Full text of Economic Review (Federal Reserve Bank of San Francisco) : Fall 1976

View original document

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

The Federal Reserve Bank of San Francisco’s Economic Review is published quar
terly by the Bank’s Research and Public Information Department under the supervision
of Michael W. Keran, Vice President. The publication is edited by William Burke, with the
assistance of Karen Rusk (editorial) and Janis Wilson (graphics). Subscribers to the
Economic Review may also be interested in receiving this Bank’s Publications List or
weekly Business and Financial Letter. For copies of these and other Federal Reserve
publications, contact the Public Information Section, Federal Reserve Bank of San Fran
cisco, P.O. Box 7702, San Francisco, California 94120. Phone (415) 544-2184.

The spirit of Friedrich Hayek walks through
these pages, explicitly in the quotation leading
off the first article and implicitly in the approach
followed by the authors of all three articles. The
Nobel laureate spent his career expounding the
efficacy of the classical adjustment process and
the need for moderation in government-policy
decisions. He would never have thought of himself as a revolutionary, but many of his intellectual descendants are now participating in a
revolution which is just as impressive as the
Keynesian Revolution of the 1930's. For example, Keynes in his research assumed a closed
economy, but important differences are now
seen to arise when we deal with an open economy. Again, Keynes assumed away the classical
adjustment process, but that approach is now
coming back in the guise of rational-expectations. Here then are some new perspectives
based on the new theories, new facts and new
tools of the 1970's.
Michael Keran and Michael Riordan test
some of the new facts in the light of new theories, in the process of pointing out the dangers
to the u.s. economy of expansive stabilization
policies pursued in the rest of the world. They
argue that expansionary policies abroad were a
major cause of the 1973-74 acceleration in U.S.
inflation-and that another shift toward expansion in 1975-76 could set the stage for some reacceleration of domestic inflation in 1977.
Keran and Riordan show that the world
money stock-a summary measure of world
stabilization policies-is an important factor in
explaining price fluctuations of internationally
traded goods. A simultaneous expansion of industrial countries' aggregate demand, resulting
in synchronized monetary expansion, can lead
to a more-than-proportional increase in world
prices. And this price rise can significantly
affect the domestic rate of inflation, even though

it has less influence in this regard than domestic
monetary policy. The analysis suggests an important policy implication: any single country
can follow domestic stabilization policies to offset the effects of imported inflation, but the result could be disastrous if all countries followed
such a policy simultaneously.
In earlier decades, industrial countries largely
conducted their stabilization policies in isolation
from each other; typically, when one was in an
easy-money phase, others were in a tight-money
phase, and vice versa. But the early 1970's were
different, as most countries followed either
tight-money or easy-money policies in a uniform pattern. World money growth accelerated
in 1970-73 as the old system of fixed exchange
rates broke down. As foreign central banks
monetized the dollar inflows caused by large
U.S. balance-of-payments deficits, they expanded their holdings of international reserves
and their domestic money supplies. Then, with
the abandonment of the fixed exchange rate
regime in early 1973, they began to regain control of their domestic money supplies and money
growth decelerated. These sharp fluctuations in
world money were accompanied (with about a
two-year lag) by an instability in prices of internationally traded goods.
In the model developed by Keran and Riordan, a 12-percent growth rate of the world
money stock in 1975-76 suggests a reacceleration of world inflation in 1977. For the U.S.,
this could mean more inflation, depending on
the size of the increase in prices of internationally traded goods and on the share of such
goods in the domestic price index. The result
could be a 7- to 8-percent rise in U.S. wholesale
prices next year, compared to the 4-percent increase in the first nine months of 1976.
In a second article, Kurt Dew analyzes a recent theoretical controversy in stabilization pol-

be increasingly disappointed throughout the
period when the government-spending program
was in effect.
In a third article, Rose McElhattan conducts
an incomes policy experiment on the U.S. economy-specifically, a government wage-restraint
policy designed to improve the tradeoff between
unemployment and inflation, apart from the
traditional instruments of stabilization policy.
She analyzes this new tool not in the spirit of
advocacy, but rather in a pragmatic attempt to
define its structure and to demonstrate the sensitivity of model results to different assumptions.
The policy tested is one which sets the average
increase in wages equal to the 3-percent trend
rate of growth in labor productivity. Its economic impact is estimated through simulations
of the U.S. economy over the 1967-70 period.
The simulation results suggest that a program
which controls wage-rate increases can temporarily control the rate of increase in domestic
nonfarm prices. But although the program
might limit inflation, real-income growth during
the control period might be insuflicient to maintain employment at desired levels. Additionally,
it should be emphasized that the simulation results are dependent upon the behavioral structure of the MPS model as well as the assumptions underlying the analysis.
McElhattan notes four important caveats
needed to round out the analysis. First, under
equally feasible alternative assumptions to those
chosen, the impact of wage controls upon output and employment can differ considerably.
Second, the monetary growth rate assumed in
the analysis is too high to support for very long
the low rate of inflation implied by the wagegrowth assumption. Third, implementation of
an incomes-policy can distort the pricing mechanism, because in trying to achieve a desired
growth in the average growth rate, policymakers would be forced to exert some control
over wages in individual sectors or industries.
Finally, policy implementation similarly can
distort the distribution of income between capital and labor, thus destroying the program's
political acceptability.

icy, in contrasting the different policy approaches that go under the names of "optimal
control" and "rational expectations." Advocates of optimal control tend to be activists,
while those utilizing rational expectations tend
to believe in the greater efficacy of passive policies. Optimal-control theorists suggest that policy-makers could improve upon passive policies
through the utilization of econometric models
as well as mathematically-derived rules for policy adjustment. Rational-expectations theorists,
in contrast, tend to question the basic assumptions of the optimal-control approach, arguing
that households and firms do not form their
expectations of future events in the way that
most economists think they do. Indeed, if
households and firms form expectations in a
"rational" way- using all instead of just a part
of the available evidence-policy-makers have
little if any latitude to exert a beneficial impact
on economic welfare.
Dew asks: "Can the policy responses that are
generated by optimal-control rules overcome
the uncertainties regarding future economic behavior that are created by rational expectations?" As an example, he discusses the aggregate consumption decision, showing how the
outcome of a particular policy can be adversely
affected through a misinterpretation of the
means by which the consumer forms his expectations. He analyzes the effects of a given
increase in government expenditures upon consumption, assuming first that the policy-maker
believes that consumers form expectations adaptively-the standard method in most models of
the economy-and then contrasts the expected
policy outcome with the actual outcome when
expectations are rational.
The author notes that the rational consumer
is concerned about future income, not past income, so that past policies only matter to him if
they affect future income. The fiscal stimulus
of an increase in government expenditures
would be greater at the beginning of the spending program and would thereafter decline. The
opposite would be true for a policy-maker using
an adaptive-expectations forecast, who would

Michael Keran and Michael Riordan':'
"Whatever our views about desirable behavior . .. they can never legitimately be applied
to the situation of a single country which is part of an international economic system, and
any attempt to do so is likely in the long-run and for the world as a whole to be an additional source of instability."
F. A. Hayek, 1937.

One of the more painful lessons that economists, businessmen and policymakers learned in
the first half of the 1970's was that their domestic economic environment is influenced not only
by stabilization policy at home, but also by developments in other countries. Inflation in the
U.S. in the 1973-74 period was higher because
inflation was a worldwide phenomenon. The
recession in 1974-75 was more serious because
it was a worldwide phenomenon. In short, the
synchronous development of inflation and recession has made the U.S. economy more unstable
than it was in the 1950's and 1960's.
The purpose of this article is to explain how
stabilization policies in the rest of the world
could impact on the U.S. economy to frustrate
domestic stabilization goals. According to one
interpretation, "supply shocks" associated with
the oil price rise, agricultural shortfalls, and
other natural calamities have contributed to the
inflation, but this is only a partial explanation.
Rather, expansionary stabilization policies in
the rest of the world were a major source of
the acceleration in U.S. inflation in 1973-74,
and expansionary policies abroad in the last year

and a half could set the stage for some reacceleration of U.S. inflation in 1977. 1
In Section I, we develop a simple model
which provides a consistent explanation of how
expansionary stabilization policies in the rest of
the world can lead to higher domestic inflation
and lower domestic real output growth than
would be warranted by domestic policy alone.
In Section II, we present empirical evidence in
support of this model. In Section III, we evaluate the potential impact of recent world and
domestic monetary policies on U.S. inflation in
1977 and beyond, and in a concluding section
consider the implications of this analysis for
making national stabilization policies in a world
environment. A technical appendix provides a
mathematical exposition of the model and its
application to the U.K. and Germany.
The term "stabilization policy" refers here to
traditional monetary and fiscal policy tools of
aggregate demand management. In both the
theoretical and empirical sections of this article,
monetary policy will be given the primary consideration. This focus does not imply that fiscal
policy is an inferior policy tool. However, over
the historical period considered here, monetary
poliey has shown the greater variance and hence
its impact on aggregate demand is more amenable to empirical testing.

*Keran is Vice President and Director of Research, and
Riordan is Research Assistant. Federal Reserve Bank of
San Francisco. The authors wish to thank David Condon
for his research assistance.

I. The Basic Model
We can model the impact of world and domestic stabilization policies on short-run domestic equilibrium by drawing together elements of
a monetary analysis of world inflation and a
Keynesian analysis of national-income determination. Specifically, the model is based on
the following relationships:

for internationally traded goods, and a rise in
the world price of traded goods. In the absence
of exchange - rate adjustment, traded - goods
prices will rise even in the country which has not
participated in the expansionary policy.
Why, under flexible exchange rates, doesn't
exchange-rate appreciation offset the world
price rise? The simplest explanation is that
world inflation will increase import and export
prices proportionately, and hence will have no
initial substitution (terms of trade) effect.' With
no substitution effect, only the income effect will
operate on the exchange rate. The higher domestic price level will reduce the real money
stock, inducing a decline in real aggregate demand for goods, including imports. The exchange rate will appreciate in response to the
resulting trade surplus. Only at this point will
the movement in the exchange rate operate to
offset the impact of the rise in world prices on
domestic prices. While this income effect is
powerful, most studies suggest that it requires
a year or more to unfold. In addition, it operates only after the higher world price is passed
through to higher domestic prices.
If nothing else happened, the unusual rise in
domestic prices would be temporary. A trade
surplus would eventually emerge, and the exchange rate would appreciate to reduce the domestic relative price of traded goods to nontraded goods. However, this result will not
occur if there is an induced increase in money
wages during the period when domestic prices
are higher. Higher money wages will shift the
aggregate supply curve upward, in the sense
that for each level of output, the unit cost of
production or supply price will be higher. The
resulting equilibrium of aggregate demand and
supply will be at a lower level of real output
and a higher level of prices than would be expected from domestic monetary policy alone.
In this case the relative price of traded and nontraded goods is reestablished by a rise in nontraded goods prices.
Thus expansionary policies in the rest of the
world-' can lead to hoth a higher inflation rate

I. Domestic nominal aggregate demand is
determined by domestic monetary and fiscal
policy. according to the standard Keynesian
LM/IS analysis. Given constant stocks of capital and labor, and a constant level of technology, real aggregate supply is determined by the
price level and the nominal wage rate. Internal
balance is achieved when aggregate demand is
equal to aggregate supply. Abstracting from
capital flows, external balance is defined as the
equality of exports and imports. Thus the equilibrium conditions for internal and external balance are identical.
2. Each country faces an elastic world demand for the internationally traded goods it
produces, with the price level determined by the
aggregate of world monetary policies." We
assume that no single country has a significant
impact on the world demand for internationally
traded goods. Furthermore, each country is
assumed to have effective control over its domestic money stock, both under fixed and flexible exchange rates."

3. By the "law of one price," the domestic
price of internationally traded goods is jointly
determined by world prices in foreign currency
and the exchange rate between domestic and
foreign currencies.
4. Money wages are rigid downwards but adjust upwards with a rise in the domestic price
level. Hence, if exchange rates are "sticky" in
the short run, for reasons to be explored below,
world price inflation will be translated into domestic price inflation and then into domestic
wage inflation.
According to this model, if a number of major
countries uniformly follow expansionary monetary polices, the result will be increased demand

internationally traded goods directly (if temporarily) adds to domestic inflation, and thus
can explain how the Phillips curve, which relates inflation to output (or unemployment),
can break down under certain circumstances.
(3) The response of domestic wages to domestic inflationary pressures (which can be
viewed as a cost-push element) determines the
extent to which world inflation is permanently
or only temporarily translated into domestic
inflation.
Four empirical propositions underlie this
analysis.
1. The aggregate of stabilization (largely
monetary) policies of the major industrial countries determines the prices of internationally
traded goods.
2. The domestic price index (which is a
weighted average of traded and non-traded
goods prices) responds to both domestic and
world monetary-policy actions.
3. The permanency of the impact of world
prices on domestic prices depends upon the
wage response to increases in domestic prices.
4. Domestic monetary and fiscal policies determine domestic aggregate demand but not the
split between prices and output.
We shall now proceed to investigate the
validity of each of these propositions.

and a higher unemployment rate in the U.S.
than would have otherwise occurred. By the
same token, when other countries stop following expansionary policies, the U.S. inflation rate
will decline to a level consistent with strictly
domestic demand and supply considerations.
The deceleration of traded-goods prices will reduce domestic inflation. Then, with a given
aggregate demand, real income will rise, leading
to an increase in real demand for output and a
decline in unemployment.
This explanation of the interaction between
world and domestic stabilization policies helps
to reconcile a number of apparently conflicting
explanations of the current inflation. It provides
a method to link the traditional aggregate-demand analysis with both world-monetary and
cost-push explanations of inflation, along the
following lines.
( 1) Domestic monetary and fiscal policies
explain shifts in nominal aggregate demand if
there is a stable relationship between nominal
monetary and fiscal policies and nominal income. However, the split between prices and
output cannot be explained by aggregate-demand analysis alone, but must also take into
account the shifts in aggregate supply.
(2) The rest of the world's monetary policies
can explain the process whereby inflation in

II. Testing the Model
Standard regression techniques were used to
estimate the functional relationships implied by
the propositions developed in Section I. The
source for all raw data, except where otherwise
noted, was International Financial Statistics
published by the International Monetary Fund.
The data covered the period 1960.2 through
1975.3, which included a period of fixed exchange rates until 1973.1 and largely floating
rates thereafter.
All equations were estimated in quarterly
percentage-change format. This should be kept
in mind in evaluating the quality of the statistical
results. With this type of computation, the percent of explained variance (R") will be in the
40-to-80 percent range rather than the 90-to-99

percent range common to equations estimated
in level form. By using change rather than level
data, we omit the variance to be explained by
trend (thus reducing R2), leaving only the
cyclical and random component. Actually the
random component is magnified, because the
change data add the random element in the two
adjacent level observations. As the equation is
not expected to explain random movement, this
further reduces the R 2 • The superior measure
of "good fit" in this case is the standard error
(SE) which has the same meaning in both level
and change form.
World Inflation and World Money
Our key hypothesis states that the price level

of internationally traded goods is a positive
function of the nominal world money supply,
which serves as a summary measure of the
world's monetary stabilization policies. To test
our hypothesis, we regress percentage changes
in world prices on percentage changes in nominal world money.

