[ (,^ + s + p) — p fra + T ^ + c + k ) ]
(X — p) (r(l +

(2) M t =

+e+k>

(n * +

s +

Specifically, we assume that Yt = Yt + E t , where E(et) = 0. Thus,
E(Yt) = Yt, i.e., the Federal Reserve correctly forecasts nominal
income on average. This forecast introduces another source of error
into the model which is ignored in this paper for convenience;
however, this does not affect the qualitative conclusions. The
Federal Reserve has to forecast the parameters of the model as
well. This problem is usually ignored.

Digitized for24
FRASER

p )(r(l +

+ e)

+ k)

Yt
+ (rp. + 8 + p)

NBRt

+ (rp, + 8 + a)

+ P
(X -

p )(r(l +

T) +

1 + k

4- (rp. + 8 + a)

aX
(X -

p )(r(l +
1+ k

+ e)

idt

+ (rp, + 8 + a)

(3) M , = p f t + Xit

The expected value of the equilibrium money stock
under interest rate targeting (equation 3) is simply the
demand for money. This reflects the well-known fact
that the quantity of money is completely demanddetermined if the Federal Reserve chooses an interest
rate target.6
While these equations appear somewhat compli
cated, they merely represent expressions for the ex
pected value of the equilibrium money stock, repre
sented graphically by M* in figure 1. For example,
both the base and income appear in the equilibrium
equation 1, because the money supply is conditional on
the level of the base under base targeting and the
demand for money is conditional on the income level.
This is illustrated in figure la. Therefore, the equilib
rium money stock depends both on the level of income
and the base under a monetary base target.

Control of the money stock through each of these
targets can be analyzed by evaluating the expressions
for the expected value of the equilibrium money stock
obtained by treating each of these variables as exoge
nous. (The Federal Reserve must forecast the level of
income Yt in order to control the money stock. Thus,
the forecasted value Yt replaces Yt in the model.)5 The
expected value of the equations for the equilibrium
money stock under monetary base, nonborrowed re
serve and interest rate targeting, respectively, are:

(X -

1 + k

The model requires that three variables be exoge
nous. Two of the exogenous variables are the discount
rate (idt) and nominal income (Yt). The remaining ex
ogenous variable is determined by the operating pro
cedure. If the Federal Reserve chooses to target on the
market interest rate (it), it would be treated as exoge
nous; in this case, the monetary base (Bt) and nonbor
rowed reserves (NBRt) would change to whatever
levels are necessary to achieve the interest rate target.
If the Federal Reserve targets on nonborrowed re
serves, the monetary base and the interest rate would
move endogenously to achieve levels consistent with
the nonborrowed reserve target. The same would be
true of the interest rate and nonborrowed reserves if
the Federal Reserve chose a monetary base target.

( x ~ p )(r ^

1 + k

The model initially assumes CRA. This assumption
will be changed later to analyze the implications of
lagged reserve accounting (LRA) for short-run mone
tary control and to analyze the effects of the Board’s
proposal for CRA. Initially, the deterministic form of
the model is considered. This is achieved by taking the
expected value of the endogenous variables (the ex
pected value is denoted with a hat, e.g., E(X) = X).
The full model is taken up in the final section, which
deals with the variance of money and interest rates
under CRA and LRA.

(1) M '

T) +

The discount rate appears in the money stock equa
tion when nonborrowed reserves are exogenous, but
not when the monetary base is exogenous. The reason
for this is simple. Changes in the discount rate alter the
spread between it and the market interest rate and,

6In this case, the money supply curve would be perfectly horizontal,
as the Federal Reserve simply accommodated the public’s demand
for
) B| money at some nominal rate. Of course, it is well-known and
widely accepted that the Federal Reserve cannot “peg” the nomi
nal rate in an inflationary environment without continuously
accelerating the growth rate of money. See Milton Friedman, “The
Role of Monetary Policy,” American Economic Review (March
1968), pp. 1-17. More recently, however, McCallum has shown
that the price level is determinant if the Federal Reserve smoothes
rather than pegs interest rates. See Bennett T. McCallum, “Price
Level Determinancy with an Interest Rate Policy Rule and Ration
al Expectations,” Journal o f Monetary Economics (November
1981), pp. 319-29.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Figure l

E q u ilib riu m M o n e y Stock U n d e r Base and N o n b o rr o w e d R e serve T a rg e tin g
In te re s t
rate

Ms |N B R t = N B R ,id t = id

M
dlY,=Y

A
*

s tock

(a)

thus, the level of bank borrowing.7 If the monetary
base were the control variable, changes in borrowings
would be offset through open market operations in
order to maintain the base at its target level. Changes
in the discount rate would have no effect on the
equilibrium money stock under monetary base target
ing. Thus, the discount rate does not appear as an
exogenous variable in equation 1.
This is not the case for nonborrowed reserve target
ing. Changes in the discount rate would produce
changes in depository institutions’ borrowing, the
monetary base and the money supply unless the
Federal Reserve simultaneously changes its target
level of nonborrowed reserves.8 Thus, the money
there were a significant “announcement effect” of a discount rate
change on market interest rates, there would be an endogenous
movement in the money stock even under base targeting. For
more details on the relationship between the discount rate and
market interest rates, see Daniel L. Thornton, “The Discount Rate
and Market Interest Rates: What’s the Connection?” this Review
(June/July 1982), pp. 3-14.

7I f

8If the Federal Reserve changed its nonborrowed reserve target in
light of these changes, it would, ipso facto, be targeting on the base

Money
s to ck

(b)

supply schedule is conditional on both nonborrowed
reserves and the discount rate under nonborrowed
reserve targeting, as illustrated in figure lb.

MONETARY BASE VS.
NONBORROW ED RESERVE
TARGETING
The controversy over base versus nonborrowed re
serve targeting depends critically on the stability of the
link between each of these reserve aggregates and
nominal money. The stability of this link, in turn,
or total reserves, not nonborrowed reserves. It is sometimes
argued that the Federal Reserve is essentially base or total reserve
targeting because it first determines a total reserve path and, from
that, its nonborrowed reserve path given an initial borrowing
assumption. It would be total reserve targeting, however, only if it
changed its nonborrowed reserves every time it recognized it was
off its total reserve path. Recently, Gilbert and Trebing have shown
that the Fed often knowingly stuck with its nonborrowed reserve
path despite the fact that it correctly projected it would be off the
total reserve path necessary to hit its short-run money target. R.
Alton Gilbert and Michael E. Trebing, “The FOMC in 1980: A
Year of Reserve Targeting,” this Review (August/September 1981),
pp. 2-16.

