The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
[COMMITTEE PRINT]
THE IMPACT OF THE FEDERAL RESERVE SYSTEM'S
MONETARY POLICIES ON THE NATION'S ECONOMY
STAFF REPORT OF THE
SUBCOMMITTEE ON DOMESTIC MONETARY POLICY
OF THE
COMMITTEE ON BANKING, CURRENCY AND HOUSING
HOUSE OF REPRESENTATIVES
94th Congress, Second Session
DECEMBER 1976
This report has not been officially adopted by the Subcommittee on Domestic Monetary Policy
and may not therefore necessarily reflect the views of its members
Printed for the use of the Committee on Banking, Currency and Housing
U.S. GOVERNMENT PRINTING OFFICE
WASHINGTON : 1976
COMMITTEE ON BANKING, CURRENCY AND HOUSING
HENRY 8. REUSS, Wisconsin, Chairman
ALBERT W. JOHNSON, Pennsylvania
LEONOR K. (MRS. JOHN B.) SULLIVAN,
J. WILLIAM STANTON, Ohio
Missouri
GARRY BROWN, Michigan
THOMAS L. ASHLEY, Ohio
CHALMERS P. WYLIE, Ohio
WILLIAM S. MOORHEAD, Pennsylvania
JOHN H. ROUSSELOT, California
ROBERT G. STEPHENS, JR., Georgia
STEWART B. McKINNEY, Connecticut
FERNAND J. ST GERMAIN, Rhode Island
HENRY B. GONZALEZ, Texas
JOHN B. CONLAN, Arizona
JOSEPH G. MINISH, New Jersey
GEORGE HAN SEN, Idaho
FRANK ANNUNZIO, Illinois
RICHARD T. SCHULZE, Pennsylvania
THOMAS M. REES, California
WILLIS D. GRADISON, JR., Ohio
JAMES M. HAN LEY, New York
HENRY J. HYDE, Illinois
PARREN J. MITCHELL, Maryland
RICHARD KELLY, Florida
WALTER E. FAUNTROY,
CHARLES E. GRASSLE Y, Iowa
MILLICENT FENWICK, New Jersey
District of Columbia
RON PAUL, Texas
LINDY (MRS. HALE) BOGGS, Louisiana
STEPHEN L. NEAL, North Carolina
JERRY M. PATTERSON, California
JAMES J. B LAN CHARD, Michigan
CARROLL HUBBARD, JR., Kentucky
JOHN J. LAFALCE, New York
GLADYS NOON SPELLMAN, Maryland
LES AuCOIN, Oregon
PAUL E. TSONGAS, Massachusetts
BUTLER DERRICK, South Carolina
PHILIP H. HAYES, Indiana
MARK W. HANNAFORD, California
DAVID W. EVANS, Indiana
CLIFFORD ALLEN, Tennessee
NORMAN E. D'AMOURS, New Hampshire
STANLEY N. LUNDINE, New York
PAUL NELSON, Clerk and Staff Director
WILLIAM P. DIXON, General Counsel
MICHAEL P. FLAHERTY, Counsel
GRASTT CREWS II, Counsel
ORMAN S. FINK, Minority Staff Director
GRAHAM T. NORTHUP, Deputy Minority Staff Director
SUBCOMMITTEE ON DOMESTIC MONETARY POLICY
STEPHEN L. NEAL, North Carolina, Chairman
JOHN B. CONLAN, Arizona
JOSEPH G. MINISH, New Jersey
GEORGE HANSEN, Idaho
MARK W. HANNAFORD, California
WILLIS D. GRADISON, JR., Ohio
JAMES J. BLANCHARD, Michigan
LEONOR K. (MRS. JOHN B.) SULLIVAN,
Missouri
CLIFFORD ALLEN, Tennessee
NORMAN E. D'AMOURS, New Hampshire
SUBCOMMITTEE STAFF
ROBERT E. WEINTRAUB, Staff Director
CARL A. MINTZ, JR., Research Statistician
LINDA L. LORD, Research Assistant
DAVIS O'CONNELL COUCH, Counsel
MIRIAM W. RAKOW, Secretary
(II)
LETTER OF TRANSMITTAL
DECEMBER 1,
1976.
Transmitted herewith for use by the Subcommittee on Domestic Monetary
Policy, the full Committee, and the Congress is a staff report entitled, "The
Impact of the Federal Reserve System's Monetary Policies on the Nation's
Economy."
When I became Chairman of the Subcommittee last spring, I asked the
staff to determine as precisely as possible the effects that the Federal Reserve's
money supply policy has on our economic performance, and to analyze also the
parts played by fiscal policy, import prices and other factors. Although this
research will be refined and extended in the future, the results which have been
obtained to date are sufficiently important to be made public. They provide new
validated insights into the determination of yearly inflation rates and yearly
changes in our real economic growth. In the next Congress, we intend to add
to the understanding which we have now acquired about the inflation process
and cycles in real GNP, and to broaden the analysis to cover employment
trends. In addition, we plan to investigate credit flows and interest rate trends
and patterns. Our aim is to provide background information on the behavior
and determination of the variables which are central to the legislative and
oversight functions of the full Committee and its several Subcommittees.
Now, let me review the results of our research to date. It was carried out
in two phases. Phase one looked at inflation. Phase two studied recessions and
expansions in real GNP.
As a first step, the staff graphed post Korean War percentage changes in
the Consumer Price Index (CPI) and money supply, contemporaneously and
at lags up to 36 months to find the best inflation/money supply relationship.
Money supply was measured conventionally by M-l, which consists of publicly
held coin, currency and demand or checking deposits. Percentage changes were
computed between the same months from one year to the next. This initial
research showed that we can track and predict inflation in 12-month periods
fairly well just by looking at the percentage change in money supply occurring
23 months earlier in the same 12 months. For example, inflation from June
1975 to June 1976 is predicted fairly closely by the July 1973 to July 1974
percentage change in M - l .
The research was then extended to cover the full post World War II period.
Yearly averages were used instead of monthly data and the analysis was
broadened to take into account factors other than expansion of money supply
which are commonly believed to affect inflation. This completed the first phase
of the analysis. Phase two analyzed recessions and yearly percentage changes
in real or constant dollar GNP during the 1953-1975 period. The effects of the
Federal Reserve's money policies and various other factors, including past
inflation, were studied.
The results now in hand explain much (about 90%) of both year to year
percentage changes in the CPI in the 1947-1975 period and real GNP in the
1953-1975 period. In summary—
ON INFLATION
• On average, consumer prices rose 1% two years after each yearly
increase of VA% in M-l money supply. More than 70% of the rise in the CPI
that we experienced in the inflationary 1966-1975 decade can be attributed to
excessively rapid money growth. But money supply is not all that matters.
• Inflation can be triggered and fueled by velocity increases. A buying
spree that started right after North Korea invaded South Korea on June 25,
1950, caused consumers' prices to jump sharply in 1951 over the 1950 average.
(in)
IV
Apart from the Korean War years, however, changes in velocity were not a
significant inflationary factor in the 1947-1975 period.
• Inflation can be imported from abroad. Increases in import prices contributed substantially to inflation in the 1973-1975 period. The staff estimates
that increases in import prices raised U.S. consumer prices by 1%% in 1973,
5% in 1974 and 1% in 1975.
• In 1949, large increases in agricultural and raw material supplies appear
to have contributed to the fall in consumer prices. But there is no evidence
that special supply factors affected inflation in any other year.
• Nor is there evidence that the state of the economy—whether measured
by unemployment rates or changes, or by production trends—affected the
inflation process in the period since World War II. The staff explains this as
follows. When production rises and unemployment falls, prices tend to rise
because markets tighten. However, offsetting this is that increased supplies
must compete for the same money supply. This tends to lower prices. Vice
versa, in recessions, markets are looser but supplies are down, and the price
effects of the two cancel one another.
• On average, Federal Government deficits did not have significant
inflationary impact during the post World War II period. Though this finding
is certain to be controversial, there is a plausible explanation. Big swings in the
deficit are largely induced swings. Thus, the deficit is as much or even more
effect as it is cause. Moreover, with a given money supply, financing deficit
spending by selling bonds causes private spending to fall roughly in proportion,
dollarwise, to the increase in Government spending. Any residual spending
effect is too small to be a significant inflationary factor.
The staff also investigated the inflationary impact of the High Employment deficit. It, too, was found not to be significant. A plausible reason is that
to the extent the economy is stimulated by High Employment deficits, government revenues rise and the deficit vanishes. This happens very quickly in
inflation because of our income tax structure. In other words, High Employment fiscal stimulus erodes before enough time passes for it to generate inflation.
• Finally, and perhaps most important, I want to stress that the evidence
which the Subcommittee staff has assembled shows that our economy's pricing
system is not explosive. It is stable. On average, since World War II, changes
in the rate of inflation one year have carried over to the next year with between
yA and y2 power. This means that a 4% rise in consumer prices this year would
bring further increases in the cost of living of l % - 2 % next year and % % - l %
two years from now. After two years, however, no matter how large the initial
jump, the momentum dissipates and the subsequent changes in inflation cycle
around zero.
The carry-over effect reflects that administered pricing and collective
bargaining are importantly influenced by last year's price trends. However,
though this influence is significant, it is far less than what it would take to make
our economy's pricing system explosive. There is a momentum in inflation, but
it is not explosive, cumulative, or even constant. It comes to nothing after
two years. Rather, no matter what triggers it and however rapid it becomes,
money supply changes are decisive to whether inflation accelerates or is checked
and subsides.
RECESSIONS
Both money supply and velocity play important parts in recessions and
recoveries. Specifically—
• Slowing money growth sharply for any prolonged period weakens the
economy and increases the risk of recession. For a time, however, velocity
increases sustain expansions even after money growth slows. But inevitably if
money growth is retarded sharply for very long, confidence is undermined, the
trend in velocity turns down, and recession follows.
• Once underway, recessions are aggravated by continuing to retard
money growth.
• Money supply expansion during and immediately after recession promotes recovery. Recession tends to last as long as the trend in velocity is down.
But if money growth is kept commensurate with our production potential,
inevitably confidence builds, the velocity trend turns up, and recovery builds.
REAL
GNP
The staff also quantified how yearly percentage growth of real or constant
dollar GNP was affected by contemporaneous changes in money supply, past
inflation and other factors as here described during the 1953-1975 period.
It was found that the springboard for this year's real growth performance
was built on past inflation performance and unemployment. Specifically
• For each 1 % unemployment a year ago, real GNP increased during the
1953-1975 period, on average, by 1.2% in the current year. In other words, in
the post Korean War period, there has been a natural tendency for our economy
to rebound from recessions.
• A jump in the rate of inflation primed small increases in real GNP two
and three years later. But the stimulus provided by inflation jumps was minor
and dissipated quickly. We can't inflate our way to full employment. To the
contrary, the results warn that trying to do so can make things worse. For each
1% rise in the CPI in the current year, on average, real GNr decreased 1% in
the next year.
In this connection, the report tested whether real GNP is affected differently by increases in import prices than by increases in domestic prices. The answer
is "not significantly." Increases in import prices cause the CPI to rise concurrently, and a year later real growth falls the same as it would following a rise
in domestic prices.
• During the period which the staff studied, 1953-1975, both monetary
and fiscal policy affected our real growth performance contemporaneously.
(1) For each 1% increase in money supply, real GNP increased on average
by -73% the same year. This does not mean, however, that we can increase real
growth in any permanent sense by printing money. Because of feedback from
money's inflationary effects, the staff found that the rise in real GNP growth,
which initially accompanies increased money growth, recedes quickly after
three years and doesn't differ significantly from zero after five years.
(2) The High Employment deficit was found to have only a small and
marginally significant impact on real GNP. On average, a deficit equal to 1%
of potential GNP generated a K% rise in real GNP the same year. Today,
this means $15-$20 billion stimulus is required to raise real growth just K%
Let me turn now to the question of how well the Subcommittee's results
fit actual experience. The answer, as shown by Exhibits 5 and 8, is "very well."
Exhibit 5 charts actual year to year inflation rates from 1947 to 1975 and
inflation as predicted by the cornerstone statistical equation of our research.
Exhibit 8 does the same thing for actual and predicted yearly percentage changes
in real GNP. For convenience, Exhibits 5 and 8 are duplicated here. There
would appear no need for me to elaborate on what they show.
Only time will tell whether our results will continue to closely track
changes in inflation and real GNP. Given the information that we now have
about the variables which were used to estimate changes in the CPI and real
GNP in the research, the staff points out that it is a 2 to 1 bet that in 1976 the
CPI will rise 3.1%-5.5%, and also 2 to 1 that real GNP will increase 3.9-5.7%.
Further, in 1977, conditional upon import prices rising only moderately, it is
a 2 to 1 bet that inflation will decline further to 2.6%-5.0%. Finally, if 1977's
M-l growth and High Employment deficit repeat 1976's money growth and
fiscal policy, real GNP will repeat this year's predicted 3.9%-5.7% growth.
In regard to these predictions, however, let me stress that they assume
that the M-l statistics now published for 1975 and 1976 are correct. A recent
Federal Keserve study indicates that M-l growth may be understated by
.6% in 1975 and 2.0% in 1976. To the extent that this is true, the forecasts
would have to be revised upwards, as follows: 1977 inflation by .4% and 1976
real GNP growth by 1.5%.
CURRENT POLICY IMPLICATIONS
We have two major policy tools for attempting to affect economic performance. One is monetary policy, the other is fiscal policy. The staff report
finds that fiscal stimulus does not have significant impact on the course of
inflation, perhaps because the stimulus itself dissipates quickly; but does have
VI
a small, marginally significant impact on real GNP, perhaps because in the
ast we have achieved such stimulus largely by cutting taxes, which spurs
oth investment and work. Monetary policy, on the other hand, powerfully
influences real growth in the short run and inflation in the long run. Given the
current need both to check and reduce inflation and to revive and sustain the
faltering recovery, what can be done?
• With respect to monetary policy, the report demonstrates that in the
context of economic conditions as experienced since World War II, faster money
growth generates both increased real GNP growth and accelerated inflation.
The good news comes quickly, in the same year as money growth is increased.
Inflation accelerates only after a lag. Worse, the rise in inflation causes real
GNP to recede; and five years after money growth accelerates, the growth of
GNP, for all practical purposes, is back where it started. Worse still, we are
left with the problem of reducing inflation. To do it requires reducing money
growth, which risks recession.
Thus, although we now need additional stimulus to put new life into the
recovery and reduce unemployment, the staff report indicates that there is
little point and considerable risk in providing for substantial extra monetary
stimulus. In this regard, I recognize that many people argue that our capacity
to produce is so underutilized that we needn't fear that faster money growth
will lead to higher inflation. They advise more rapid money growth on the
ground that, at the present time, only good news would come from stepping up
money growth. I can't agree. There are awesome bottlenecks which warn
against following such advice. We have no spare capacity in food, in energy, or
in medical care to cite three important sectors where increased money growth
surety would cause higher prices.
Fortunately, the evidence which staff has assembled indicates that we can
fight inflation and unemployment at the same time. It shows that for each 1%
drop in the rate of inflation, we can achieve now, we can expect, on average, a
1% rise in the rate of real GNP growth next year. Thus, bringing inflation
down will increase production and employment and reduce unemployment. In
fact, initially a 1% drop in the rate of inflation has a larger effect on output
and employment than a 1% rise in M-l growth. Thus, there is no need to
choose rapid money growth; moderate growth will achieve a better economic
performance all around.
Specifically, the report recommends 4%-6% M-l growth at the present
time—or about the same as the Federal Reserve currently plans to provide, i.e.,
4%%-6%% between the summer of 1976 and the summer of 1977. Together
with further deceleration of inflation, this policy will provide the foundation for
sustained economic expansion.
• Filially, the staff report recommends $15-$20 billion of new fiscal
stimulus to give the economy a small, extra push now. This recommendation is
made in recognition that the recovery has slowed down and, however firm the
underlying monetary foundation is for sustaining expansion while reducing
inflation, economic growth now needs a bit of a prod. In recommending fiscal
stimulus, the staff report points out that such stimulus dissipates quickly and,
therefore, is not likely to be inflationary. The report does not choose between
tax cuts and additional spending. It points out, however, that tax cuts can
stimulate production by providing new incentives to work and investment. At
the same time, the staff emphasizes that structural labor market problems,
which have been growing for years and persist even during vigorous expansions,
cannot be solved by tax cuts. They require judicious spending programs to
upgrade the skills of the hard-core unemployed.
The report was prepared by Robert Weintraub, Staff Director, with the
assistance of Linda Lord and Carl Mintz.
E
STEPHEN
L. NEAL, Chairman, Subcommittee on Domestic Monetary Policy.
EXHIBIT 5
CONSUMER PRICE
INDEX
12 .a
YEfiR TO YEflR PERCENT CHRNGE
I ACTUAL 4 FORECAST;
K643 M1YP (t-2) -I-6Dummy 4-1 .
PIMYPW+.2A8[pPIYP (t-1)-CPIYT (t-
10.0
3.0
B.O
-2.0
»E7 *58 ' S 3 f SO
Lines from the time axis delineate recession periods.
EXHIBIT 8
GROSS NRTIONRL PRODUCT ( I N-; CONSTRNT $)
PERCENT CKflNGE YEflR TO YERR
flCTUfil
RiNO P R E D I C T E D
F o r e c a s t « - 2 . 5 3 + .728 M1YP + . . 5 5 2 BY' + 1.217 UNY(t-l)
- 1 . 0 1 1 C P I Y P ( t - l ) + .393 [ C P I Y P ( t - l ) - CPIYP(t-2)J t - 1
•7.0-
-7.0
fl.O-
-6.0
6-0
-6.0
-4.0
-3-0
UJ
U
-2.0
•1-0
-0.0
- —1.0
-Z.0--
-z.o
63 I 64 i65 r60^57 '
J
* 04 U S * OD
:|
07
l
00
YEftRLY OflTR.
Lines from the time axis delineate recession periods.
l
00
J
70
J
71
-3.0
TABLE OF CONTENTS
Page
Background and summary
Part I: The Money Supply Roller Coaster
The Roller Coaster
Controlling Money Growth
Part I I : Inflation
Initial Analysis
The Initial Experiment
Further Analysis
Results
.
Table 4
.
Korean War Buying Trends
Table 5
Subperiods
Table 6
Table 7 . .
Tables
Predictions
Contributions
Part III: Recession
Velocity
Determinants of Changes in Real GNP
Series 1
Series 2
Inflation Feedback
Predictions and Contributions
Predictions
Policy Implications 1977-1978-----Some Long-Run Policy Implications
Part IV: Policy Under H. Con. Res. 133
Before the Resolution
Policy Since March 1975
Closing Remarks
Importance of Controlling Money Growth
.
1
3
3
5
7
8
0
17
18
18
19
21
22
22
23
25
27
31
33
33
35
37
37
38
40
40
41
41
45
45
47
47
49
TABLES
TITLE
Number
1 Unemployment Rates and Changes in the Consumers7 Price Index
(CPI), Annual Observations, 1947-1975
.
2 Regression Results
3 M-l Percentage Growth Between Corresponding Quarters of
Adjacent Years, 1961-1965
4 Regression Coefficients and Standard Errors
5 Regression Coefficients and Standard Errors
6 Regression Coefficients and Standard Errors
7 Regression Coefficients and Standard Errors
8 Regression Coefficients and Standard Errors
9 Predicted and Actual Inflation Rates Using the 1947-1975 Series
3 Table 4 Regression
..10 Actual Inflation Rates and Predictions Using the Adjusted 19471970 Table 5 Regression
_.
(IX)
80-401 O - 76 - 2
1
9
17
18
21
23
24
25
28
30
Number
Page
11 Contributions to Inflation, 1947-1975
12 Recession Dates and Major Characteristics, 1947-1975, and the
Great Depression of the 1930's
13 Regression Coefficients and Standard Errors
14 Direct and Feedback Effects of 1% Yearly M-l Growth
15 Real GNP Predictions and Contributions
31
33
37
39
40
EXHIBITS
TITLE
1
2
3
Yearly Averages of M-l Percentage Changes, 1945-1975
.
Yearly Averages of M - l and CPI Percentage Changes, 1916-1934.
Scatter Diagram of Percentage Changes in CPI Occurring 2 Years
After Changes in M-l Growth
4
Yearly Money Supply Changes and Yearly Changes in the Cost
of Living Occurring 23 Months Later, 1954-1976
4(a) Year to Year Percent Changes in CPI and M-l, 1954-1976
5
Year to Year Percent Changes in CPI, Actual and Forecast,
1947-1975
.