~ Log PW t

12
27. 1 +~3.59 ~ Log MW 12
t(3.9) (5.1)

= -

R2/SE

DW/DF
1. 53/48

.40/9.49

The world price level is defined empirically
to be the index of export prices of industrial
countries expressed in U.S. dollars. There are
several reasons for this definition. Since industrial countries predominate in world trade, they
collectively determine the price level of internationally traded goods and therefore the world
inflation rate. Secondly, data for industrial
countries are the most complete and most reliable data available. Finally, a definition in
terms of U.S. dollars removes the need to introduce the exchange rate explicitly into the empirical analysis, although the measure of world
prices will obviously be influenced by exchangerate movements.

Log Pw

'c~c

Log Mw

Percent change in international traded goods
prices, measured by export prices of industrial countries.
Percent change in world money supply, measured by weighted average percent change of
!vh in 10 industrial countries.

The Durbin-Watson statistic indicates the
presence of serial correlation, which suggests
that an important explanatory variable may
have been omitted from the estimation. Wc
hypothesized that world prices would depend
positively on nominal world money supply and
negatively on real world income or output, but
we do not have a suitable proxy for the lattcr
variable.
In view of the unusually large quarterly variation in our measure of international prices, thc
predicted value of world prices differed substantially from the actual value on a quarter-toquarter basis. For the period from 1973.3 to
1974.3, for example, errors ranged from a high
of 32 percent to a low of 4 percent.

The world money supply is defined to be the
weighted average of the money supplies of ten
major industrial countries. The money supply
of each individual country is defined in terms of
M I , currency in circulation plus demand deposits. To convert different M 1 values into a
common denominator, quarter-to-quarter rates
of change are computed and a weighted sum of
the changes calculated for each quarter. The
weights represent each country's proportionate
share in 1975 of the total trade of industrial
countries. G These weights are adopted because,
in view of the distinction between traded and
non-traded goods, the impact of domestic demand-management policies on world prices will
depend on the extent to which the country engages in trade.

Period

Actual
Value

Predicted
Value

Error

1973.3
1973.4

41
6

27
26

14
+20

1974.1
1974.2
1974.3

17
52
14

24
20
18

+ 7
-32
+ 4

Average

However, the average rate of increase in international prices over this period was 26 percent,
and the equation performed reasonably well by
predicting an inflation rate which averaged 23
percent over the period. Thus, the average
error for the total of the five quarters was only
3 percent-less than any quarter-to-quarter
error because of the tendency of quarterly
errors to offset each other. '
Our estimation indicates that a one-percent
increase in world money-supply growth will,
over a period of about th ree years, add 3 1/ l

A second-degree polynominal distributedlag equation was estimated, linking quarterly
percentage changes of world prices with quarterly percentage changes of world money supply. World money supply was lagged over 12
quarters. The results are summarized below
(t-statistics in parenthesis).

When the equation is estimated in nominal
form, the sum of coefficients on nominal money
supply is statistically significant. The sum of
coefficients on government expenditure has the
right sign but its t-statistic is insignificant; however, the inclusion of this variable improves the
overall fit of the equation. Only 34 percent of
the observed variance of nominal income is explained by the equation. However, as income
is measured on a quarter-to-quarter basis, the
"random element" is magnified relative to the
cyclical element, thus reducing the amount of
total variance which the independent variables
are expected to explain. The relatively low
standard error (around three percent) indicates that the fit is reasonably good, and the
fact that the Durbin-Watson statistic is close to
two in value suggests that all the systematic
movement in income has been accounted for.
When the eCluation is estimated in the real
form, the coetficients remain approximately the
same (supporting the hypothesis of homogeneity) but their statistical significance increases.
The equation explains 60 percent of the variance in real aggregate demand, and the standard error of the estimates is slightly lower than
in the nominal version. Dividing the equation
through by a common price index which exhibits a large systematic variation increases the
percentage of total variance which is systematic
and reduces the percentage which is random.
The fit of the equation improves as a result.

percent to world price inflation, or an average
of just under one percent a year. The very
high negative constant term implies that a
world money-supply growth rate of approximately 7Y2 percent is necessary to achieve
rough world price stability (27.1/3.59 ~= 7.5). A
world money-supply growth rate less than 7Y2
percent would lead to world price deflation.
This agrees with what we observe empirically.
Between 1960 and 1970 world
averaged between 7 and 8 percent, and during
this period world prices remained relatively
stable (Charts 2 and 3).
National Income and Domestic MonetaryFiscal Policies

The model presented in Section I assumed
that domestic aggregate demand, measured by
nominal national income, is positively related
to domestic nominal money supply and nominal government expenditure. If the nominal
aggregate demand function is linear homogeneous, then aggregate demand in real terms,
measured by real income, would be positively
related to the real money stock and real government expenditures. We test these two relationships by estimating percentage changes in
nominal (real) income as a second degree
polynominal distributed lag of nominal (real)
money stock and government expenditure. In
each estimation, explanatory variables are
lagged over four quarters.

Our estimates indicate that income is approximately proportional to the money supply.
That is, a one-percent increase in money supply
leads to about a one-percent increase in aggregate demand. The factor of proportionality
(income velocity) is approximately 2.5 whether
the equation is estimated in real or in nominal
form. Moreover, the relatively short lag with
which changes in money-supply growth influence the growth of aggregate demand, indicates
that the demand side of the economy adjusts
very quickly, i.e. the economy moves along the
aggregate-demand curve rather than off it.
The assumption of unit elasticity for the
aggregate-demand curve implies that nominal

4
4
.d Log Yt = 2.28 + 2.:1.04 .dLog M _ + 2.:0.3 LlLog G
t
4
(1.23)
(3.85)
(0.23)
2
R /SE

DW/DF

.34/3.17

.d Log Yt

1.99/46
4
4 .
2.52 + 2.: 0.97 .JL09 M
+ I 0.04 L'1Lo9 G*
4
(5.18)
(6.92)
(0.36)

2
R /SE

.60/2.81

DW/DF

1.97/46

'" Log Y = percent change in nominal income, measured
by GNP in current dollars
'" Log M
percent change in money supply, measured by
Ml
!l Log G
percent change in government expenditures in
current dollars
An asterisk (") indicates a real variable, defined by dividing the nominal variable by the GNP price deflator.

income will be constant at any point along a
given demand curve. In other words, nominal
income cannot change unless there is an underlying shift in the demand curve. Thus we may
interpret our estimation of aggregate demand in
nominal form as a measure of the shift in the
aggregate-demand curve resulting from changes
in domestic monetary-fiscal policies. By the
same token, we interpret our estimation of the
aggregate-demand equation in real form as a
measurement of movements along the demand
curve resulting from a change in the real money
stock, with the nominal money stock held constant. To put it another way, an increase in the
price level will decrease real money supply, inducing a decrease in aggregate demand for real
goods, i.e., a movement along the demand
curve. This interpretation, which is consistent
with the evidence, allows us to distinguish shifts
in the aggregate demand curve from movements along the curve. This will prove helpful
in Section III, because it will allow us to sidestep the assumption of a stable Phillips curve
in analyzing the trade-off between inflation and
unemployment.

same results. Only the latter version is reported below (Chart 1 ) .
Chart 1

U.S. Inflation Rate
Percent

1vl

'''\

1966

1964

1968

1970

1972

1974

Source: Federal Reserve Bank of San
Francisco-IFS data.

!\log en

-3.57 - 3.35 DUM + L1.49
(3.9) (5.8)
(7.4)

L 0.16
(5.3)

R2/SE
OW/DE

World and Domestic Monetary Policies and the
Domestic Price level

lILog M -

t 12

Hog PW _
t 4
.87/1.11
1.66/41

The equation explains 87 percent of the variance in the U.S. inflation rate. For every onepercent increase in the U.S. money stock (M!)
over the current and twelve past quarters, the
U.S. inflation rate (CPl) will increase by 1.5
percent. For everyone-percent increase in international inflation over the past four quarters,
the U.S. inflation rate will increase by .16 percent. This lag reflects the time delay that is
typically observed between changes in wholesale prices (by which international prices are
measured) and changes in retail prices (by
which domestic prices are measured).
A dummy variable (DUM) was added to
the equation to account for Phases I and II of
the price-control period (August 15, 1971 to
December 1972). Our estimate suggests that,
as a result of controls, the annual U.S. inflation
rate was 3.3 percent lower over the 5lh-quarter
period than would otherwise have been the
case. Removal of controls probably pushed up

In Section I, we argued that the domestic
price level could be influenced by monetary
policy in the rest of the world as well as by
domestic stabilization policies. Two alternative
reduced-form tests can be applied to confirm
this proposition. The common independent
variable in each of these tests is the domestic
money stock. This variable measures the impact of domestic stabilization action, and its
inclusion is based on the proposition that inflation is in the long run a monetary phenomenon.
The other influence on domestic inflation (as
postulated in Section I) is world money growth,
operating through the prices of internationally
traded goods. This influence can be estimated
either by adding world money directly as an
independent variable affecting domestic prices,
or by adding internationally traded goods prices
to the domestic price equation. Both approaches were tested and gave roughly the

the U.S. inflation rate in following periods, but
this is not incorporated in our equation because
of timing problems. The adjustment did not
necessarily have to occur immediately after the
end of Phase H controls. Phase HI lasted
through 1973, and market conditions may have
caused the adjustment to extend over an even
longer period. The error pattern in the estimated equation (as shown in Chart 1) suggests
that much of the adjustment occurred in 1974.
This equation was also estimated for the
wholesale price index. The results are summarized below.

no offset to world inflation and the higher domestic prices will be permanent.
A very simple test was conducted to measure
the influence of prices on wages, the former
being measured by the CPI and the latter by an
hourly-earnings index. If wages rise at the
average rate of productivity plus some proportion of the rate of domestic inflation-and if we
assume, as a first approximation, that average
productivity growth is constant-then we can
relate a distributed lag of past price changes to
current wage changes. The number of lag quarters required for the coefficient relating prices
to wages to approximate unity is crucial. The
shorter the lag, the larger and more permanent
will be the effect of world prices on domestic
prices. The results for the U.S. are given below.

6Log WPI = 7.0 - 3.4DUM + Zl.98 6 Log Mt _8
(3.0) (1.6)
(3.7)
2

+7. 416
'( 4.8)

LogPw t _2

12
610g W= 1.75 +
(2.1)

.63/4.15

1.88/43

L .906
(5.0)

log Pus t _12
-2 _
R /SE
OW/DE

.45/c.3
2.40/43

After 12 quarters (3 years), 90 percent of the
original past inflation is reflected in U.S. wages.
Longer lags do not increase the size of the coefficient. About 60 percent of the impact is
achieved in the first year, and 80 percent by
the end of the second year.
The critical period for the impact of domestic inflation on domestic wages is one year because, as discussed above, this is the length of
the period over which aggregate-demand adjustment takes place. Over a longer period, income effects can create an export surplus, thus
causing the exchange rate to appreciate and to
offset the effects of world inflation. The share of
internationally traded goods in the CPI market
basket is approximately 25 percent-60 percent
of which is . I 5, a number very close to the. I 6
coefficient on world prices in our estimated
equation explaining domestic inflation. Similarly, 60 percent of the share of internationally
traded goods prices in the WPI is .30, which is
somewhat smaller than the estimated coefficient
of .41.

The results differ from those in the CPI equation in ways one would expect. The time lag
between changes in the money stock and world
prices and changes in the WPI are shorter, reflecting the fact that wholesale prices tend to
move earlier in the cycle than do retail prices.
The coefficient in world prices was much larger
(.41 vs. .16) because the weight of internationally traded goods is much larger in the WPI
than in the CPI.
Effect of Prices on Wages

In Section 1 and more explicitly in Appendix
I, the responsiveness of exchange rates to differential (U.S. and world) inflation rates is related to the speed and completeness of the response of nominal wages to changes in prices.
If this adjustment occurs quickly and completely-i.e. if the short run wage-price elasticity is unity-then the higher domestic prices
induced by higher international prices will be
permanent. The higher supply price of output
will prevent an export surplus from developing
and will frustrate any exchange-rate appreciation. Without such appreciation, there will be

Summary of Empirical Findings

(1) World money supply, as a summary
measure of world monetary stabilization poli-

cies, is an important factor explaining the price
of internationally traded goods. Our estimates
indicate that over a period of three years every
one-percent increase in world money supply
growth in excess of 7.5 percent will add 3 1/2
percent to international inflation. This magnified effect of world money supply on world
prices provides support to the hypothesis that a
simultaneous expansion of aggregate demand
by industrial countries, resulting from synchronized money expansion, leads to a more than
proportional increase in world prices. The implications of the hypothesis are explored in
Section III.
(2) Domestic monetary and fiscal policy is
an important determinant of domestic aggregate demand, with a relatively short lag of
about one year. An increase in domestic money
supply growth will increase domestic aggregate
demand proportionately. On the other hand.,
changes in the real money stock will induce a
movement along a given aggregate demand
schedule of unit elasticity. An increase in

prices, from whatever source, will decrease the
real money stock and lead to a proportional
reduction in the demand for real output after
about one year.
(3) Both domestic money supply and world
inflation exert significant influences on the domestic rate of inflation, although the importance of domestic monetary policy is proportionately larger. A one - percent increase in
world inflation will, over a period of one year,
add a .16-percent increase to the domestic CPI
and .41 percent to the domestic WPI. In contrast, a one-percent domestic monetary expansion will, over a period of two years, add 1.5
percent to the domestic CPI and 2.0 percent to
the WPI.
(4) Wages will respond to domestic inflation over a relatively long three-year period.
However, in the critical first year-when it influences the extent of the impact of world inflation on domestic inflation-about 60 percent
of the higher CPI will be reflected in higher
wages.

III. Implications for the Future

The previous discussion suggests that expansionary monetary policies in the rest of the
world can have perverse effects on a domestic
economy. When one country acts in relative
isolation to control domestic aggregate demand, it can, at least in the intermediate period
of a business cycle, trade along the Phillips
curve and gain an increase in output at the expense of an increase in the inflation rate. However, when many countries are operating in
concert (whether by design or accident), the
net effect on a domestic economy could be a
higher inflation rate with little or no favorable
impact on output or employment.
During the 1950's and 60's, industrial countries largely conducted their stabilization policies in isolation from others, reflecting their
lack of synchronization of policies. Typically,
when one country was in an easy-money phase,
other countries were in a tight-money phase of
the cycle. The net effect was relative stability
in the growth of world aggregate c1emand.

The environment of the early 1970's was
different. Most major industrial countries followed tight and easy monetary polices in a
uniform pattern. This increased synchronization of monetary policy is illustrated in Chart
I, which shows the weighted-average money
growth rate of 10 major industrial countries
over the 1955-76 period. In the period through
1970, the world money stock grew at a relatively stable rate in the 7-8 percent range.
While in individual countries money growth
showed strong cyclical patterns-because they
were in different phases of the cycle-the average money growth was relatively stable. Only
in the 1970's was this pattern of stability
broken. From 1971 through early 1973, world
money stock accelerated to a 13-percent growth
rate, but then it decelerated in 1973-74 to a
9-percent rate. Finally, in 1975 through mid1976, it reaccelerated to approximately a 12percent growth rate. This instability resulted
from the synchronized pattern of monetary ex12

most central banks still attempted to maintain
fixed values of their national currency in terms
of dollars. However, this policy required buying an increasingly large amount of dollar
assets in response to the widespread attempts
by private citizens (both here and abroad) to
shift their portfolios out of dollars-first into
the stronger currencies like the German mark
and Japanese yen, and eventually into the currencies of almost all other industrial countries.