FEDERAL RESERVE BANK OF ST. LOUIS

depends on (a) the exogeneity of the respective reserve
aggregate multiplier and (b) the exogeneity of the re
serve aggregate. The first of these issues can be dealt
with by considering the extent to which the money
stock is exogenous under each target.9

The Exogeneity o f Money under Base and
Nonborrowed Reserve Targeting
Consider the extent to which the money stock is
exogenous under nonborrowed reserves and monetary
base targeting— that is, determined only by the Feder
al Reserve’s control over the target variable and the
parameters of the model. This can be accomplished by
observing the conditions required to make the money
supply schedule vertical under the two regimes. Equa
tion 1 indicates that the money stock is exogenously
determined by the base if the public’s demand for
currency and time deposits and depository institutions’
demand for excess reserves are unresponsive to in
terest rate changes (|x = 8 = p = 0). If these condi
tions hold, equation 1 reduces to

«■> M- = K 1 + ‘) t ! l

+ k B-

The less interest-sensitive these factors are, the less
interest-sensitive will be the money supply. If these
factors are completely insensitive to the interest rate,
9In a recent Board study dealing with the observed historical stabil
ity of the monetary base multiplier, David Lindsey and others
discuss two types of multiplier endogeneity. First, they argue that
the variability of the observed multipliers may be biased downward
because they do not account for the possibility that the targeted
level of the reserve aggregate may have changed in response to an
unanticipated change in some other factor. If, for example, the
money supply took an unanticipated jump (because of an unantici
pated jump in the demand for money) at the beginning of the
intermeeting targeting period, the Federal Reserve might reduce
its target for nonborrowed reserves (or the monetary base under
base targeting). The result might be a larger observed multiplier if
the reduction in the reserve aggregate were larger than the reduc
tion in the money supply from its unanticipated level (of course, it
could be smaller if the money supply response was greater). This
type of reserve aggregate endogeneity error applies to all potential
reserve aggregates, but only if the Federal Reserve is actually
targeting on it. Moreover, this type of reserve endogeneity is
concerned only with the question of the observed stability of the
multiplier; it has nothing to do with the issue of monetary control.
The second type of reserve endogeneity is the traditional type, in
which factors that make up the various multipliers change with
changes in other endogenous variables (e.g ., interest rates) in the
system, as considered here. See, David Lindsey, et. al., ‘Mone
tary Control Experience Under the New Operating Procedures,”
New Monetary Control Procedures, Federal Reserve Staff Study,
Volume II (Board of Governors of the Federal Reserve System,
July 1981); Balbach, “How Controllable is Money Growth?” Johan
nes and Rasche, “Can the Reserves Approach to Monetary Control
Really Work?” and “Predicting the Money Multiplier.”

Digitized for 26
FRASER

OCTOBER 1982

the money supply schedule becomes perfectly verti
cal, and the equilibrium money stock is exogenously
controlled.
In contrast, the money stock is exogenous under
nonborrowed reserve targeting only if the above condi
tions hold and if, simultaneously, a, the interest re
sponsiveness of borrowing from the Federal Reserve to
the rate spread between the discount rate and the
market interest rate, is zero. Under these conditions,
equation 2 reduces to

Thus, the conditions necessary for the Federal Re
serve to control the money stock are more restrictive
under nonborrowed reserve than under base target
ing. Indeed, it is commonly accepted that excess re
serves are interest-insensitive, and Johannes and
Rasche recently have argued that the currency and
time deposit ratios are fairly interest-insensitive as
well.10 This is not the case, however, for depository
institutions’ borrowing. Casual observation shows a
strong relationship between borrowing and the dis
count rate/market interest rate spread. Thus, the sup
ply of money might exhibit greater interest sensitivity
under nonborrowed reserve targeting than under base
targeting.11
The importance of this for monetary control can be
seen by noting that the less interest-sensitive the
10See Johannes and Rasche, “Predicting the Money Multiplier” for
details.
11This question is more complicated in a nonlinear model. In the
nonlinear case, the relative interest sensitivity of the money sup
ply under base and NBR targeting depends on the relative magni
tude of these reserve aggregates and their multipliers. To illus
trate this, let the money supply under base and NBR targeting be
given by the expressions below.
M s = m(i; B) • B
M s = m(i; N B R ) • N B R

Here, m(i; B) and m(i; NBR) denote multipliers that are functions
of the interest rate, given either a base or NBR target. The
difference in the interest responsiveness of the money supply
under base or NBR targeting is given by
flm (i; B) • B -

3m(i; N B R ) • N B R

This expression is less than or equal to zero if
B

_ 3m(i; N B R )

NBR ~

, dm(i; B) .
di

Since the base is about four times as large as NBR, the base
multiplier must be about one-fourth as interest-responsive as the
NBR multiplier if the money supply under base targeting is to be
less interest-sensitive than under NBR targeting. Whether this
condition holds depends on the relative magnitude of the struc
tural parameters as discussed above.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Figure 2

The Effect of U n a n tic ip a te d Shifts in M o n e y D e m a n d and S upply

money supply, the less responsive the equilibrium
money stock will be to unanticipated shifts in money
demand. This is illustrated in figure 2a, which shows
the effect of an unanticipated increase in the demand
for money.
While it is more difficult to illustrate, the less in
terest-sensitive the supply of money, the more sensi
tive equilibrium money stock may be to unanticipated
changes in factors that affect the supply of money. (The
exact relationship depends on the relative magnitude
of certain parameters of the model.) This is illustrated
in figure 2b. An unanticipated decrease, say, in excess
reserves, shifts both M | and M s Br to the right.
N
Although the latter curve shifts further, the resulting
change in the money stock may be smaller. This is the
result of the effect of a reduction in depository institu
tions’ borrowing associated with the declining interest
rate on the quantity of money supplied. The above
result depends on the source of the supply-side shock.
If the supply-side shock comes from an unanticipated
change in currency, the effect on the money stock will
be larger than if it comes from excess reserves or time
deposits.

The absolute magnitude of the differential effect of
supply-side shocks under base and NBR targeting de
pends on the relative magnitude of the interest sensi
tivity of borrowing (a) and the demand for money (\).
The less interest-sensitive is borrowing and the more
interest sensitive the demand for money, the smaller
will be this differential effect. If a is sufficiently small
relative to X, the money supply would be less respon
sive to supply-side shocks under base targeting.12
Given the above analysis, we would expect base
targeting to result in more stable money growth if most
of the exogenous shocks come from the demand side. If
most shocks come from the supply side, however, base
targeting may result in less stable growth. This last

12The required condition for smaller effects of supply-side shocks
under base targeting is (\k + p)/(l + k) < a. Since it is commonly
assumed that p is small, this condition will hold only if IXI is large
relative to Ia I Most empirical evidence on the demand for money
.
suggests XI is very small. Some recent estimates, however, sug
gest a much larger value of I\ I See Daniel L. Thornton, “The
.
Long-Run and Short-Run Demand for Money: Additional Evi
dence, ’’Journal o f Macroeconomics (Summer 1982), pp. 325-38.