__6
Percent Change in M-l Money Supply Year to Year and Recession
Periods, 1948-1976
7
Money Velocity (Current GNP/M-1) Percent Change, Year to
Year, and Recession Periods, 1948-1976
8
Gross National Product (In Constant $) Percent Change, Year
to Year, Actual and Predicted, 1953-1975
9
Realized Levels of M-l and the Federal Reserve's Target Corridor, March 1975-October 1976
4
10
11
14
15
29
34
36
42
48
APPENDIXES
TITLE
1
2
3
4
5
Correlation Matrices
MI YP Lag Correlation Analysis 1945-1975
,--.--Cochrane-Orcutt Regressions
Regressions Using M-l as the Dependent Variable, 1946-1975
Inflation Regressions Using M-l Ahead as an Independent Variable,
1946-1975--..
.
6 Regressions Using M-2 in place of M-l
7 Infla tion Regression Separating the Lagged CPI YP Terms Bracketed in Other Regressions. _
53
54
55
56
57
58
59
In March 1975, the Congress passed House Concurrent Resolution 133*
Under the Resolution, the Federal Reserve discloses money growth rate targets
for the next year. These estimates are updated quarterly at hearings before
the House and Senate Banking Committees. The Resolution further expresses
the Sense of Congress that the Fed maintain long-run money growth—"commensurate with the economy's long-run potential to increase production, so
as to promote effectively maximum employment, stable prices and moderate
long-term interest rates."
This report is intended to provide the full Committee, the Congress and
the public with information and analysis of the Fed's money policies, thereby,
permitting better oversight and greater accountability of these critical
operations.
BACKGROUND AND SUMMARY
Measured by the twin goals of full employment and stable prices, our
economy has been far healthier since World War II than before it. But from
1973 to 1975, we slipped significantly.
In 1973, the cost of living, measured by the average of the consumers'
price index (CPI) for the entire year, rose 6.23% from the 1972 average. The
rise was 11.04% in 1974 and 9.13% in 1975. Unemployment followed, though
not in lock-step. Unemployment averaged 4.85% in 1973, 5.62% in 1974 and
8.48% last year.
Compared to earlier post World War II years, that record is abysmal.
With respect to inflation, the yearly CPI rose more than 6% only in the immediate post-war years, 1946-1948, and in 1951 immediately after the Korean
War began. Unemployment never averaged more than 7% until 1975. The
complete record is set forth in Table 1.
TABLE 1.—UNEMPLOYMENT RATES AND CHANGES IN THE CONSUMERS' PRICE INDEX (CPI), ANNUAL OBSERVATIONS, 1947-75
CPI percent
changes Unemployment
Year
1947
1948 .
1949
1950 .
1951
1952
1953
1954
1955
1956
1957
1958
1959 1960
1961
.
.
. . .
14.47
7.67
-.99
1.06
7.94
2.28
.77
-.35
-.26
1.47
3.40
2.73
.92
1.51
1.07
Year
3 90
3.75
6.05
5.21
3.28
3.03
2.93
5.59
4.37
4.13
4.30
6.84
5.45
5.54
6.69
1962
1963
1964
1965
1966
1967
1968
1969 .
1970
1971 .. .
1972.
1973 ..
1974
1975..
CPI percent
changes Unemployment
1 17
1.25
1 32
1.59
2 99
2 78
4 21
5.42
5.90
4.26
3.31
6 22
11.04
9.13
5 57
5.64
5 16
4.51
3 79
3.84
3 56
3.49
4.98
5.95
5.60
4.85
5.62
8.48
So far in 1976, there has been improvement. Unemployment has dropped
below 8%. The per annum rise in the cost of living has decelerated to the 6%
level.
By historical standards, these are still very high unemployment and inflation rates. It is important to ask whether the recent slip in our economic performance is permanent. Some believe that it is; that structural changes in our
economy require us, from now on, to accept high inflation or high unemployment, if not both. We recognize that there
have been important structural
changes in our economy since the 1930Js. But these changes do not require
accepting either high unemployment or high inflation as a permanent feature
of economic life. Both the current level of unemployment and rate of inflation
are unacceptably high. We can do better on both counts. To do so, we have
only to recognize how the economy went wrong in the past and be willing to
learn from those errors.
(l)
A major part of our current levels of unemployment and inflation reflect
past mistakes in public policy. Such mistakes will have to be avoided in the future
if we are to achieve full employment and stable prices on an enduring basis. We
recognize that public policies (monetary, fiscal and regulatory) are not the only
factors that shape our economy's price and employment performance. Other factors such as collective bargaining, administered pricing and import prices
always influence and sometimes dominate economic performance. But the
pattern of exogenous shocks is random, not recurrent, and their influence
dissipates in time. Thus, in general, public policies dominate our economy's
price and employment trends.
This report focuses on the part that monetary policy has played and should
play in our economy's performance. Though it isn't all that matters, monetary
policy matters very, very much. In summary, the record shows that throughout
the post World War II period, time and again excessively sharp, prolonged
cuts in money growth helped to turn economic expansions into recessions.
Between 1965 and 1973, a new twist was added. Sustainable expansions were
turned into unsustainable inflations before monetary policy moved to cut them
down. In this regard, a full Committee staff report issued in August 1973, before
the oil price boost, has proved to be sadly prophetic. That report warned:
Beginning early in 1972, money supply growth was again accelerated to 8 percent per
year and in the latest quarter, Spring 1973, has zoomed to nearly 11 percent per year.
Inflation, which had tapered off, has resumed and prices are now advancing faster than
before. Interest rates are skyrocketing. Financial disintermediation is again underway. The
future is even more bleak. Soon the Federal Reserve, in order to end the inflation it has fueled,
will again cut back the growth of the money stock too sharply and too long. Interest rates
will rise to still unseen heights. Credit will become nearly unavailable to home buyers, consumers, small business and local governments. Economic activity will decline. Unemployment will rise.
We know now that this bleak monetarist scenario came true. Money growth
was decelerated beginning in the second half of 1973, and sharply so after mid1974. Caught in a vice between accelerating inflation and decelerated money
growth, interest rates soared. In the summer of 1974, the three-month Treasury
bill rate averaged 8.3%, the Federal funds rate 12.1%, the rate on Moody's
Aaa corporate bonds 9.0% and the FHA new home mortgage rate 9.9%. Then,
in the six months from September 1974 to March 1975, industrial production
fell more than 15 percent. In May 1975, unemployment reached 8.9%.
Despite such monetarist scenarios coming true, monetary policy remains
a mystery to most Americans. As stated by Representative Stephen L. Neal,
Chairman of the Subcommittee on Domestic Monetary Policy in the hearings
held by this Subcommittee described below,
Most Americans do not know how important money policy is to the prices of the
goods they buy, the interest rates they pay, their job opportunities, wages and profits.
Understanding these relationships is essential for objective analysis and
successful oversight of current monetary policy. To increase public understanding, the Subcommittee held hearings on June 8, 9, 10 and 24, 1976 to
determine
how the Federal Reserve's policies affect our economy's performance.1 These hearings have been printed and are available for public distribution upon request. In addition, Chairman Neal asked the Subcommittee
staff to analyze the data on the relationships between monetary policy and
economic performance. The results have now been compiled. Together with
the hearings' testimony, they provide a record which should prove helpful in
the monitoring and oversight of the conduct of monetary policy. The record
is reviewed in this report, divided into four parts. Part I discusses money
growth trends. Part II presents the results of our computer analyses of U.S.
inflation. Part III focuses on recessions and unemployment. Part IV evaluates
monetary policy under House Concurrent Resolution 133.
i See, "The Impact of the Federal Reserve's Money Policies on the Economy."
PAKT I
T H E MONEY SUPPLY ROLLER COASTER
Some observers judge monetary policy by trends and patterns in interest
rates, others by money supply. We use the basic money supply, M-l, as our
measuring rod. Monetary aggregates are more reliable indicators of the thrust
of monetary policy than interest rates. "Interest rates," as Federal Reserve
Board Governor J. Charles Partee testified, "are particularly exposed to the
influence of many variables external to the scope of monetary policy." In
consequence, using interest rates as a yardstick involves a large risk of measurement error.1
Among the several monetary aggregates, M-l appears to be at least as good
a yardstick as any other. M-l equals the sum of publicly held coin, currency
and commercial bank demand (check) deposits. It measures very closely the
stock of media that can be exchanged for goods and services. Close identification
with our stock of exchange media provides a powerful reason for choosing M-l
to measure the thrust of monetary policy. In the final analysis, our concern is
with the impact of Federal Reserve actions on spending and thereby on prices,
production and employment. Other stores of value such as savings deposits
must be exchanged for coin, currency
or check deposits before they can be
used to purchase goods and services.2
THE ROLLER COASTER
Exhibit 1 graphs the yearly percentage growth in M - l from 1945 to 1975.
Each bar indicates the percentage change in M-l from the previous year,
computed by dividing last year's average money stock into this year's average
and subtracting 1. The exhibit shows that the growth of our M-l money supply
from World War II until now has moved up and down like a roller coaster.
Furthermore, from 1945 to 1953, the money supply roller coaster was angled
sharply downward. Since 1963, it has been tilted upward. Both the "tilts" and
the "rolls" had important effects on our economy's performance during the
past 30 years. Broadly speaking—
• The sharp downward tilt in M-l growth after World War II broke the inflation that followed the termination of wartime price and wage controls in 1946.
• The upward tilt after 1963 helped to produce the two waves of inflation
which we experienced since then.
• We suffered recessions in the months surrounding the bottoming of M-l
growth in 1949,1953,1957,1960,1967 (a mini-recession), 1970 and 1975.
These effects are discussed in more detail in Parts II and III.
* It is easy to be fooled if interest rates are used as the yardstick. During inflations, interest rates tend to rise and
money growth tends to be rapid. The former because credit demand rises and lenders demand inflation premiums. The
latter because banks use their reserves relatively intensively. Corrective policy would call for slowing money growth which
would lead to further interest rate increases for a time. If money growth doesn't slow, money watchers will know the Fed
hasn't been acting aggressively enough to slow it. But those who watch interest rates can be fooled because interest rates
can rise in inflation even though money growth hasn't slowed.
In the same way, during recessions, mone y growth is usually slow. Coirective action in this case would be to accelerate
it, and money watchers will conclude that the Fed is acting to stimulate the economy only if money growth speeds up.
They can't be fooled. Interest rate watchers, on the other hand, can be fooled into thinking the Fed isfightingrecession
when it isn't. This is because correct anti-recession policy calls for decreasing interest rates and decreases can occur even
though the Fed has done nothing to bring that about.
» Another reason for picking M-l as our yardstick is that its GNP velocity or turnover (GNP/M-1) is at least as stable
as other GNP monetary velocities. Since 1960, the GNP velocity of M-l hasfluctuatedless around its trend than the
velocities of M-2, which includes savings and time deposits other than certificates of deposit, and the other aggregates,
havefluctuatedaround their trends.
(3)
16
8-
a
EXHIBIT 1
M-l MONEY SUPPLY, NSfl
YEfiR TO YEflR PERCENT CHflN&E
7-
6-
4LU
#> tmm
I
•-1 •
46 *46 T 47 T 48
i
60 3 61 '62 »63
>68
YEflRLY FIVERfiGE OF MONTHLY DflTR
l
7E l 73 J 74 l 76
CONTROLLING MONEY GROWTH
Some argue that money growth moved up and down with the economy.
They observe that M - l growth acclerated during expansions and inflatioary
periods, and slowed in recessions. They hypothesize that our economy's cycles
cause money growth to cycle. The argument contains a grain of truth. This is
because historically, the Fed has accommodated money growth to the economy's
cycles. The pro-cyclical pattern of money growth did not, however, result
from conscious decisions to put M-l growth on a roller coaster in tune with
the economy's cycles. Rather, it happened this way. In expansions, loan demand rose and banks sold Treasury bills and bid aggressively for reserves in
the Federal funds and other money markets. As a result interest rates rose.
The Fed often tried to resist these rises in interest rates. In doing so, they
supplied the reserves that accommodated increased money growth.
Conversely, in recessions, with loan demand weak and banks buying Treasury
bills and selling Federal funds, interest rates fell. In turn, the Fed, satisfied
because interest rates were falling, slowed the supply of reserves and allowed
money growth to fall. Thus, as Professor William Dewald of Ohio State University has written, "Monetary growth rates have been above average during
booms and below average after booms peak out and recessions develop."
Although past money growth has rolled with the economy, it needn't
have done so. Ours is a managed money supply system. Economic cycles can't
cause cycles in money growth unless the Federal Reserve lets it happen.
The Federal Reserve has ample powers to control money growth reasonably closely. To accelerate money growth, the Fed can increase its open market
purchases of Treasury securities, reduce discount rates or lower reserve requirement ratios, or employ a combination of these alternatives. If one degree
of policy change doesn't work, the Fed need only push the levers harder.
Some degree of open market purchases and (or) reductions in discount rates
and reserve requirements will do the trick. The same can be said if a slowing
of money growth is desired.
The fact that the Federal Reserve has the power to control money growth
has been attested to by many Federal Reserve officials. In a 1964 colloquy
between the present chairman of the House Banking Committee Representative
Henry S. Reuss and Federal Reserve Board Governor George Mitchell, Mr.
Reuss asked, "It is a fact, is it not, that the Federal Reserve System can control
the money supply?" 3 The ensuing dialogue follows:
Mr. MITCHELL. Well, let me say first that the purpose or the objective of
both fiscal and monetary policy is to effect the growth of the economy and try
to moderate economic fluctuations. The whole discussion of public policy, as
it relates to monetary policy and fiscal policy, is to effect the rate of growth
and to minimize cyclical fluctuation. * * * Well, I think monetary policy
can make a contribution in this direction; yes.
Mr. REUSS. Agreed. Is it not also true that monetary policy can control
the money supply? Whether that, in turn, will produce the beneficent results
on employment and production that we all want is another matter, but you
can control the money supply, can you not?
Mr. MITCHELL. Well, it is at times difficult. The money supply grows
in a very lumpy fashion, and this is because the initial action which we take,
in providing reserves, does require a response from the banking system and
from individuals—
Mr. REUSS. But the banking system is either going to make loans or buy
investments—
Mr. MITCHELL. YOU can eventually do something with the money supply
but in the course of doing this you may cause some things to happen that you
believe are undesirable and, therefore, the goal of just making the money supply
move, if followed relentlessly, could have an impact upon credit conditions
and expectations that might be totally undesirable.
3 Source: "The Federal Reserve System After Fifty Years," Volume II, 1964, House Ranking and Currency
Committee,
Mr. RETJSS. Yes, but you must answer my question, which was not
whether it is good to make the—
Mr. MITCHELL. Oh, I think I am answering your question.
Mr. REUSS. (continuing). Not whether it is good to make changes in the
money supply the sole goal of monetary policy. There you have indicated you
disagreed with Friedman, who says it is, but my question is whether you
can—
Mr. MITCHELL. Ignoring all other consequences, absolutely.
Precisely the same point was brought out in'the Subcommittee's recent
hearings when Chairman Neal questioned Governor Partee.
Mr. NEAL. YOU state that without an increase in the money supply during
these years, [1953-1975], we could not have increased our Gross National
Product and so on. There is no question about that. But could the Fed have
kept the growth of money supply on a more even keel if it so desired?
Governor PARTEE. Yes. It could have been done; that is, in abstraction.
Further, in interviews conducted by House Banking Committee staff in
1973 with the 12 Federal Reserve Bank presidents and 5 of the 7 Federal
Reserve Board Governors, all agreed that they had ample power to control
money growth. Excerpts from 3 of the responses are provided below: *
Meserve Board Governor Robert Holland, asked whether there would be any
difficulty controlling money growth from year to year, responded,
If we were giving that top priority,, if we decided that the goal of hitting that target
overrides everything else we could do, I think we could probably come fairly close from one
year to the next in achieving the M-l growth that we decided we wanted to create in the
economy.
President Darryl Francis of the St. Louis Reserve Bank, asked if a year
were long enough to control money growth, answered,
Well, certainly a year, and of course, I believe we can do it effectively on a quarterly
basis.
President Bruce MacLaury (Minneapolis), asked how money supply could
grow, "just because the economy wants it to grow, unless the Federal Reserve
supplies the base [reserves] for it," answered,
The way you put that, it could not. It could not. We have the ultimate control and the
Question is, of the growth of the monetary base. I agree with you on that.
Chairman Burns has not explicitly agreed that the Federal Reserve can
control year-to-year money growth. However, on occasion he has stated that
the Federal Reserve would not allow money to grow at certain specified rates.
For example, in 1975 hearings before the Senate Banking Committee, he said,
"Maybeyou want us to increase the money supply, as you may define it, at an
8 or 10% rate.
We have no intention of doing that, not as long as I am the
Chairman." 5 In stating that the Federal Reserve will not allow money to
grow at such and such rates, the Chairman is, of course, substantiating that the
Federal Reserve h^ts ample power to control money growth.
Finally, we would point out that it would be a mockery for the Federal
Reserve to set money growth targets as it does pursuant to House Concurrent
Resolution 133, if its officials did not believe that they could stay within those
ranges, and would do so unless new economic developments emerged that required changing the targets.
In summary, the Federal Reserve, not the economy, controls the pattern
of money growth. The Fed may choose to accommodate and reinforce an
economic trend. But it doesn't have to do so. The roller-coaster pattern of money
supply growth shown in Exhibit 1 need not have occurred. The role it played in
shaping our economic experience over the last thirty years is detailed in Parts
II, III, and IV. Our inflation experience is considered in Part II, recessions in
III, and our experience under H. Con. Res. 133 in Part IV.
* Source, "The Federal Reserve and Inflation and High Interest Rates/' hearings before the House Banking Committee, July and August 1974.
* Source: "Monetary Policy Oversight," on Senate Concurrent Resolution 13, February, 1975, Senate Banking
Committee
PART II
INFLATION
There may be as many theories about inflation as there have been inflation episodes. Events and actions that cause inflation can be classified by
whether they result in (1) increased spending or (2) a larger part of current
dollar or nominal GNP being allocated to prices and a smaller part to production. In turn, events and actions that increase spending can be classified
by whether, directly, they cause (a) accelerated money growth or (b) increases
in the velocity at which money circulates (measured as current dollar gross
national product or GNP divided by M-l).
The Federal Reserve affects the rate of inflation because its policies and
actions, like it or not, change the rate at which money supply grows. Fiscal
policy (government deficit spending financed by the sale of securities) might
enter directly by bidding up goods and services prices or indirectly by changing
velocity. Incomes policies enter by affecting the division of GNP between
production and prices. This part of the report identifies the roles played in
our inflation history since 1947 by public policies—
Monetary (Federal Reserve).
Fiscal.
Incomes, and also, the following non-policy developments:
Korean War buying trends.
Velocity changes outside the Korean War period.
Import prices.
Inflation momentum released in administered pricing and wage
settlements.
Monopoly wage and price setting not reflected in inflation momentum.
Unemployment and production.
Changes in agricultural and raw materials supplies.
IN SUMMARY
In regard to money growth.—-On average, the rate of inflation rose 1% two
years after each l}{% rise in money supply from one year to the next. Put
another way, two years after the Fed increased the rate of money growth 1%,
measured year to year, the CPI rose on average by .64%. Changes in money
supply, moreover, were decisive to whether inflation endured however virulent
it began and whatever the original cause.
In regard to velocity.—A buying spree, which started right after North
Korea invaded South Korea, on June 25, 1950, and shows up in historically
extraordinary increases in the velocity at which money circulated, was primarily responsible for the brief but virulent Korean War inflation; and though
this may seem a paradox, for the lull that followed. Apart from this period,
changes in velocity have not played a major role in our inflationary experience
since 1947.
Government deficits do not appear to have had significant inflationary
impact independent of money growth. (Though, unquestionably, large deficits
created imposing money supply management problems for the Federal Reserve.)
In regard to the division of GNP between production and
prices.—Com-
modity supply reductions do not appear to have been a significant inflationary
factor at any time in the period since World War II. On the other hand, large
increases in agricultural and raw material supplies appear to have played a
significant part in the fall of the consumers' price index (CPI) in 1949.
Our results point also to these conclusions:
(7)
80-401 O - 76 - 3
8
• Our economy's pricing system is not explosive. It is stable. On average,
since World War II, changes in the rate of inflation for each year have carried
over to the next year with only about % power. That is, a 4% rise in the current
inflation rate was associated, on average, with a further 1% rise the next year.