Chart 2

World Money
and
30

Reserve Growth
(Year-tn-Year Changes)
Percent

The international reserves of industrial countries grew at a relatively stable 7-percent rate
from 1958 to 1970, but from that point to early
1973, as foreign central banks monetized dollar
inflows, they expanded their holdin&s both of
international reserves and of domestic money
(Chart 2). Yet when most central banks collectively abandoned the fixed-rate regime in
March 1973, there was an abrupt deceleration
in the growth of international-reserve holdings.
Within a short time, central banks returned the
growth rates of their domestic money stocks to
more normal levels. However, just as the
acceleration in money led to a worldwide business cycle boom and inflation, so deceleration
in money led to a worldwide business recession
and a rise in unemployment to the highest levels
since the 1930's.

-5
1965
1960
1970
1955
1975
Source: Federal Reserve Bank of San Francisco-IFS data.

pansion of industrial countries during the
1970's.
Why Synchronized Monetary Cycles?

Central banks have usually responded in the
past to high unemployment rates by following
an easy monetary policy, especially when there
were no balance-of-payments constraints to interfere with that goal. HJ The same was true in
1975-76, as most countries reacted to high unemployment with a simultaneous domestic
monetary expansion which then brought about
another expansion in the world money stock.

The acceleration in world money growth in
the 1970-75 period was associated with the
breakdown of the Bretton Woods fixed exchange-rate system." Previously, countries other than the U.S. maintained the international
value of their national currencies at a fixed rate
by buying and selling dollars in the foreignexchange market. This policy worked well for
most countries in the period through the mid1960's. But then the U.S. inflation rate accelerated, and the resulting balance-of-payments deficit increased the supply of dollars to the rest
of the world relative to demand. The problem
came to a head in August 1971, when the U.S.
suspended convertibility of the dollar into gold
for foreign central banks. While this event led
to an exchange-rate adjustment (the so-called
Smithsonian agreement in December 1971),

Consequences of a Synchronized Cycle

This instability in the average growth of
world money was accompanied, with an average lag of about eight quarters, by greater instability in the prices of internationally traded
goods. This relationship is seen in Chart 3.
The broad - gauge measure of international
prices was relatively stable in the 1950's and
'60's, accelerated sharply in late 1972 and early
13

reason is that the current rise in materials prices
was preceded by a rise in the world money
stock, as was the case in 1972, but not in 1968.
If the lags between world money and international prices are stable, we would expect that a
rise in prices of internationally traded goods
will occur in 1977.
The duration of the rise in international
prices depends upon the future growth in the
world money stock. Given the lags in the system, however, a 10- to 14-percent rate of international price inflation in 1977 seems to be
already determined by past monetary policies.
This is in contrast with the negative rate of
inflation in 1976 and the 26-percent inflation
rate in 1974.
What do these international price developments imply for U.S. inflation in 1977? The
model suggests that-contrary to general economic opinion-flexible exchange rates may
not completely isolate a country from international inflation. The size of the inflation will
depend upon the extent of the rise in internationally traded goods prices and upon the share
of such goods in the domestic price index. On
the basis of the estimates of Section II, the
Wholesale Price Index (WPI) would rise 7 to
8 percent in 1977 versus a 4-percent annual
rate in the first nine months of 1976. The Consumer Price Index (CPI) would rise 6 to 7
percent in 1977 versus a 51;2 percent annual
rate in the first nine months of 1976.
Compared to the 1973-74 experience, the
impact of international inflation on U.S. inflation is expected to be relatively modest in 1977.
However, the impact would be large by the
standards of the 1950's and '60's, and could
increase in later years. The table below summarizes the impact on U.S. prices of alternative
world money growth rates.

Chart 3

World Materials
and General Industrial Prices
(Year-to-Year Changes)
Percent

!'-,,1aterials Pilee Index

60
40

-20

l.u.U.u-Li!
1965

! I

I I

! I

I! I I I!

1970

I I

1 !

I I

1975

Source: Materials Price: The Economist commodity price ($) index.
Genera! Industrial Price: IFS index of export prices of industrial countries.

1973, decelerated sharply in 1975, and continued low in the first half of 1976.
Meanwhile, sensitive materials prices tended
to move ahead of the general body of international prices. Materials prices generally move
in a more dramatic fashion than the prices of
other goods, reflecting the relatively inelastic
short-run demand and supply for such goods.
This greater price variance can, at times, give
misleading signals as to the future course of
other international prices. For example, sensitive materials prices increased 15 percent between late 1968 and late 1969, but were not
followed by an equally substantial rise in other
international prices. Eventually, the materials
price increase was reversed. However, in 1972
this index provided a reliable leading indicator
of a rise in the broader index of international
prices-and in 1974 it foreshadowed the decline in general international prices. In each
case the lead was about one year.

U.S. Inflation Forecast
(Assume U.S. MJ = 5 percent)
World MJ = 8%
World M,

The current rise in materials prices raises the
question of whether it is a prelude to another
bout of international inflation or a random
gyration soon to be reversed. The model developed here suggests that international inflation will, in fact, accelerate again in 1977. The

1978
1979
1980
14

= 12%

CPI

WPI

CPI

WPI

5.5
4.5
4.0

5.5
3.0
3.0

6.0
6.0
6.5

8.5
8.5
9.0

therefore to higher rates of imported inflation.
To avoid the unemployment consequences predicted by the Phillips curve would require even
further monetary ease, which on the aggregate
of all countries would further aggravate the inflation rate.

On the assumptions of a 5-percent growth
in domestic M, and an 8-percent growth in
world M 1 (the average of the 1960's) world
prices would be stable and the U.S. inflation
rate would be dominated by strictly domestic
considerations. By the end of the decade, the
CPI would rise at about a 4-percent rate and
the WPI at about a 3-percent rateY If world
M 1 growth continues at 12 percent, then the
longer-term trend in the u.s. CPI will be in the
6- to 6 h -percent range and the WPI in the
8 Vz - to 9-percent range. '2

In this context we have a "game theory"
solution. If anyone country took action in
isolation it could achieve an unambiguous improvement in its situation by following an easy
monetary policy. However, if all countries engaged in that practice, all would be worse off
than before, in the sense that they would experience a higher rate of inflation with the same
rate of unemployment. Thus, the optimal stabilization strategy in this world context would
be for all governments to agree to follow moderate monetary policies. While this course
would prevent any country which acts alone
from improving its unemployment rate as fast
as might otherwise be the case, the world as a
whole would experience a deceleration in inflation with a given unemployment rate.

Conclusion

What implications can policymakers derive
from this analysis? One possible implication
is that policymakers should respond to imported inflation by expanding domestic aggregate demand sufficiently so that the higher
prices will not induce a fall in real output. Even
if a country cannot control its domestic inflation rate, it can at least control its domestic
unemployment rate with appropriately stimulative stabilization policy.
This policy inference, however, is incomplete and therefore misleading. While it may
be desirable for anyone country to follow domestic stabilization policies to offset the effects
of imported inflation, if all countries followed
such a policy simultaneously the result would
be explosive inflation. The expansion in world
money supply would lead to an even larger
world business-cycle expansion and pressures
on internationally traded goods prices-and

This clearly suggests that in an interdependent world, with its close ties in real and financial markets, the degree of monetary independence expected under flexible exchange rates
has been circumscribed in a significant way.
This provides a rationale for increased international cooperation, along the lines attempted
at the economic summit conferences in Puerto
Rico in June of this year and at Rambouillet
in November 1975.

FOOTNOTES

1. This argument was presented by one of th('l authors in

3. This assumption contrasts with that of the monetary
theory of the balance of payments. It may be justified as
follows: In a world of flexible exchange rates, monetary
independence is achieved automatically through the absence of central-bank intervention in the foreign-exchange
market. In a world of fixed exchange rates, this is achieved
by central banks imposing capital market controls to such
an extent that domestic and foreign financial markets are
independent of each other over the relevant period of
analysis, i.e. the business cycle. It may be argued that the
experience of the early 1970's makes this assumption
unrealistic. However, we would argue that when loss of
domestic monetary control became a significant problem,
central banks took actions to retain their monetary independence by moving to a flexible-exchange rate regime in
March 1973.
4. An alternative explanation is that real net exports do
increase, but that this is offset by a disproportionate

another context in the Spring 1975 issue of the Review.
See M. W. Keran "Toward an Explanation of Simultaneous
I nflation-Recession_"
2. This relationship may be derived from the monetary
theory of the balance of payments, which states that world
prices are determined by the stock equilibrium condition
of equality between world demand and supply of money
balances. The excess of the growth rate of nominal world
money supply over the growth rate of real world money
demand determines the world rate of inflation. H. J.
Johnson, "The Monetary Approach to Balance-of-Payments
Theory," Journal of Financial and Quantitative Analysis.
March 1972. J. A. Frenkel and H. G. Johnson, eds., The
Monetary Approach to the Balance of Payments, University
of Toronto Press, 1976, and R. Dornbusch, "The Theory of
Flexible Exchange Rate Regimes and Macroeconomic PoliCY," Scandinavian Journal of Economics (forthcoming).

change in import and export prices. The latter explanation
is perhaps more realistic in the case of those developed
countries which import raw materials and export finished
products. Since raw materials have a relatively less elastic
short-run supply response to changes in demand than do
finished products, raw-material prices will rise before finished-product prices. The temporary adverse shift in the
terms of trade could offset the real export surplus and
leave the nominal trade balance temporarily unchanged.
5. More precisely, in the case being considered, world
stabilization policies lead to an expansion of world aggregate demand which exceeds the expansion of domestic
aggregate demand. This is the most relevant case in terms
ofihe U.S. in the 1970's. For formal analysis of this, as
well as of the reverse case, see Appendix I.
6. The countries and their respective weights are: Belgium
(.060), Canada (.071), France (.108), Germany (.166),
Italy (.074), Japan (.115), Netherlands (.071), Switzerland
(.027), United Kingdom (.099), United States (.213). The
weights do not sum to one due to rounding. The weights
are derived from export and import trade shares in 1975.
7. These observations represent an extreme example of a
fairly common phenomenon in estimating "change" equations where the random element is large. These equations

are more successful in forecasting the average value of
the dependent variable than in forecasting the individual
quarters.
8.• An F-test indicates that the improved fit gained from
the inclusion of this variable is significant at the .05 level.
9. See M. Keran, "An Appropriate International Currency:
Gold, Dollars, or SDR?," Federal Reserve Bank of St. louis,
Review, August 1972 (Reprint 78), and D. I. Meiselman,
"WOrldwide Inflation: A Monetarist View," The Phenomenon
ot Worldwide Intlation, American Enterprise Institute, 1975.
10. Balance-ot-payments constraints were eased in late
1974 tor two reasons: (1) most industrial nations had an
historically high level of international reserves at that time,
and (2) flexible exchange rates were seen as a device to
avoid traditional balance-ot-payments consequences of easy
money.
11. The differential reflects the lower productivity growth
in the service industries (which have a larger weight in the
CPI) than in the goods industries (which have a higher
weight in the wpn.
12. The differential is due to the CPl's lower internationaltrade component, which more than offsets the lower productivity growth in the service industries.

APPENDIX I
Stabilization Policy in a World Context

This appendix formalizes the model presented
in Section 1. Since this paper concentrates on
monetary variables, the model is simplified by
assuming that both domestic nominal government
expenditure and real world income are constant.
To further facilitate exposition, it is assumed that
all goods are internationally traded, although this
assumption does not alter qualitative results.

specification implies that the elasticity of the aggregate demand curve is unity. Equation (4) expresses aggregate supply as a function of the price
level relative to the nominal wage. This particular
specification is based on the assumptions of an
aggregate Cobb-Douglas production function and
profit maximization by competitive firms. Equation (5) is implied by an assumption of flexible
exchange rates. Abstracting from capital flows
and central bank interventions, net exports must
be zero, in which case domestic aggregate demand
and supply are equal." Finally, Equation (6)
states that wages are flexible upward, rising with
prices, but rigid downwards. The term 00 may be
thought of as the initial wage level. As a result of
downward wage rigidity, the labor market does
not necessarily clear and output may vary over the
short run.
Upon differentiating, substituting, and rearranging, we arrive at the following relationships (lower
case letters are logarithmic differentials):

The model is summarized by a system of six
equations in six unknowns:
Pw
P

o i1w

0' 0:::>0

(1)

EPw

(2)

So 11
Q

IA = 11AX

/W)

B > 0
·0

(3)

Yo' y>o

(4)
(5)

{olo,

0o}

00>0, 0::'0::.1

(6)

n = J..-. ('!l- + w)
l+y

where Mw and M are exogenous world and domestic money supplies, respectively, Pw and Pare
world and domestic price levels, E is the exchange
rate in domestic currency, Y is domestic nominal
income, Q is real domestic output, and W is the
nominal wage.

q =

(m - w)

= p -

(7)

(8)

(9)

"Under fixed exchange rates or a managed float, (5) is
replaced by

Equation (1) expresses the world price level as
a positive function of world money supply. In
Equation (2) world and domestic price levels are
linked by the exchange rate due to goods arbitrage. Equation (3) states that nominal income is
determined by domestic monetary policy. This

prJ - Y

(5')

where R is a reserve flow (a policy parameter). SiI:ce
positive net exports add directly to income, equation (3)
then becomes
y"

(PQ-,!) "

(3')

equations (7) - (II) were solved simultaneously
for all endogenous variables, the solution would
be exactly the same as the case of a simple domestic monetary expansion, except that the exchange
rate would depreciate less or possibly even appreciate due to world price inflation.

(10)
w

~1AX

(11 )

(op,o)

Equations (7) and (8) indicate that domestic
equilibrium price and output depend on domestic
demand management policy, measured by nominal
money growth, and also on nominal wage movements. The former causes a shift in the aggregate
demand curve and the latter a shift in the aggregate supply curve.
We go on to consider the cases of first a simple
domestic monetary expansion, and then a simultaneous domestic and world monetary expansion.
Domestic and world money supplies have been
assumed to be independent of each other.

(11 ')

Substituting (11') into (7) and (8), and solving
yields for m > 0:
p ,

(13 )
(14)

e ' P

A domestic monetary expansion will have a positive inflation ·effect and a non-negative output
effect. Domestic inflation is offset by proportional
exchange rate depreciation. An interesting result
here is that the closer the wage-price elasticity ( 0)
is to unity, the higher will be the inflation-depreciation effect and the lower will be the real output
effect. In the limiting case where, 0 = 1 the output effect is zero.
A short-run relationship between the change in
output and inflation is derived by combining (12)
and (1 3) to yield:
q ,

(1-0)

y p

(14')

The extent of wage inflation in the economy will
therefore depend on the interaction of world and
domestic monetary ·policies. If world price inflation exceeds the domestic inflation rate warranted
by domestic monetary policy, then wage inflation
will be proportional to world price inflation. If,
on the other hand, the world inflation rate is less
than or equal to that warranted by domestic policy, then wage inflation will be proportional to
domestic price inflation. World price inflation is
given by equation (10), while domestic inflation
warranted by domestic monetary policy is given
by equation (12). The appropriate wage adjustment equation for a simultaneous domestic and
world monetary expansion is therefore

(12)

'(lX~li~l))

If, on the other hand, endogenous variables adjust at differeDt speeds to shocks to the system,
then these conclusions could be altered. In particular, suppose that wages and prices adjust
quickly relative to the exchange rate. '" Then world
price inflation would have no initial impact on the
exchange rate, but would be translated into proportionately higher domestic prices through (2)
and then higher wages through (6). The initial
domestic price inflation is a disequilibrium phenomenon and will eventually be partially offset by
exchange-rate appreciation as the real sector begins to adjust to the price rise. Wages, however,
will remain permanently higher because of the
assumption of downward rigidity.