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

statement must be tempered by the fact that supplyside shocks resulting from exogenous changes in de
pository institutions’ borrowing will not affect money
under base targeting, but will affect it under NBR
targeting. Thus, if borrowing is unstable, money may
be less responsive to supply-side shocks from all
sources under base targeting.13

Endogeneity of the Monetary Base
The above analysis explicitly assumes that both non
borrowed reserves and the monetary base can be con
trolled exogenously at any desired level. This is gener
ally accepted to be true for nonborrowed reserves,
though not for the monetary base. In fact, a principal
objection to monetary base targeting is that the mone
tary base is endogenous.
In its most basic form, this objection argues that
depository institutions’ borrowing is functionally re
lated to the level of open market operations. Thus, any
attempt at hitting a monetary base target, by offsetting
uncontrolled borrowing through open market opera
tions, necessarily changes the level of borrowing that
the system must offset. The problem of base en
dogeneity, while important, could be handled in part
by tying the discount rate to the market interest rate.

TYING THE DISCOUNT RATE TO
MARKET RATES
The interest responsiveness of borrowing could be
substantially reduced by tying the discount rate to a
market interest rate (i.e., idt = it + A, where A is a
positive or negative constant).14 If this were done,
borrowing would be unresponsive to interest rate
13The error term in the borrowings equation is not present in the
error term for the reduced-form money stock equation under base
targeting. This, of course, assumes that information on depository
institution borrowing is available very quickly (it is currently
available the next day). If the borrowing equation is unstable, as
reported by a recent Board study, then base targeting may still be
less sensitive to supply-side shocks. See “Impact of Discount
Policy Procedures on the Effectiveness of Reserve Targeting,”
New Monetary Control Procedures, Federal Reserve Staff Study,
Volume I (Board of Governors of the Federal Reserve System,
February 1981).
14W e do not say “eliminated” because it is unlikely that tying the
discount rate to any one market rate would eliminate all interestresponsive borrowing. This is due to the fact that different deposi
tory institutions may have portfolios of different assets that reflect
their opportunity cost of borrowing. Hence, tying the discount
rate to one asset may not suffice for every institution. Tying the
discount rate to the federal funds rate, however, would probably
reduce the interest responsiveness of borrowing for most deposi
tory institutions.

Digitized for 28
FRASER

changes. This would substantially reduce the en
dogeneity of the base and, to this extent, make it much
easier to hit and maintain a monetary base target.
Some have argued that the discount rate should be a
penalty rate, that is, A > 0. This concern is probably
overstated. The problem with depository institutions’
borrowing is their interest sensitivity, not their level.
Given the administration of the discount window, the
level of aggregate borrowing will move inversely with
A. The value of the spread between the market rates
and the discount rate will have little impact on its
interest sensitivity.15
Tying the discount rate to market rates also has
implications for nonborrowed reserve targeting.
Under a tied discount rate, equation 2 could be rewrit
ten as
(2") Mt =
f li

~ P)(r(l +
1 +

T) +
k

(X - p)(r(l +
1 +

NBRt
+

(r|jL +

p ( r ( l + T) + e

P(i> + 8)+

1+ k
+

(rp . +

Yt.
8)

The multiplier in this equation is similar to the mone
tary base multiplier of equation 1. Thus, one might be
tempted to conclude that controlling the money supply
by targeting on the base or nonborrowed reserves
essentially would be the same if the discount rate were
tied to market interest rates. This is not the case.
Changes in depository institutions’ borrowing unre
lated to interest rate changes will continue to affect the
money supply under nonborrowed reserve targeting.
This would not be true under base targeting since all
changes in borrowing would be offset through open
market operations.16
15The word “little” is used here because Polakoff has shown that
borrowing increases at a decreasing rate as the rate spread widens.
Making the discount rate a penalty rate might be a consideration if
one believes Polakoff s “reluctance elasticity” is strong or if one
believes that the administration of the discount window is highly
variable. See M. E . Polakoff, “Reluctance Elasticity, Least-Cost
and Member-Bank Borrowing: A Suggested Integration, "Journal
o f Finance (March 1960), pp. 1-18. For a recent discussion of the
role of the discount rate as a penalty rate, see Economic Report o f
the President (1982), pp. 67-68; and Bryon Higgins and Gordon
H. Sellon, Jr., “Should the Discount Rate be a Penalty Rate?”
Economic Review, Federal Reserve Bank of Kansas City (January
1981), pp. 3-10.
16In this regard, if borrowing is unstable, tying the discount rate
may not produce greater control under NBR targeting.

FEDERAL RESERVE BANK OF ST. LOUIS

Furthermore, this argument fails to take account of
the likely response by depository institutions to a
monetary base operating procedure. A policy of offset
ting all changes in borrowing to maintain a monetary
base target may, at times, result in high levels of the
federal funds rate; however, there is an upper limit to
the federal funds rate that is established by depository
institutions’ credit demand. These institutions would
be unwilling to pay an interest rate on short-term
reserve adjustment funds in excess of the rate on mar
ginal short-term loans for any extended period. Thus,
depository institutions might increase their holdings of
excess reserves if the Federal Reserve adopted a base
targeting procedure. This response might permit the
Federal Reserve to hit a base target, but would not
guarantee better short-run monetary control. Any im
provement in short-run monetary control in this in
stance depends critically on the volatility of excess
reserves under a base targeting procedure. Thus,

If the time subscript, t, is understood to be the end
of a reserve maintenance week, the explicit effects of
CRA and LRA for monetary control can be seen by
comparing the model with the following required re
serve equations:
CRA

(4) RRt = r(Dt + T D t)
(5) RR,
(6) RRt

This argument is important; however, it has implica
tions for short-run monetary control under both base
and nonborrowed reserve targeting. If the target level
for nonborrowed reserves is inconsistent with deposi
tory institutions’ required reserves, it produces a
short-run change in borrowing, total reserves and,
hence, money.

LAGGED VS. CONTEMPORANEOUS
RESERVE ACCOUNTING AND
MONETARY CONTROL

While tying the discount rate to the market interest
rate will reduce the endogeneity of the base under a
system of CRA, this need not be the case under LRA.
Under the present system of LRA, required reserves
in the current week are determined by the level of
deposits held two weeks previously. Thus, any dis
crepancy between the amount of reserves supplied by
the Federal Reserve and the amount of reserves that
depository institutions are required to hold must be
made up either at the discount window or through
changes in desired levels of excess reserves. Since
historically excess reserves are relatively interestinsensitive, the discrepancy between the amount of
reserves supplied and the amount of reserves required
will likely be made up at the discount window. If the
Federal Reserve attempted to hit a level of the mone
tary base that was inconsistent with the level of re
quired reserves (given by deposit levels from the pre
vious two weeks), the result would be an immediate
change in the level of depository institutions’ borrow
ing and a movement of the monetary base from its
target level. Thus, it is argued, LRA precludes the
Federal Reserve from hitting a short-run monetary
base target.

rather than being an objection to base targeting, this
argument is simply an indictment of monetary control
under LRA.

ll
it
C

A Tied Discount Rate and Lagged Reserve
Accounting

OCTOBER 1982

LRA

-2 + T D t_

Federal Reserve’s
proposal for CRA

+ T D t_ 2)