However, the average may be misleading. Pricing behavior during the Korean
War years reduced the 1947-1975 average carry-over significantly. Since the
Korean War period, changes in the rate of inflation one year have carried over
to the next year with (rounding up) % power. Using this measure of momentum,
a 4% jump in the base inflation rate this year (whatever the source) causes
jumps of 2% next year, and 1% a year later. After that, the momentum of
inflation reverses
and the carry-over effect cycles around zero with decreasing
amplitude.1
The existence of a carry-over effect indicates that administered pricing and
collective bargaining in the current year are importantly influenced by price
trends in the preceding year, and in turn play important parts in the determination of the current year's prices. Though this influence is significant, it is far
less than what it would take to make our economy's pricing system explosive.
There is a momentum in inflation, but it is not cumulative or even constant.
Rather, it reverses after two years, then cycles around zero and gradually
plays itself out.
• Apart from the carry-over effect, administered pricing and collective
bargaining do not appear to have been significant factors in our post World
War II inflationary experience. However, the wage-price freeze in the second half
of 1971 appears to have dampened inflation that year, and the Kennedy-Johnson
guidelines may have done so from 1963 to 1967.
• Increased import prices were very important in 1973, 1974 and 1975.
They caused consumers' prices to rise about \%% *n 1973, 5% in 1974, and 1%
in 1975. Increases in import prices weighted by the value of imports relative to
GNP, were reflected roughly percent for percent in increases in the CPI.
There was no significant multiplication of domestic prices when import prices
increased.
• Finally, neither unemployment nor production appear to have had
significant independent influence on inflation trends during the post World
War II period. Nor did unemployment trends and levels significantly affect
the part that money supply played in our inflation experience since World
War II. Unemployment had affected this relationship in the middle and late
loan's.
We used econometric analysis to quantify the inflation effects of money
growth, deficits Korean War buying trends, velocity changes outside that period,
import prices, inflation motnentum, unemployment and production. Of this group,
variations in money growth, import prices and the momentum of past inflation,
and buying trends in the Korean War, proved to be statistically significant.
Together they explain 90% of year-to-year changes in the CPI in the 1947-1975
period. Money growth was the dominant persistent factor.
Inflation which could not be explained by our econometric analysis was
analyzed qualitatively to determine the separable roles played by incomes
policy, droughts and other uncontrollable events and monopoly pricing and
collective bargaining which cannot be related to past inflation.
INITIAL ANALYSIS
Prior to World War I and up to the mid-1930's, the relationship between
money growth and inflation was concurrent. As shown in Exhibit 2, when M - l
growth accelerated, the rate of inflation increased. Conversely, when money
growth decelerated, the rate of inflation decreased; and when the volume of
money fell, prices also fell.
Since the mid-1930's, there have been important structural changes in our
economy which have had the effect of delaying the impact of money growth
on inflation. First, storable goods, which can be marked up and down easily as
1 Thefirstyear after the 4% jump in base inflation, actual inflation exceeds base inflation by 2%. The second year, the
difference from the base year is only 1%. Thus momentum reverses in the third year, since the change in the rate between
years one and two is minus 1%.
EXHIBIT 2
YEAELY AVERAGES OF M-l AND CPI PERCENTAGE CHANGES, 1916-1934
MIVP
CPlYP
Yearly Average. Mi# percent change
Yearly Average* C P If percent change
|
r
16.0
/ \
\
t
\
\
A* •
\ 1
m
/
\\
0
\
-8.0
\
/
1
V
/
/
1
/
/
/>
/
/ \\
/ \
- \ \^
\
\
-16.0
16 17 18 19 80 81 22 83 84 85 86 87 88 89 30 31 38 33 34
YEARLY AVERAGE OF MONTHLY DATA
10
economic conditions change, have become less important as the economy has
become increasingly service oriented. Second, the spread of collective bargaining
has' delayed wage adjustments by putting the wages of large numbers of
workers under 2- and 3-year contracts. Similarly, the spread of contractual
arrangements in raw materials and energy supplies has acted to delay price
adjustments in these industries. In turn, the spread of wage, materials, and
energy price contracts has operated to slow price changes in manufacturing
industries characterized by oligopoly. Firms in concentrated industries tend to
ignore demand shifts and change prices only when costs change. Thus, factors
which delay the impact of changes in money growth on labor, materials and
energy costs have the effect of delaying also adjustments in final product prices.
Finally, new government policies have acted to delay price adjustments in the
period since the mid-1930's. The spread of price regulation has had this effect
directly. Adoption of counter-cyclical monetary and fiscal policies has generated
delay indirectly. Because government now is expected to act counter-cyclically,
businessmen have found it prudent to wait before changing prices in response
to cyclical demand changes.
There is, therefore, good reason to assume that prices now respond to
money growth with a lag. We analyzed the year-to-year percentage chang es in
monthly CPI and M-l data in the 1953-1976 period to find the lag that best
reflects the association between present inflation and past money growth. 5
We found that the correlation peaks when inflation is measured 23 months
after M - l growth changes. This does not mean that money growth changes
influence prices only after a 23-month lag. Rather, it means that the association between current inflation and lagged M-l growth is best at 23 months.
There is moreover, very little difference in the values of the
correlation statistic s
at lags between 16 and 30 months, and the peak value.3
THE INITIAL EXPERIMENT
Using data for the 1953-1976 period, we mapped on a scatter diagram,
shown here as Exhibit 3, all points that pair inflation measured from 4 the same
month a year ago and M-l growth in the year ending 23 months ago. We then
used computer analysis to find the regression or trend line that best fits this
scatter. The straight line that fits best passes through the origin of the diagram
where both inflation and money growth are zero, and rises so that for every
increase of 1 percentage point in money growth (e.g., from 4 to 5%), inflation
rises by 93 one-hundredths of a percent (e.g., from 3 to 3.93%).
TABLE 2.-REGRESSI0N RESULTS
[Dependent variable: CPIP, percent change in the CPI between corresponding months of adjacent years. Independent variable: M1P,
percent change in M-l between corresponding months of adjacent years, lagged 23 mo. Period: January 1953—April 1976.]
Constant
M1P
Coefficient
Adjusted
R*
Coefficient
T
SE
.930
19.63
.995
.018
.09
.08
1.81
.603
.601
.912
.880
.804
.810
.847
.896
1.041
1.114
1.097
1.024
.961
.886
5.35
4.88
4.60
4.43
4.68
5.02
5.94
7.23
7.20
6.81
6.57
5.39
.992
.956
.874
.871
.928
.973
1.111
1.170
1.145
1.065
.998
.963
.028
.149
.431
.444
.239
.092
-.380
-.584
-.504
-.227
.008
.125
.04
.19
.55
.55
.30
.12
-.51
-.90
-.78
-.35
1.61
1.69
1.65
1.80
1.94
1.87
1.66
1.34
1.49
1.76
1.75
1.45
1.98
2.05
2.07
2.09
2.01
1.94
1.79
1.52
1.54
1.60
1.67
1.97
.589
.543
.515
.495
.536
.570
.650
.733
.732
.709
.694
.592
.568
.521
.490
.470
.511
.547
.631
.719
.718
.694
.678
.572
Range
.804
1.114
4.43
7.23
.871
1.170
1.34
1.94
1.52
2.09
.495
.733
.470
.719
Mean
.939
5.68
1.004
1.67
1.85
.613
.593
Ml P (month)
CPI P (month)
All
January
February
March
April
May. _ . . .
June
July
August
September..
October
November...
December
_
February
March
_. April
May
June
-.
July
August
September
October . . .
November
December
January
-. _
T
.01
.17
DW
SE
R3
2
We u s e d the smallest s u m of squared errors b e t w e e n actual percentage changes in the C P I and estimates of these
changes m a d e o n the basis of percentage changes i n M - l to find the best association. T h i s s u m approaches zero as t h e
correlation between the actual and estimated data rises.
s T h e peak value of the correlation coefficient is .776. A t 16 months, it is .733. A t 30 months it equals .732.
* T h e earliest observation point pairs M - l percentage growth from February 1953 to February 1954 and the percentage
change in the C P I between January 1955 and January 1956.
EXHIBIT 3
SCATTEE DIAGKAM OF PERCENTAGE CHANGES IN CPI OCCUEEING 2 YEAES AFTEE CHANGES IN M-l GROWTH
YEAR TO YEAR PERCENT CHriNGE* OF MONTHLY VMH
* CPIP(Tt23)
12. 0
t.
i
*
'
I
*
*
t
i
•—*
NGE
X
%
$
'"X
^
j
t sf *
1—
. UJ
3. 0
o
a:
a
?
*** Aw * * '
-
* •
\
j—
T
*
1 * ^k*
we
-
1 ^
...
6
-3.
3.0
PERCENT CHANGE-IN Pll
6.
9.0
12.0
12
The regression results for this initial experiment are set forth in full in the
top row of Table 2, labelled "All." They show in addition to the statistics discussed above, that the regression line explains 60% of the variance of observed
inflation. To state this another way, the results show that on average, year-to. year, changes in money growth 23 months ago.exglainJ50% of_current year-toyear changes in inflation, The major part of 1955-1976 inHation changesT^wliicli
cannot be attributed to past money growth, occurred in 1973-1975 when inflation first rose and then fell substantially faster than could have been expected
from M - l growth 23 months earlier. Most of what happened then that can't
be explained by M-l growth 23 months earlier, can be explained by changes
in import prices, as will be discussed later.
!
Exhibit 4 maps the fit for the full 1953-1976 period. It puts the scatter
diagram in chronological order, overlaying for every possible 12-month period
from 1953 to 1957 percent changes in the CPI on yearly percent changes in M - l
occurring 23 months earlier. Because the CPI and M-l percentage changes
were computed for all months of the year from the same month a year ago;
January to January, February to February, etc., the exhibit provides an uninterrupted look at inflation and lagged (23 months]) M-l growth trends. The
exhibit shows the two series generally rising and falling togetherT
Checks.—As revealed by the Durbin-Watson statistic (DW) in row 1 of
Table 2, there is substantial common trend when overlapping periods are
analyzed. What happens in the 12 months ending in, say, April 1976, is very
closely related to that happened in the 12 months ending in March 1976. The
two periods overlap for 11 months, and so there is bound to be a lot of common
trend.
I
To test whether the overlap significantly affects our resiilts, we ran the
same experiment (regressing yearly percentage changes in the CPI on yearly
percentage changes in M—1 lagged 23 months) using, however, only one month
of each year. We did it for all possible pairings of the CPI and M - l , that is,
January to January CPI changes with February to February M-l; data (lagged
2 years), February to February price data with March to Marqh M - l data
(lagged 2 years), etc. This procedure eliminated the overlap in the data. The
results are presented below the top row in Table 2. Other than wjth rep§ect to
the DW statistic, they do not differ significantly from the results! of the .Initial
experiment reported in the top row.
\
Exhibit 4 (a) provides graphic illustration that the relationship between
current inflation and M-l growth two years ago is not due to common trend.
This exhibit graphs current percentage changes in yearly average^ of monthly
M-l data and percentage changes in yearly averages of the CPI pushecj back
two years. Only one observation—the; average—is used for each series each year.
As in the case of Exhibit 4, this exhibit shows the two series (yearly percentage
changes in CPI and yearly percentage changes in M-l two years earlier)
generally rising and falling together.
During the Subcommittee's June hearings, the witnesses were asked to
comment on our results, as shown in Exhibit 4. Excerpts from; the dialogue
are given below.
Chairman NEAL. Let me ask you, if I can, to comment on the chart specifically. Do you have any problem with that relationship between money jpolicy
and inflation?
,
"
;
Mr^ JORDAN (Vice President, Pittsburgh Natipnal Bank). No; I don't at
all. The basic relationship is one that I think is correct.
Mr. FIEDLER (Vice President, National Industrial Conference Board). The
relationship in the long run that you are referring tfr, it seems to me, is very well
established.
•
^
There is a casual relationship, that is, a more rapid rate of increase in
money over any decade is clearly going to be reflected on the average over
that decade in a more rapid rate of price inflation.
I think most of the testimony you have -heard here- today, is. consistent ^
ith that long-term kind of relationship . . . .
EXHIBIT 4
YEARLY MONEY SUPPLY CHANGES
Percentage Changes From One Month in One Year to the Same Month in the Next Year
YEARLY CHANGES IN THE COST OF LIVING OCCURRING 23 MONTHS LATER
1954-1976
Percent Change
14
f
/
-
-* MONEY
/ A SUPPLY
i**
A*/
AT
\ COST.OF
V LIVING
IK '•
/
01
l
l
I
l
l
l
l
l
l
'
l
l
lriuliiliiluliili* . i i l n l i i l i i l h l i i l n h i l i i l i J
I l l l l l l l i l i l l u l i l l l i l l l l l i l l i l l l l ill. .1.J i l l . . ! , l . i ^ l l i . l n l . i l . l l .
1954 '55 '56 '57 ' '56 '59 '60 '61 '62 '63 '64 '65 '66 '67 '68 '69 '70 '71 '72 '73 '74 '75 '76
CO
EXHIBIT 4A
YEflR TO YERR PERCENT CHRNGE
BfiR CHflRT IS C P I PUSHED BflCK 2 YEflRS
LINE I S Ml MONEY SUPPLY
1954-1975
•u
-13
-IE
-It
-10
-a
-B
-7
-B
-5
-4
-3
-2
-1
-0
64 ' 5 5 I EB ' B7 ' 58 > 69 ' 0 0 1 6 1
r
62 ' 03 ' 64 ' 65 ' 60 ' 67 ' 68 ' 69 ' 7 0 « 71 ' 72 ' 73 ' 74 ' 7 5
vcooi v cn/FRonF n r
MnkiTm v
IIDTO
15
It seems to me that the 23-month lag that you have built into the chart
is something that you should not view as a highly specific, accurate, precise
kind of a thing. Certainly the lag is variable. In one business cycle, it would
be longer, and in another business cycle, it would be shorter, depending on a
lot of different things.
There is no magic in that relationship.
Mr. PERRY (Senior Fellow, Brookings Institution). If you put the real state
of the economy into an equation, and you do a conventional attempt to explain
inflation—relying on how tight the market is, what is the history of inflation—
you don't add to the explanation by adding the past money supply.
If Mr. Perry had stopped here, his testimony would have to have been
regarded as contradicting our results. But he did not, and it is difficult to
know how to interpret his assertion that "you don't add to the explanation
by adding the past money supply" when it is viewed together with his further
remarks on the question. Moreover, as will be disclosed, our further regression
runs show that past money growth not only adds to the explanation of inflation
in the context of the state and momentum of the economy, it is the single most
important factor by far.
Mr. PERRY. (Continuing) Since money matters very much in determining
GNP, you will still get a relationship between monetary growth and the subsequent inflation rate but you will not get a relationship that, as sometimes is
suggested or implied, operates without affecting the real economy.
I would like to make this point as clear as I possibly can. Monetary policy
matters very much. If monetary policy is very expansionary and pushes the
economy very far very fast, the consequences of that will be more inflation.
But the inflation arises because the real economy has been pushed too far
too fast, not because of any magical property of the money supply.
Later, Mr. Perry added:
The point I am making is not that Mi is not very important in this process.
The periods when we have run ourselves into inflationary situations have
usually been periods when Mt has grown rapidly. In growing rapidly, it pushed
the real economy up fast and far and ran us into inflation.
We have no quarrel with the thrust of these further remarks by Mr. Perry.
Mr. Pierce (Professor of Economics, University of California, Berkeley) made
essentially the same point. The role played by money supply should be analyzed
and evaluated in the context of the state and momentum of the economy.
More on this later.
Representative HANNAFORD. (Questioning Mr. Pierce). We have talked
about the lags and we have talked in general terms about a 2-year lag in monetary policy and its effect.
Mr. PIERCE. The relationship that is being looked at is an average. It
depends on the circumstances in the economy and how long that policy is
pursued.
Later, Mr. Pierce told the Subcommittee:
Charts are simple things, and they have to leave out a lot of complexity,
but the relationship is there.
There is a great deal of empirical evidence that says that there's a
relationship between monetary policy and inflation, and that's borne out by
the chart. There's also a great deal of evidence in the statistical literature
which says that economic time series tend to have steady patterns in their
behavior, so-called serial correlation in them, and that two series can appear
to be related even though behaviorally
they are not. I think there is some of
that problem in that chart.5
5 As already indicated, if only one observation a year is used, the chart pattern remains virtually the same as in Exhibit
3, and serial correlation is not a serious problem.
80-401 O
76 - 4
16
But I would have to say that the chart dramatizes something that is
v e ^ important to dramatize. There is a relationship between monetary policy
and inflation.
All I would counsel you against doing is inferring that every time there is
an expansion in money growth, that as night follows day we are going to have
inflation 2 years from now. I don't know of any empirical evidence that allows
one to make that assertion, because all we have is average relationships. On
the average, the economy tends to be fairly highly employed, and if the economy is fairly highly employed and we have an expansion in monetary policy,
then the only place that that increase in demand in the economy can go is in
terms of bidding up prices. So, on average, there should be a relationship
between accelerations in money growth and acceleration in prices, with a lag,
because the economy tends toward full employment.
When the economy is not fully employed, then it doesn't follow that if
we have a relatively rapid money growth for, say, two quarters, if we wait
2 years, we will observe prices going up. I reject that proposition theoretically
and I reject it empirically. The important lesson is, though, don't keep it up.
Chairman NEAL. Have you had a chance to look at our exhibit?
Mr. MELTZER (Professor of Economics, Carnegie-Mellon University). Yes;
1 have looked at the chart.
Chairman NEAL. DO you see any problems?
Mr. MELTZER. I think in general the relationship is correct. It is a careful
reproduction and extension of earlier work. What the chart shows is there is
about a 2-year lag with no absolute precision or uniformity about the timing.
Each time there is a ripple in mone}^, as Dr. Pierce has said, it doesn't show up
in price changes. However, it is certainly true that the main movements are
shown on the chart. When the rate of monetary growth expands, on average
2 years later we have an increase in the rate of price change, and when the rate
of monetary growth slows down, about 2 years later we have a slowdown in
the rate of inflation. We are now experiencing the results of that process. We
should be careful to avoid saying, and I think the paper accompanying the
chart does, that the only thing that has any effect on the rate of price change
is the rate in the growth of the money stock. A very large fraction of the rate
of inflation is explained by past changes in the growth of the money stock.
Chairman NEAL. What is emerging here is fascinating to me. Our chart
maps the increase in the yearly changes in the cost of living occurring 23 months
after fluctuations in money supply—Mi. A consistent relationship is shown
here, indicating that the sharp growth or contraction in Mi is reflected in the
level of prices.
Governor PARTEE. Mr. Chairman, I do not disagree with the thrust of
that chart. I am sure that you will understand when I sa}^ that it is not an
elaborate analytic chart. It is simply an arithmetic exercise. If one, say, were
to have plotted the change in unit labor costs and compared them with the
Consumer Price Index, you would find a high correlation also.
There are some possible statistical difficulties with the chart. But I think
that it is true that a rate of increase in the money supply above a minimal rate
that would be associated with real growth in the economy, given our limited
capacity to increase output in real terms in the economy over the years will in
time bring an inflation. . . . But I would point out that there is an entire
process involved. If in looking at this chart, you tend to conclude that had the
money supply not gone up we would not have had inflation—everything else
would have been just the same and we would not have had inflation—that is a
wrong conclusion. . . .
What would have to happen in order to reduce that rate of inflation 23
months hence is that wage increases, cost increases, price increases, and profit
shares for companies would have had to be held down. That process is very
difficult to deal with in our economy.
17
Chairman NEAL. YOU mentioned wage increases several times as being
highly inflationary. Why were wage increases relatively moderate during the
early sixties? What would explain that?
Governor PARTEE. That is a very interesting question. I have thought
long and hard about how to reestablish those conditions. I think as a matter of
fact that the economy is closer to doing so now than it has been for the whole
decade that has gone by.
My view of the matter is that we had a period of very slow growth in the
economy in the latter part of the fifties. We had a number of recessions—in
1953, 1954, and in 1957, 1958 and in 1960 and early 1961. It was a painful
process—and I might say that monetary policy was very widely blamed for
the poor performance of the economy in the latter half of the fifties. But, as a
result of that process, expectations and aspirations did tend to fall back and
expectations of inflation tended to moderate so that we found ourselves in the
early sixties with moderate growth rates in wages, quite slow increases in consumer prices, good increases in profits based not on rising profit margins but
on rising volume—a very decent situation. I think that is what we are trying
to restore at this time.
Chairman NEAL. (Contrasting the early sixties to other periods.) Also
during this period you had a very moderate and steady monetary policy.
Table 3 gives M-l growth rates between corresponding quarters of adjacent
years from 1961 to 1965.