Domestic Monetary Expansion
Since domestic monetary expansion places upward pressure on the price level, Equation (11)
becomes
w ' op

e ' p - Pw '

vi - t1AX (o~mw'

T+YrJ=6T} ,

for

ril

w' m,:-o

( 16)

Let 8 = l/(~+~y(l-o»).. In the case where mw:'0m
world inflation is less than or equal to that warranted by domestic policy-equilibrium is defined
by equations (12), (13), and (14'). The more
interesting case is where
m\/ 8m
world inflation exceeds that warranted by domestic policy.

(15)

As long as wages do not completely adjust to
prices (0:,..6< 1) there is a short-run trade-off between inflation and output growth. The slope of
this trade-off is related positively to y , labor's
share in output.

Then combining (7)-(10) and (16) yields for
mw, m > o.

Simultaneous Domestic and World Monetary
Expansion
Now consider a simultaneous monetary expansion at both the domestic and world level. If

-Y- ( ~ + ocrm)

-Y- ( m - o~m )
l+y
w

l+y

> 0

(17)
( 18)

"For an explanation of why the exchange rate might he
sticky, see Section 1, p. 4.

e=_l_(m_m w )
l+y
0·

(19)

monetary expansion, or in the case where mw .5- om.
The exchange rate will unambiguously appreciate.
If the wage-price elasticity (6 ) is sufficiently close
to unity, the output effect will be negative.

Domestic inflation is higher and domestic output
expansion is lower than in the case of a simple

APPENDIX II
Evidence from Other Countries: The Case of the
U.K. and Germany

The model developed above suggests that the
impact of international inflation on the domestic
price level depends upon. ( I) the share of internationally traded goods in the domestic price index
and (2) the responsiveness of domestic wages to
domestic prices. Both influences work to increase
the impact of international inflation on domestic
prices. 1

These equations are estimated over the period
1960.2 to 1976.3. In the case of the U.K., it takes
two quarters for the full effects of a rise in the
domestic price level to be transmitted to nominal
wages. When longer lags are estimated, the coefficient rises above unity. The minimum standard
error for this equation is achieved with a lag of
eight quarters and a coefficient value of 1.39. This
implies that money wages in the u.K. respond on
average by more than the rate of inflation! For
Germany the coefficient relating domestic inflation
to domestic wage rates was insignificant for all
lagged periods investigated. The result presented
was the lagged pattern (t-8) with the minimum
standard error.

Wages and Prices
In the body of this article the response of wages
to prices was estimated using U.S. data. The results
suggest that over a period of 3 years every I-percent
increase in consumer prices leads to a .9-percent
increase in wages. Sixty percent of the wage increase occurs in the first year. Similar equations
are estimated for the U.K. and Germany. Nominal
wages are determined primarily by the growth in
lab'(;r productivity and the rise in consumer prices.
If productivity grows at a relatively constant rate,
then an equation estimating changes in wages in
response to changes in prices would as a first-order
approximation give information on both sets of
determinants. The constant term would show the
growth in wages with respect to productivity, and
the coefficient on the price variable would show the
growth in nominal wages with respect to the rate
of inflation.
On the basis of postwar economic history, one
would expect that the coefficient relating wages to
prices to be larger in the case of the U.K. than in
the case of Germany. In the U.K. the strong laborunion movement has exercised its power systematically to protect the real wages of its members
from the effects of inflation." In Germany, on the
other hand, the labor unions have been less militant.
While German wages have risen faster than German prices in the last J 5 years, labor unions apparently have not tried to use their power to vary
wages substantially in response to inflation. The
estimated equations for the U.K. and Germany are
given below.~'

These results suggest that international inflation
is almost completely transmitted to U.K. prices,
but that it has a much smaller impact on German
domestic prices. We can confirm this conjecture
directly by estimating domestic price equations for
both countries.
U.K.
6109 CPI

3.0 + i 1.01
(1.4) '(5.9)

,\109 Pt-2

(0.7)

6109 M _ +
t 8

.61

6109 PW _
t 4

.5)

.65/4.76
2.10/38

Germany
12
6109 CPI

-1.0 +
(0.6)

Z·38

6109 M +
t 12
(2.4)
2

R /SE
O,J/OE

Z .15

(3.6)

6109 PW _
t 4

.38/2.44
1.61/42

The equation for each country is identical to
the specification in the test for the U.S. inflation
rate. For the u.K. the international inflation
variable has a large and significant coefficient. For
every l-percent increase in international inflation
(measured in a dollar-denominated index) over
the previous four quarters, U.K. CPI inflation will
increase .61 percent. In the German case a I-per-

.39/7.49
1.67/45

= 8.3 + 1.018 6109 Pt-8

(4.3)

Z 0.15

Germany
.\109

3.1 +
(2.2)

R /SE
OW/DE

U.K.
Ll09 W

(.04)

Germany was 54 percent, and for the u.K. 56
percent.
These results are consistent with the wage equations presented above. Since German wages are less
sensitive to domestic inflation than U.K. wages, a
such smaller proportion of the international inflation will "stick" in Germany.

cent increase in international prices leads to a .15
percent increase in domestic prices. Germany is
almost as open an economy as the u.K. but its
world inflation coefficient is much lower (.15
versus .61). A rough measure of the share of international goods in a country is the share of exports plus imports to GNP. For 1975 the share for

APPENDIX FOOTNOTES
1. A third influence is the speed of response of real income
to changes in real money balances. Earlier work by one
of the authors suggests that such response is relatively
uniform across countries in the range of one to one and a
half years. M. W. Keran, "Selecting a Monetary Indicator," Federal Reserve Bank of St. Louis, September 1970
(Reprint No. 59).
2. Sir John Hicks, "What's Wrong with Monetarism,"
Llovds Bank Review, October 1975.
3. P is the consumer price index for both countries. W is
an index of average monthly wages for the U.K. and an

index of hourly earnings for Germany. International
Monetary Fund, International Financial Statistics (com-

puter tape).
4. This is explained partially by institutional factors. During 1973-74 the combination of domestic price control and
threshold wage agreements in a world inflationary situation caused real wages to rise on trend. For a fuller
explanation, see Marcus H. Muller, "Can a Rise in Import Prices be Inflationary and Deflationary? Economists
and the U.K. Inflation, 1973-74" American Economic
Review, September 1976.

Kurt Dew*

Two new themes-( 1) optimal control and
(2) rational expectations - have arisen recently in the economic-policy literature, ~nd
each of them promises to have a dramatic impact upon future analyses of appropriate policy-making. First, the literature on optimal
control deals basically with the use of imperfect
econometric models in forming policy decisions. The literature emphasizes the development of efficient rules for responding to the
errors that would otherwise lead policymakers
away from their economic goals. Optimal-control research seems to suggest that with a reasonably careful utilization of an econometric
model and the use of mathematically derived
rules for policy adjustment, the policymaker
can improve upon alternative policies such as
the constant money-growth rule proposed by
Nobel laureate Milton Friedman. Some research even suggests that the adjustment process derived from optimal-control techniques is
so efficient that the policymaker who uses the
wrong model (i.e., one that doesn't describe
the economy's behavior as well as other available models) may still improve upon Friedmanesque inactive policy by responding quickly
to his mistakes. 1 Optimal-control results thus
seem to provide arguments for activist economic policies.
In contrast, the rational-expectations literature tends to discredit activist policies because
of a different interpretation of one of the fundamental issues in policy-making-the nature of
the public response to economic-policy decisions. Since the time of Keynes, economists
have made the reasonable assumption that eco-

nomic agents - households and firms - cope
with an uncertain future by making forecasts,
and that these forecasts play a key role in determining eventual future levels of economic
activity. Yet the rational-expectations literature suggests that households and firms do not
form their expectations of future events the way
that most economists presume they do. Furthermore, if economic agents form expectations
in a way that is "rational" (i.e. using all the
available information rather than just part of
it), the latitude of policy-makers to exert a
beneficial impact upon economic welfare is reduced or even eliminated. Policy-makers in
such a world may not improve the expected
future levels of economic activity, but they
may, by informing consumers ahead-of-time
about future policy, reduce consumer uncertainty about future variations in economic activity. Therefore, rules such as Friedman's constant money-growth rule, having the twin advantages of simplicity and clarity, are good
policy prescriptions. The rational-expectations
assumption thus tends to nullify the implications of optimal-control analysis and leads to
the conclusion that passive policies are the
most appropriate ones.
Consequently, a policy-maker's choice between active policies of "leaning against" the
economic winds or passive policies such as
Friedman's constant money-growth rule comes
down to this: Can the policy responses that are
generated by optimal-control rules overcome
the uncertainties regarding future economic behavior that are created by rational expectations? To provide some insight into this issue,
we describe two alternative methods used by
economists to analyze the formulation of house-

*Economist, Federal Reserve Bank of San Francisco.

hold expectations. By considering the particular example of the consumption decision, we
show how the outcome of a particular policy
can be adversely affected through a misinterpretation of the means by which the consumer
forms his expectations.
When the policy-maker writes his decisions
in stone, a mistaken notion of the method of
formulating consumer expectations can be dis-

astrous. But the more interesting case is one
where the policy-maker is more flexible, using
the method of optimal control to adjust his
policies to his initial errors. In this instance,
we show that the policy-maker can improve his
policy results by using optimal-control techniques, but that his misjudgment of the means
by which consumers forecast future income
nonetheless adds instability to the economy.

Formation of Expectations: Adaptive

The current standard approach to the modeling of expectation formulation assumes that
economic agents form expectations adaptively.
The adaptive expectations hypothesis is developed by induction. In the case of consumer
forecasting of future income, for example, the
consumer is presumed to begin with a forecast
of levels of personal disposable income (PDI)
in each future period. The consumer knows
that he will commit forecasting errors, and he
believes that these errors are related, in the
sense that a low forecast in one year indicates
that all his predictions may be .too low. He
reflects this knowledge by revising his future
estimates upward when his current forecast is
too low. We might suppose that a $10-billion
under-estimation of PDI this quarter will cause
the consumer to revise his next year's forecast
upward by 20 percent of the $lO-billion error.
If so, this year's forecast will differ from last
year's attempt by $2 billion, 20 percent of $10
billion.

itself would become less and less important,
and that the actual levels of PDI in periods
after the original estimate would become more
and more important. If so, we could safely
state that the forecast of PDI for 1977 (or
1978, or any subsequent year) depends upon
actual past levels of PDI. We might write this
hypothesis in the form of an equation

Where the symbol
E197G(PDI1977) represents the estimate (E)
in 1976, of personal disposable income (PDI)
in 1977,
and k 1 , k 2 , k", ... the weights used to project
past levels of PDI into 1977.
We might further suppose that the estimate is a
weighted average of the past values of PDI, so
that k 1 + k 2 + k a + ... = 1.

But by the same token, last year's. estimate
was the result of a revision of the forecast of
two years ago, which was increased or lowered
depending upon whether it had been an underestimate or an over-estimate. So we may think
of this year's expectation of 1977 PDI as a
forecast made in 1974 that has been subsequently revised in light of the errors in 1975
and 1976 income-or proceeding backward,
may even think of it as a forecast originally
made (say) in 1970 and adjusted for the errors
made in PDI in each subsequent year. We
might thus expect that the original estimate

The hypothesis that expectations of future
values of an economic variable are weighted
averages of past values of this variable is
known as the hypothesis of adaptive expectations. Macroeconomists utilize adaptive expectations to help explain the sluggishness of the
economy's response to external shocks. If the
adaptive expectations hypothesis is correct,
consumption (for example) would be set by
the consumer at a level proportional to his
estimate of the value of his own future income.
But because the consumer forms his expecta21

lower PDI of earlier years also would have an
effect on his forecast. Thus, he raises his estimate of expected future income more slowly
than the rate of increase in present earnings.

tions adaptively, he is sluggish in revising his
estimates and therefore sluggish in revising his
consumption. Last year's PDI might lead him
to expect an increase in his future PDI, but the

Formation of Expectations: Rational

current value the consumer places upon the
income he expects to earn throughout his lifetime. Suppose, for example, the consumer were
certain that he would earn $10,000 in personal
disposable income for each coming year in perpetuity. To determine his life cycle income, he
uses the relationship

Some economists have challenged the adaptive-expectations assumption because the consumer under this approach tends to ignore
some important information about the future
path of key economic variables. The consumer
may well have some notions of the intentions
of public policy-makers, which it would be
"rational" for him to include in his forecast of
the growth of future disposable income. 2
How would a consumer forecast the effect of
government expenditures upon PDI? He might
estimate the present value (PV) to him of the
stream of future government expenditures:
(2)

G1976 +

l~P

(4)

PDI

1976

G1977 + n4;-2 G1978 + ...

PDI

1977

POI
POI
1978
1979
k(POI 1977 + - 1 - - + " -+
+P
(1+p)2

Here "k" is a constant chosen so that the sum
of the weights of each of the yearly forecasts of
PDI is one, and A is a weighted average of the
PDI's. For example, the weight of PDI ,D7 , is k;
the weight of PDI ,D78 is k/l +p; for PDI 'D7 !l'
1</(1 +pF and k + 1</1 +p + 1'/(1 +p)2 +
.
= 1. Since in this case PDI ,D77 = PDI 1D78 =
.
= $10,000 and the weights on the PDI's sum
to one, A = $10,000. The consumer is interested in the behavior of income over his lifetime because he prefers to minimize the year~
to-year variation in his rate of consumption.
The consumer's income may fluctuate as time
goes on, but he attempts to mitigate the effects
of his fluctuating income upon the level of consumption, and pays attention primarily to the
average level of income he expects to receive
over the long haul.

Where G jD ,,;> G 'D77 • • • are the intended levels of government expenditure in the future,
and p is the consumer's internal rate of discount
of future disposable income, added to reflect
the fact that a dollar received now is of greater
value to the consumer than the same amount
received later.
Or if the consumer were more sophisticated,
he might analyze the full impact of government
spending by estimating the value of the added
disposable income accruing from the intended
government expenditure.
(3)

Aggregating consumption over the population of consumers, we may characterize the
life-cycle hypothesis in two equations

~2 PDI 1978 + ...

where PDI 1ll7 ,,, PDI ,D ,7' • • • are the added disposable income in future years resulting from
the intended government expenditure program.

(5)

C
= aA
1977

The Consumption Function

According to the widely accepted life-cycle
hypothesis, consumption expenditures during
any particular period of time depend upon the

A=ntPT

[

POI

1977

POI 1979
POI1978
+ --+ --+
1+p
(1+p)2

...J

where C I077
a

=
=

consumption in 1977
proportion of life-cycle income
consumed in each period.

Chart 1

Income Generated under Adaptive
Expectations and Rational Expectations

These equations describe how consumption is
related to expected life-cycle income when A,
the value of life-cycle income, is known to the
consumer.

Change
($ Billions)

Disposable income generated
with adaptive

Consumption Function with Adaptive
Expectations

expectations~

When we cease to assume that the consumer's life-cycle income, A, is known beforehand, we must replace it with an estimate, denoted by E(A). If expectations of future PDI
are adaptive, expectations take on a form

Quarters after beginning of program

(6)

1976

CA) ~ k o PDI 1976

+ k 1 PDI 1975 + ...