These equations represent simplified versions of pure
CRA, pure LRA (the present system) and the Federal
Reserve’s proposal for CRA, respectively.17 The analy
sis begins with a comparison of the model using equa
tion 4 with the model using equation 5, and ends with
an analysis of the likely implications of the Federal
Reserve’s proposal.
To this point, the analysis has assumed equation 4,
so the effects of LRA can be seen by substituting equa
tion 5 into the model. This modification affects the
model in two ways: it weakens the contemporaneous
link between the reserve aggregate and the equilib
rium money stock, and it makes the model dynamic.18
These changes are illustrated by the following equa
tions for the equilibrium money stock under base and
NBR targeting when equation 5 replaces equation 4:
(7) Mt =

(X - p) (k + e)
+ (8 + p)
1+ k
r(l +

t )A

(\ -p )(k + e)
(1 + k)

(8 + P)

17For a discussion of CRA and LRA as implemented, see R. Alton
Gilbert, “Lagged Reserve Requirements: Implications for Mone
tary Control and Bank Reserve Management,” this Review (May
1980), pp. 7-20. The Board’s proposal for CRA has a lag of two
days. The reserve maintenance period ends on Monday, two days
prior to the end of the reserve settlement week.
18Actually, the excess reserve equation may change with CRA or
LRA as well. Current levels of excess reserves would be related to
deposits of the previous two weeks and the current market in
terest rate. This change is ignored for convenience.

FEDERAL RESERVE BANK OF ST. LOUIS

rpA
(\ -p )(k + e)

(1 + k)

+ (8 + p)

(\ -p )(k + e)
+ (8 + p)
1+ k
(8) Mt =

— NBR,
(X -p )e
+ (8 + a)
1+ k
r(l + t)X
pi ~ p)e + (8 + a)
1+ k

D t_ 2

rpA

Pi~p)e + (8 + a)

It — 2

1+ k

a\
Pi~p)e + (8 + a)

id,

1+ k

[ ( 8 + a) -

P i~p)e
+ (8 + a)
1+ k

The first of these changes can be seen by noting that
neither the reserve ratio (r) nor the time deposit ratio
(t ) appears in the contemporaneous multipliers for the
base and NBR. Indeed, NBR are contemporaneously
linked to money, through non-interest-rate effects,
only via excess reserves and borrowings. The reserve
aggregates provide a link to current money creation
primarily through their link to current deposits. LRA
severs part of this link. (It should be noted, however,
that LRA does not eliminate completely the contem
poraneous link between the money stock and either
\1Q
reserve aggregate.)

19It is sometimes argued that there is no contemporaneous link
between deposit creation and reserves independent of its effect on
market interest rates, because depository institutions are free to
create all the deposits they wish in this period without any consid
eration about the current level of reserves (e.g., LeRoy, “Mone
tary Control Under Lagged Reserve Accounting”). This argument
clearly ignores the role of currency and excess reserves in estab
lishing a contemporaneous link between the reserve aggregate
and the money stock. It is easy to show that current deposits are
related to curren t base under LRA, even if the interestresponsiveness of currency and excess reserves are zero. From the
appendix: if p = 8 = 0, then M, = (1 + k)/(k + e)Bt. Pesek and
Saving have made this point in a simple money multiplier model
when there were no reserve requirements. See Boris P. Pesek and
Thomas R. Saving, The Foundations o f Money and Banking (Mac
millan Co., 1968), pp. 76-78.

Digitized for30
FRASER

OCTOBER 1982

The explicit dynamics of the money stock under
LRA are seen by noting that lagged deposits and in
terest rates are included in the equilibrium money
stock equations. NBR and the monetary base not only
influence the level of money immediately, but have
lagged effects via their influence on deposits and in
terest rates. These variables, in turn, affect future
money. It can be shown that if the dynamic system is
stable, the long-run base multiplier is identical to the
static base multiplier of equation 1. Thus, if the Feder
al Reserve were to achieve and maintain some target
level of the base, the expected value of the long-run
equilibrium money stock would, ceteris paribus, be
the same as that obtained under CRA.20

Short-Run Movements in Money and
Interest Rates
Despite the fact that the return to CRA has no
consequences for long-run monetary control, it does
have some implications for short-run movements in
money and interest rates. The following analysis can be
carried out in terms of either base or NBR targeting;
however, the results are presented only for base
targeting.
The analysis begins with a simple graphic presenta
tion of the model under base targeting in figure 3. M£
and Mf denote the money supply schedules under
CRA and LRA, respectively. Under fairly reasonable
conditions, the money supply schedule is flatter under
LRA.21 Again, the money demand equation is drawn
for a fixed income level. The curves are drawn to
intersect for ease of illustration.

20This point has been made by Laufenberg, although in a slightly
different context. See Daniel E. Laufenberg, “Contemporaneous
Versus Lagged Reserve Accounting, ” Journal o f Money, Credit
and Banking (May 1976), pp. 239-45. This result is more logical
than it first appears. For example, one would not expect the
introduction of LRA to have any impact on money creation if
depository institutions had not previously adjusted their required
reserves during the current week under CRA. Thus, we might
expect LRA to affect interest rates and money only to the extent
that it forces the system to follow a different aggregate reserve
adjustment path than it would have followed under CRA.
21The condition is that
>

(1 +

(e +

(8 +

p) <

Since (1 + T )/ (e + k) is greater than one, this condition is likely
to hold. If the interest sensitivity of time deposits, p., is sufficient
ly negative, however, the LRA could be steeper than the CRA
curve. If this were the case, the results of this section would be
reversed. Because of the simplicity of his model, Laufenberg
obtains the above result by ignoring this condition. See Laufen
berg, “Contemporaneous Versus Lagged Reserve Accounting.”

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

Figure 3

Figure 4

Effect of an In c re a se in the R e se rv e A g greg a te
Under LRA and CRA

Effect of an In cre a se in M oney Demand
Under LRA and CRA

The effect of changes in Federal Reserve actions on
the money stock can be illustrated via figure 3. For
example, suppose that the monetary disturbance
comes through an increase in the base. The increase in
the base shifts both M* and Mt to the right. Because the
s
multiplier is larger under LRA than under CRA, the
MJ curve shifts further and the new equilibrium level
of money is larger.22 As a result, both the initial in
crease in the money stock and the initial decrease in
the interest rate are larger under LRA.
A different result is obtained if the monetary dis
turbance occurs through a change in the demand for
money, as illustrated in figure 4. The increase in the
demand for money results in a larger initial increase in
the money stock and a smaller initial increase in the
interest rate under LRA. If all money shocks are associ
ated with changes in money demand due to unantici
pated changes in the level of income, the move back to
CRA would result in less short-run money stock and
more interest-rate variability. If all money shocks are
associated with changes in the policy control variable,
22The condition required for the equilibrium money stock to be
larger is that

The same condition is required for both base and NBR targeting.