TABLE 3 . - M - 1 PERCENTAGE GROWTH BETWEEN CORRESPONDING QUARTERS OF ADJACENT YEARS, 1961-65
1961
Quarter:
1st
2d
3d
4th..
_._.
1962
1.28
2.29
2.06
2.82
1963
2.99
2.60
1.82
1.44
1964
1.79
2.34
3.52
3.97
1965
3.62
3.59
4.29
4.48
4.56
4.39
3.88
4.35
FURTHER ANALYSIS
In the simple linear regressions that comprised our initial tests, factors
other than past M-l growth were not considered explicitly. The initial tests,
thus, should be viewed as providing only a rough first approximation of the
inflation/money relationship. These results could be misleading. Changes in
lagged money growth might affect inflation only if there is full employment, for
example. Also, the measured impact of money growth on inflation could be
magnified spuriously if inflation is gaining momentum, or if import prices are
increasing concurrently. Further analysis was required to check these possible
complications and obtain a more complete understanding of inflationary processes in general, and the inflation/money nexus in specific.
We used multiple linear regression analysis to analyze the inflation effects
of changes in money growth together with other factors that are widely held to
be inflationary. In essence, this standard econometric procedure permits quantif}'ing the partial and combined inflation effects of the diverse factors that are
assumed to cause inflation. Using this standard procedure, we quantified the
inflationary impact of mone}^ growth together with the parts pla}7ed b}^ the
following factors.
• Buying trends that were set off by Korean War.
• Velocity trends outside the Korean War period.
• Import prices.
• The momentum of past inflation.
• Production.
• Unemployment.
• Federal budget surpluses and deficits.
Specification of these variables is discussed below in conjunction with our
results.
We analyzed CPI data covering the 1947-1975 period. The data were
assembled in yearly series by averaging intra-year data. Where monthy data
18
were available, they were used. Quarterly data had to^be used in the cases of
import prices, the deficit, and constant dollar GNP. Use of yearly statistics
avoids most seasonal adjustment problems as well as the overlapping data
problem. On the other hand, it should be recognized that when data are blocked
off in years, relationships that tend to be continuous can be distorted. Our
results thus should not be judged by how well they explain isolated years.
Rather they must be looked at as statements of average tendencies over the
long haul, and judged by their over-all or average predictive power.
RESULTS
The results of our futher analyses are summarized in Tables 4-8.
TABLE 4
Table 4 presents three series of analyses, with the second and third building
on the first. In all series, the basic period of CPI changes covered was 1947-1975.
Three subperiods also were covered: 1951-1975, 1956-1975, and 1947-1970.
The first series relates yearly percentage changes in the CPI to (a) yearly
percentage changes in M-l two years ago 6 and (b) Korean War buying trends.
The second series adds percentage changes in import prices to the list of causal
factors (weighted by the ratio of current dollar imports to current dollar GNP).
The third adds the momentum of recent inflation (as determined by the change
in the CPI inflation rate last year).
TABLE 4
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: Yearly percentage change in the CPI computed using
yearly averages of CPI levels, denoted CPIYP.
Independent variables: M1YP (t—2); yearly percentage change in M - l
(two years ago). Dummy; 1951 = 1, 1952= —.02, 1953 = — .38, 1954= —.48,
1955= —.28. PIMYPW; yearly percentage change in import prices weighted
by imports as a percent of GNP. [CPIYP {t-1)-CPIYP
(t—2)]; the change
in the yearly inflation rate last year.
"
Series 1 :
Constant
,
MlYP(t-2)_____
_
Dummy: coefficient...
Adj R2
SE
_
DW
_
Series 2:
Constant
_
_
___
_
MlYP(t-2)
Dummy
______
_
_
_
_
_______
PIMYPW
Adj R2.._._____
SE
DW.....
______
Series 3:
Constant—
MlYP(t-2)
___
_
_
_
___
PIMYPW
_
_
___
_
_
_
DW
1951-75
(n=25)
1956-75
(n=20)
1947-70
(n=24)
1.017
(.515)
2.916
(.102)
»8.864
(1.474)
.771
1.723
1.959
i —. 162
(.553)
* 1.012
(.137)
»9.233
(1.280)
.757
1.421
1.667
i —.137
(.674)
»1.010
(.156)
_-__—.682
1.570
1.665
* .105
(.448)
*.827
(.094)
a 8.514
(1.279)
.805
1.487
2.457
».116
(.410)
a.767
(.089)
i .470
(.422)
a.691
(.121)
».535
(.510)
2.677
(.137)
i .105
(.448)
2.827
(.094)
2 7.284
(1.235)
2 1.413
(.351)
27.047
(1.037)
2 1.367
(.296)
2 1.392
(.329)
».535
(.367)
a.643
(.084)
2 6.000
(1.109)
2 1.403
(.2%)
2 .248
(.074)
.898
1.153
.856
1.368
2.185
_
Dummy
rCPIYP(t-l)-l
lCPIYP(t-2) J
Adj R2
SE
_
_•
1947-75
(n=29)
___
2.523
.874
1.024
1.606
.836
1.127
1.594
.805
1.487
2.457
1.554
(.406)
2.645
(.118)
*6.328
(1.073)
t.836
(.439)
2.538
(.125)
_._
t .461
(.410)
2.714
(.091)
2 7.364
(1.188)
2 1.347
(.283)
1.166
(.095)
.885
.978
21.296
(.277)
2 .465
(.162)
.885
.943
___
2 .218
(.086)
.843
1.285
1.935
2.139
1 Coeficient is not significant.
2 Coeficient is significant.
Note: Standard errors are in parentheses below the coefficients.
• This means that the 1947 percentage change in the CPI was related to 1945 M-l growth, etc.
2 8.514
(1.279)
________
2.623
19
KOREAN WAR BUYING TRENDS
The buying trends set off by the Korean War were put into the equations
by Using a sd-called dummy variable. In econometric research, dummy variables are used to measure the impact of unique, unexpected exogenous shocks.
The invasion Of Sduth Korea on June 25, 1950, was such an event. Its effects
on the U.S. domestic economy flowed, most importantly, from a buying spree
which started almost immediately and continued into the second quarter of
1951, when wage and price controls became effective.
To understand the necessity for using a dummy to cover Korean War
buying trends, and its specification, we have to go back to World War II.
During World War II, tne necessity of allocating enormous resources to the
War effort caused major shortages to develop. Demand arising from wartime
monetary expansion was "pent-up" until after the war by rationing programs.
Termination of wartime price and wage controls and rationing in mid-1946
Released it, Prices were bid up 29% between June 1946 and June 1948. When
the Kotfe&ii Waf began, memories of shortages, controls, rationing and subsequent huge price increases were of sufficiently recent vintage to impel an
immediate buying spree which production could not match. So in contrast to
World War II, when buying, had to be postponed until after the war, the
Korean War was marked by an immediate buying spree. The GNP velocity of
M-l jumped phenomenally in the next year. It increased 10% in the summer or
third quarter of 1950 over the year-earlier figure, and 14% in each of the next
three quarters from year-earlier velocities. These increases exceed the highest
increase that can be expected to occur with 95% probability, based on velocity
changes between the same quarters from one year to the next from 1948 to
1976. Using yearly averages, the increase in velocity was 10.47% in 1951, a
rise that also exceeded what can be expected by chance 95% of the time.
There can be no doubt that the outbreak of the Korean War caused an unusual
spending spree in the United States.
The first effect of the buying spree was to quickly bid up prices. Using yearly
CPI data, this effect shows up in 1951. A second and quite different effect followed. In subsequent years, there was less buying than would otherwise have
occurred. Just as pent-up demand must sooner or later lead to a buying spree,
as after World War II, so too must a buying spree which begins without warning, as at the outset of the Korean War, lead later to a buying lull. Put another
way, when buying is postponed, demand later increases. When there is forward
buying, demand later decreases.
To capture both effects, we constructed a weighted dummy for the 19511955 period.7 Our hypothesis is that the invasion of South Korea caused an
immediate buying spree, followed by a period during which household and
business inventory levels attained during the buying spree were maintained
at above-normal levels in fear that the war would yet cause shortages. Finally,
when the war ended in mid-1953, demand subsided and a buying lull occurred.
The weights we used in constructing the dummy are
1 for 1951, —.02 for 1952,
— .38 for 1953, —.48 for 1954 and —.28 for 1955.8
7
Because we used a weighted dummy rather than separate ones for each of the years we weighted, the number of
degrees of freedom implied by the number of observations minus independent variables is overstated by four in all regressions where the dummy is used. But this is not crucial. Inspection of the results shows that the roles played by our independent variables does not depend on whether significance is tested with the number of degrees of feeedom implied by
the tabulated data, or four less.
Use of the weighted dummy rather than separate dummies for each year, 1951-1955, allowed us to keep the tabulation
of the results simple.
* To determine the exact weights we regressed the yearly percentage change in prices on yearly percentage M-l growth
lagged two years and separate dummies (each equal to 1) for 1951,1952,1953,1954 and 1955, and used the resulting regression
coefficients. In specific, we arbitrarily set 1951=1 and constrained the 1952-1955 weights to equal their coefficients divided
by the 1951 coefficient.
We also computed weights from a regression which included import prices and inflation momentum as independent
variables. These weights are 1 for 1951, -.40 for 1952, -.06 for 1953, - . 5 1 for 1954 and - . 4 1 for 1955. They differ significantly
in 1952 and 1953 from the weights used in the regressions reported in Table 4 and subsequently. However, for 1947-1975
they yield results that are not very different, actually providing a better statistical fit, than those in Series 3 of Table 4.
Using the alternative weights, C P I Y P =
Constant
+ MlYP(t-2) +
Dummy
+
P1MYPW + [ C P I Y P ( t - l ) - C P I Y P ( t - 2 ) ]
.799*
.577*
5.660*
1.390*
.352*
(.326)
(.072)
(0.898)
(.270)
(.064)
Adj R2=.914; SE=1.056; DW=2.409
* Coefficient is significant.
Standard errors are in parentheses below the coefficients.
Despite the marginally better results, we chose to continue using the weighted dummy which we constructed from the
coefficients on the 1951-1955 dummies in the regression of CPI YP on only lagged money growth and the 1951-1955 dummies,
without including either import prices or inflation momentum. We did so because it makes more sense for the 1952 weight
to be closer to zero than the 1953 weight. We note, however, that the procedure we have used collects the carry-over of
inflation changes in the dummy variable which lowers the coefficient on the carry-over variable.
20
Series-1.—The Series 1 Table 4 results indicate that changes in the CPI
during the 1951-1955 period were powerfully affected by the buying spree
that accompanied the invasion of South Korea and the subsequent, corollary
buying lull. That period aside, the inflation rate was not affected by velocity
changes but was powerfully affected by money growth lagged two years.
The impact of lagged money growth is shown by its coefficients in the Table 4
data. That velocity played no further role is shown by inspection of Exhibit 7
(page 36) which graphs percentage changes in M-l's GNP velocity quarterly
from 3^ear earlier quarters in the 1948-1975 period. Moreover, adding velocity
to the independent variables in the 1947-1975 regression did not provide added
explanatory power.
The full result of this regression is: C P I Y P =
Constant
+
i.525
(.689)
MlYP(t-2)
2.906
(. 101)
f
Dummy
+
J 7.838
(1.698)
Velocity
1.160
(.134)
» Coefficient is not significant.
a Coefficient is significant.
Nt)te: Adj R*=0,771; SE = 1.725; DW=1.545. Standard errors are in parentheses below the coefficients.
Series-2 and 3.—Series 2 of Table 4 introduces the role played by percentage changes in import prices. As noted above, import price changes were
weighted by the ratio of current dollar imports to current dollar gross national
product or GNP. Weighting was required because imports have fluctuated relative to GNP, tending on average to rise. The same percentage increase in import
prices today is more important than in years past.9
Series 3 of Table 4 introduces the effect of the momentum of past inflation
on present inflation. The momentum of past inflation is measured by the change
last year in the CPI inflation rate from the year before.
Both of these new variables have a significant impact on inflation. As
shown by the Series 3 results, in combination,
• M-l growth lagged 2 years,
• Korean War buying trends,
• last year's change in the rate of inflation, and
• this year's change in weighted import prices
explain 90% of year-to-year changes in the CPI in the 1947-1975 period,
89% in the 1951-1975 and 1956-1975 periods and 84% in the 1947-1970 period.
The Series 3 regressions also quantify the separate contributions of the
independent or explanatory variables. In regard to 1947-1975,
• On average, a 1% per year increase in money growth increased the
rate of inflation by .643% two years later. Put another way, on average, a
1/2% increase in money growth increased inflation by 1% per year two years
later.
• The Korean War buying spree added 6% to the CPI in 1951. The subsequent corollary buying lull damped inflation by 2.3% in 1953, 2.9% in 1954,
and 1.7% in 1955.
• The change in the rate of inflation one year tended to be carried over to
the next year with roughly % power. For example, the 4.8% jump in the rate of
inflation in 1974 generated a further rise in the CPI of 1.2% in 1975.
• On average, import price rises (weighted by imports as a percent of
GNP) increased the CPI inflation rate by a multiple of I A, However, subtracting one standard error reduces the coefficient to 1.1. The result, in other words,
does not refute the hypothesis that percentage changes in import prices
(weighted by the ratio of imports to GNP) cause roughly proportional changes
in U.S. consumers' prices—percent for percent.
21
These results do not depend on our viewing the Korean War buying spree
as later reversed. Using a dummy=l in 1951 and 0 in all other years, i.e.,
ignoring all secondary effects of the Korean War buying spree, changes the
1947-1975 equation only marginally, as follows, CPI Y P =
Constant
+
1.343
(.446)
MlYP(t-2)
2.615
(.095)
+
Dummy
2 6.121
(1.517)
4
PIMYPW
rcpiYP(t-i)-]
LCPIYP(t-2) J
+
* 1.510
(-337)
2.310
(.083)
1 Coefficient is not significant.
2 Coefficient is significant.
Note: Adj R2 = .861; SE = 1.342; DW = 1.945. Standard errors are in parentheses below the coefficients.
TABLE 5
To further check the inflationary impact of changes in import prices, we
constructed an index of consumers' prices purged of import prices and related
yearly percentage changes in this index 10
to the independent variables of
Series 3. The results are reported in Table 5.
TABLE 5
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: DCPIYP; yearly percentage change in domestic
prices.
Independent variables: As specified in Table 4.
Series 1 (DCPIYP):
Constant
MlYP(t-2)
Dummy
PIMYPW
rCPlYP ( t - l ) - l
LCPIYP(t-2) J
AdjR2_____SE
DW
1947-75
(n=29)
1956-75
(n=20)
1947-70
(n=24)
1.485
(.435)
3.701
(.099)
2 6.309
(1.314)
i.O52
(.350)
3.280
(.088)
.828
1.366
2.214
1.665
(.560)
3.630
(.159)
».436
(.384)
2.685
(.086)
2 6.493
(1.116)
i—.217
2.212
SIP
1.204
2.204
(
:8°!P
1.202
2.283
1
Coefficient is not significant.
2 Coefficient is significant.
Note: Standard errors are in parentheses below the coefficients.
i° AvS a matter of logic, if the proportionality hypothesis is valid, domestic prices on average, will be unaffected by
changes in import prices. The only other possibilities are that domestic prices will (1) fall or (2) rise. The rationale for
asserting they will fall is that, by assumption, our demand for imported input such as sugar and oil is insensitive to price,
and hence when import prices rise, we have less to spend on domestic goods and so their prices must fall. The usual assumption underlying the more widely believed hypothesis that domestic prices will rise is that prices equal costs plus a standard
percentage mark-up. Thus as imported input are processed into final goods, prices rise higher and higher percentagewise
at every stage in the process.
The cost plus thesis is illustrated by the following example. Panel A shows initial costs, percentage of mark-ups and
sales prices in a four stage production process. Panel B shows costs and sales prices, using the same mark-ups, following a
$2 rise in costs at Stage 1. Under cost plus pricing, a $2 rise in costs at Stage 1 generates a $7.50 price rise in final goods.
Panel and Stage Number
Costs
Panel A:
1
2
3
4
$10
15
20
30
1
2
3
4
$12
18
21
36
Panel "B:""
Percent
markup
50
mi
50
25
50
50
25
Sales price
$1.5.00
20.00
30.00
37.50
18.00
24.00
36.00
45.00
22
Our test tends to confirm the proportionality hypothesis. As reported in
Table 5, domestic prices are unaffected by changes in import prices. This conclusion follows since none of the coefficients on the percentage change in import prices differs significantly from zero. It also is noteworthy that the coefficients on the other variables in the Table 5 regressions
are about the same
as the corresponding coefficients in Series 3 of Table 4.11 The significance of the
latter is that import prices might best be regarded as exogenous and used only
to explain that part of our inflation experience which other variables cannot
explain.
TABLE 6
SUBPERIODS
The Table 4 Series 2 and 3 regressions for the subperiods provide two
noteworthy results. First, the coefficient on the change in last year's inflation
rate did not differ significantly from zero in 1951-1975. Second, in the 19561975 period, the coefficient on this same variable was relatively high and that
on M - l growth relatively low. The 1956-1975 statistics are the more interesting.
The insignificance of the coefficient on inflation's momentum in 1951-1975
reflects only that last year's change in inflation played little role in wage and
price decisions during the Korean War period. As already noted, decisions
then were based on memories of World War II, rather than on occurrences of
the most recent past.
The 1956-1975 coefficients on the other hand, are suggestive of a structural
change affecting inflation momentum and the timing and/or force of the impact
of money growth on inflation. Such a change would be disturbing if it has occurred. It woud mean the full period (1947-1975) coefficients are unreliable
forecasters. There is, however, an alternative explanation for the 1956-1975
results. Inspection of Exhibits 4 and 4 (a) reveals large offsetting errors in inflation rates predicted by 1959 and 1960 money supply changes. These misses
are not totally unexpected because the swings in money supply in that period
were unusually sharp and short, distorting the yearly statistics more than usually. The sharp run-up in M-l growth reflected in the 1959 statistic, started in
April, 1958, and was over a year later. The second half of 1959 was marked by
negative money growth—a decrease in money supply. The fall in money supply
continued into the first
half of 1960, after which there was another sharp reversal—this time up.12
Inspecting Exhibits 4 and 4(a), it would appear that what happened to
prices in 1961 and 1962, viewed together, or averaged, is reasonably well predicted by average M-l growth in 1959 and 1960. To further test this hypothesis, we computed regressions for both 1947-1975 and 1956-1975 in which a
dummy was used equal to 1 in 1961 and —1 in 1962. The results are set forth
in Table 6. They provide added support for the view that the 1959 and 1960
M-l statistics are predictably bad forecasters of later inflation because of the
sharp intra-year changes in M-l growth that occurred in 1959 and 1960. As a
corollary, the Table 6 results also indicate that we can reject the hypothesis
that there has been a structural change in the relationship of inflation to money
supply. Taking into account the ephemeral nature of the sharp zig-zag in money
growth in 1959-1960, the 1956-1975 coefficient on lagged money growth is
shown to be almost identical to the 1947-1975 coefficient.
11
Twelve of the 13 coefficients in Table 5 are within one standard error of their values in the corresponding time period
in Series 3 of Table 4. The exception is the coefficient on the inflation momentum variable for 1956-1975. This is 1.2 standard
errors larger in Table 5 than in Table 4.
» In summary, M-l rose sharply from $136.4 billion in April, 1958, to $143.3 billion a year later, reached $144.9 billion in
July, 1959, and then dropped to $143.4 billion in December, 1959, and $142.6 billion in May, 1960. After May, it again increased and reached $144.2 billion in December I960.
23
TABLE 6
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: CPIYP.
New independent variables 1961-1962 dummy; 1961 = 1, 1962= —'1.
Constant
-
MlYP(t-2).__—___ — .
Korean war dummy.
PIMYPW
—
__-_—_
.
—
1956-75
<n=20)
1.445
(.355)
a.664
<.08l)
* 6.107
(1.063)
21.382
i .546
(.394)
2.630
(.113)
...
_____..„„_
_______
1947-75
(n=29)
..
;
______
(.283)
3
fCPlYP Ct-1)—I
LCPIYP <t-2) J
-244
(.071)
1961-62 dummy____.__
____
t -1.516
(.793)
AdjR*
SE
OW
.906
1.104
2.321
a 1.179
(243)
2.457
(.139)
» -1.553
(.608)
.915
.811
1.461
i Coefficient is not significant.
' Coefficient is significant.
Note: Standard errors are in parentheses below the coefficients.