In other words, the current (1976) estimate of
life-cycle income depends upon past disposable
income. To illustrate the effect of adaptive expectations upon the consumption decision, we
use a variant of the expression for the consumption function used by Modigliani. 3

when expectations are rational. Although the
new government expenditures increase income
from the very outset at a rate in excess of $10
billion per year, the consumer initially has only
a single quarter of higher income to offset his
past experience of income at a lower level. He
is thus slow to revise his estimate of life-cycle
income upward, so that consumption at first
rises by only a relatively small amount. However, as time goes on and the Government continues to spend at the higher rate, the consumer
becomes increasingly convinced of the permanence of the additions to income. This increasing certainty leads to higher levels of consumption and therefore to steadily increasing
levels of income, over and above the $10 billion
per year in added income produced directly by
the government expenditures. 4 If in general we
define .6PDIt as the difference between PDI
with and without the added government expenditures in period t, then the government-spending multiplier becomes
. Chart 1 indicates the path that income takes when the
estimation of permanent income is based upon
adaptive expectations.

Since we assume that consumers form estimates of future income on the basis of knowledge of the past levels of income, we may conclude that they respond to a change in public
policy as they would to any other economic
shock, revising their expectations of future income only slowly as income increases. As a
result of this sluggish response, a change in government expenditure increases the level of consumer expenditures quite slowly. To demonstate this point, we use the consumption function
from Modigliani to display the effect of adaptive
expectations upon a critical variable,
the income multiplier of an increase in government expenditures at any point in time.
Consider the case of a $1 O-billion increase in
real government expenditures sustained over a
three-year horizon. We assume at first that the
policy-maker believes that consumers form expectations adaptively, and then contrast the expected policy outcome with the actual outcome

The increase in the multiplier
through
time is due to the increased consumer estimate
23

of life-cycle income, which in turn results from
expectations of future increases in government
expenditures and higher estimates of consumptiondue to multiplier effects. In Table (II), the
second column represents the consumer estiI11ate of aqditional life-cycle income resulting
from the three-year government expenditure
program and its multiplier effects upon consumption. Similarly, the third column shows the
portion of the higher life-cycle income due directly to government expenditures, i.e., without
multiplier effects upon consumption.
Table I
Multiplier Effects Under Adaptive Expectations
Quarter

o
1
2
3
4
5
6
7
8
9
10
11

Increase in life-cycle POI
Total
Due to G

1.74
3.16
4.76
6.38
7.98
9.55
11.09
12.56
13.95
16.40
16.40
17.41

1.56
2.99
4.28
5.44
6.46
7.35
8.12
8.75
9.25
9.63
9.88
10.00

Thus, consumers' forecasts of the future government-spending contribution rise throughout the
period, until at the end of three years the consumer expects to receive $10 billion per year in
perpetuity.
There is a disturbing aspect to this adaptiveexpectations approach. At the end of the threeyear period, when it is public knowledge that
the $1 O-billion government-expenditure program will be curtailed, the adaptively forecasting consumer is expecting the government to
continue spending the $10 billion in perpetuity.
It takes three more years without the $10 billion
to disabuse him of this notion.
24

Consumer Knowledge of Policy-maker's Intentions

How would the consumer value the same
3-year, $lO-billion per year program of increased government expenditures if, contrary to
the policy-maker's belief, he were to perform
according to a rational rather than adaptive
scheme of expectations? This question can be
answered by reference to equation (5b). As
this equation suggests, the rational consumer is
concerned about future income, not past income, so that past policies only matter to him if
they affect future income. He will thus react to
the government-expenditure program by evaluating its future effects. As the three-year period
approaches its end, the program will have very
little further impact on his consumption decision, because it will only affect his income for a
few remaining quarters.
The policy's impact on life-cycle income instead will be maximized at the outset, because
government expenditures are expected to continue for twelve quarters into the future. Consequently, consumption out of life-cycle income
-and the income multiplier-also will be
greatest in the beginning of the program.
Chart I displays the difference between the
income generated by a government program
based on adaptive expectations and the income
generated based upon rational expectations. In
the latter case, the fiscal stimulus is greater at
the beginning of the period and thereafter declines-just the opposite of what would be ex-

Table II
Multiplier Effects With Rational
Quarter

o
1

2
3
4
5
6
7
8
9
10
11

Expecta~ions

Increase in life-cycle POI
Total
Due to G

6.50
5.90
5.31
4.86
4.15
3.60
3.06
2.52
2.00
1.48
0.98
0.48

5.27
4.86
4.44
4.02
3.60
3.17
2.73
2.29
1.84
1.40
0.93
0.47

pected by a policy-maker using an adaptive-expectations forecast, who would be increasingly
disappointed throughout the three-year period.
Since the rational consumer expects future
consumption to be increased as a result of the
higher government expenditures, the value to
him of the government-spending program exceeds the value of the expenditures themselves,
including the value of the added consumption
induced by those expenditures. In either case,
the value of the income multiplier of the spending program declines as time goes on, whereas
the multiplier associated with an adaptive-expectations approach increases as time goes on.

Policy-maker's Mistaken Assumption of
Adaptive Expectations

As Chart 1 shows, the policy-maker who mistakenly assumes that the consumer forms expectations adaptively would find his policy multipliers becoming increasingly incorrect over the
three-year period. This is not an unusual turn
of events for the macroeconomist. Indeed, substantial empirical evidence suggests that policy
multipliers are subject to massive uncertainty.
Carl Christ has pointed out that estimated values
of policy multipliers differ widely among the

major econometric models, and has suggested
that this divergence of opinion seriously damages economists' ability to give policy advice. c,
The lack of certainty about the effect of public policies would clearly be of serious concern
if the policy-maker were required to write his
decisions in stone. Arguing against the Christ
conclusion, Gregory Chow has shown that a
more flexible policy, which is revised when
short-run errors occur, can under some circum25

ers~'. Such a policy-maker would begin by looking at his income goal for the first quarter. In
the present example, suppose that the policymaker wishes to raise the level of GNP by $10
billion above the level it would otherwise attain
in each of the next four years,8 and then reduce
government expenditures to their old levels. At
the beginning of the four-year program, the
poiicy-maker might well announce-utilizing
his adaptive expectations assumption-the desirable levels of increased government expenditures throughout the entire period.
We will assume that policy is revised once
each year over the four-year period. We will
also assume that the policy-maker does not
"forgive" himself for policy mistakes, i.e., he
intends to add the same $40 billion to the level
of GNP throughout the period as a whole regardless of his year-to-year performance." We
can then determine the outcome at the end of
each year, and, show how the policy-maker

stances be quite effective in offsetting multiplier
errors. (; Chow analyzes the policy-maker's behavior and his impact upon the economy when
he receives conflicting signals from two major
econometric models and erroneously follows the
incorrect one. To evaluate his argument and
obtain a realistic picture of the impact of publicexpenditure decisions when based upon a mistaken understanding of consumer expectations,
we must allow the policy-maker the latitude to
adjust his decisions.
We will imagine a world in which the policymaker believes that the consumer is an adaptive
forecaster, but the consumer is actually "rational" in the sense that he bases his forecasts
of disposable income upon his knowledge of the
announced path of government policy. When
the policy-maker discovers that he has made a
forecasting error, he will adapt his policy to this
mistake, revising his planned expenditures and
announcing his revised intentions to consum-

Chart 2

Intended and Actual Expenditures
Actual

Intended
$ Billions

$ Billions

Billions

Year II

Year I
,0

6
II

III

Year III

III

Year IV

o '--------'

OL------

III

would revise his subsequent policy whenever he
errs in hitting his GNP target.

Chart 3

Actual and Intended Changes
in Disposable Income

Chart 2 pictures each of the strategies that
the policy-maker constructs in this case. After a
false start in the first year, he realizes that his
policy did not incorporate sufficient stimulus in
the last half of the four-year period. This turn
of events results inevitably from his belief that
the stimulus provided by past policies is the
dominant concern of the consumer, when in fact
the consumer is quite rationally concerned with
the future stimulus the policy-maker intends to
provide.

Assumption of Adaptive Expectations
Change
($ Billions)

The policy-maker then makes three revisions in his forecast:

20
/

1.) Reduces the planned additions to income in the remainder of the period by the
amount of the initial overshoot.

~Actual

Change

2.) Revises upward the consumption multiplier. The policy-maker mistakenly assumes
that part of the added income was due to a
shift in consumer's preference to a higher rate
of consumption out of life-cycle income.

o "'-__--'
1917

--'1978

--L

1979

-'-_-'

1980

icy-maker from making the necessary adjustments to target income more closely is the temporary nature of government policy. Even
though the policy-maker boosts expenditure
levels by hefty amounts, the temporary nature
of the program tends to offset these increases as
the end of the program moves closer in time.
If the policy-maker intended a permanent increase in annual income, the optimal-control
procedure would provide him with much greater
success.

3.) Increases the estimate of life-cycle income. The policy-maker correctly detects
that the consumer's estimate of life-cycle income was higher in the current period than
he had expected, but he incorrectly concludes
that the estimate of the next period's lifecycle income will also be higher. These revisions in the policy-maker's forecast result in
a change in the planned policy over the following three years.

Table III

The policy-maker then revises his 'intended
expenditures downward in an attempt to bring
his projection of more rapid income growth
back to desired levels.

Income·Expenditure Patterns with Given
Expenditure Information

Chart 3 pictures the effect of the four-year
policy upon income, and contrasts this with the
policy-maker's intended levels of income. In
the end, the increased government expenditures
produce $4.58 billion less PDI than intended, as
Table III demonstrates.

Year

Expenditures

Income

1
2
3

7.63
3.85
10.34
6.03

10.68
5.52
12.34
6.88

Total
27.85
Ratio of income to
expenditures-35 .42/27 ,85

35.42

The fundamental factor preventing the pol27

1.27

Conclusion
While consumer behavior is an important
concern of the policy-maker, policy decisions
are an equally important concern of the consumer. Since policy is announced ahead of time
while other events affecting economic growth
are not, it is reasonable to suppose that households and firms will be affected by policy decisions in a qualitatively different way than they
are by other economic shocks.
This possibility serves to emphasize the importance of the distinction between the response
of an economy to an unforeseen turn of events
and the same economy's response to a predictable change in policy. While the existence of
prolonged changes in economic growth is indisputable, the policy-maker's ability to offset these
divergences in the short run is still open to question.
What we have shown is that announcing economic policies ahead of time may create serious
difficulties for the policy-maker. This announce-

ment can .affect consumer expectations in ways
that are difficult to forecast. But the governmentcan take two measures that would reduce
the extent ,of this problem.
(1). If the policy-maker did not announce
his policies ahead of time, the consumer would
have no information about future policies and
would therefore form his expectations adaptively.
(2). If government-spending programs
brought about permanent, rather than temporary, increases in future disposable income, the
government-expenditure multiplier under rational expectations would not decline so rapidly
through time, and control of income would become easier.
However, these options are, to our good fortune, not available to elected policy-makers.
The benefits of a government elected by the
people are, like most benefits, not without economic costs.

FOOTNOTES
1. Gregory C. Chow, "Usefulness of Imperfect Models for
the Formulation of Stabilization Policies." Princeton University: Econometric Research Program, Memorandum
#199 (1976).

= $1.56 billion. Since the change in consumption and
income associated with the government-spending increase
is found from the joint solution of
bC

2. Critics of the rational-expectations argument note that
the conclusion that policy is impotent is based upon the
very special assumption that economic agents forecast
prices only. R. J. Gordon, "Recent Developments in the
Theory of Inflation and Unemployment." Journal of Monetary Economics, (April 1976), pp. 185-219. If economic
agents forecast quantities-e.g. consumers' forecast of
future income in the commonly accepted life-cycle consumption hypothesis-then policymakers may still have a
beneficial impact upon such goals as income, employment
and price stability. Advocates of ratiOnal expectations disagree, as in R. E. Lucas, "Econometric Policy Evaluation,
A Critique," Journal of Monetary Economics, (January 1976
supplement), pp. 19-46. Despite the possibility of a beneficial short-run impact on employment and income, if expectations are rational, policy-makers will still misjudge
the long run impact of their decisions.

I-'Je

and

"PDl

have

~PDIo

= .663 [.1564.:l.PDl)
= IIC O + 10.

= the

change in income resulting from the first
quarter of ~ncreas2d government expenditures

= 1.16 + 10 = $11.15 billion, r;;easured at an
annua 1 ra te.

5. Carl F. Christ, "Judging the Performance of Econometric Models of the U.S. Economy." International Economic Review, (February 1975), pp. 54-74.
6. Chow, op. cit., p. 23-24.
7. The policy-maker assumes the error to be partially
carried forward through the lagged adjustment term,'-l
in the consumption function, and through the effect of
the unexpected level of income upon the adaptive-expectations estimate of life-cycle disposable income.

3. Franco Modigliani, "Monetary Policy and Consumption:
Linkages via Interest Rates and Wealth Effects in the MPS
Models." Consumer Spending and Monetary Policy. Conference Series No. 5 (Boston: Federal Reserve Bank of
Boston, 1971).

8. The four-year period is chosen because it is the length
of the political cycle determined by presidential elections.

4. In the first quarter of the expenditure program, for
example, the increment provided by increased government
spending to consumers' life-cycle income is 0.1564(10)

9. This presumption is consistent, for instance, with the
wording of the Humphrey-Hawkins bill.

APPENDIX I
Derivation of government expenditures and their
impact on DPI

In an adaptation of the Modigliani model, the
policy-maker's model of income generation in the
period i quarters ahead of the present period, -r ,
under his four-year policy of generating $10 billion
per year of added PDI, can be expressed as

method for forming his expectations, however. Instead of using announced government expenditures
to estimate future income, the consumer simply
makes his own valuation of an added $ 10 billion
per year in PDI for the second through the fourth
years:

LlC

i
b.LlPDI.+10
L b.] + (.74l i - T e_ 1
j=O J
J
j=T+1 J

10
(10 + ~ + (1+p) 2 )
l+p

=.663[L

LlC.1 + LlG.1 =

{I0- e 1
-

= T

+1, ... T+4

::::

+ 5, ... , i

and adds this to income generated in the current
year. By this method. the consumer considers himself wealthier in the first year than he did in the
earlier example, primarily because he is unaware
of the low level of intended government expenditures in years 2 through 4. The result is a relatively high level of consumption, $12.58 billion.
somewhat above the $ I0.58 billion in added income shown in the earlier example in the text.