the return to CRA would reduce the short-run variabil
ity of both money and interest rates.
In either case, however, the short-run money stock
initially overshoots its long-run equilibrium. In the
former case, the overshoot is due to the fact that in
terest rates fall too far in response to an increase in the
policy aggregate, while in the latter, it is due to the fact
that they do not rise enough in response to an increase
in money demand. This initial overshooting of money
may have repercussions in subsequent periods as de
pository institutions attempt to obtain reserves to sup
port the current overexpansion of deposits. The effect
of this dynamic response on the variability of money
and interest rates is an empirical question.23
23See Laufenberg, “Contemporaneous Versus Lagged Reserve
Accounting. ” As noted, the introduction of LRA makes the model
dynamic and, thus, the model with LRA follows an adjustment
path toward the long-run equilibrium. This path differs with the
dynamic structure of the model. A comparison of one dynamic
adjustment path with another is relevant only if the dynamic
structures are well-specified. In this regard, the conclusions con
cerning the variability of money and interest rates reached by
Laufenberg, based on a comparison of the dynamic adjustment of
the CRA and LRA models, are misleading. See Daniel L. Thorn
ton, “Lagged and Contemporaneous Reserve Accounting and the
Variance of Money and Interest Rates,” unpublished paper,
Federal Reserve Bank of St. Louis (1982). The oscillatory nature of
the adjustment of money and interest rates to their long-run
equilibrium under LRA noted by Laufenberg is discussed in the
appendix.

FEDERAL RESERVE BANK OF ST. LOUIS

F igure 5

The Money Supply Under the Board's
Proposal ior CRA

OCTOBER 1982

money and may increase the variance of interest rates,
depending on the relative magnitudes of the variances
associated with the factors that affect supply and de
mand. While the basic intuition that leads to this con
clusion is correct, it fails to account for possible Federal
Reserve reaction to random changes in time deposits.
Under both CRA and LRA, random changes in time
deposits, TD, affect the demand for required reserves
and, hence, the equilibrium money stock and the in
terest rate. Under LRA, however, the effect of a ran
dom change in time deposits does not manifest itself for
two periods. If such changes were identified with 100
percent accuracy and if the Federal Reserve made
temporary, compensatory changes in the exogenous
reserve aggregate, random fluctuations in time de
posits two weeks previous would have no effect on
either the current money stock or the current interest
rate. If the variance of the time deposits is sufficiently
large, the variance of money and interest rates may be
less under LRA than under CRA. The critical issue is
the extent to which the Federal Reserve correctly
identifies and offsets random shifts in time deposits.24
This result is illustrated as follows: Let V(M^) and
V(Mjj) denote the variance of the money stock under
CRA and LRA, respectively. If the Federal Reserve
correctly identifies random changes in time deposits, it
is easy to show that

The effect of the Federal Reserve’s proposal for CRA
can be seen by noting that the money supply schedule
under the Board s proposal, denoted Mb, lies between
the CRA and LRA curves, as illustrated in figure 5.
Furthermore, the curve will shift further than the CRA
curve, but not as far as the LRA for a given change in
the monetary base. If most money shocks are associ
ated with changes in the policy control variable, the
Federal Reserve’s proposal should improve the shortrun stability of both money and interest rates. If most
shocks are associated with unanticipated changes in
the demand for money, the result will be more stable
money and less stable interest rates.

V(Mf) = erf + a 2 T
M

and
V(M{) = of,

The Variance of Money and Interest Rates
under CRA

where cjmT is the variance in money associated with the
demand for time deposits, e rf is the variance associated
with the other random components of the money stock
under CRA, and o f is the variance of money under
LRA. It can be shown that o f > crf . Thus, the variance
of the money stock would be larger under LRA, all
other things constant. The loss of the variance of time
deposits [omX does not appear in the expression for
V(Ml )], however, makes the variance of money under
t
LRA smaller. The reduction in variance associated
with the removal of this source of variation could more
than offset the increase in variance due to other

Given the above analysis, one might be tempted to
conclude that random variations in the factors that
affect the money supply will cause both money and
interest rates to be more variable under LRA, while
random variations in the demand for money will cause
money to be more variable and interest rates to be less
variable. Thus, one might suspect that the movement
from CRA to LRA would increase the variance of

^Furtherm ore, if individual depository institutions anticipated the
effects of their collective actions on interest rates, the systematic
deposit overexpansions that we have noted need not occur; conse
quently, the greater money and interest rate volatility associated
with differences in the slopes of the short-run money schedule
under LRA need not materialize. This point has been made by
Edgar L. Feige and Robert T. McGee, “Federal Reserve Policy
and Interest Rate Instability,” The Financial Review (May 1982),
pp. 50-62.

Digitized for32
FRASER

OCTOBER 1982

FEDERAL RESERVE BANK OF ST. LOUIS

factors.25 Thus, contrary to the common belief, no
general conclusion can be reached about the relative
variance of money under LRA and CRA. It is an empir
ical issue.26
This analysis can be extended to the Federal Re
serve’s proposal for CRA. If the variance of the money
stock under the Board’s proposal is denoted as
V(Mj’), it is easy to show that V(Mj’) < V(Mt).2'
The question of the relative variance of interest rates
under LRA or CRA has an ambiguous answer. If the
reduction in variance associated with correctly iden
tifying random changes in time deposits is ignored, the
variance of interest rates will be smaller under CRA if
the variance associated with money demand is small
relative to the variance associated with the factors that

25This result is even stronger when it is recognized that variances of
borrowings and excess reserves may be affected by the reserve
accounting structure. There is evidence that both of these error
structures changed with the move to LRA in 1968. See Albert E.
Burger, “Lagged Reserve Requirements: Their Effects on Feder
al Reserve Operations, Money Market Stability, Member Banks
and the Money Supply Process,” unpublished paper for the
Federal Beserve Bank of St. Louis (1971); and Board of Gov
ernors, “Impact of Discount Policy Procedures on the Effective
ness of Reserve Targeting. ”
“ See Poole, “Federal Reserve Operating Procedures: A Survey and
Evaluation of the Historical Record Since October 1979;” LeRoy,
“Monetary Control Under Lagged Reserve Accounting;” and
David S. Jones, “An Empirical Analysis of Monetary Control
Under Contemporaneous and Lagged Reserve Accounting,”
Federal Reserve Bank of Kansas City Working Paper 82-002
(March 1982).
Furthermore, the empirical work to date does not provide an
unambiguous answer to this question. Early empirical work by
Burger, Coats, Poole and Lieberman suggested that the move
ment to LRA in 1968 resulted in greater instability of the federal
fiinds rate. Recent work by Feige and McGee, however, suggests
that the movement to LRA actually reduced the funds rate
variability. Poole and Lieberman report somewhat less stable
money growth under LRA; however, in another study, Feige and
McGee report slight increases in money and reserve predictabil
ity under LRA during the reserve-targeting period, and a substan
tial increase in the predictability of the federal funds rate. See
Burger, “Lagged Reserve Requirements: Their Effects on Feder
al Reserve Operations, Money Market Stability, Member Banks
and the Money Supply Process;” Warren L. Coats, “Lagged Re
serve Accounting and the Money Supply Mechanism, "Journal o f
Money, Credit and Banking (May 1976), pp. 167-80; Feige and
McGee, “Federal Reserve Policy and Interest Rate Instability,”
pp. 50-61; Edgar L. Feige and Robert T. McGee, “Has the
Federal Reserve Shifted From a Policy of Interest Rate Targets to
a Policy of Monetary Aggregate Targets? An Application of E x
ogeneity Test Procedures, "Journal o f Money, Credit and Bank
ing (November 1979), pp. 381-404; Edgar L. Feige and Robert T.
McGee, “Money Supply Control and Lagged Reserve Account
ing,” Journal o f Money, Credit and Banking (November 1977),
pp. 536-51.
27The reader is cautioned that this analysis of the variance of money
and interest rates ignores variability through time associated with
the dynamic structure of the model.