We also tested whether the inflation-money supply lag structure has
changed by computing regressions for 1947-1975 and 1956-1975 using MlYP
at (t-1) and (t-3) as well as at (t-2), together with the other variables of the
Table 6 regressions. The coefficients on money (MlYP) are set forth below
together with their standard errors. They show that the average lag remains
2 years. We conclude then that there has been no change in force or timing of
the impact of money growth on inflation.
1947-75
M1YP(M)
-—.~_
1.161
(. 137)
> .470
(.148)
1.124
(.090)
MlYP(t-2)
MlYP(t-3)
1956-75
1.242
(. 150)
».375
(.180)
1.094
(.144)
i Coefficient is not significant.
> Coefficient is significant.
Note: Standard errors are in parentheses below the coefficients.
In regard to the structure of inflation momentum, however, the Table 6
results confirm that for the period since 1956, the carry-over of last year's
change in the inflation rate to the current year is closer to % than to %. Our
guess is that this reflects only that the full period result, % was reduced by
behavior during the Korean W_ir> though we cannot dismiss the possibility that
the scope of administered pricing and collective bargaining has increased in
recent decades.
TABLE 7
The regressions reported in Table 7 add so-called real variables to the
equations reported in Series 3 of Table 4. They were used singly, not in combination.
Series 1 of Table 7 adds the yearly percentage change in the industrial
production index from the preceding year, in the 1947-1975 and 1956-1975
periods, and alternatively, the yearly percentage change in real GNP in the
1947-1975 period. No additional predictive or explanatory power was gained
by adding these variables to the analysis. The coefficients on the new variables
did not differ significantly from zero. They were dropped.
0-401 O
76 - 5
24
Series 2 added the change in the yearly average unemployment rate from
a year ago, in the 1947-1975 and 1956-1975 periods, and alternatively, the
yearly average tmemployment rate itself in 1947^1975; Th& tabulated results
show that inflation was not related to current levels or changes in unemployment
in our sample period. Nor did lagging unemployment help explain inflation.
TABLE 7
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: CPIYP.
New independent variables: IPYP; yearly percentage change in the industrial production index. GNPPCON; yearly percentage change, in GNP
measured m constant dollars. UNY; yearly average of unemployment. UNYUNY(t-l); change last year to this year in the average unemployment rate.
1947-75
(n=29)
1956-75
(n=20)
1947-75
(n=29)
Series 1:
Constant
MlYP(t-2)
Dummy
PIMYPW
[CPfYP(t-l)-y
LCP»YP(H) J
IPYP
OKPPCON
Adj R C — '
SEl.
DW.__
.
J.361
(.403)
2.641
(.083)
2 5.795
(1.125)
2 1.424
(.2%)
2.265
(.076)
.038
(.036)
1.151
2.313
il.OOl
(.577)
25&
(129)
2 1.320
(.289)
1.388
(.264)
J —. 029
(.063)
.879
.968
2.167
i.36O
(.615)
2.655
(.092)
2 5.863
(f. 193)
* 1.423
(.306)
2.255
(078)
i.036
:J§8
J§8
1.175
2.430
Series 2:
Constant
MlYP(t-2)
Dummy
PfMYPW
fCPIYP(t-l)~l
lCPIYP(t-2) J
UNY-UNY (t-1).
1.500
(.374)
2.656
(.087)
2 5.775
(1.164)
2 1.403
(.299)
2.250
(.075)
-.142
(.199)
2 1.306
(.297)
i.446
(.226)
.038
(.302)
1.165
2.373
.974
2.151
1.848
(.465)
2.534
(.132)
UNY
Adj R*
SE
DW............
1.900
(.957)
2.640
(.085)
2 5.929
(1.142)
2V1.426'
(.306)
>.247
(.076)
*—.074
(.179\
1.174
2.479
i Coefficient is not significant.
* Coefficient is significant
Note: Standard errors are in parentheses below the coefficients.
Moreover, our results indicate that levels and changes in unemployment
and production did not affect the relationship of inflation to money supply in
the post World War I I period. The coefficients on M-l growth lagged 2 years
in the regressions reported in Table 7 are virtually the same as in the regressions
reported in Series 3 of Table 4. It would appear, as Mr. Pierce indicated in his
testimony, that during the post World War II period, our economy has been
close enough to full employment on average for money growth changes to be
absorbed (after a lag) primarily in changes in the rate of inflation. We would
caution, however, as the testimony did, that with sufficient unemployment,
the relationship breaks down. Experience in the middle and late 1930rs indicates
that when unemployment exceeds 15% of the labor force, annual M - l growth
can average nearly 10% for 4 to 5 years without substantially increasing prices.
We would like to know more about these trade-offs but (fortunately) our
economy, excepting possibly in 1975 and 1976, has operated too close to full
employment since the 1930's to permit gaining more insight.
25
Finally, it is worth briefly exploring why we failed to find a significant
relationship between inflation and the "real" economy. The reason, we believe,
is that the real economy affects prices in two distinct but opposite ways. First,
when production and employment increase, prices tend to fall as increased
supplies compete for the same monetary resources. Offsetting this effect, however, is that when production and employment increase, labor and materials
niarkets tighten and prices tend to rise because the same monetary resources
must compete for increasingly scarce supplies of labor and materials.
Similar offsets occur in recessions. Directly, reduced production and
supplies tend to increase prices. Indirectly, reduced production loosens labor
and materials markets, which in turn, acts to cause prices to fall.
The Table 7 results with respect to output and unemployment reflect these
ojtsets. In both the full 1947-1975 period and the 1956-1975 subperiod, the
downward impact on the CPI of increases in the supply of output tended to be
offset by the fact that markets tighten and costs and prices rise as output
expands. Conversely, the inflationary impact of output decreases were offset
by deflationary effects of markets loosening in recessions as nroductktti falls.
TABLE 8
Table 8 reports the results of regressions which added the deficit to the
explanatory variables of Series 3 Table 4. Following suggestions made by the
Federal Reserve, we used the so-called "High Employment deficit" measured
as a % of potential GNP. Scaling was necessary because $1 billion deficit today
is not the same as ten and twenty years ago. We also tried the actual deficit
as a % of potential GNP. The High Employment deficit series was provided
to us by staff of the Federal Reserve for 1952-1975. We restricted our analysis to
this subperiod.
The results that were obtained using the High Employment deficit lagged
two years in respect to CPI changes are given in Series 1. The results that were
obtained using the actual deficit (also lagged two years) are tabulated in Series
2* Neither of the deficit variables was significant. We also tried the High
Employment deficit concurrently and at lagjs of one and three years in respect
to CPI changes. The results of these experiments differ only marginally from
those reported in Table 8 Series L
TABLE 8
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: CPIYP.
New independent variables: BY(t-2),
of potential GNP two years ago. Bl(t-£):
current dollar GNP, lagged two years.
entered as positive numbers, surpluses as
the High Employment deficit as a %
the actual Federal deficit, as a % of
Following convention, deficits were
negative numbers.
Series 1:
Constant
MlYP(t-2)
„__
Dummy
PIMYPW
rcpiYP(t-i)-i
LcPIYP(t-2) J— " -—
"
Adj R
SE->._-_—_
DW-...
-—
-
2 469
gf
1
BY(t-2)
2
1964-76
(n=££)
1. 087
(.491)
2
. 520
(122)
2
7. 290
(2.141)
__ 21.203
(.278)
2
.%85
.
(.326)
_. — _-_-__„____ — ___.._.._ .897
___ —
.-_..__ .915
__-___„____________ 2.336
26
TABLE
8—Continued
Series 2:
Constant
MlYP(t-2)__
___
Dummy....
PIMYPW__
.
_____
_
( ) ]
2
LCPIYP)t-2) J Bl(t-2)_
1954-75
(n=££)
*. 794
(.427)
. . . 2. 5 3 6
(.123)
2
6. 227
(1.908)
__ 21. 192
(.299)
:
Adj m
__
—
.__-_
1
__-_-_
SE
_ _
___
__•
,_•
_
__
DW..__.
_--__._
1
Coefficient is not significant.
2
Coefficient is significant.
Note: Standard errors are in parentheses below the coefficients.
2
. o2^
' i. 194
.242
—
.893
_
.932
2. 124
In analyzing the effects of public policy on inflation, attention usually is
focused on the role played by deficits. It often is asserted and is widely believed
that deficits are the principal cause of inflation. For example, Federal Reserve
Board Chairman Arthur Burns, testifying before the full Committee last
April declared:
The fact is that the inflation started in the mid-1960's and was mainly caused by the
large deficits, continued year after year, in the Federal budget. As a result of the excess
demand created by a persistently loose fiscal policy, a spiral of wages
and prices got under
way in the private sector and the rate of inflation began to quicken.18
Our results refute the hypothesis that deficits, actual or High Employment,
directly cause inflation. This is not surprising in the case of actual deficits for
actual deficits reflect induced as well as purposeful elements. This clouds the
relationship of actual deficits and inflation. As Mr. Perry testified—
Big swings in the budget are largely induced swings. They are not changes that represent
active movements in the budget designed to move the economy in a particular way. Rather,
they reflect the automatic stabilizers of the budget. . . . So the budget deficits as actually
experienced mirror the health or lack of it in the economy . . . they don't show the sort of
correlation the public frequently imagines between inflation and deficits.
A different reason why actual deficits do not show up concurrently or later
in higher inflation is that, given the stock of money, financing Government
deficit spending by selling securities tends to cause compensating changes in
private deficit spending. New Government borrowing "crowds out" private
spending, and conversely, surpluses "fund" the private sector. Actual Government deficits will raise consumers1 prices only to the extent that Government
borrowing is financed abroad, or finances more consumption relative to investment than the private borrowing it replaces.
Neither the crowding out or automatic stabilizer arguments can be used,
however, to explain why we did not find a significant inflationary impact in
the case of High Employment deficits. The High Employment deficit is computed by eliminating induced elements from the actual deficit. Its relationship
to inflation should not be clouded by the automatic tendency to run deficits
in recessions and surpluses in inflation periods. Similarly, because actual
borrowing bears little relationship to the High Employment deficit, the
crowding out thesis can't be applied to the High Employment deficit. In 1974,
for example, the High Employment budget showed a surplus of $25.4 billion
while the actual budget was $11.7 billion in deficit. The puzzle is further complicated because, as will be discussed in the next part of the report, the High
Employment deficit expressed as a percent of potential GNP, has a small,
marginally significant effect on real GNP in the same year. Why then isn't
the High Employment deficit a significant inflationary force? M Frankly, this
M Honorable Arthur F. Burns, Statement before the Committee on Banking, Currency and Housing, April 9, 1976
p. 5.
i* As shown by the results in Series 1 in Table 8, the coefficient has the right sign but is less than twice its standard
error, and hence must be regarded as not statistically different from aero. Deducting one standard error yields an inflationary effect only .06% for each 1% the High Employment deficit is of potential GNP.
27
puzzle remains to be solved. At the moment we can only provide a plausible
speculation. To wit, deficits that would be run were the economy operating at
high employment, are self correcting. Immediately, they can stimulate production either directly by purchases (to order) of new goods and services, or
indirectly by reducing existing tax disincentives to work and investment.
However, to the extent that the economy would be stimulated by these High
Employment deficits, government revenues will rise and they will change into
surpluses* Moreover, under the current tax rate structure, the reversal will
happen very quickly in inflation. The result is that High Employment fiscal
stimulus, holding the money supply constant, peters out before enough time
can pass for prices to respond.
Actual deficits financed by creating new money are another matter. In
this case, there is crowding out not only in financial markets, but in goods and
services markets. Prices are bid up directly. We have, therefore, some sympathy
for Chairman Burns' remarks. Deficits can create enormous problems for the
Federal Eeserve, and be transmitted to the CPI through Federal Reserve
money supply policies. Definitely, this has happened at some points in the
past. As Governor Partee told the Subcommittee:
I think we must recognize that the Federal deficit's inflationary effect may well show
up in monetary policy, in the monetary aggregates, because a large Federal deficit will,
other things being equal, put substantial upward pressure on interest rates and it will tend
to induce a faster rate of monetary expansion just in order to hold the system together.
Therefore, a large Federal deficit may—not as a deliberate matter, and not because the
Secretary of the Treasury says it must be so—it may give you a higher monetary growth
rate, and then the monetary growth rate is later reflected in inflation.
The point, in short, is that M-l growth is our economy's inflation conductor. To the extent that large deficits (actual) have compelled the Federal
Reserve to accelerate M-l growth in the past, as Governor Partee states,
then, in some ultimate or first-cause sense, inflation is "mainly caused by the
large deficits/' as Chairman Burns asserts. However, the chain of causation
from deficits to accelerated money growth to inflation contains weak links
both as a matter of history and logic, as discussed next.
As a matter of historical record, years of large M-l growth are not always
years when deficits are large (relative to GNP). For example, in 1973, M-l
growth was 7.46%, higher than any other year in the 1952-1975 period, although the deficit was only % of 1% of GNP. (For the entire 1952-1975 period,
the deficit averaged 1% of GNP.) In 1969, M-l growth w;as 6.02%, fourth
highest in the 1952-1975 period, although the budget was in surplus. Moreover, large deficits are not always associated with rapid M-l growth. By far the
largest deficit occurred in 1975 when M-l growth averaged only 4.25%.
Another large deficit occurred in 1958 when M-l growth was only 3.05%.
Thus there is strong evidence both that rapid money growth can occur in the
presence of moderate and small deficits and even surpluses, and that large
deficits need not lead to rapid money growth. Finally, as a matter of logic,
there is no reason why the Federal Reserve's money policy should be shaped
in any substantial way by the size of the Treasury's deficits. There is no statute
requiring that the Fed accommodate or ease the Treasury's financing problems,
and the evidence strongly indicates it shouldn't.
PREDICTIONS
The behavior of economic variables is determined by the joint and simultaneous operation of a number of economic relations. Our model, though admittedly a simplification of the complexities of reality, captures the crucial
features of the inflation process. From the understanding that the model gives
of the system, we may predict the future course of inflation, and possibly also
control it to improve economic performance and welfare. For example, the
regression that covers 1947-1975 which is reported in Series 3 of Table 4 may
be applied. It closely tracks our inflation experience, year to year, throughout?
the post World War II period using only four explanatory variables, viz.
• M-l growth 2 years earlier,
• Korean War buying trends,
• Import price changes, and
• The momentum of past inflation.
28
Exhibit 5 graphs year-to-year percentage changes in the CPI predicted
by this equation and actual CPI percentage changes during the 1947-1975
period. The predicted and actual CPI inflation rates also are tabulated in
Table 9 along with the differences between them from year to year.
TABLE 9—PREDICTED AND ACTUAL INFLATION RATES USING THE 1947-75 SERIES 3 TABLE A REGRESSION CPI YP i
Predict'
Diff a
Diff 2 *
14.47
7.67
.99
1.06
7.94
2.28
.77
.35
.26
1.47
3.40
2.73
.92
1.51
1.07
1.17
1.25
1.32
1.59
2.99
2.78
4.21
5.42
5.90
4.26
3.31
6.22
11.04
9.13
13.65
6.94
1.63
-.87
7.75
3.63
-.52
.59
.29
1.48
3.12
1.52
.65
.92
3.09
.22
2.00
2.11
2.49
3.33
3.68
3.55
3.68
6.00
4.94
3.24
6.47
11.69
7.76
.82
.73
-2.62
1.94
.19
-1.35
1.30
-.24
-.55
-.01
.27
1.21
.28
.59
-2.02
.95
-.75
-.79
-.90
-.34
-.90
.66
1.74
-.10
-.68
.06
-.25
-.65
1.37
.67
.53
6.88
3.75
.04
1.81
1.68
.06
.30
.00
.07
1.47
.08
.34
4.08
.90
.57
.63
.82
.11
.81
.44
3.03
.01
.47
.00
.06
.42
1.88
-2.62
1.94
-.00
1.07
.00
6.88
1.10
1.56
Year
1947
1948....
1949
1950
1951
1952
1953
1954
1955
1956
1957.
1958
1959
1960
1961
1962
1963
1964
19S5
1966
1957.
1968.
1969.
1970
1971
1972
1973
1974
1975
1976
Minimum
Maximum
Mean
Standard deviation.
i Yearly average, CPI, percent change.
a.535+0.643xMlYP(t-2)+6Xdummy+1.403XPIMYPW+0.248xlCPIYP(t-l)-CPIYP(t-2)l.
»CPIYP—Predict.
< DiffXDiff.
The single largest error or difference occurs in 1949 when the equation
overpredicts inflation by 2.62%. The reason for the overprediction is that 1949
was marked by sharp increases in agricultural and raw materials supplies which
generated substantial exogenous and unpredictable downward pressure on the
food and apparel and upkeep components of the CPI. These components fell
4% and 3.8% respectively, in 1949.
Further examination of the differences between actual and predicted
inflation rates reveals no other years when exogenous supply changes played a
major role in our post World War II inflation experience. Nor does such anatysis
provide evidence that administered pricing or collective bargaining exerted
upward pressure on prices separate from the roles they play in generating inflation momentum, which is captured by the regression coefficient on last
year's change in the inflation rate. On the other hand, the wage-price freeze in
1971 undoubtedly helps to account for overpredicting inflation that year. Also
the persistent pattern of overpredictions from 1963 to 1967, averaging .75% per
year, is consistent with the view that the Kennedy-Johnson Administration
guidelines served to damp (or repress) inflation. We would caution, however,
that the apparent success of the guidelines may have been due to the relatively
low inflation base in the immediately prior years (1959-1962). Guidelines may
be more difficult to implement successfully today because recent inflationary
experience probably has made all of us less receptive to White House requests
I/O hold the line on wages and prices.
As observed earlier, our results should not be judged by how well they
explain individual years. They must be looked at as statements of average
tendencies over the long haul, and judged by their overall or average predictive power. Only time will tell whether our regressions will predict and track
future inflation as well as they approximate past inflation. Naturally, our bet is
that they will.
EXHIBIT 5
CONSUMER PRICE
YEAR
TO YERR
INDEX
P E R C E N T CHflNGE
• F°RAcnsT ^
.403 PIMYPW + .248[CPIYP(t:-l)-CPIYP(c-2j]
54 '55 '56 '57 '58 '59 '60 '61
•63 '64 '65 '66 '67 '68 '69 '70 '71
YEflRLY DflTfl
Lines from the time axis delineate recession periods.
30
In this regard, it is important to stress that although our regressions can
be used to predict ahead, any forecasts made with them are conditional and
subject to a known range of error. Using the 1947-1975 Series 3 Table 4 regreS'sion to predict the CPI inflation rate in say 1980 we must know the percfcntagfc
change in weighted import prices that year, the change in the QP1 inflation
rate between 1978 and 1979, and the percentage growth in M - l in 1978. Fot
1976, we need M-l growth of 1974 and the change in the inflation rate betweto
1974 and 1975, both of which are available now, and the change in weighted
import prices in 1976 which is not. For next year (1977), only M - l growth now
is in hand. In using our regressions to forecast next year's inflation rate, however, we would also caution that the M-l growth statistics now published for
1975 are likely to be revised upwards by about .6%.
Using the 1947-1975 cornerstone Series 3 Table 4 regression, and guessing
that weighted import prices will rise %% this year over 1975, our bold or precise
estimate is that the CPI will average 4.3% higher in 1976 than in 1975. We are,
moreover, 95% confident that the increase will range between 2% and 6.6%
and would bet 2 to 1 that the rise in prices will range between 3.1 and 5.5%.15
At the moment, it appears that the increases will be near the top of our 2 to 1
confidence interval. Assuming that this year's increase turns out to be 5.5%,
and import prices (weighted) rise 1% in 1977, our 2 to 1 bet for 1977 is for
inflation to taper off further to 3.8% plus or minus 1.2%.
Other equations might be used to predict future inflation. One is an adjusted version of the 1947-1970 Table 5 regression. The adjustment involves
adding the change in weighted import prices to the regression forecast to obtain
the final forecast.
The rationale for using this procedure is that, as previously discussed, our
results tend to confirm the hypothesis that on balance domestic prices are
not substantially affected by changes in import prices. This being true, it
follows that import price changes can be regarded as exogenous and operating
to cause the CPI to change exactly in proportion to the change in weighted
import prices. We now know that import prices can affect the CPI (though not
domestic prices). Using this knowledge together with the adjusted 1947-1970
Table 5 regression in the 1971-1975 subperiod, provides a post sample test of
the usefulness of our results for forecasting purposes. Predicted inflation rates
using this procedure are given in Table 10 along with actual CPI inflation for
the 1971-1975 post-regression period. Also given are forecasts for 1976 and
1977, the former assuming weighted import prices rise only %% this year, and
the latter assuming a 1% increase in 1977, and that 1976 inflation equals 5.5% v
Inspection of the 1971-1975 predictions reveals that this equation is useful for
forecasting purposes
now that we know substantially the role played by
import prices.16
TABLE 10.—ACTUAL INFLATION RATES AND PREDICTIONS USING THE ADJUSTED 1947-70 TABLE 5 REGRESSION
1971
1972
1973....