The consumer's model of income generation i years
in the future is

lie.1

Planned Government Expenditures-Alternative
Strategy

The consumer treats this problem as a dynamic
programming problem with E given.
These two sets of relationships highlight the difference between consumer and policy-maker. The
consumer is choosing present consumption based
upon his estimate of future income, but the policymaker is forecasting consumption from information contained in past levels of personal consumption.
Suppose that instead of announcing his planned
expenditures the policy-maker announced planned
levels of income. While this may seem more devious than announcing planned expenditures, in the
light of the earlier example it may be more informative to the consumer to release income goals,
since the policy-maker's planned expenditures will
not be realized anyway. In this case, the government would simply announce its intention to increase income by $10 billion in each of the next
four years, stating that expenditures would be at
whatever level is required to produce this outcome.
This change in announced intentions would have
no effect upon the policy-maker's procedures for
forecasting future consumption and determining
future expenditure intentions, but it would affect
the consumer's valuation of future government
policy and hence the final outcome both in terms
of income and expenditures. For example, in the
first year the policy-maker would proceed as before, choosing expenditure levels that produce $10
billion in income under the assumption that consumers forecast by means of an adaptive-expectations scheme. The consumer would use a different

To spend
in year

Planned in year
2
3

7.63
4.91
3.58
3.37

2
3
4

.97
3.45
3.55

9.67
3.79

5.10

The table below shows the income resulting
from the policy-maker's two alternative approaches
to providing information to the consumer. The
results are quite similar, but by revealing income
intentions rather than expenditure intentions the
policy-maker gains a marginal increase in the fouryear ratio of income generated to expenditures1.48 with the income policy and 1.27 with the expenditure policy. At the same time. he loses some
control over income generated-$5.48 billion away
from the $40-biIlion target with income policy, and
$4.58 billion away from target with the expenditures policy.
Income-Expenditure Patterns with Given
Information

Expenditure Information Income Information
Year Expenditures Income Expenditures Income
I
7.63
10.68
7.63
12.58
2
3.85
5.52
.97
3.75
3
10.34
12.34
9.67
12.40
4
6.03
6.88
5.10
5.83
Total

27.85

Income-expenditure
ratio

35.42
1.27

23.37

34.56
1.48

APPENDIX II
Consumer forecasts-prices only
A second type of rational-expectations model
may be developed where economic agents forecast
price rather than quantity variables. In this model.
economic decisions might depend upon the divergence of actual prices from price forecasts. Consumers might, for example, adjust planned expenditures in an attempt to hold them constant in real
(price-adjusted) terms, but would treat an unanticipated deviation of price from expectation as a
temporary change in the relative price of present
goods vis-a-vis future goods-perhaps postponing
current consumption outlays when prices are unexpectedly high.
The consumption decision in 1977 would therefore depend upon

prise change in prices. If prices increase above
expectations, the consumer will believe the increase
to be temporary, and therefore will find it advantageous to defer some present consumption to the
future if the substitution effect of high present
prices is dominant.
By the same token, if prices fall below expectations, present consumption will be increased at the
expense of lost future consumption. When consumption is subjected to a shock, however, time is
required for the consumer to readjust his consumption to equilibrium levels. Thus we have
Ct = a 1Ct _1 + aZC t _Z + a 3 [Pt - E _ (P )]
t 1 t

PH'" - E 1D ," (PIa")
which is the deviation of 1977 prices from the level
expected in the previous year. This forecast of
1977 prices might be based upon all available information, such as the expected path of future government expenditures and the impact of these expenditures upon personal disposable income.
In the standard version of the life-cycle consumption model, the level of current consumption is
assumed to depend upon the consumer's valuation
of his future lifetime income. The problem generally is viewed as one of judging the approximate
value of an uncertain quantity, in other words a
forecasting problem. But suppose that the consumer did not have to forecast the future level of
income but instead had a large measure of choice
in the matter, given some "natural" restrictions
such as his current holdings of capital goods and
the value of his labor services. Then he could determine the value of his future DPI at his own
discretion-depending, for example, upon whether
he greatly preferred leisure to added wages. One
of the determinants of his decision to work would
be the price level. If prices were high relative to
expectations, he might decide to increase his present
offering of labor services, hence increasing his Iifecycle income. At the old level of life-cycle income,
the same consumer-laborer might respond to price
increases by reducing present consumption, waiting
for future periods when prices would be closer to
his expectations to "catch up." However, unexpectedly higher prices would increase his life-cycle
income as well, since he offers more labor services
at higher prices. Therefore, the decision to reduce
or increase consumption would depend on the relative magnitude of income and substitution effects.
In this model, then, consumers may be assumed
to forecast prices only, and to consume on the
average at a fixed rate. They will revise their
planned consumption only if presented with a sur-

Et _1 (P t )

E (PtIIt-1) aI' a Z

O,a 3

The second of these two equations states the rule
by which the consumer forecasts future prices.
E t _ 1 (P t ) is the price level that the consumer expects to find in period t from a perspective one
period in the past. This forecast is conditioned on
all the information, I t _1 , which is available to the
consumer in the earlier period. The consumer may
have some knowledge of the policymaker's plans,
and he may have some knowledge of the behavior
of the economy as well. These equations suggest
that if the policy-maker informs the consumer in
advance of his policy intentions, he cannot hope to
influence the consumer's behavior. This is because
the consumer is able to form price forecasts identical to those of the policy-maker when both possess
equal levels of information. Consider, for example, the impact of an increase in government expenditures, G. In this case, the consumer revises
his forecast of future prices:

The policy-maker may estimate the effect on consumption Et _1 (C t ) = ao ct _1 + a c _
1

t 2

+ Et _1 [P t - Et _1 (PtIG)]

butsince

E _1 (P ) = E _ (PtIG)
t
t
t 1

i.e. consumer and policy-maker have the same information

and an announced increase in government expenditures has no effect upon consumption.

Rose McElhattan'::

tied to increases in output per manhour.~ This
paper applies an approach of that type to the
U.S. economy. We consider the impact upon
U.S. aggregate economic activity of an incomes
policy which sets the average increase in wages
equal to the trend rate of growth in labor productivity. The economic impact is estimated
through simulations of the U.S. economy with a
version of the MPS model (Massachusetts Institute of Technology/University of Pennsylvania/Social Science Research Council) over
the 1967-1971 period. The policy analyzed
here differs from the policies actually adopted
in the 1971-73 period because it is concerned
only with the rate of growth of wages. One advantage of such an approach, as we will discuss
later, is that by controlling prices indirectly
through wages, we can avoid direct cumbersome
controls on final prices. We have analyzed the
1967-7 I period because we lacked sufficient
data to analyze the period of the late 1970's,
and because that earlier period produced unemployment and inflation problems that our conventional tools of monetary and fiscal policy did
not-or could not-solve.
The period beginning in 1967 was one of

The growth in output since the beginning of
the recovery in early 1975 has matched that of
previous upturns, but unemployment and inflation remain abnormally high. The twin problems of high unemployment and inflation may
remain with us for several years; forecasts are
remarkably similar on that point, with the jobless rate between 6 and 7 percent and the inflation rate around 5 to 6 percent in 1980.' The
persistence of these twin problems helps explain
why incomes policies are again receiving consideration by the press, economists and policy
makers.
The term "incomes policy" refers to a wide
range of government measures which supplement the traditional instruments of fiscal and
monetary policy and which are designed to improve the tradeoff between unemployment and
inflation. An incomes policy can incorporate
direct government controls to hold down prices
and freeze wages, as well as milder policies such
as "jawboning." It can also include measures
that would tend to promote competition in labor
and products markets in order to keep prices
down, such as more vigorous antitrust action.
In general, incomes policies are de.signed to
bring about a lower level of prices than would
otherwise exist at a given level of unemployment. They are aimed at affecting prices through
the supply side, such as by constraining costs of
production or decreasing monopoly power.
European experiences with incomes policies
in the post-World War II period mostly have
involved a wages guideline, with wage increases

phenomenal increases in unit labor costs and
soaring prices. Arthur Okun has suggested that
we had a "second chance" in mid-1967 to stem
the inflationary climb, which would have been
successful if we had taken advantage of traditional fiscal and monetary policy tools." But
studies of that period, using a variety of monetary- and fiscal-policy mixes, indicate that we
could not have avoided a substantial rise in innation in late 1968 without suffering a high cost

"'Economist, Federal Reserve Bank of San Francisco. The
author wishes to acknowledge the assistance of David
McKinnis and Miriam Ciochon.

several other variables. The paper points out
the effects but ignores the costs of an incomes
policy, such as administrative costs or possible
resource misallocation. In summary, our experiment indicates that a wage-directed incomes
policy will have an ambiguous effect on output
and employment, depending on the assumptions made, and that the labor share of income
will be less than otherwise would have been the
case. The major benefit of the incomes policy
would be a temporarily lower rate of inflation.
How long this would last cannot be determined
in the model. The paper does not consider the
possibility of supplementing the incomes policy
with monetary or fiscal measures, but attempts
to retain historical monetary and fiscal policies.
The next section of this paper describes the
price equation in the MPS model. This is followed by the simulation results of a wage-control policy, and then by a discussion of the
meaning and applicability of the econometric
results.

in terms of lost jobs and output. 4 The present
study is designed to determine what impact an
incomes policy would have had in the 1967-71
period. We do not consider the difficult problem
of policy administration, and during the model
simulations of proposed wage constraints, we
have made a number of assumptions which significantly affect the final results. We ask the
following question:
What is the result of imposing restraints
solely on wages with respect to major economic measures such as unemployment,
prices, real income and income shares?
This article describes an experiment in incomes policy. It is called an experiment because
the results presented cannot be considered a
forecast of what actually would have happened
if such a policy had been implemented. Rather
the paper considers the way in which the incomes policy would be analyzed, through the
use of an economic model to determine the policy's impact on inflation, unemployment and

Inflation and the Price Equation in the MPS
Model

sector variables such as employment and real
output. In the longer run, the model has very
classical properties, with real output being left
unaffected by the change in the money stock,
and with the rate of inflation being determined
entirely within the monetary sector. In the
shorter run, however, an acceleration in the rate
of money growth initially stimulates the demand
for goods and services as well as for the labor
to meet that demand. The additional pressures
in labor and other factor markets affect wages
and other costs of production; in this sense,
changes in business costs are the proximate but
not the fundamental cause of inflation.
The price equation in the MPS model, as in
most large scale econometric models, is a cost
mark-up equation. It has been shown by Nordhaus that pricing behavior for a profit-maximizing firm results in an optimal price (net of
indirect taxes) based on factor costs. These
costs include the prices of capital services,
labor, raw materials and a trend component to

Incomes policies are essentially efforts to reduce the rate of inflation at a given level of unemployment. Policy measures which associate
wage and productivity increases are designed to
control the rise in unit labor costs, the predominant price-raising factor in the short-run in
most large structural economic models.
The price equation used here ignores the role
of money in price determination, which may
appear at odds with monetary explanations of
inflation. If inflation is fundamentally a monetary phenomenon, why is the money supply excluded from the direct determination of prices?
The answer is that the price equation is embedded in a larger model in which prices and
money are indeed tied together in the determination of income, output and employment. The
channels in the MPS model through which
changes in money affect prices and output have
been discussed in detail elsewhere. 5 In brief, a
change in the rate of growth of the money supply will have only transitory effects on real-

Chart 1

M ajor Economic V a ria b le s-----Com parison of H is to ric a l Values
and W age-C ontrol Model Sim ulations
Real G N P
B illio n s o f 1958 $

U n e m p lo y m e n t R ate
P e rce n t

In fla tio n

R ate

(G N P Im p lic it P ric e D e fla to r)
P e rc e n t

capture the advance of productivity through
time. The level of prices can then be determined
by the level of these costs, a term representing
productivity, and a scale factor, which repre
sents the mark-up fraction.0

P= k*W/Q

(1)

where
P = Price deflator for nonfarm domestic
business product
W = Employee compensation per manhour
in nonfarm domestic business
Q = Output per manhour
k = Mark-up factor
W/Q = Unit labor costs

Numerous econometric efforts to find the im
pact of the price of capital services have been
unsuccessful, so that this cost is generally as
sumed to be estimated in the constant term of
the price equation. In addition, the price index
which is estimated by the basic behavioral price
equation is the deflator for nonfarm domestic
business product. Also, it is a value-added con
cept, which means that the cost of raw materials
to the nonfarm business sector does not enter
directly in the determination of the price index.
Raw-materials inputs to the business sector con
sist mainly of farm products and imports, so
that any increase (say) in their prices will raise
prices in the aggregate nonfarm business sector
with a delay, as each price increase is passed on
to final consumers. With these considerations
in mind, we can represent the basic price equa
tion ^s follows:

Price determination in the form of equation
(1) means that if business profit margins (k)
remain constant, then price changes will be
strictly labor-cost determined. In this simplest
representation, a rise in unit labor costs (W/Q)
will be matched by a proportionate rise in
prices. Thus the rate of price inflation is deter
mined solely by the “pass-through” of laborcost increases into prices.
Equation (1) is an oversimplified version of
the MPS pricing equation.7 Four major adjust
ments convert it into a form suitable for shortrun price estimation in the MPS model. First,
the mark-up is assumed to vary with demand
pressure. Second, terms in current productivity
33

and trend productivity are included. Third, the
rates of change of farm and import prices are
added to capture initial adjustment effects.
Lastly, prices are assumed to adjust with a lag
to cost and mark-up changes.
It is assumed that the mark-up fraction (k)
depends upon the level of excess demand. Firms
which possess some short-run monopoly power
might raise their mark-up margins to a high level
during a boom and shade their prices when demand weakens. Demand pressures are represented by the ratio of unfilled orders (OUPD)
to shipments of producers' durable equipment
(EPD), specifically:

P
=

.7099 In( WI) + .07331
(7.23)
(2.83)

.00746
H.13)

01258 In( 31.91 PWf1 + 68.09 PFM )
(~.53)
31.91 PWM~I + 68.09 PFM~I

In(~~~~)~i

.00109 JS2

(~.99)

+ .00011 JS4
(.09)

.00016 JS3
(~.13)

TIME

.11742
(~3. 60)

(2)

1954:1 to 1968:IV

.9993; SE

.0028; DW

1.93; DF

where
P = Price deflator for nonfarm domestic
business product
W = Employee compensation rate in nonfarm domestic business
OUPD = Unfilled orders for producers durabIes
EPD Expenditures on producers durables
XBNF = Nonfarm domestic business product
and produce of households
LMHT Manhours in nonfarm domestic business sector, including proprietors
PWM = Raw materials prices, imports
PFM Raw materials prices, farm
JS2 Seasonal dummy variable for the second quarter
JS3 = Seasonal dummy variable for the
third quarter
JS4 Seasonal dummy variable for the
fourth quarter
Time = Time with 1947.1 = 1,1968.4= 88

k=bJ+b"(OUPD/EPD)t-b)(OUPD/EPD)t~l

The negative coefficient (bJ implies a rate-ofchange variable which is intended to capture the
effect on the mark-up of expectations of demand
change.'
Secondly, it is assumed that firms base their
estimate of the rate of technical change both on
long trends and on more recent movements of
average labor productivity. The value (Q) is
replaced with two terms: a term representing
current productivity (measured as an eightquarter average to remove some of its cyclical
movement) and a time variable to capture longrun trend movement. Thirdly, since the price
index is a value added-deflator for the nonfarm
domestic-business sector, the index will initially
decline when farm or import prices rise if these
costs are not immediately passed on to final
consumers. In order to capture this temporary
effect, a fixed-weight average of farm and imported-materials prices was added to the equation. Its coefficient should be negative. Finally,
it is assumed that cost and mark-up factors influence prices with a distributed lag through
time. To incorporate lagged adjustment, the
price index was included with a one-period delay on the right-hand side of the estimated
equation.
Equation (1) was estimated with these four
modifications, subject to the constraint that the
long-run elasticity of prices to wages be unity.
The result was as follows (numbers in parenthesis are t values) :

The most notable features of this equation
are the following:
( 1) The lag structure of wages means that
any wage change is almost entirely passed
through to prices in a little over two years. A
one-percent increase in wages will result in .75
percentage-point increase in the rate of inflation within one year and about .95 percentagepoint increase by the end of two years.')
(2) The trend rate of growth in productivity
results in a steady decline in prices each year of

control of wage-rate increases can successfully
control domestic nonfarm prices, without di
rectly controlling the latter. Incomes policies
which fundamentally rely on wage restraints
have an important advantage in that they obvi
ate the need to create a cumbersome adminis
trative apparatus to control final prices. Inter
esting examples of two such policies are the
tax-based incomes policy advocated by Henry
C. Wallich and Sidney Weintraub,11 and an in
comes policy recently updated by Vijaya G.
Duggal and Lawrence R. Klein.1The importance of the term representing de
mand pressures illustrates the potential signifi
cance of monetary- and fiscal-policy effects
upon prices. An increase in aggregate demand
which is initiated, say, by expansive monetary
and fiscal polices can act to increase demand
pressures and thereby prices. An incomes pol
icy which holds down unit labor costs will not
be able under such circumstances to stop the
inflationary rise due to demand-pull pressures,
and will not survive when monetary and fiscal
stimulus becomes excessive.

about .60 percentage points, while a one-per
cent increase in the sum of current and lagged
estimates of productivity leads to an additional
decline of about .36 percentage points.
(3) Prices respond positively to demand
pressures, even when unit labor costs are held
constant. This is demonstrated by the esti
mated coefficients of unfilled orders to ship
ments. On average, during post-Korean War
cycles, the demand effect is estimated to have
raised the rate of inflation between trough and
peak quarters about 2.5 percentage points,
assuming no changes in the other price deter
minants in equation (2). This result is con
sistent with Gordon’s recent work on the im
pact of excess demand on prices, which suggests
that on the average, demand pressures have
added about 2.8 percentage points (trough to
peak) to the inflation rate during post-World
War II business cycles.10
The pricing behavior estimated by equation
(2) has major significance for stabilization pol
icy. It suggests that an incomes policy which
can restrain the rise in unit labor costs through

C h a rt 2

Returns to Labor and C a p ita l-----Comparison of H isto rica l Values
w ith W a g e -C o n tro l Model Sim ulations
Real D isposab le In c o m e /

Real Labor Income

Per M em ber o f the La bor Force
$ T h o u sa nd s

Real C o rp o ra te P ro fits and IV A
$ B illio n s

$ B illio n s

C ontrolling Unit Labor Costs Through an
Incomes Policy— A Simulation Experiment

inal wage rates, which is equal to the long-term
growth rate in output per manhour. The his
torical and assumed wage rates are shown be
low.