affect the money supply.28 The same conclusion holds
for a comparison of the Federal Reserve’s proposal for
CRA with LRA.
We have shown that if we account for a possible
reduction in the variance of interest rates under LRA
associated with correctly identifying random changes
in time deposits, the variance of interest rates under
LRA could be smaller than under CRA even if money
demand were less variable than the money supply.
This conclusion would not hold for the Federal Re
serve’s proposal, however, since time deposits enter
LRA and it in the same way.
Thus, while an analysis of this model suggests that
the Federal Reserve’s proposal for CRA may reduce the
variance of money compared with the present system
of LRA, its impact on the variance of interest rates is
ambiguous.29

CONCLUSIONS
This article reviews three frequently suggested
changes in Federal Reserve operating procedures to
achieve more stable short-run monetary control:
monetary base targeting, tying the discount rate to the
market interest rate and adopting a system of contem
poraneous reserve accounting.
Since the conditions necessary for money to be ex
ogenous are less restrictive for base than for nonbor
rowed reserve targeting, adopting a base targeting
procedure would likely result in greater short-run
monetary control if most of the shocks to the money
supply come from the demand side. If most of the
shocks come from the supply side, then base targeting

“ In this regard, there has been a great deal of concern about the
stability of money demand in recent years. See, for example,
Michael J. Hamburger, “Behavior of the Money Stock: Is There a
Puzzle?" Journal o f Monetary Economics (July 1977), pp. 265-88;
G. S. Laumas and David E. Spencer, “The Stability of the De
mand for Money: Evidence from the Post-1973 Period,” Review o f
Economics and Statistics (August 1980), pp. 455-59; and R. W.
Hafer and Scott E. Hein, “Evidence on the Temporal Stability of
the Demand for Money Relationship in the United States,” this
Review (December 1979), pp. 3-1 4 ; and “The Shift in Money
Demand: What Really Happened?” this Review (February 1982),
pp. 11-16.
29The Board has also considered a provision for staggered-reserve
accounting. The effect of this controversial provision on short-run
variability of money and interest rates remains a question. For a
discussion of the effects of staggered-reserve accounting, see
Michael L. Bagshaw and William T. Gavin, “Stability in a Model
of Staggered-Reserve Accounting,” Federal Reserve Bank of
Cleveland, Working Paper 8202 (August 1982); and William T.
Gavin, “The Case for Staggered-Reserve Accounting,” Federal
Reserve Bank of Cleveland Economic Review, (Spring 1982), pp.
30-36.

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

may result in less stable monetary growth. This conclu
sion depends critically on the relative interest sensitiv
ity of the demand for money and depository institution
borrowing. The more interest sensitive is the former
and less interest sensitive is the latter, the more stable
money growth will be under base targeting relative to
NBR targeting. Furthermore, if depository institution
borrowing is highly variable, base targeting may be
more stable, even if the shocks come from the supply
side, since money will be unresponsive to fluctuations
in borrowing under base targeting.
Monetary control under base targeting could be en
hanced further by tying the discount rate to a market
interest rate. This would reduce substantially the in
terest sensitivity of borrowing and make it easier to hit
a monetary base target. Adopting a system of contem

poraneous reserve accounting also should make it
easier to hit a base target, since it would no longer be
necessary for borrowing to respond to differences be
tween a predetermined level of required reserves and
an amount of reserves consistent with the base target,
as under the present system of LRA.
It is not certain whether the return to a system of
contemporaneous reserve accounting on all deposits
would increase or reduce the variance of money and
interest rates. The Federal Reserve’s proposal for con
temporaneous reserve accounting, however, will like
ly reduce the variability of money. Furthermore, it will
likely reduce the variability of interest rates, if the
variance of the money demand schedule is sufficiently
small relative to the variance of the money supply
schedule.

APPENDIX
The purpose of this appendix is fourfold. First, it
presents the system of reduced-form equations for
base and NBR targeting under CRA. Second, it pre
sents the reduced-form equations under LRA. Third,
the long-run base multiplier under LRA is derived and
discussed. Finally, analytical results for the variance of
the endogenous variables for base and NBR targeting
and for CRA and LRA are presented and discussed. In
what follows, we will denote the expected value of the
variable with a hat, e.g., E(M t) = M .1

The Expected Value of the Reduced Form
fo r Base Targeting under CRA

...
M,

n r /
-

„

— Bt +
AO
1 „

** ~ A
o

I -

p l >

+ 8 + p) ~ ------------- l + k ----------

(*-~ p )(r (l+ T ) + e)

+ (rjjL + 5 + a)

( l + k)

N BR,

A
o

ft[(r(l + T) + e) (a — p) + k(rp. + 8 + a)]
l + k

A
o
aid t
a

Ut —

(X —p)/(l + k)

i
w h e r e Ao =

B, +

ft(rp. + S + p)/(l + k) ,

(X —p)(r(l + t ) + e + k)
i+~k

^ +

pW1 + t ) + e + k)l
A0

J Y,
,

The Expected Value o f the Reduced Form
fo r NBR Targeting under CRA

P W l + T) + e + k)/(l + k),.

---------------A
A
o

rtl /
i
\ P(r(1+ T )
e)
p| (rp + -8 i + a) --------- — +---- "I
'The symbol Y, is used here in explicit recognition that the Federal
Reserve must forecast aggregate income in order to control the
nominal money stock.

M, =

----- N B R , +

FEDERAL RESERVE BANK OF ST. LOUIS

OCTOBER 1982

f (X -a )k + (p -a )"|

rtx L

iT k— J

P ((8 + a )( k + e )
(X — p ) ( r ( l +

) +

e +

(r|jL +

8 +

NBR,

Ai
e )(g — p)

(r p . + 8 + c t)k ]

(1 + k ) A j

T (X. a

p )(r(l + t ) + e + k)
I ---------------------------L+ k

(rp, +

8 +

(A. — p ) / ( l + k )

“i

l + k

p)
' id ,

0 ( r jx

a )/ (l+ k )

A
i

o t(X — p ) / ( l + k )
id ,

w h ere A! = ( X- p) [—

]

(rp , +

a ).