1974
1975.
1976
1977
Mean square error
Actual
dieted
Difference
4.26
3.31
6.22
11.04
9.13
4.96
3.18
6.15
10.08
7.45
M.3
3 3.6
-.70
.13
.07
.96
1.68
.85
i The 95 percent confidence limits are 2 to 6.6 percent It is a 2 to 1 bet that the CPI increase will range between 3.2 and 5.4 percent.
a The 95 percent confidence limits are 1.2 to 6 percent The odds are 2 to 1 that the increase will range between 2.4 and 4.8 percent.
is The 95% confidence range equals the precise estimate (4.3%) plus and minus two regression standard errors (1.15%).
The 2 to 1 limits=4.3% plus and minus 1.2%.
w As was discussed earlier, the momentum of a change in inflation last year probably carries over to the current year
with greater force than the .212 estimated by the 1947-1970 Table 5 regression. We are confident that the true value of the
carry-over coefficient lies between .2 and .5, and would give odds that it is between .3 and .4. In most years, it does not
matter very much whether the carry-over coefficient is .2 or .5. In recent years, however, the inflation rate has jumped
up and down enough for it to matter very much whether momentum equals .2 or .5. Using .35 to estimate the carry-over,
for example, increases predicted inflation rates in 1971, 1974 and 1975 to 5.03, 10.49 and 8.12 percents respectively, and
decreases them to 2.96, 6.02, 4.00 and 2.63 percents in 1972,1973,1976 and 1977. For the 1971-1975 period, the mean square
error is .42, which is substantially better than obtained using the adjusted 1947-1970 Table 5 regression. However, the
comparison may not prove so favorable to the higher carry-over assumption after the results for 1976 and 1977 are known.
Time and further research are needed to find a carry-over coefficient whose value we can be confident about. •---"
31
CONTRIBUTIONS
Finally, we set forth in Table 11 the specific yearly contributions to inflation in the 1947-1975 period of lagged M - l growth, import prices, the momentum of past inflation and the Korean War buying spree as assigned by the
cornerstone regression covering 1947-1975, which is reported in Series 3 of
Table 4, and the adjusted Table 5 regression for 1947-1970. These data show
that M - l growth is the key to understanding and controlling inflation. In
specific,
• M - l growth in excess of our economy's potential to increase production
can cause inflation by itself.
• Virulent inflation can't endure without corresponding M-l growth.
• Inflation will persist and gain momentum no matter how it starts if it
is accommodated by M - l growth in excess of our economy's potential to
increase production.
During the 1947-1975 period, we experienced four separate waves of inflation—1947-1948, 1951, 1966-1971, and 1973-1975. The first had its roots in
World War I I . It subsided rapidly because M - l growth declined rapidly after
1945. The second was triggered by the Korean War. It lasted only a year because M - l growth was not rapid enough to sustain it. The third wave of
inflation began to build up with escalation of the Vietnam War in the middle
1960's. It gained momentum throughout the late! 1960's, peaked in 1971 and
quickly tapered off. The course of the Vietnam War inflation was almost
entirely predictable by looking at past M - l growth.
TABLE 11.—CONTRIBUTIONS TO INFLATION, 1947-75
Dummy
M1YP (t-2)
Year
1947
1948
1949
1950
1951
1952
1953
1954
1955 . . . .
1956
1957
1958
1959
I960
1961
1962
1963
1964
.
1965
1966
1967
1968
1969
1970
1971,.,,....
1972
".I
1973
1974
1975
<D
10.49
4.42
3.05
11,15
4.99
3.43
.30
.32
-.66
1.73
2.86
3.24
1.59
-.71
1.84
3.05
3.45
1.70
1.04
2.18
.98
2.05
.77
.34
.76
.82
.37
.81
2.46
-.07
1.36
1.41
1.87
2.56
2.75
3.00
2.55
4.60
3.87
2.48
4.29
4.55
4.80
2.62
-.07
1.44
1.50
1.99
2.73
2.93
3.19
2.72
4.90
4.12
2.64
4.57
4.84
5.11
Import prices
(3)
(2)
(4)
0
0
0
0
0
0
0
0
6.00
-.13
-2.25
-2.89
-1.73
6.49
-.14
-2.44
-3.13
-1.87
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(5)
1.09
.50
-.27
.44
Inflation carryover
(7)
(8)
0.78
1.54
1.49
-1.69
-2.15
1 31
1.27
-1.44
—1.84
1.71
-1.40
-.37
-.10
—.15
1.44
-1.20
—.32
-.09
—.13
.06
-.17
r-.45
-.14
-.38
-.03
-.09
.14
.12
-.11
-.09
.02
.02
.02
.07
.35
.02
.02
.01
.06
.30
-.05
-.05
—.41
-.24
—.35
—.20
1.19
1.02
.36
-.19
.32
.97
1.37
-.21
-.27
-.15
-.19
-.01
-.01
-.25
-.07
—.18
-.05
.09
.13
.10
.08
-.05
-.13
.08
.14
.07
.16
.03
.08
.23
.58
.42
.63
1.87
5.89
1.24
{
(6)
.06
.09
.07
.06
.10
.05
.11
.02
.06
.17
.41
.30
.45
1.33
4.20
.88
.51
.43
.48
.36
.30
.12
.72
.44
.37
.41
.30
.26
.10
.62
Note: (1) .643XM1YP (t-2); (2) .685XM1YP (t-2); (3) 6XDummy; (4) 6.493XDummy; (5) 1.403XPIMYPW; (6) PIMYPW; (7) .248
XJCPIYP (M)-CPIYP (t-2)J; (8) .212X[CPIYP (M)-CPIYP (t-2)].
32
As shown in Table 11, actual inflation and inflation predicted by M - l
growth moved together from 1966 to 1972 excepting only in 1969.17
The fourth wave of inflation began in 1973. It became exceedingly virulent
in 1974 and began to subside in 1975. The pattern of this latest inflation episode
(through 1975) has been dictated primarily by changes in import prices, and
to a lesser extent by its own momentum. The underlying trend, on the other
hand, throughout the 1973-1975 period was based on past M - l growth. From
1971 to 1973, yearly M - l growth ranged between 6.7% and 7.5%. These rates
will accommodate so-called "steady state" or continuing yearly CPI increases
averaging 4.6% to 5.1% and ranging up to 6.7%.
It is widely believed today that our underlying or base inflation rate is 6%7%, and that other factors will increase inflation next year and in 1978 to the
§%-9% range. Gur results require us to disagree on both counts. The base
inflation rate may have been 6%-7% last year. Such a range can be obtained
by adding the precise estimates of inflation expected as a result of 1973 M - l
growth (5.0%-5.3%) to the expected contribution from the 1974 acceleration
of inflation (1.0%-1.2%) and the constant term (.4%-.5%). But that was last
year. The momentum is now down not up, and furthermore, M - l growth was
reduced after 1973. Using the relevant M - l growth (1974-1975) and taking
into account that the momentum is now down, we believe that our base domestic
inflation rate is now 3%-5% and will drop to the 2%-4% range in 1977. However, for 1978, using M-l statistics now published for 1976, our analysis
leads us to forecast an increase in base inflation back to the 3%-5% range
because M-l growth accelerated a little more than 1% in 1976. To the extent
that currently available 1976 M-l statistics are revised upwards, our 1978
inflation forecast will be low. Present indications are that M-l will be revised
upwards l % - 2 % this year, which would require raising the prediction of base
inflation in 1978 by %% to 1%. Looking beyond 1978, our forecast is for gradual
reduction of the base domestic inflation rate to the l % - 3 % level, if, in contrast to 1965-1973, the Federal Reserve follows a non-inflationary course,
gradually reducing money growth from the present 4K%-6%% range to 2%-4%
beginning gradually in 1977 as unemployment recedes.
" T h e relevant data are as follows:
Inflation attributable to
Actual lagged M-l growth (percent)
inflation •
(percent)
Column 11
Column 2 *
Year:
1966....
1967....
1968
1969
1970..
1971
1972....
,
^
^
Average
Percent of the average actual inflation rate attributable to lagged
M-l growth
» Source: Col. (1) of table 11.
a Source: Col. (2) of table 11.
2.99
2.78
4.21
5.42
5.90
4.26
3.31
2.56
2.75
3.00
2.55
4.60
3.87
2.48
2.73
2.93
3.19
2.72
4.90
4.12
2.64
4.12
3.12
3.32
75.7
81.0
PART III
RECESSION
Since 1947, this country has suffered through six recessions and one minirecession. The first of these episodes began in November 1948 after the rapid
and severe inflation of 1946-1948 had been broken. It ended eleven months
later in October 1949. The latest episode began in November 1973, picked up
force after mid-1974 and ended in March 1975. The dates and major characteristics of U.S. recessions in the 1947-1975 period are given in Table 12. To
show these episodes in proper context, we also have set forth the dates and
major characteristics of the great depression of 1929-1933.
Sharp prolonged decreases in money growth preceded each recession
episode. Moreover, sustained monetary expansions helped to promote recovery
in each case. These fundamental facts are shown in Exhibit 6. This exhibit plots
percentage changes in M-l growth between the same months from one year to
the next from 1948 through September 1976, and delineates recession periods
with lines from the time axis. It provides powerful graphic evidence that trends
in economic activity are closely related to trends in money growth. Specifically,
(1) slowing money growth sharply for any prolonged period weakens the economy and increases the risk of recession; (2) once underway, recessions are
aggravated by continuing to retard money supply growth; and (3) monetary
expansion in and immediately after recessions, promotes recovery. Money
supply, however, is not the only determinant of economic activity.
TABLE 12.-RECESSION DATES AND MAJOR CHARACTERISTICS, 1947-75, AND THE GREAT DEPRESSION OF THE 1930's
Largest 4-quarter decline in GNP
Duration
(months)
Dates
November 1948 to October 1949 _.
July 1953 to Way 1954
July 1957 to April 1958
.
April 1960 to February 1961
October 1966 to October 1967
December 1969 to November 1970
November 1973 to March 1975
,_
11
10
9
10
12
11
16
Percent
Terminal
quarter
-1.35
-3.29
-2.5
-.55
1949-4
1954-2
1958-1
1961-1
-.8
-5.63
1970-4
1975-1
Peak unemployment
Rate (percent)
Date
7.9 October 1949.
6.1 September 1954.
7.5 July 1958.
7.1 May 1961.
4.0 October 1967.
6.1 December 1970.
8.9 May 1975.
i Growth slowed but did not fall. It slowed from 7.70 percent for the 4 quarters ending 1965-4 to 4.29 percent in the 4 quarters ending
1966-4, and to 2.56 percent in the 4 quarters ending in both 1967-1 and 1967-2.
Duration
(n.onths)
The Great Depression
August 1929 to March 1933
1
Largest calendar year decline
43
Yearly average
Percent
Year
Rate (percent)
Year
1-13.49
1932
24.9
1933
i This fall followed decreases of 9.26 percent in 1930 and 7.48 percent in 1931.
VELOCITY
Changes in money's velocity also play an important part in recessions.
Exhibit 7 charts yearly percentage changes in M-l velocity between the same
quarters from one year to the next from 1948 to 1976. As noted earlier, velocity
equals current dollar GNP divided by M - l . The chart shows the extraordinary
rise in velocity during the early stages of the Korean War, and a roller-coaster
pattern much like that exhibited by M-l growth thereafter. However, changes
in velocity did not tilt upwards as did money growth in the late 1960's and
early 1970's, when inflation was relatively high. As was observed earlier, aside
from the Korean War period, price trends were not related to velocity trends.
The impact of velocity changes showed up rather in changes in economic
activity.
(33)
EXHIBIT 6
12.0
PERCENT CHRNGE IN M-l MONEY SUPPLY
-12.0
YERR TO YEflR
AND RECESSION PERIODS, 1948-1976
10.0'
-10.0
8.0
8.0
6.0UJ
o
UJ
4.0-
2.0--
-2.0
0.0-
-o.o
»47 »48
T
61 '62 >G3 I 6 4 '65 (6B
MONTHLY DRTF1
70*71 l 72 »73 »74 «7B »7B
-2.0
35
As depicted in Exhibit 7, changes in velocity have been pro-cyclical. The
trend in velocity moved sharply down together with economic activity in
every recession, though the downturns in 1969-1970 and 1973-1975 were not
nearly as severe as in earlier recessions. Also in every episode, the trend in
velocity changed from down to up very nearly coincident with recovery. The
upturn was especially powerful in 1975, and appears to have been the dominant
force underlying recovery from the 1973-1975 recession.
To'obtain a more comprehensive picture of the forces at work in economic
cycles, changes in velocity must be looked at in conjunction with money supply
changes. Several conclusions emerge from studying Exhibits 6 and 7 together.
In specific—• Year to year money growth slowed well in advance of the recession
periods, while changes in year to year velocity peaked in the same quarter
that economic activity did in 1957, 1966, 1969, and 1973; and one quarter
earlier in 1953 and I960.1 It would appear from these data that for a time rising
velocity will sustain an expansion even in the face of a slowdown in money
growth. Inevitably however, a sharp prolonged slowdown in money growth
undermines confidence, which in turn causes the trend in velocity to peak and
a recession to follow.
• Year to year money growth and velocity accelerated together at the
same time that ^^onomic activity turned up in 1949, 1954 and 1975. In 1958,
money growth accelerated one quarter before the trend in velocity turned
from down tc up and economic activity rose. In 1961, 1967 and 1970, money
growth turned up well in advance of the trend in velocity and economic
activity, It would appear from these data that recessions endure as long as
the trend in velocity is down. However, if money growth is maintained commensurate with the economy's production potential, inevitably confidence is
restored, the velocity trend bottoms, and recovery begins.
In summary, during the 1948-1975 period, the effects of money supply and
velocity changes on economic activity usually have been procyclical. To their
credit, Federal Reserve policymakers usually acted quickly to accelerate
money growth in recessions. By repairing the economy's monetary foundation,
they increased the likelihood of a strong recovery occurring when, inevitably,
the trend in velocity turned up, whether by chance or because confidence
inevitably returns with resumption of monetary growth. On the other hand,
seven times in the past thirty years, the Fed caused or failed to prevent money
growth from slowing sharply for a period long enough to undermine the
economy's monetary foundation and increase the risk of recession. Moderate,
stable money growth might not have prevented these recessions, but it most
certainly would have greatly diminished their severity.
DETERMINANTS OF CHANGES IN REAL
GNP
To further our understanding of how recessions develop and end, we used
linear multiple regression analysis to quantify the effects on real or constant
dollar GNP of concurrent changes in money supply and the following other
factors:
• Percentage changes in the CPI, and alternatively in domestic and
import prices; all lagged one year.
• The change in the inflation rate last year, lagged a year, and alternatively, entered concurrently.
• Unemployment lagged one year.
• Concurrent changes in the high employment Federal deficit scaled by
potential GNPT
As in the case of our inflation regressions, the data were assembled in yearly
series by averaging intra-year data. Changes were measured as percentages
from the year earlier average. The results are given in Table 13.
i Data limitations prevent determining the timing of the change in velocity's trend in relation to the 1948-1949 recession.
EXHIBIT 7
11.0
ll-.O
MONEY VELOCITY (CURRENT GNP / Ml)
PERCENT C H R N G E P YEflR TO YERR
9.0
LLJ
n I
1.0
-3.0
47 '48^49 '50 »6l
'62
E '63 »B4 ! 65 «6B "67 !B8
QURRTERLY
f
!
'69 70 '71 72 73 '74 '75
'76
3.0
37
Series 1
The Series 1 Table 13 regressions relate yearly percentage changes in constant
dollar GNP to (l) concurrent yearly percentage changes in M - l ; (2) the yearly
average unemployment rate lagged a year; (3) yearly percentage changes in the
CPI lagged a year; (4) inflation momentum as measured by the change in the
inflation rate last year, entered concurrently tod alternatively lagged a year in
relation to constant dollar GNP; aiid (5) the High Employment Federal deficit
scaled by potential GNP and entered concurrently.
Series 2
The Series 2 Table 13 regressions replace changes in the CPI with changes in
the domestic and import components of prices. The sole purpose of this series
was to test whether increases in import prices cause recession as well as inflation.
TABLE 13
REGRESSION COEFFICIENTS AND STANDARD ERRORS
Dependent variable: GNPJfCON, as already defined.
Independent variables: As already defined, MIYP, CPIYP(t—l), [CPIYP
{t-l)—CPIYP{t-2)]{t-l)y
UNY(t-l), and BY in Series 1, and DCPIYP
(t-1) and PIMYPW{t-l) in Series 2.
1953-^75
(n=23)
Series 1:
Constant
22. 550
(.769)
2.806
(• 199)
MIYP
2-.870
(.174)
CPIYP(t-l)
1953-75
(n=23)
1953-75
(n=23)
1953-75
(n-23)
1953-75
(n=23)
1-2.293
(1.421)
2.747
(.156)
2 - . 850
(. 135)
2-2.880
(1.210)
2.656
(.134)
2 - . 888
.114
2-3.046
(1.239)
2.640
(. 137)
2 - . 805
(.155)
1.148
(.180)
2-2.528
(. 896)
2.728
(.101)
.. 2-1.011
(.090)
rCPIYP(t-lH
LCPIYP(t-2) J
a.393
(.098)
rcpiYp(t-i)-i .
LGPIYP(t-2) J - l
BY
UNY(t-l)
Adi R2
SE
:"
DW
::_
.548
1.803
2.510
Series 2:
Constant
12.663
(1.564)
2.738
(. 173)
2 - . 780
(.257)
2-1.234
(.409)
MIYP
DCPIYP(t-l)
PIMYPW(t-l)
BY
21.094
(.293)
UNY(t-l)
rCPIYP(t-l)-l
LCPIYP(t-2) J - l
Adi R2
s aj
21.033
.27b
.727
1.402
2.309
.
DW
.722
1.414
2.408
21.003
(.336)
21.325
(.251)
. 807
1.177
2.674
2-3.514
(1.316)
2.614
(.148)
2 - . 727
(.212)
2-1.437
(-342)
21 043
(.335)
21.437
(.265)
.812
1.162
2.807
2.875
(.374)
21.317
(.253)
.803
1.190
2.567
2.552
(.272)
21.217
.187
.895
.868
2.846
2-2.650
(.983)
2.749
(.113)
3-1.022
(.171)
a-1.312
(.251)
i 530
(276)
21,243
(. 199)
2.402
(.101)
.900
.847
2.825
» Coefficient is not significant.
2 Coefficient is significant.
Note: Standard errors are in parentheses below the coefficients.
Last year's unemployment rate is especially important for analyzing
changes in real (constant dollar) GNP. Inclusion of this variable in the analysis
substantially improves our ability to explain changes in real GNP, as shown by
comparing R2 values in columns 1 and 2 of Series 1. Confirming that since
World War II, the economy has not imploded, the higher unemployment was a
year ago, the more, on average, real GNP rose in the current year. Vice versa, the
lower unemployment was a year ago, the less it rose in the current year. Specifically, for every percentage point of unemployment a year ago, the regressions
show that output increased, on average, in the current year by 1.2%. For example,
38
with a base past unemployment rate equal to 4%, real GNP will tend to grow by
4.8% in the current year, abstracting from other influences. Higher base unemployment means that, on average, the economy will grow even faster.
The Table 13 regressions also throw substantial light on how year to year
changes in real GNP are affected by changes in money supply, inflation, the
Federal deficit, and import prices. In specific—
(1) Increases in import prices (weighted by GNP) cause real GNP to fall
a year later. The percentage fall almost exactly equals the percentage rise
that occurs in the CPI the year that import prices rise. As estimated by the
regressions in Series 2 of Table 13 and Series 3 of Table 4, for each percent
yearly increase in weighted import prices, on average, the CPI rises the same
year by 1.3%-1.4% and a year later real GNP Falls 1.3%. Neither change
differs significantly from 1%.
Thus our results confirm that increases in import prices cause both inflation and recession in equal degree. The CPI rises roughly proportionally
to import prices weighted by imports as a percent of GNP in years when import
prices rise. Real GNP decreases roughly proportionally a year later.