From this discussion of the direct determi
nants of prices, we can conclude that a change
in wages equal to the change in productivity
will result in unchanged prices after a two-year
adjustment period, assuming no change in ex
cess demand. If prices are held down by an
incomes policy, a given amount of aggregate
demand should, at least in the short run, result
in greater real output and employment, since
part of the demand does not become dissipated
in higher prices. On the other hand, because
of adjustment lags between wages and prices,
holding down wages may initially hold down
real income and could adversely affect employ
ment. To determine the impact of wage con
trols upon economic activity, we turn to simula
tion experiments with the MPS model over the
period from 1967.2 to 1971.2.
These experiments may be identified as Wage
Control-Alternative 1, and Wage Control-Al
ternative 2. One common element is present
in both experiments—the assumption of a con
stant 3-percent annual rate of increase in nom-

Hourly Wage Rates in Nonfarm Private
Domestic Business
Annual Rate of Increase, 1967-1 - 1971.2

Assumed Value in
Historical Value Wage Control Alternatives

1967.1
.2
.3
.4
1968.1
.2
.3
.4
1969.1
.2
.3
.4
1970.1
.2
.3
.4
1971.1
.2

C h a rt 3

Labor Share of National Income
Percent

1967

[

1969

[

2.5
6.5
6.7
4.8
10.7
6.0
7.5
8.2
4.5
7.5
6.2
8.7
6.4
6.1
10.2
2.6
8.0
7.5

2.5
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0

The difference between the actual historical
value and a wage-simulation result is a mea
sure of the impact of controlling the increase in
wages.13 But, if this difference is to reflect
accurately the impact of wage controls, it is
necessary that the structure of the MPS model
allow all significant reactions to the wage
change to take place. In several ways, the MPS
model is not structured to capture the economic
response to the wage changes we have made.
We have attempted to adjust the model for
some of its structural shortcomings, and we
acknowledge that the adjustments, which are
our “best guesses,” are somewhat arbitrary. It
is for this reason that we present two wagecontrol alternatives and detail the judgmental
changes imposed upon the results.

1971

In Wage-Control-Alternative I, we not only
assume a constant 3-percent average wage increase, but also adjust the cost of capital for
producers durables to reflect the expected drop
in inflation rates due to this 3-percent wage
constraint.

provement over historical values for real output, employment and prices (Table 1). Real
output increases steadily and by 1970, GNP is
$19.7 billion greater than historical estimates.
The unemployment rate declines in response to
the greater growth in real output and remains
below 4.0 percent over the 1967-70 period; by
1970 it is 3.7 percent, compared with the 5.0percent historical value. The rate of inflation
responds to the drop in wages, falling in 1967
to 2.6 percent from its historical rate of 3.2
percent, and remaining below 2.0 percent over
the following three years. By 1970, the inflation rate is 1.7 percent, compared with the historical rate of 5.5 percent.

The cost of capital depends in part upon real
interest costs, which are determined by the difference between a nominal long-term interest
rate and the expected rate of inflation. The
expected inflation rate is estimated by a distributed lag on past inflation rates. In the wagecontrol simulation, the mechanical application
of the distributed lag on past rates of inflation
results in a very low expected inflation rate,
and hence a very high real rate of interest and
cost of capital. The high cost of capital worked
in our initial simulations to reduce investment
substantially in business durable equipment, so
we then adjusted the cost-of-capital term so
that it would not go above the rates experienced in the first half of the 1960's, when prices
and interest rates were similar to their simulated valuesY

Real disposable income per member of the
labor force declines relative to historical values
for the first two years of the simulation, but
then increases substantially, reaching $6,231 in
1970, or $151 greater than its historical value.
The changes in this measure mostly reflect the
time required in the model for prices to adjust
fully to changes in wages. Because of this delayed reaction, the real purchasing power of
wage income will initially fall while profits,
calculated residually, will show improvement.
We may therefore expect some drop in labor
income (relative to its historical value), as
prices adjust slowly to the lower rate of wage
growth. In fact, labor income does not increase
relative to historical values until 1970, the last
year of the simulation. The earlier increase in
real disposable income represents an increase
in the purchasing power of non-labor components: property and proprietor income, and
transfer payments. Alternative I thus suggests
that labor income may not show any marked
increase (relative to historical values) when an
incomes policy is initially instituted. 'G In Table
2, we show the effect of Alternative I upon real
GNP and its components, along with the differences between these results and historical
values. The decline shown here in personal
consumption is related to the drop in real disposable income in the initial periods of the
simulation. The relative decline in business
fixed investment reflects the higher real interest

Business expectations of future prices are an
important determinant of investment plans and
expenditures. Undoubtedly, an incomes policy
can influence these expectations to a major extent. Our assumption regarding price expectations, and hence the cost of capital, implies that
the business community believes the incomes
policy will be successful in holding down unit
labor costs through its constraints on wage-rate
increases. If the business community believes
otherwise, any number of alternative possibilities could emerge. For example, if business
felt that controls would lead to bottlenecks in
certain raw-materials areas, a sizable increase
in demand could take place, creating price
pressures which otherwise would not exist.
Our price expectations assumption also implies that market participants build their inflation expectations on more information about
the effects of wage controls on prices than simply extrapolating past price changes into the
future. '5
Wage Control-Alternative 1 results in an im37

TABLE 1
Historical Values and Wage Control Alternatives
for Selected Economic Variables, 1967·70
Wage Control
Alternative 1

Historical
Value

Change from
Historical Value

Wage Control
Alternative 2

Change from
Historical Value

(1) Real GNP (Billions of 1958 dollars)

1967
1968
1969
1970

$675.2
706.6
725.6
722.3

1967
1968
1969
1970

3.8
3.6
3.5
5.0

1967
1968
1969
1970

3.2
4.0
4.8
5.5

1967
1968
1969
1970

$5,849
5,983
5,984
6,080

1967
1968
1969
1970

$404.2
428.0
449.3
456.0

1967
1968
1969
1970

$ 68.1
70.0
63.3
52.2

$675.3
708.1
728.2
742.0

.1
1.5
2.6
19.7

$675.1
704.0
715.5
718.7

-.1
-2.6
-10.1
-3.6

(2) Unemployment Rate (Percent)

3.8
3.4
3.2
3.7

.0
-.2
-.3
-1.3

3.8
3.7
4.1
5.4

.0
.1
.6
.4

2.5
1.0
1.3
1.8

-.6
-3.0
-3.5
-3.7

(3) Inflation Rate (GNP Implicit Deflator)

2.6
1.0
1.2
1.7

-.6
-3.0
-3.6
-3.8

(4) Real Disposable Income Per Member ofthe Labor Force

$5,828
5,962
6,043
6,231

$ -21
-21
59
151

$5,824
5,937
5,981
6,145

$ -25
-46
-3
65

$401.6
420.1
440.1
449.1

$ -2.6
-7.9
-9.2
-6.9

(5) Labor's Total Real Income ($ Billions)*

$401.6
422.0
446.2
461.0

$-2.6
-6.0
-3.1
5.0

(6) Real Corporate Profits and I.V.A. ($ Billions)**

$ 70.4
77.6
70.5
69.4

$ 2.3
7.6
7.2
17.2

$ 70.3
75.6
64.7
59.6

2.2
5.6
1.4
7.4

(7) Relative Income Shares of Corporate Profits (Percent)t

1967
1968
1969
1970

16.8%
16.3
14.1
11.4

1967
1968
1969
1970

71.5
72.4
73.9
75.5

17.5%
18.4
15.8
15.0

.7%
2.1
1.7
3.6

17.5
18.0
14.7
13.3

.7
1.7
1.6
1.9

71.1
71.3
73.0
73.8

-0.4
-1.1
-0.9
-1.7

(8) Labor's Share of National Income (Percent)

71.1
71.1
72.6
73.2

-0.4
-1.3
-1.3
-2.3

'Employee compensation deflated by consumer price index.
"Corporate profits and inventory valuation adjustment deflated by consumer price index.
tCorporate profits divided by employee compensation.

the model, are adjusted to change only in response to changes in real income.

rate in the wage-control run as compared to its
historical estimate. Nominal interest rates in
the wage-control simulation do not decline as
rapidly as prices, because price changes take
some time-through adjustment delays in the
model-to alter other economic and financial
variables.
A substantial amount of the change in real
GNP in this wage-control simulation is due to
increases in real net exports and government
expenditures. These factors may partly reflect
some shortcomings in the basic model, for
which adjustments are made in our second
simulation.
Wage Control-Alternative 2 includes several
of the same assumptions underlying the first
alternative. We assume as before a constant
3-percent annual increase in nominal wages, and
we continue the same adjustment to the cost of
capital for producers' durable goods. But we
also make two further adjustments. First, certain fiscal variables which are exogenous and
fixed in current dollars are changed to reflect
the lower prices generated by the incomes policy. Second, real exports, which are exogenous
and thus not determined by the model, are kept
at their historical values, while real imports,
which are estimated by behavioral equations in

Both Federal grants-in-aid to state-and-Iocal
governments and Federal transfers to persons
(other than unemployment-insurance benefits)
are exogenous variables in the model. The
amount of these expenditures is, in some way,
affected by current prices, and allocations could
decline somewhat in the wage-control situation
as a result of the significantly lower prices.
Consequently, we adjusted the amounts of
these two variables so that real expenditures
equaled their historical real magnitudes. In
Alternative 2, our adjustment implies that federal fiscal policy is determined in real rather
than current dollar magnitudes, which is the
reverse of the assumption under Alternative I.
With Federal grants reduced, Alternative 2
shows smaller values than Alternative 1 for
state - and - local government expenditures,
amounting to about $1 billion in 1969 and $2
billion in 1970 (Tables 2 and 3). Restricting
Federal transfers to their historical real-dollar
magnitudes results in relatively smaller real disposable income and hence relatively smaller
consumption expenditures. In Alternative 2,
real transfers to persons (other than unemployment insurance) are reduced by $1. 3 billion in
1968, $2.9 billion in 1969 and $5.2 billion in
1970.
Export and import prices, as well as export
volume, are exogenous variables in the model,
while import volume is assumed to respond to
relative prices and real income. Consequently,
the decline in U.S. prices associated with an
incomes policy should lead to substantial increases in net exports, since current-dollar exports remain at their historical levels and imports decline dramatically as U.S. prices fall
relative to fixed foreign prices. But because of
uncertainty regarding the world response to a
decline in U.S. prices, we assumed that the incomes policy would be relatively neutral in its
impact upon net exports, with no price effects
on the quantities of internationally traded
goods. Thus, there would be no gain in the
U.S. foreign trade balance from the price ad-

Chart 4

Profits Relative Income Share*
Percent

13
*Corporate profits

divided by employee compensation

o'(

1967

1969

1971

TABLE 2
Real GNP and Components-Historical Values and Values
Under Wage Control-Alternative 1
(Billions of 1958 dollars)

1966

1967

1968

1969

1970

Historical Values

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
Residential Structures . . . . . . . . . . ..
Inventory Change
Net Exports
Exports
Imports
Government Purchases
Federal
State & Local

.
.
.
. ..
.
.
.
.
.
.
.

658.1
418.1
74.1
21.3
13.9
4.2
40.2
36.0
126.5
65.4
61.1

675.2
430.1
73.2
20.4
7.7
3.6
42.1
38.5
140.2
74.6
65.6

706.6
452.8
75.6
23.2
6.5
0.9
45.6
44.7
147.7
78.1
69.6

725.6
469.1
80.1
23.7
6.7
0.2
48.4
48.3
145.8
73.4
72.4

722.3
477.4
77.3
22.2
3.9
2.2
52.2
50.0
139.3
64.4
74.9

Wage Control-Alternative I
(Simulation period 1967-1970)

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
Residential Structures
Inventory Change
Net Exports .,
Exports
Imports
Government Purchases
Federal
State & Local

.
.
.
.
.
.
.
.
.
.
.

658.1
418.1
74.1

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
Residential Structures
Inventory Change
Net Exports .,
Exports
Imports
Government Purchases
Federal
State & Local

.
.
.
.
.
.
.
.
.
.
.

0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0

21.3

13.9
4.2
40.2
36.0
126.5
65.4
61.1

675.3
429.7
73.1
20.4
7.6
4.1
42.1
38.0
140.3
74.6
65.7

708.1
452.5
74.0
23.6
5.3
4.1
45.6
41.5
148.7
78.1
70.6

728.2
470.2
74.5
24.5
4.1
6.5
48.4
41.9
148.4
73.4
74.9

742.0
485.7
71.0
24.9
5.9
10.6
52.2
41.6
144.0
64.4
79.6

Change From Historical Values

0.1
-0.3
-0.1
0.1
-0.2
0.5
0.0
-0.5
0.1
0.0
0.1

1.5
-0.4
-1.6
0.4
-1.1
3.2
0.0
-3.2
1.0
0.0
1.0

2.6
1.1
-5.6
0.8
-2.5
6.3
0.0
-6.3
2.5
0.0
2.5

19.8
8.3
-6.2
2.7
2.0
8.3
0.0
-8.3
4.7
0.0
4.7

on the other hand, remains substantially lower
than historical rates; for example, it is 1.8 percent in 1970 compared with the actual 5.5percent rate.
These results suggest that a wage-control
policy may be successful in controlling the rate
of inflation. But while reducing inflation, the
growth in real incomes may be insufficient to
maintain employment at historical levels during
much of the period of controls.

vantage which could occur due to relatively
lower domestic prices and the time delay involved in exchange-rate adjustments. Because
of this assumption, net exports in Alternative 2
are close to historical values.
As a result of these adjustments, real GNP
generally remains below historical values
throughout the simulation, and the unemployment rate is higher than historical values from
1968 until 1971. (Table 1.) The inflation rate,

Evaluation of the Incomes Policy Simulations

Our analysis has focused upon the consequences of a change from historical experience
in average hourly wages. The findings are dependent upon the behavioral structure of the
MPS model as well as the assumptions we have
imposed along the way. If the model results
are to have any applicability, they should be
carefully interpreted within that context. To
round-out our analysis, we should also consider
the consequences of alternative assumptions, as
well as other factors which could modify our
conclusions.

sizable increase in net exports and thus a large
stimulus to domestic activity. In the final analysis, Wage Control-Alternative 1 resulted in
lower inflation than we actually experienced
over the 1967-70 period, as well as a lower
unemployment rate and a higher level of income.
In Alternative 2, by contrast, we assumed
that Federal policy with regard to grants and
transfers was determined in real terms rather
than nominal. We adjusted the model so that
these expenditures equalled their historical
real-dollar magnitudes, and, in doing so removed a good deal of economic stimulus. We
also assumed that the incomes policy would be
relatively neutral in its impact upon net exports,
so that internationally-traded goods would not
be affected by price changes brought on by the
wage policy. This assumption kept net exports
in Alternative 2 close to historical values. As a
result, prices were kept lower, but unemployment rose and real income and output declined
relative to their historical values.