The Expected Value of the Reduced Form
fo r Base Targeting under LRA
X
M , =

—

r (l + T )X

(X -p )(k + e)

where A2 = ----- — ;------- 1 8 + p
)

A,
Ut -

Several interesting observations should be made
about the above. First, the discount rate enters the
reduced-form under base targeting only in the NBR
equation. This merely reflects the fact that if the
Federal Reserve is to hit a base target, it must change
NBR proportionally to the change in the discount rate.
Thus, changes in the discount rate cannot alter policy
with respect to either the money supply or interest
rates independent of the base target. This would not be
true of NBR targeting. Under NBR targeting, changes
in the discount rate produce effects on money and
interest rates independent of the target level of NBR.
Second, it is easy to show the conditions under
which the initial change in the equilibrium money
stock is larger under LRA than under CRA. Note that
A M ,

A M ,
D ,_ 2

A
2

— a id ,

p [ ( r ( l + T) +

e ( 8 + p ))/ (l + k )

( l + k)
B,

—

A B,

LRA

CRA

> 0,

implies
rp.X .
i t - 2 ------------------------------: ---------------------------Y ,

> 0.

(X — p ) ( e + k )
.
D t

(X -p )/ (l + k)
-

r(l +

r p .(X -p )/ (l + k )

)( X — p ) / ( l + k )

“

P (k + e )/ (l + k )

a2
A

) + e + k)
+

(r p . + 8 + p)

l + k

A ,

r (1

This condition will hold if

p ( p + 8 )/ (l + k )

J_ R _
A
A9 1

8 + p
K

l+ k
Dt

lt —

(X — p ) ( r ( l +
+

l^t-2

•

»2
A *t-2

< ~

p )(l +

)

;------ ■

l + k

The term on the right-hand side of this inequality will
be positive on the reasonable assumption that I X I> I p I.
.
The condition for the change in the money stock under
the Federal Reserve’s proposal to be less than under
LRA is

A2
p. <

—

(^ ~ P ) T
l+ k

N BR ,

+ (8 + a )
l + k
v
-------------------------------------- B ,

'I .

The Long-Run Monetary Base Multiplier

^ r(x _ a )k+(p_ a )i

r(1+T L
) ------ITiT- J

D ,-

The above system based on LRA is dynamic in that
lagged endogenous variables, Dt_ 2 and it~ 2>appear on
35

FEDERAL RESERVE BANK OF ST. LOUIS

the right-hand side of the reduced-form .2 If we
assume, for simplicity, that Yt = 0 for all t, then it is
easy to show that Mt can be expressed as the following
distributed lag by successive substitution.
\
Xip
M, = — B, + —
Z2
A
tA
£

H
2

where tp = _ [~ + T)(^~p + r^(1 + k) ~
r(1
^
|
(1 + k )

Now if we let the base be constant for all t (i.e., Bf =
Bt - 2 = B t - 4 = • • • = B), we get
M, = —— [l + ip + < 2 + < 3 + . . .]
p
p

Using the formula for summing a geometric series, and
taking the limit, we have
M, = lim

BA.

BX
A-2

Some Observations on the Variability of
Money and Interest fo r Base vs NBR
Targeting and LRA vs CRA
The question naturally arises about the variability of
the endogenous variables under alternative operating
procedures and under different institutional arrange
ments. Unfortunately, the model does not lead to con
clusive answers to these questions. We begin by deal
ing with the question of the variability of the money
stock under base and NBR targeting.
The error terms for the money stock under base and
NBR targeting are, respectively:

fell

The sum on the right-hand side of the above equation is
finite if cp
n-»-0 as
Under this condition,
M, =

slowly or rapidly depending on whether IcpI is close to
one or zero. Furthermore, under the general condition
that < < 0, the system will oscillate toward its long-run
p
equilibrium, as indicated by Laufenberg.3

Xtp2
+ "7~ B,_4 + . . .

b ,_ 2

OCTOBER 1982

X (r(l+ x) + e - l )

+ e + k)

X(r(l + T ) + e)
“ T T T T T

a-----(1 + k) A!

X ,
u«

7 “

(u bt -

rU rt -

U et)

A2 | t i r(l + t)(X —p) + r p . ( l
1+
(1 + k) A2

+ k) j

A2 (1 + k) + r ( l + t )(X - p) + (rp,)(l + k)

Substituting in for A2, we get
M, = B

(1 + k) A0

L l-ip .

a n d

= B X

7“ (ruT + uet)
t
A0

(l + k)(rp, + 8 + p) — p(r(l +

fed

Substituting in for cp, we get
Mt

X ,

----7T77T7
(1 + k) Ao

+ k) (rp. + 8 +

- p(r(l + t ) + e)

(1 + k) A,

If we denote the variance of money under base and
NBR targeting by V(Mt ) and \ (Mt ), respectively,
B
/ N
and if we assume the individual error terms are inde
pendent of each other and through time, we get

(X —p ) [ r ( l

) + k

+ e]
+

- D l 2

( 8 + p + r p .)

1 + k

Comparing this multiplier with the base multiplier
under CRA, we see that they are identical. Thus, LRA
makes the system explicitly dynamic but does not
affect the long-run equilibrium.

-----(1 + k) A,0
J

[ “t ] (2?+
r<
T

j^(l + k)(rp, + 8 + p) —p(r(l +
(1 + k )

The above solution, however, does require the sta
bility condition I(pi < 1. The system may converge

t ) + e + k ) "1 " ,
--------------J Cn
T

and
V(M,

2The inclusion ofit_ 2 in these equations is based on the assumption
that changes in time deposits in period t —2 induced by changes in
the market interest rate affect current required reserves. This is a
highly questionable assumption (see below). Therefore, it might be
more reasonable to use the reduced-form equations obtained by
letting p. = 0.

3See Daniel E. Laufenberg, “Contemporaneous Versus Lagged
Reserve Accounting, ’’Journal o f Money, Credit and Banking (May
1976), pp. 239-45.

FEDERAL RESERVE BANK OF ST. LOUIS
T (1 + k)(rp. + 5 + a) - p(r(l + t ) + e) "I ~

(1 + k)

J C
T
n

A number of interesting observations can be made
from the above. First, random shocks to currency, uct,
move money in opposite directions under base and
NBR targeting if 0 < t( 1 + t) + e < 1. Under this
condition, a random increase in currency will reduce
the money stock under base targeting and increase it
under NBR targeting. Furthermore, the magnitude of
this shock on the money stock will be larger under base
targeting. This can be seen by noting that the absolute
value of the coefficient on uct under base targeting will
be larger than under NBR targeting if r(l + t ) + e < 1/2
and if IAo I < IAj I. Thus, the variance of money associ
ated with random changes in currency will be larger
under base targeting.
The variance of money associated with other supplyside shocks will be smaller under NBR targeting if
IA0 1< IAx I. This condition requires (\k + p)/(l + k) >
a. Standard estimates of these parameters suggest that
this condition will hold. Thus, supply-side shocks will
have a greater impact on the money supply under base
targeting. Note, however, that if the discount rate
were tied so that, effectively, a = 0, the above condi
tion would not hold, and more stable monetary control
could be achieved through base targeting.
Second, the variance of money associated with ran
dom shifts in money demand can be seen to be larger
under NBR targeting if the money supply schedule is
flatter. This can be seen by denoting the slopes of the
money supply under base and NBR targeting by 1/iJjb
and l/vJiN respectively, where
,
>B =
I<

p(r(l +

Third, random shocks in depository institutions’
borrowing have no impact on money under base
targeting, but they do under NBR targeting. If the
variance of these shocks are larger— relative to the
variances of currency, excess reserves and time de
posits— the net effect of supply-side shocks from all
sources could be larger under NBR targeting. Thus, no
general conclusion about the variability of money
under base and NBR targeting can be reached. (The
outcome depends on the magnitudes of o f , o f ,and o f
relative to o f , and on the magnitude of o f ,.)