(2) The High Employment Federal deficit scaled by potential GNP was a
marginally significant determinant of real GNP in the 1953-1975 period. In
years when the deficit equalled 1% of potential (current dollar) GNP, real
GNP tended to rise by .55%.
(3) In regard to the impact of changes m money supply: Using the column
5 regression in Series 1 of Table 13, for each 1% that money supply grows
yearly, real GNP increases on average by .728% concurrently. In subsequent
years, however, the rise in output which is induced by accelerated money
growth recedes sharply. Based on our results, it levels off at .06% after eleven
years during which it averages .21%. The fall-back comes largely from feedback
from the inflation effects of money growth.
INFLATION FEEDBACK
In our model, yearly real GNP growth is affected by inflation in two ways.
It falls very nearly at the same rate as the CPI rose the previous year, and it
rises the year after the inflation rate rises. These effects are quantified using
the regression results in column 5 of Series 1, Table 13, together with the cornerstone regression of Series 3 Table 4. As shown by that equation, the rate of
inflation rises on average by .643% two years after a 1% rise in yearly money
growth. The results given here in Table 13 show that for each 1% rise in the
CPI in the current year, real GNP falls on average by a little more than 1%
(1.011%) next year. Because of this feedback, the initial .728% rise in real
GNP which results from accelerating M-l growth by 1% recedes by .65%
three years later.
The scenario becomes quite complicated from then on. This is because
inflation has momentum, and the carry-over has complicated feedback. On
the one hand, as the Table 13 results show, the momentum of inflation provides
a spur to output, boosting it two years later by .393% for each 1% jump in
the rate of inflation. On the other hand, increases in the inflation rate this year
cause a further rise in inflation next year which, in turn, causes new depressing
effects on output the year after that. Both these effects, however, are ephemeral.
In time, inflation carry-over reduces to zero and we are left with only the
direct effects. For a 1% yearly increase in M-l, the direct effects add to a
yearly increase of .078% in real GNP (.728%-.650%). We estimate that it
takes eleven years to reach the equilibrium. During this period, real GNP
rises yearly, on average, by .26%. Table 14 summarizes the model's direct
and inflation feedback effects from a 1% increase in yearly M-l growth.
Further downward adjustments both in the equilibrium and average rate
of rise in real GNP are required to take into account that unemployment is
decreased by output increases and, in turn, this constrains future output
increases. We estimate the unemployment adjustments reduce the yearly
average increase in real GNP to .21% and the final equilibrium rate of rise to
.06%.
T A B L E 14
DIRECT AND FEEDBACK EFFECTS
OF.1% YEARLY M-l GROWTFI
Effect of
Effect of
Year
t
t+1.
t+2
t+3
t+4
t+5
Change
Added
Direct Effects Effect of Direct
in
Inflation
CPIYP
CPIYP Change on
on
CPIYP
Carry-over
on
GNPPCON
GNPPCON
GNPPCON CPIYP
• 728
• 643
-.650
t+7
t+8
t+9
t+10
.159
.039
-.030
-.017
.003
.005
-.161
-.040
+ .030
+ .017
-.003
-.005
' .643
.159
-.120
-.069
.013
.020
.002
,
.25 3
.063
-.047
-.027
.005
.008
.001
Unemployment
Adjustment
GNPPCON on GNPPCON
Sum
.728
.728
.728
.078
.170
.101
.061
.068
.080
.081
.078
-.295
-.176
~. 224
+ .060
-.094
-.002
-.024
-.018
-.026
-.022
.26
Average
CPIYP
Change on
GNPPCON
Explanatory notes:
(1) The direct effects are given by the column 5 regression of Series 1 Table 13 and the
cornerstone regression in Series 3 Table 4.
(2) The effect of CPIYP equals -1.011 times .643, where -1.011 is the CPIYP coefficient in
the Series' 1 Table 13 column 5 regression.
(3) In (t+3)/ CPIYP carry-over equals .248 times the increase in inflation in (t+2).
This
Was .643/ the direct effect of increasing yearly money growth 1%. Thus, the carry-over equals
•159 and the incremental inflation from time zero rises to .802. In.(t+4), carry-over equals
.'248 times the jump in inflation in (t+3).
This was [[.802 - .643^ or .159. Thus, the
carry-over equals .039 and incremental inflation from time zero drops to £.64 3 + .039^ or
•682.
In (t+5), carry-over equals .248 times the change in inflation in (t+4). This was
£.682 - .8023 or -.120.
Thus the carry-over equals -.030 and incremental inflation from
time zero drops to £.643 - . 0303 or .613. Etc.
(4) The effect of added inflation equals -1.011 times the estimate of added inflation a
year ago.
(5) The effect of the change in CPIYP equals .393 times the change in CPIYP two years ago,
where .393 is the column 5 Series 1 Tab?e 13 estimate of the spur to GNPPCON from a change
in the rate of inflation.
(6) The unemployment adjustment equals minus one-third sum GNPPCON in the prior year times
1.217, the unemployment effect on output.
40
It is interesting finally that in our model, abstracting from carryover, the
effects of which dissipate in time,
1% per year money growth generates .77%
yearly inflation and .24% 3rearly real GNP growth. The inflation result is
taken from the 1947-1975 regression in Series 2 of Table 4. The estimated
change in real GNP is taken from the column 1 regression in Series 1 Table 13.
PREDICTIONS AND CONTRIBUTIONS
Year by year percentage changes in real GNP, as assigned by the column
5 regression of Series 1 Table 13 to our independent variables, individually
and in combination, are set forth in Table 15, along with the 1953-1975 history
of percentage changes in real GNP. Actual and predicted changes (using the
regression in its entirety) are graphed in Exhibit 8. The data in Table 15 indicate both the significance of last year's unemployment rate in the sense that
the worse things are the better they are likely to become, and the important
part played by past inflation which, on net, suggests that the worse things are
the worse we have to expect them to become before they get better. These data
also show that past changes in real GNP have been related strategically to
money supply changes and marginally to the High Employment deficit.
PREDICTIONS
Our real GNP regressions can be used to forecast future percentage changes
in constant dollar GNP. As in the case of the inflation regressions, such forecasts necessarily are conditional and subject to error. This year's forecast requires knowing this year's macro-policy parameters—money growth and the
High Employment deficit, as well as past inflation and unemployment rates.
Past inflation and unemployment provide the base on which this year's money
growth and deficit build. Using the column 5 Series 1 regression of Table 13,
the stage for 1976 was set as follows.
Source
— 1.011 times 1975 inflation (9.13%)
.393 times jump in inflation 1974 (4.82%)
1.217 times 1975 average unemployment.
Constant
Net change
Predicted
GNPPCON
_. - 9 . 2
_-.—
1.9
;„,.
10.3
—2. 5
=
T_
.___ = __
= _______
=
.
_-,.
^_^I_J
.5
G4
G5
TABLE 15.-REAL GNP PREDICTIONS AND CONTRIBUTIONS
Year
1953
1954.. . .
1955
1956.
1957
1958
1959
1960
1961
1962
.....
1963
1964
.: ..
1965.
1966
1967
1968
1969
.....
1970.
1971..
1972
1973 .
1974
1975
Gnppcon
3.89
-.30
6.70
2.14
1.81
-.21
6.02
2.28
2.51
5.80
3.95
5.26
5.89
5.95
2.72
4.38
2.57
-,32
2.99'
5.74
5.46
-1.70
-1.84
Predict
Diff
Gl
4.08
-.63
5.25
3.23
-.19
-.67
l.,45
-1.09
1.16
-.91
-.13
.42
-.33
-.17
-.38
-.55
.51
1.06
.13
-.91
1.29
.36
.10
-.38
-.04
-1.32
.41
1.80
1.11
2.32
.87
.65
.70
6.15
1.86
2.84
5.97
4.34
5.81
5.38
4.89
2.59
5.28
1.27
1-.68
2.90
6.12
5.41
-.39
-2.25
.39
.86
2.78
-.08
1.53
1.60
2.12
2.90
3.11
3.39
2.89
5.21
4.38
2.81
4.86
5.15
5.43
3.97
3.10
G2
.73
.24
-.40
-.52
-.50
-.11
-.42
-.98
-.68
-.39
-.67
-.21
-.18
.11
.53
.18
-.57
-.30
-.02
.38
-.32
-1.06
.36
G3
*-2.31
-;78
-.36
.26
~l t 49
. -3.43
-2,76
•" -?93
-1.52
-1.08
-1.18
-1.26
-1.33
-1.60
-3.02
-2.81
-4.26
-5.48 *
-5.96
-4.30
-3.34
-6.29
-11.16
2.70
- 2 . 23 :
r-.59
r-. 17
-.24
.68
* .76
-"26
-.71
.23 ;
-.17
.04
:
•
1
.11
.55
-;08
.56
.19
< .48
-.64
-.37
*1U5
3.6«
3.56
6.8l
5.31
5.02
5.23
8.33
6.63
6.74
8.14
6.77
6.87
6.28
5.49
4.61
4.68
4.33
4.25
6.06
7.24
6.82
5.90
6.84
Note: (1) Gnppcon, Pct(AGNP,l); (2) Predict, -2.528+.728*Miyp+.552*BY-l.bll*CPIYP(t-l)-f.393*A(t-l)4l.
4l.217*UNY(t-l);
(3) Diff. Gnppcon-Predict;(4) Gl, .728*M1YP;(5)G2, .552*BY; (6) G3, -1.011*CPi YP(t-l); (7)G4, .393*A(t-l); (8)G5,1.
51.217*UNY(t-l).
A=lCPiYP(t-l)-CPIYP(t-2)l.
With 1976 M-l growth now forecast to be about 5.5% and the High
Employment deficit about .5% of potential GNP, our precise 1976 forecast
is for 4.8% real GNP growth. This year's M-l growth contributes 4% and the
High Employment deficit contributes . 3 % . We are 95% confident that the
41
increase in real GNP this year will range between 3.1% and 6.5%. The odds are
2 to 1 that the increase will range between 3.9% and 5.7%. A further cautionary
note is required, however. There are indications that this year's M - l statistics
are understated by as much as 2%. If true, our precise forecast becomes 6.3%
and the limits of the forecast ranges also rise, both bottom and top, by 1.5%.
We are likely to start 1977 in much the same position as we began 1976.
The push which this year's estimated 7.6% average unemployment will provide
to real growth next year will be almost fully offset by the depressing effects
of this year's inflation, the fall in inflation a year ago, and the constant term.
Thus, if 1977 M-l growth and High Employment deficit repeat 1976's monetary
and fiscal policies, our analysis forecasts that real GNP growth in 1977 will
repeat the forecasted 1976 rise; i.e., 4.8% plus or minus 1.7% (at the 95%
confidence level). Additional stimulus will be required to achieve higher growth
if that is desired.
As in the case of our inflation analysis, only time will tell whether our
analysis of changes in real GNP captures in significant and substantial degree
the essentials of the underlying macro-economic processes. We shall know by
future experience. If real GNP changes are tracked well, on average, over the
next 5, 10, etc., years by our regressions, we can be reasonably confident that
we have found the heart of these processes. There can be individual years when
actual experience departs substantially from what our analysis forecasts. The
essential point is that the forecasts do well on average.
POLICY IMPLICATIONS
1977-1978
With unemployment above 7%, it would appear highly desirable to aim
for real growth higher than 5% next year. To achieve it requires accelerating
M-l growth and/or increasing the High Employment deficit.
A permanent tax decrease aggregating $15-20 billion would provide mild
stimulus.2 Perhaps most importantly it could reduce present high disincentives
to work and investment, and thereby increase output by increasing incentives
to produce. Alternatively, or even additionally, spending on job-creating
programs can be considered. A major advantage of such a program would be
that it would begin long-needed structural reform on the supply side of the labor
market. A major aim would be to upgrade skills and work habits among younL
persons now unwilling (in part because they have little future without skills
to take work, or unable (in part because they lack even minimal skills) to find it.
In regard to monetary policy, our analysis warns that faster M - l growth
could prove a risky and illusory cure. Our analysis indicates both that faster
money growth this year tends to increase inflation two years from now, and
that higher inflation largely erodes the immediate gains in real GNP from current money growth. Thus we would recommend an upper limit of 6% on M-l
growth in 1977, and aim at 5%.
SOME LONG-RUN POLICY IMPLICATIONS
Our results have important implications for long-run economic policy.
To begin with, our results show that fluctuations in prices abroad are destabilizing domestically. Increases in import prices raise the CPI concurrently
and operate to decrease real GNP the following year. To the extent that policies can be designed to avoid increases in import prices, the problem of achieving economic stability—full employment with minimal inflation—will be greatly
eased. At first glance, this might seem to suggest returning to a fixed exchange
rate regime in international trade. But this does not follow at all. On the contrary, a fixed exchange rate system requires us to be "price takers" in importing
3
Given that our results show that the High Employment deficit (scaled by potential GNP) has a significant, albeit
small and marginal, impact on production but not on prices, it is a fair question to ask why taxes shouldn't be reduced
$60 billion, or even more. There are two reasons why. One is that the actual as well as the High Employment deficit would
rise, creating mammoth pressures and problems in financial markets. The second is that a decrease of this order of magnitude would put the High Employment deficit, as a % of potential GNP, substantially outside the range of past experience. From 1952 through 1975, there have been only eight High Employment deficits, and the highest was only 1.4% of
'" ' GNP,
~ N P , in 1953. A $60 billion High Employment deficit would equal nearly 4%
i% of 1977 potential GNP. It is wise
potential
to move slowly, if at all, where we nave not been before. This is especially true here, since our results provide some warning
that a large High Employment deficit relative to potential GNP could be inflationary. As reported in Table 8, the coefficient of the High Employment deficit has the right sign though it is statistically insignificant.
EXHIBIT 8
GROSS N f l T I O N f l L PRODUCT ( I N CONSTRNT
PERCENT C H f l N G E . YEflR TO YEfIR
$)
RCTURL flNO PREOICTEO
j
-8.0
8.0Forecast
-2.53 + .728 MIYP + .552 BY
-1.011 CPIYP(t-l) •+ .393 C
7.0-
1.217 UNY(t-l)
) _ CPIYP(t-2)J t-1
-7.0
6.0-
-6-0
6.0-
-5-0
4.0-
•4.0
3.0-
-3-0
UJ
UJ
-2-0
i.O-
-1.0
o.o-
-o.o
-t.o
-t.o-
-2.0
-3.0-
53 • 54 ' 55 • 56 • 57 ' 68 f 59 ' 60 ' 61 ' 6S ' 63 ' 64 • 65 ' 66 ' 67 ' 68
YEflRLY DflTfl
Lines from the time axis delineate recession periods.
{
-3.0
69 • 70 ' 71 ' 7£ ' 73 ' 74 ' 75
43
goods from abroad. If for whatever reason, prices rise abroad, the prices of the
goods that we import must rise commensurately as long as exchange rates
are fixed. With flexible rates, however, increases in prices abroad will tend to
be offset by decreases in the rates at which foreign currencies exchange for
U.S. dollars. The effect of England's ongoing inflation on prices we pay for
English goods and services, for example, has been largely, if not fully* offset by
a continuing fall in the dollar price of the pound. Under flexible exchange
rates, England cannot export the problems its monetary policies have been
creating. Under fixed exchange rates, it could.
Flexible exchange rates, moreover, have the further advantage of allowing
gradual adjustments to underlying disequilibrating trends. If U.S. labor productivity increases more slowly than productivity abroad, the dollar will
slowly depreciate with flexible exchange rates and the prices that we pay for
imports will rise only gradually. Under fixed rates, such trends are treated
with "malign" neglect until they no longer can be ignored. Sooner or later,
however, the pressures of such trends can no longer be contained. Exchange rates
and import prices change all at once—sharply and traumatically.
To avoid sharp traumatic increases in import prices, it is essential that we
pursue monetary and fiscal policies that avoid inflation. Such policies cannot of
course prevent foreign cartels such as OPEC from raising import prices. Only
the development of alternative supplies can do that. Monetary and fiscal policies can, however, prevent our having to pay higher import prices because of
falling demand for dollars. This is strong reason for understanding what monetary and fiscal policies we have to put in place to avoid fueling inflation in the
future.
In regard to monetary'policy',the lesson of our analysis is that the Federal
Reserve must provide for moderate and steady M - l growth. Our analysis,
however, sheds little light on the optimal rate. Based on our regression results,
on average, 64% of annual M-lgrowth was absorbed in price increases during
the period since 1945. Only 26% served to finance current output increases.
To some this will suggest that the optimal equilibrium long-run growth rate
for M - l is between 0 and 2% per year on the ground that, on average, real
GNP increased 4% per year during the post World War II period and .26 of
4% equals 1%, half way between 0 and 2%, However, we would regard this
range as well below optimal.
If money growth is limited to 0-2% per year, utilization of the normal
expansion of our capacity to produce would require price deflation unless
velocity increases commensurately. Since World War II, velocity has trended
up, so that we could have had substantially lower yearly money growth on
average, than we had without price deflation. But rising M - l velocity is by
no means a permanent fixture in U.S. economic life. From the Civil War to
World War II, velocity trended down not up, and this happened despite the
spread of checkbook money after 1865. Thus from the historical perspective,
it would not appear prudent to bet that M-l velocity will continue to rise
in the indefinite future. Moreover, it is far from far-fetched to believe it won't
despite existing interest rate incentives to economize on holding currency and
demand deposits. This is because holding currency and demand deposits is a
"luxury" which can be indulged in more easily as incomes rise. Moreover, if
in the future Congress repeals the prohibition against paying interest on demand deposits, this would tend to decrease M - l velocity by reducing the
interest rate incentives which now impel minimizing holding currency and
checking deposits. We do not and cannot know whether M - l velocity will
tend to rise or fall over the next 25 years. Our point is that we cannot count
on velocity continuing to rise, and hence in full employment should keep M - l
growth 2%-4% per year, "commensurate with the economy's long-run
potential to increase production," as Congress resolved in H. Con. Res. 133.
This range of money supply growth will provide the monetary framework for
sustained full employment real growth without inflation.
Finally, our results indicate that in the context of past deficits, fiscal
policy is not a very powerful macro-policy tool. As Mr. Meisselman (Professor of
Economics, Virginia Polytechnic Institute and State University at Reston)
testified, "when allowance is made for the quantity of money, the short-run
44
stabilization impact of changes in fiscal policy, that is variations in the size of
the budget or in the schedules of tax rates have been either minor or absent."
The impact of deficits, independent of monetary policy, is marginal at
most—statistically insignificant in respect to prices, small and barely significant
in respect to the nation's output or real GNP. Our results cast no light on
whether a tax cut or spending is the more efficient and effective way to increase
real GNP. But however created, our results indicate that today a $30 billion
High Employment deficit is required to raise real GNP 1%. Trying to spend
our way out of recessions thus would appear to be extremely difficult. Spending
programs of a size that could change the deficit by enough to have a substantial
impact on GNP cannot speedily be legislated and implemented. If one with
merit, for example, one that would upgrade work habits and skills, is on the
table, or can be speedily designed and, if passed, implemented, it should, of
course, be adopted. On the other hand, if quick fiscal stimulus is desired,
and there is no meritorious spending proposal on the table or in the wings,
a tax cut can be speedily legislated and implemented.
Our results also indicate that those who would put the blame on deficit
spending for inflation are misplacing their emphasis. Deficits generate significant inflation only to the extent that they are financed by increasing money
supply. In this regard, there is no law that requires the Federal Reserve to
monetize deficits and our results indicate it is risky to do so. Whatever the
deficit is next year, it would be a mistake to allow M-l to grow more than 6%.
Over the longer run, as we return to full employment (not after), annual
money growth must be gradually reduced to the 2%-4% level, commensurate
with our potential to increase production at full employment regardless of
fiscal policy.
In summary, our results indicate that a complete change in attitude about
Government spending and deficits is in order.
(1) The Federal budget should be analyzed as an instrument for allocating
the nation's resources, broadly speaking, between unmet family and unmet
public needs. It should not be viewed as a macro-policy tool. Each new spending
proposal, and indeed all existing programs, should be evaluated, funded, beefedup, cut or eliminated based on their intrinsic merit, not by some perceived need
to stimulate (or restrain) the economy.
(2) Tax cuts can be legislated if it is desired to give the economy a quick
jolt and no meritorious spending proposal is on the table.
(3) Finally, money supply growth should be completely divorced, in
practice as well as principle, from the Treasury's financing needs. M - l growth
should be planned and pursued without regard to the size of the deficit to be
financed next week, next month, and next year.