Sensitivity to Alternative Assumptions

The model results suggest that a program
which controls wage-rate increases can, for a
time, control the rate of increase in final prices
without direct price intervention. The impact
of wage controls upon real output and employment, however, remains uncertain. Under
equally feasible alternative assumptions, the
impact of wage controls upon output and employment can differ considerably.
In Alternative 1, we assumed that Federal
grants-in-aid and transfer payments were determined in terms of nominal dollars. Under this
assumption, we obtained a sizable boost in
such expenditures in terms of the real goods
and services they commanded, because prices
were considerably lower than in the real-life
situation. Again, we assumed that a price advantage would occur in the U.S. relative to
foreign-priced products, and this resulted in a

Linkage Between Money and Price Changes

In the MPS model of the U.S. economy, the
percentage change in prices over the long-run
tends to equal the percentage change in the
money supply. Price reactions to changes in
money begin with changes in interest rates, and
these lead to changes in the cost of capital and
in the demand for real output. Changes in
demand for final products lead to changes in
41

TABLE 3
Real GNP and Components-Historical Values and Values
Under Wage Control-Alternative 2
(Billions of 1958 dollars)

1966

1967

1968

1969

1970

Historical Values

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
. . . . .. .
Residential Structures. . . .
Inventory Change . . . . . . . . . . . . . . . . . . .. .
Net Exports
Exports
Imports. . . . . . . . . . . . . . . . . .. .
Government Purchases
Federal
State & Local

.
.
.
.
.
.
.
.
.
.
.

658.1
418.1
74.1
21.3
13.9
4.2
40.2
36.0
126.5
65.4
61.1

675.2
430.1
73.2
20.4
7.7
3.6
42.1
38.5
140.2
74.6
65.6

706.6
452.8
75.6
23.2
6.5
0.9
45.6
44.7
147.7
78.1
69.6

725.6
469.1
80.1
23.7
6.7
0.2
48.4
48.2
145.8
73.4
72.4

722.3
477.4
77.3
22.2
3.9
2.3
52.2
49.9
139.3
54.4
74.9

Wage Control-Alternative II

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
Residential Structures
Inventory Change
Net Exports
Exports
Imports
Government Purchases
Federal
State & Local

.
.
.
.
.
.
.
.
.
.
.

658.1
418.1
74.1
2\.3
13.9
4.2
40.2
36.0
126.5
65.4
61.1

675.1
429.6
73.1
20.4
7.6
4.1
42.1
38.0
140.3
74.6
65.6

704.0
451.1
73.7
23.6
4.9
2.5
45.6
43.1
148.3
78.1
70.2

715.5
465.6
72.9
24.3
2.9
2.6
48.4
45.9
147.3
73.4
73.8

718.7
476.5
67.5
24.9
4.1
3.9
52.2
48.3
141.9
64,4

77.5

Change From Historical Values

Gross National Product
Personal Consumption Expenditures
Business Fixed Investment
Residential Structures
Inventory Change
Net Exports . . . . . . . . . . . .
Exports
Imports
Government Purchases
Federal
State & Local

.
.
.
.
.
.
.
.
.
.
.

0.0
0.0

0.0
0.0
0.0
0.0
0.0
0.0

0.0

-0.1
-0.4
-0.1
0.1
-0.2
0.5
0.0
-0.5
0.1

-2.6
-1.7
-1.9
0.4
-1.6
1.6
0.0
-1.6
0.6

0.0

0.1

0.6

-10.1
-3.5
-7.2
0.6
-3.8
2.4
0.0
-2.4
1.4
0.0
1.4

-3.6
-0.9
-9.7
2.7
0.1
\.6

0.0
-1.6
2.6

0.0
2.6

labor demand, and these bring about changes
in wages which in turn become the major determinant of prices. The imposition of controls
on wage increases thwarts the major channel by
which monetary changes lead to price changes.
The blockage of this channel, however, will not
eliminate the pressures upon prices precipitated
by the initial change in the money supply. If
an incomes policy is to have any long-run success in keeping prices down, long-run monetary
growth rates must be consistent with the price
objectives of the incomes policy. In our simulations, the historical M, money supply grew at
a 5.5-percent average annual rate over the
1967-70 period. This rate is too high to support for very long the zero or one-percent rate
of inflation implied by our wage-growth assumption.

movement in the structure of wages to guide
resources efficiently. Although this intention is
clear in principle, its implementation presents
a formidable task and constitutes one of the
greatest problems facing incomes policy.
The results of our model simulations depend
upon the policy-maker's success in dealing with
problems of implementation, which have been
recognized by supporters of wage guidelines,
such as Sidney Weintraub and Robert Solow:
A proper policy would maintain the average wage movement within the average
improvement norm. Simultaneously it
would seek to achieve a strong measure of
equity between wage earners of similar
skills. It must also aim to direct labor
into industries, occupation, and geographical areas of most urgent need. . .. Policy
implementation is difficult though the general principles are less recondite.'T

Policy Implementation and Resource
Allocation

We have not touched upon the difficult problem of implementing the intended incomes policy, because our econometric model simply
assumes that the intended policy is successful.
The important point to bear in mind is that the
average wage is not a policy variable which can
be manipulated by policy-makers, being unlike
tax rates, federal expenditures, or discount
rates in this respect. To achieve a desired
growth in the average wage rate, policy-makers
must exert some control over individual sector
or industry wages and, unavoidably, their actions in doing so will change the relative price
structure which would otherwise exist. Such
interference in the marketplace can ~istort the
operation of the pricing system which, even in
markets characterized by large power groups,
has proved to be a relatively efficient means of
allocating resources in a complex society.

The guideposts are intended to have an
effect on the general level of money wages
and prices, not on relative wages and relative prices. Most of the things we expect
free markets to accomplish are real things,
more or less independent of the price
level. Ideally, the guideposts should permit markets to allocate resources freely,
insuring only that the price level does not
drift up in the process. . . . In practice,
the guideposts will operate unevenly; relative prices and resource allocation may
thus be affected. . . . One can hope that
the uneven effects of guideposts will be of
second order. . .. This inevitable unevenness in operation strikes me as the main
weakness in the guideposts. '"

Implementation of an incomes policy apparently will have to be flexible enough to consider
individual cases, in order to allow markets to
allocate resources freely. Policy-makers in this
situation try to insure that the average wage
level does not drift upward, while allowing

Our experience with incomes policies sug~
gests that it will be impossible to maintain any
form of wage and/or price programs unless the
policy is generally regarded as equitable.'" As
we indicated, the real purchasing power of
labor income is likely to fall relative to profits

Income Distribution: Profits and Wages

occurred between 1967 and 1970. But in doing
so, we have implicitly assumed that the resulting income distribution was acceptable to both
labor and profit recipients, and to the extent
that it is not true, the policy has little chance of
success.

when a wage-restraint policy is initiated. This
unequal burden may stand in the way of successful policy implementation. cO
Again, we have assumed restraints only on
wages, largely because it seemed realistic to
accept the faUing-profits trend which actually

Summary and Conclusions

We have tried to answer the question: What
impact would an incomes policy have upon
U.S. economic activity? Our simulated incomes
policy involved restricting the growth of the
average wage rate in domestic nonfarm business to 3 percent a year, equal to the trend rate
of growth in output per manhour in that sector
since the late 1930's. We analyzed the impact
of this policy on the U.S. economy from
1967.2 - 1971.2, employing simulation techniques in a version of the MPS model.
We presented results of the proposed wagecontrol program under two alternative sets of
assumptions. The results generally suggest that
a program which controls wage-rate increases
can for a time control the rate of increase in
domestic nonfarm prices without any direct
intervention in prices. However, a wage-control
program can have ambiguous effects on output
and employment. The model results are sensitive to assumptions regarding the foreign sector's reaction to lower U.S. inflation, and re-

garding fiscal policy's impact on allocating funds
(in either real or nominal terms).
Both alternative wage-control simulations
suggest that labor's real income may decline
relative to its historical value for some time
after the institution of an incomes policy which
restricts wage growth. In addition, labor's share
of total income is likely to fall relative to the
share of total income going to corporate profits.
Under both simulations, however, business
plant-equipment expenditures were somewhat
less than their historical values, because nominal interest rates did not fall as rapidly as final
prices while the real cost of capital remained
higher than its historical cost. Finally, we
should emphasize that we maintained the his··
torical money growth and federal tax rates in
our simulations, so that the results would reflect only the impact of keeping wage growth
within the limits set by the long-run productivity trend.

FOOTNOTES

6. Nordhaus, William D., "Recent Developments in Price

1. For example of forecasts, see Sustaining a Balanced
Expansion (U.S. Congress: Congressional Budget Office,
Washington, D.C., August 3, 1976), and the article, "UCLA
Gives Forecast on Ford or Carter Economy," los Angeles
Times, September 24, 1976, p. 111-13.
2. Ulman, Lloyd, and Flanagan, Robert J. Wage Restraint:
A Study of Incomes Policies in Western Europe. Berkeley:
University of California Press, 1971.
3. Okun, Arthur M. The Political Economy of Prosperity.
New York: Norton, 1970.
4. Rasche, Robert H. "Simulations of Stabilization Policies
for 1966-1970," Journal of Money, Credit and Banking,
February 1973, pp. 1-25.
5. De leeuw, Frank and Gramlich, Edward M. "The Channels of Monetary Policy," Federal Reserve Bulletin, June
1969, pp. 472-491, and Ando, Albert, "Some Aspects of
Stabilization Policies, the Monetarist Controversy, and the
MPS Model," International Economic Review, October
1974, pp. 541-571.

Dynamics," The Econometrics of Price Determination,
Conference sponsored by the Board of Governors of the
Federal Reserve System and Social Science Research Council, October 30-31, 1970, Washington, D.C. pp. 16-49.
7. For a detailed description of the price equation in the
MPS model see George de Menil and Jared J. Enzler,
"Prices and Wages in the FR-MIT-PENN Econometric
Model," The Econometrics of Price Determination, op. cit.,
pp. 277-308.
8. Consider a simplified adaptive expectations model in
which expectations of a variable (x e ) are updated for each
period by a fraction (a) of the discrepancy between the
current observed value of the variable (xd and the value
that had been accepted (xV
(1)

. 1

(1)

Assume that the expected value of the variable, x~ , is
some function of known variables: for simplicity, assume
that

The difference between the last two simulations is the
measured effect of the change in the last simulation.
14. Simulation results which leave ca pita I costs unadjusted
are shown below. The table provides the change from historical values for the various components of GNP, and
may be compared with the results shown in Tables 2 and 3
in the text. For the following simulation, the only change
from history which we imposed upon the model was to set
the rate of growth of wages at 3 percent annually. The
inflation results were similar to Alternatives 1 and 2, and
the unemployment results were 3.8% (1967); 3.5%
(1968); 3.5% (1969); 4.3% (1970).

(2)
Then substitute (2) in (1),

(3)
9. We may simplify and rewrite equation (2) in order to
emphasize the lag patterns associated with the key pricedeterminant variables.

Real GNP-Change from Historical Values
(Billions of 1958 $)

1n Pto. 2901 1nW + .
.0014 TIME - .0075

(3)

GNP
Pers. Cons. Exp.
Business Fixed Inv.
Resid. Structures
Inventory Change
Net exports
Exports
Imports
Gov't. purchases
Federal
State & loca I

+ .7099 In p t -!

The presence of the lagged dependent variable (P
)
means that we may replace that term by its equivalent,
i.e., equation (3) lagged one period. This substitution will
produce a variable in prices lagged two periods (P
)
which may also be replaced with equation (3) lagged 2
periods, and so on. In this way, the distributed lag for
each independent variable in the equation may be determined.
The coefficients associated with the lagged values of the
natural logarithm of wages are as follows. let InW == w
.290h't
.2509w_

1967
0.1
-0.3
-01
0.1
-0.2
0.5
0.0
-0.5
0.1
0.0
0.1

1968
0.7
-0.4
-2.0
0.4
-1.6

3.3
0.0

-3.3
1.0
0.0
1.0

1969
-2.3
0.6

1970
9.4
7.0

-8.5

-13.3

0.8
-4.5
6.7
0.0
-6.7
2.6
0.0
2.6

2.8
-1.1
9.1
0.0
-9.1
4.8
0.0
4.8

15. See article in this Review by Kurt Dew, elaborating
upon the formation of expectations.
16, It should be noted that this reduction in labor's income
share is the result of the incomes policy assumption as
well as the particular mix of monetary and fiscal policy
that existed in the historical period. If policy-makers do
not desire such a shift away from labor income. they could
try to redistribute income toward labor through a tax or
transfer program.
17. Weintraub, Sidney, Keynes and the Monetarists. New
Brunswick. N.J.: Rutgers University Press. 1973.
18. Solow, Robert M.• "The Case Against the Guideposts,"
Guidelines (Schultz, George P. and Aliber. Robert Z., editors). Chicago: University of Chicago Press, 1966.
19. On this point, see Anne Romanis Braun. "What is Incomes Policy and What Can it Achieve?" Finance and De·
velopment, April, 1975.
20. For this reason. a policy which restrains wages may
be accompanied by some rules to restrain the growth in
profits. A "Fair Shares" incomes policy has been suggested by Duggal and Klein (footnote 12). Under their
proposal, nominal wages and profits both would increase
at the same rate. which is equal to a trend rate of growth
in productivity. This policy has the effect of stabilizing
labor and profit shares in the distribution of national income. In particular, the extra funds collected from corporations would be plowed back (through federal programs)
into the income stream, where they could generate additional income and jobs. Thus, Federal programs have the
ability to counter the initial adverse effects on real wages
created by incomes policies.

10. Gordon, Robert J., "The Impact of Aggregate Demand
on Prices." Brookings Papers on Economic Activity, No.3,
1975.
11. Wallich, Henry C. and Weintraub. Sidney, "A TaxBased Incomes Policy," Journal of Economic Issues, June
1971, pp. 1-19.
12. Duggal. Vijaya G. and Klein, lawrence R., "An Approach to Disinflation," Wharton Quarterly, Winter 1974.
13. The simulation programs used with the MPS model
enable us to generate historical values. make any changes
from actual values we wish, and then compare the two
simulation results to obtain an estimate of the impact of
the change. The program first simulates each of the model
equations separately. using historical values as independent variables, and records the errors for each equation.
Next. the equations are simulated simultaneously in the
full model, using values of the variables generated by the
model and adding the error made by that equation which
was previously calculated. Each equation, simulated in
this way. will reproduce the actual values except for some
rounding errors. In the next simulation a change is made
to some variable. such as wages, and the single equation
errors previously calculated are added to each equation.

Full text of Economic Review (Federal Reserve Bank of San Francisco) : Fall 1976

FRASER