The Variance of Money and Interest Rates
under CRA and LRA
Now consider the variance of money and interest
rates under CRA, LRA and the Board’s proposal for
CRA. Only the case of base targeting will be dealt with;
however, the results hold for NBR targeting as well.
For simplicity, assume that the individual structural
errors are independent (relaxing this assumption
makes determining the relative variability under CRA
and LRA more difficult). It is easy to show that the
error terms for the equilibrium money stock under
CRA and LRA, respectively, are
\ (r(l + T) + e — 1)
X
------,, , u . -------- U ct- ~7~ (rUrt + Uet)

(1 + k) A0

(1 + k)(rp, + 8 + p) —p(r(l + t ) + e + k)

(1 + k) A„

and
X(e - i )

X
Uct -

~ r (m T t-2 + Uet)

(1 + k) A* ct

+ e + k) — (1 + k) (r[x + 5 + p)

(1 + k)(8 + p) —p(e + k)

r (l + T) + e + k

(1 + k) A*

If we let V(Mt) and V(Me denote the variance of
t)
money under CRA and LRA, respectively, we obtain

and
p (r(l+ T )

OCTOBER 1982

+ e) —

( l + k)(rp.

r (l +

+ e

Now note that the money supply schedule is flatter
under NBR targeting if i| < iJiN, which, in turn,
iB
p(r(l + t) + e) —k(r|x + 8)
requires ----------------------------------> a. We now note
r(l+ T ) + e + k
that the coefficients on the <jfrl term of the above expressions are simply

(1 + k)(rp. + 8 + p) —p(r(l +

I 2

^ ^ ■ , i = B and N. It is
j
i

easy to see that the general conclusion about the sensi
tivity of the money stock to random changes in the
demand for money hold if i)jb < viN.
|

[£]
[

(1 + k)

+ e + k)

and
+ ca

FEDERAL RESERVE BANK OF ST. LOUIS

| ( l + k)(S + p ) ~ p(e + k)~|
(1 + k) A ,
Jm
-

It is easy to see that V(M,) < V(M|). This can be
done by noting that (r(l + T ) + e - 1 ) 2 < (e —l)2 for
(r(l + t ) + e) < 1, and that (A0)2 > (A2)2. Furthermore,
it is clear from our previous analysis that the coefficient
on the variance of money will be smaller under CRA if
the money supply schedule is flatter under LRA than
under CRA.4
Note that this conclusion depends on including
u-rt — in the error term for the equilibrium money stock
2
under LRA. If the Federal Reserve could identify ran
dom changes in time deposits and make corresponding
compensatory changes in the reserve aggregate,
changes in uTt_ 2 would not affect the current money
stock. Thus, the variance of money under LRA would
be smaller than indicated above by5

OCTOBER 1982

Clearly (A3)2 > (A2)2. Furthermore,
[(1 + k)(8 + p) —p(r + e + k)]2 < [(l + k)(8 + p)-p(e + k)]2.
Thus, V(M’’) < V(M{). The move from the current
t
system of LRA to the Board’s proposed system of CRA
should reduce the variance of the money stock. This
conclusion is true whether o f is included or excluded
from these expressions.

The Variance of the Interest Rate
Denote the variances of the interest rate with re
spect to CRA, LRA and the Board’s proposal as V(it),
V(iJ) and V(iJJ), respectively. Then,
r(r(l + T + e + k - l ) ] 2 ,
)

V/-1

V ■L
,">

L
r

In this instance, it is impossible to say whether V(M?)
is larger or smaller than V(Mf). Furthermore, the
structure of other equations might change with a
change in reserve accounting.6
Ifwe denote the variance of money under the Feder
al Reserve’s proposal as V(Mj), it is easily shown that

J *

+ [ r r f i l +
r

[f]‘*

0 +1.4,

v
(i,';

~ [r(l + ) + e + k]~| 2 ,
(1 + k)A0
J Cm
T
t

,i2

(1 + k)A2J 1

-(1 + k)A2-

2
< ?)
T

—(e + kfl
r u,
(1 + k)A2J1 u'
r+e - l l
r u.
(1 + k)A3J i °

V(if>

-d + k )A 3

x; + a,2)

—(r + e + k) "I
(1 + k)A3

V(M?

M-) = [ t t t S x j ] ^ + [ i

] (r2a?+o;2

(1 + k) (S + p) —p(r + e + k) ~j ~
d + k )a,
J

(Jn
'

where

(X —p)(r + k + e)
=
r+k—
+ (8+p)'This condition is simply that
„

As with the comparison of the variances of money, a
comparison of V(iJ) and V(iJ) depends on whether the
term

(1 + t ) (8 + p)

(e + k)

is included or excluded under LRA. With respect to
the variance of interest rates, however, there is an
additional complication; it is not immediately clear
whether

•

5While the general conclusion that V(MC need not be less than
,)
V(M|) will remain valid under alternative specifications of this
model, this specific result may change. For example, this expres
sion would change if there were different reserve ratios on time and
transactions deposits and if excess reserve holdings depended on
both time and transactions deposits.
6This need not be the case, however. If there were sudden changes
in reserves due to deposit shifts AD, the amount of market opera
tions necessary to restore a bank’s reserve position would be (1 —r)
AD under CRA and AD under LRA. This need for greater market
activity under LRA could offset any effect on excess reserves associ
ated with greater certainty of required reserves.

-r

L (1+ k)A0 J
( l + k)Ao

— (r (l + T) + e +

(1 + k)A„

k)~|2 > [ - ( e + k )]2
J < L (l + k)A2J

A little algebra shows that the condition requires
(1 + t ) (8 + p)

(e + k)

This is the condition required for the slope of the
money supply schedule to be steeper or flatter under
CRA than under LRA. This adds another element of

ambiguity to the determination of the relative variance
of interest rates under CRA and LRA.
A similar problem makes the determination of the
relative magnitudes of V(il> and V(i't) ambiguous. This
t)
is seen by noting that V(i[’) < V(it) if

r-(r+ + )T . . r ~ +k i22
ek ,
(e )
~
L (l + k)A3 J

L(l + k)A2J

Following the same procedure as above, it can be

shown that this condition will hold only if
r ( 5 + p) >

Given the restrictions on the signs of these parameters,
this cannot hold. Therefore, the last term in the ex
pression for V(i^) is strictly larger than the last term in
the expression for V(i*). Thus, it is indeterminate
whether the Board’s proposal for CRA will increase or
reduce the variability of interest rates from the present
system.