PART IV
POLICY UNDER H. CON. R E S . 133
Testimony received by the Subcommittee at its June hearings often was
critical of Federal Reserve policies up to the 1973-1975 recession. At the same
time, the testimony was relatively laudatory in regard to policy since March
1975. Coincidentally, March, 1975, marks both the bottom of the 1973-1975
recession and the month when the Congress passed House Concurrent Resolution 133. The resolution expressed the sense of Congress—
That the Board of Governors of the Federal Reserve System and the Federal Open
Market Committee^(1) pursue policies in the first half of 1975 so as to encourage lower long term interest
rates and expansion in the monetary and credit aggregates appropriate to facilitating prompt
economic recovery; and
(2) maintain long run growth of the monetary and credit aggregates commensurate with
the economy's long run potential to increase production, so as to promote effectively the
goals of maximum employment, stable prices, and moderate long term interest rates.
Pursuant to this resolution, and taking into account the international flows of funds and
conditions in the international money and credit markets, the Board of Governors shall
consult with Congress at semiannual hearings before the Committee on Banking, Housing
and Urban Affairs of the Senate and the Committee on Banking, Currency and Housing of
the House of Representatives about the Board of Governors' and the Federal Open Market
Committee's objectives and plans with respect to the ranges of growth or diminution of
monetary and credit aggregates in the upcoming twelve months.
B E F O R E THE RESOLUTION
Referring to an exhibit showing the rolls and tilts of M - l growth since the
Korean War, Chairman Neal asked the witnesses who appeared before the
Subcommittee on June 8 and 9, "If you think that Fed policies have been well
administered over these years?" The question was put first to Senator Buckley.
His response was unqualified.
Senator BUCKLEY. My answer would have to be no.
Mr. Jordan and Mr. Meisselman were similarly critical.
Mr. JORDAN. I think that money policy in the post-war period was procyclical making the inflations worse than they would have been, making the
recessions and the unemployment worse than they would have been.
Mr. MEISSELMAN. I am in general agreement with Mr. Jordan's analysis.
I would like to add that we now have a great deal of inflation already built into
the system because of poor public policies in the past.
Congressman Adams singled out the early 1960's when M-l growth ranged
from 1.5% to 4.5% as years which could serve as future norms. He told the
Subcommittee:
If you will look at the period between 1960 in the middle of your chart and continue up
through the beginning of the Vietnam War, which was February of 1965, you will see that we
had a rate of inflation that ran below 3 percent. Our unemployment rate was coming
down . . . we were trying to come out of the recession period of 1958. You also had at that
time a relatively low deficit and you had a low interest rate. I am saying that that condition
of low unemployment and low interest rates can be repeated again and that is my personal
goal.
Mr. Fiedler and Mr. Perry were less critical of past monetary policy.
(45)
46
Mr. FIEDLER. My judgment on the monetary policy of the past two decades is that it was very much better than the previous two decades. Monetary
policy still left quite a bit to be desired and did, as Mr. Jordan suggested, on
occasions make some things worse than they otherwise would have been.
Monetary policy has been part of the process over the past decade where
we have had a continuous acceleration in this base, built-in, rate of inflation
where we have gotten up from the early 1960's rate of 1 percent, or virtually
zero, and now we are up to 5 percent or 6 percent.
That is unfortunate, but perhaps we should have some sympathy for policy
authorities who are dealing with very difficult processes which are not well
understood.
Mr. PERRY. I would say that monetary policy has been conducted reasonably well over the period you have charted. Its worse marks probably come
during periods of recession when the Fed has allowed money growth to slow very
sharply.
This has been a characteristic of all the recession periods we have had. I t
could have been fighting harder against downturns. In recent years, it has been
confronted with impossible problems.
It has been asked to deal with inflation that it could not possibly deal
with and the results have been very mixed. And, depending on how one views
what is important, you can end up thinking the Fed has done very well or
very poorly.
Our research, which was not completed at the time of the hearings, indicates that the increase in the "base, built-in rate of inflation" to which Mr.
Fiedler alluded was in major part due to stepped up monetary growth. Nonetheless, we have some sympathy for Mr. Fiedler's "let's not be too critical"
view. The Federal Reserve does not operate in a vacuum. As Mr. Jordan
stated:
There is a risk of asking too much from monetary policy—asking more than it can
deliver in terms of offsetting the mistakes inherited from others, or trying to correct for
some of the real shocks to the economy, such as agriculture, quadrupling of oil prices, and
some other things.
You can ask [only] that monetary policy actions not be destabilizing as they have been
in the past. They have caused worse recessions and more unemployment than we otherwise
would have had. I think it is possible to avoid destabilizing actions and Congress should
use its oversight to hold the Fed accountable on that ground.
In specific, 1974 presented difficult if not "impossible problems." Our
results indicate that the inflation then raging would by itself act to sharply
reduce real GNP in 1975. At the same time, our results also show that by
accelerating M - l growth sufficiently the Federal Reserve could prevent the
decline and allow the economy to grow the normal 4%. But if this were done,
the day of reckoning with the depressing effects of inflation would only be
postponed. Moreover, the problem would become more difficult to deal with
as time passed because stepped-up money growth now generates still higher
inflation later on. While recognizing the difficulty of the choice, we believe
that, the deceleration of M - l growth, particularly after mid-1974, was unnecessarily rapid. Deceleration was in order. But it was overdone. Difficult
problems are best handled by gradual adjustments of the policy levers.
Happily, Federal Reserve policymakers are not now unaware of the
pitfalls in jumping quickly from one rate of money supply growth to a substantially different one even though the latter is a desirable long-run goal.
Mr. Hannaford pursued this question with Governor Partee on June 10.
Mr. HANNAFORD Our testimony has indicated the general feeling that
there is too much fussing done with monetary policy and short-term adjustments. . . #- Do you think we do a bit too much of that fits-and-starts kind
of adjustment?
Governor PARTEE. I would have to say in retrospect that there has been
more variation in the rate of monetary growth over the post-war period than
I think would have been desirable. That is, as I look back,*if I^could do it
47
over and I were the dictator, I would do it a little differently
be more stability than the chart that has been before us througn tnese 6 days
of hearings has demonstrated. I would rather have a little more stability than
that shows.
POLICY SINCE MARCH
1975
On May 1, 1975, Chairman Burns testifying for the first time under
House Concurrent Resolution 133 told the Senate Banking Committee that
"The Federal Reserve System is presently seeking a moderate rate of expansion
in the monetary and credit aggregates. We believe that the course we are
pursuing will promote an increase in M-l of between 5 and 7% percent over
the 12 months from March 1975 to March 1976." In appearances before the
House and Senate Banking Committees since then, Chairman Burns has
disclosed slight modifications of the target range. Exhibit 9 charts actual
monthly M-l levels from March 1975 to October 1976 against targeted levels
which were computed by multiplying M-l at the beginning of each target
period by the percentage growth targeted for the period.
The Exhibit reveals three distinct trends in money supply since March
1975. The first ran from March to August 1975 and achieved the "expansion
in the monetary and credit aggregates appropriate to facilitating prompt
economic recovery," which the Resolution called for during tho first half of
1975.
The second trend lasted from August 1975 to January 1976. It involved
the necessary slowing of money growth from the relatively rapid rate in the
March to August period. But the retardation of M-l growth was overdone.
As shown by Exhibit 9, there was virtually no growth from August 1975 to
January 1976. The pause in the economic recovery that began in the spring
quarter of 1976 was very likely exacerbated if not triggered by the excessive
slowing of M-l growth in the second half of 1975.
Finally, from January to October 1976, M-l has grown along or zigzagged around the lower limit of the target range. Most recently, in October,
it has spurted up to the middle of the range. The steadiness of recent Fed
money policy is commendable. As indicated earlier, we would be inclined to
limit M-l growth to 6 percent per year this year and next (aiming at 5%), and
to drop it to 2%-4% per year very gradually as we return to full employment.
CLOSING REMARKS
Most importantly, our research results stress the importance of controlling inflation and the critical role that money supply policy must play in
the effort. Few of us have to be reminded that inflation erodes the purchasing
power of our incomes and wealth. In addition, though it is not so widely understood, inflation causes interest rates to rise. As Governor Partee told the
Subcommittee, "a high rate of inflation over a sustained period, and the expectation that there will be future inflation, does affect interest rates simply because
lenders are not willing to see the real value of their wealth decline, and therefore they want an interest rate that will compensate for inflation. Borrowers
are willing to pay the rate because they feel that what they invest in with the
proceeds is going to go up in value, too."
These reasons are reason enough to make controlling inflation a high
priority economic policy goal. But, moreover, our research shows that current
inflation causes production to fall and unemployment to rise next year. Contrary to the conventional view that we can increase production and reduce
unemployment by accepting more inflation, our analysis indicates that the
way to promote and sustain recovery is to eliminate inflation.
Finally, there are fiscal policy reasons for bringing inflation under control
as quickly as possible. A recent Congressional Budget Office study conducted
for the Joint Economic Committee's Fiscal Policy Subcommittee entitled,
"The Effect of Inflation on Federal Expenditures," documents that inflation
increases both federal expenditures and tax revenues. These inflation induced
changes make it more difficult both to achieve fiscal stability and to plan
fiscal policy ahead.
EXHIBIT 9
SOT-
M-l MONEY SUPPLY (Sfl)
-390
flCTUflL ^ TflRGET LIMITS
TflRGETS ORE QUflRTERLYi BCTUflL IS MONTHLY
-920
320-
CO
CO
az
-910
310-
CD
CO
CO
O
300-
-900
•—«
—J
•—*
CD
290-
280-
-290
J ' F ' M ' f l '
1975
H ' J '
J MR
' S ' O ' N ' D
J 7 F ' H
1976
' f l ' H ' J i
J .1 fl ' S ' O ' l i ' D
• J • F •• M • fl ' H • J
1977
-280
49
Many federal spending programs," including social security, civil service
and military retirement, food stamps, railroad retirement, the school lunch
program, and federal civilian pay, are inflation indexed. The CBO study
concludes:
More than 60 percent of all federal expenditures increase explicitly with prices or
implicitly through cost payments tied to the price level. Since the prices of these expenditure
items rise on an approximately one-for-one basis with the CPI, an extra one percent increase
in the price level automatically causes about a 0.6 percent increase in federal expenditures.
The effect of inflation on tax revenues is even greater. The CBO study
points out:
A 10% increase in the price level will increase expenditures by 5.8% in the short run
and 6.3% in the long run, while increasing receipts by 12.3% in both the short and long run.
IMPORTANCE OF CONTROLLING MONEY GROWTH
Inflation-induced fiscal and economic pressures must be brought under
control as quickly as possible. As the work of the Subcommittee on Domestic
Monetary Policy shows, we can bring inflation under control and keep it in
check by achieving moderate M-l growth commensurate with our potential to
increase production. With unemployment nearly 8%, our potential to increase
production is higher than normal. For the remainder of 1976 and 1977, 4 % 6% M-l growth per year is appropriate. It will promote sustainable, reasonably
rapid recovery in the months immediately ahead without adding substantially
to inflation and the interest rate, production and fiscal policy problems that
inflation generates in subsequent years.
But we must plan now to gradually reduce money growth to 2%-4% per
year after 1977 to reduce inflation to the attainable l % - 3 % rate. This is the
major policy implication of the Subcommittee's research and hearings. As put
by Mr. Meltzer:
No sustained inflation has ever been ended until the growth rate of
money has been reduced.
Senator BUCKLEY. Under the resolution [H. Con. Res. 133] the Federal
Eeserve is directed to establish targets which are in line with the economy's
long-term ability to grow. Our problem in the past is that the Federal Reserve's
monetary policy has been constructed along lines which were only haphazardly
related to this fundamentally important guideline. Monetary targeting which
is set independently of the economy's growth potential produces the kind of
erratic monetary growth policies which take us from inflationary rates to a
constricting zero level, as was the case in the latter half of 1974.
Congressman ADAMS. \ T ". What we need in money supply is a matching
of the money supply to the real growth in the economy.
Chairman NEAL. YOU would advocate a moderate and steady growth
in the money supply—•—•
Senator BUCKLEY. TO match the average growth that we project for the
economy.
As an added final note, we cite Mr. Pierce's explanation of the importance
of controlling money supply for achieving interest rate stability.
Mr. PIERCE. Historically, the Federal Reserve, like other central banks,
has exhibited a strong tendency to stabilize fluctuations in interest rates, particularly short-run fluctuations. The Fed's penchant for stabilizing interest rates
has among other things helped produce a procyclical behavior in the growth
of the money stock; that is, money grows rapidly during economic expansion
and slowly during recessions.
The reason for this is clear enough: when the economy is expanding rapidly
credit demands also expand, tending to put upward pressure on interest rates.
The Federal Reserve attempts to constrain these interest rate increases by
56
ttfoviding more bank reserves through open market operations. The increase
iii bank reserves in turn leads to expansion in the money stock and underwrites
an expansion in economic activity.
During recessions, credit demands decline and interest rates begin to fall;
the Fed attempts to constrain the decline in interest rates by selling securities
on the open market, therein reducing bank reserves and retarding growth in
the money stock. The retardation of money growth and constraint on interestrate declines tend to exacerbate the decline in economic activity. It seems clear
that attempts to stabilize interest rates can and have produced greater cyclical
fluctuations in income, production, employment, and inflation than would have
been the case had the Fed not been so concerned about interest rate fluctuations.
If the Federal Reserve adhered more closely to a target growth for the
money stock, interest rates would tend to move more quickly; that is, fall more
rapidly in recession and rise more rapidly in expansion than has heretofore
been the case. These sharper interest-rate movements would serve to moderate
the fluctuations in economic activity. If interest rates were allowed to react
more quickly, aggregate demand would be affected more rapidly and hence
would not probably fluctuate so widely. As a result, it is likely that interest
rates would themselves actually fluctuate less widely.
APPENDIXES
APPENDIX 1
Correlation Matrices
1. For the cornerstone inflation regression, Series S, Table 4,1947-1975.
Dummy
.
PIMYPW
248
344
402
.204
.224
,245
702
.353
,661
M1YP (f,-2)
Dummy
PIMYPW
rcpiYP (t-i)--i
LCPIYP (t-2) J
rCPIYP
(t-l)-l
LCPIYP (t-2) J
. 638
CPIYP
2. For the real GNP regression, Series 1, Table 13.
M1YP
BY
CPIYP (t-1) UNY(t-l)
BY
CPIYP (t-1)
UNY (t-1)
()
CPIYP (( t - l ) - i
rCPIYP
LCPIYP (t-2) J - l
252
474
093
112
.220
—.353
.418
.009
.365
-.032
GNPPCON
,271
,078
-.503
.471
Interpretation: Multi-collinearity is not a major problem.
LCPIYPit-2) J-
. 106
84
APPENDIX 2
M1YP
Lag Correlation Analysis
1945-1975
Lag in Years
0
1
2
3
4
5
6
7
8
9
Correlation
1. 00
1.00
.60
.26
.36
.07
.00
.00
.03
.04
.04
.01
.03
—.04
-.07
.18
.21
.20
-.10
-.18
Interpretation: Periodicity or cyclical regularity is not a major problem-
APPENDIX 3
Cochrane-Orcutt Regressions
1. Cochrane-Orcutt version oj the Series S Table 4 1947-1975 inflation regression.
CPIYP=
Constant
+
J
.448
(.274)
Adj W=. 917; SE=1.035
M1YP (t-2)
+
Dummy
2
.662
(.068)
+
PIMYPW
2
6.207
(.903)
+ \_CPIYP (t-2)
2
1.386
(.236)
J
.260
(.063)
g»
2. Cochrane-Orcutt version oj the Series 1 Table 13 real 6NP regression.
6NPPCON=
Constant
2
+
M1YP
2
- 1 . 785
. 749
(.804)
(.094)
2
Adj R =.868; SE=.974
+
BY
+
CPIYP (t-1)
*. 480
(.276)
2
— 1. 040
(.089)
+
UNY (t-1)
2
1. 057
(.173)
+
rCPIYP (t-1)
[_CPIYP (t-2)
2
.373
(.113)
1
2
Coefficient is not significant.
Coefficient is significant.
Standard errors are in parentheses below the coefficients.
Note: Comparison with the corresponding regressions in the text indicates that the Cochrane-Orcutt procedure
yields marginally better results.
APPENDIX 4
Regressions Using M-l as the Dependent Variable, 1946-1975
MlYP(t+2)
Constant
2
3. 790
(.768)
Dummy
1
1 . 325
(2. 233)
=
(t-l)-l
ICPIYP (it-2) J
+
PIMYPW
K 941
(2.127)
' - . 138
(.185)
+
CPIYP
» - . 142
(.261)
Adj K 2 =-.O54; SE=2.341; DW=.674
MlYP(t+l) =
Constant
2
3. 295
(.735)
Dummy
1
2. 247
(2. 030)
VCPIYP (t-l)-l
ICPIYP (t-2) J
•—. 185
(. 167)
PIMYPW
CPIYP
\ 402
(.712)
*. 041
(.204)
Adj R 2 =-.O26; SE=2.287; DW=.882
1
2
Coefficient is not significant.
Coefficient is significant.
Standard errors are in parentheses below the coefficients.
Interpretation: The chain of causation does not go to money supply changes.
APPENDIX 5
Inflation Regressions Using M-l Ahead as an Independent Variable, 1946-^1975.
CPIYP=
2
Constant
2. 462
(.707)
+
MlYP(t+2)
» —. 089
(-164)
+
1
Dummy
- . 261
(1.784)
PIMYPW
2
6. 013
1. 140
lCPIYP(t-2)
2
.452
(.115)
Adj R 2 =. 689; SE=1. 857; DW=1. 045.
CPIYP=
Constant
• 2. 660
(.832)
+
MlYP(t+l)
l
. 041
(.203)
+
Dummy
1
2. 210
(2. 025)
PIMYPW
1
2. 210
(.553)
Adj R 2 =. 592; SE=2. 279; D W = 1 . 004.
1
2
Coefficient is not significant.
Coefficient is significant.
Standard errors are in parentheses below the coefficients.
Interpretation: Nothing is gained by leading M l Y P in analyzing inflation.
rcpiYP(t-i)\jCPIYP(t-2)
2
.486
(1.39)
]
58
APPENDIX 6
Regressions Using M-2 in Place of M-l
Dependent variable: CPIYP=
Constant
M2YP (t-2)
Dummy
P1MYPW
TCPIYP ( t - l ) - l
LCPIYP (t-2)' J
Adj R2
SE
DW
1950-1975
!.998
(.540)
2
.327
(.
106)
2
3. 990
(1. 183)
2
1 . 609
(. 360)
2
.230
(. 121)
.806
1.267
1.640
1951-1975
».768
(. 576)
2
.356
(.
109)
2
4. 734
(1. 238)
2
1. 524
(. 363)
2
.240
(• H8)
.814
1.243
1.399
1956-1975
*—.614
(. 632)
2
.349
(. 109)
•1.311
(. 330)
2
.627
(• 174)
.848
1.085
1.488
Dependent variable: GNPPCON=
Constant
M2YP
BY
CPIYP (t-1)
UNY (t-1)
TCPIYP (t-i)—i
I
LCPIYP
(tr-2) J - ]
Adj R2
SE
DW
1958-1975
1
- 2 . 238
(1.338)
2
.487
(. 130)
l
.401
(.432)
2
- . 976
(. 139)
2
1. 070
(. 296)
8
.421
(. 149)
.766
1.299
2.331
1
2
Coefficient is not significant.
Coefficient is significant.
Standard errors are in parentheses below the coefficients.
Note: Comparison of these results with the Series 3 Table 4 and Series 1
Table 13 results indicates that the choice between M - l and M-2 is not crucial,
but that using M - l yields marginally better results.
APPENDIX 7
INFLATION REGRESSION SEPARATING THE LAGGED CPIYP
TERMS BRACKETED IN OTHER REGRESSIONS
Dependent variable:
Constant
1
.558
+
CPIYP =
MlYP(t-2)
2
+
.645
Dummy
2
5.979
+
PIMYPW
2
1.414
+
CPIYP(t-1)
2
.241
+
CPIYP(t-2)
2
-.255
O
(.396)
Adj. R2 = .891;
(.100)
SE = 1.187;
(1-145)
(.304)
(.096)
(.083)
DW = 2.5
-•-Coefficient i s not s i g n i f i c a n t .
C o e f f i c i e n t i s significant.
Interpretation:
The lagged CPIYP coefficients are v i r t u a l l y the same, as i s implied
s t a t i s t i c a l l y when bracketing them. However, separating them t e s t s a different
proposition than bracketing them. The momentum hypothesis requires bracketing regardless
of whether the coefficients on the lagged CPIYP terms are equal or different.
@FedFRASER