Anda di halaman 1dari 10

American Economic Association

Some International Evidence on Output-Inflation Tradeoffs


Author(s): Robert E. Lucas, Jr.
Source: The American Economic Review, Vol. 63, No. 3 (Jun., 1973), pp. 326-334
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/1914364
Accessed: 06/01/2010 12:07

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=aea.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.

http://www.jstor.org
Some International Evidence on
Output-InflationTradeofs
By ROBERT E. LUCAS, JR.*

This paper reports the results of an presumption is that inferences on these


empirical study of real output-inflation relevant, unobserved prices are made
tradeoffs, based on annual time-series from optimally (or "rationally") in light of the
eighteen countries over the years 1951-67. stochastic character of the economy.
These data are examined from the point As I have argued elsewhere (1972),
of view of the hypothesis that average theories developed along these lines will
real output levels are invariant under not place testable restrictions on the co-
changes in the time pattern of the rate of efficients of estimated Phillips curves or
inflation, or that there exists a "natural other single equation expressions of the
rate" of real output. That is, we are con- tradeoff. They will not, for example, imply
cerned with the questions (i) does the that money wage changes are linked to
natural rate theory lead to expressions of price level changes with a unit coefficient,
the output-inflation relationship which or that {"long-run"'(in the usual distrib-
perform satisfactorily in an econometric uted lag sense) Phillips curves must be
sense for all, or most, of the countries in vertical. They will (as we shall see below)
the sample, (ii) what testable restrictions link supply parameters to parameters
does the theory impose on this relation- governing the stochastic nature of demand
ship, and (iii) are these restrictions con- shifts. The fact that the implications of the
sistent with recent experience? natural rate theory come in this form sug-
Since the term "'natural rate theory" gests an attempt to test it using a sample,
refers to varied aggregation of models and such as the one employed in this study, in
verbal developments,' it may be helpful which a wide variety of aggregate demand
to sketch the key elements of the particular behavior is exhibited.
version used in this paper. The first In the following section, a simple ag-
essential presumption is that nominal out- gregative model will be constructed using
put is determined on the aggregate demand the elements sketched above. Results
side of the economy, with the division based on this model are reported in Sec-
into real output and the price level largely tion II, followed by a discussion and con-
dependent on the behavior of suppliers of clusions.
labor and goods. The second is that the
1. An Economic Model
partial "rigidities" which dominate short-
run supply behavior result from suppliers' The general structure of the model de-
lack of information on some of the prices veloped in this section may be described
relevant to their decisions. The third very simply. First, the aggregate price-
quantity observations are viewed as inter-
*
Graduate School of Industrial Administration, Car- section points of an aggregate demand and
negie-Mellon University. an aggregate supply schedule. The former
1 The most useful, general statements are those of .
Milton Friedman (1968) and Edmund Phelps. Specific
is drawn up under the assumption of a
illustrative examples are provided by Donald Gordon cleared money market and represents the
and Allan Hynes and Lucas (April 1972). output-price level relationship implicit in
326
VOL. 63 NO. 3 LUCAS: OUTPUT-INFLATION TRADEOFF 327

the standard IS-LM diagram. It is viewed is distributed unevenly over markets,


as being shifted by the usual set of leading to relative as well as general price
demand-shift variables: monetary and movements. As a consequence, the situa-
fiscal policies and variation in export de- tion as perceived by individual suppliers
mands. The supply schedule is drawn un- will be quite different from the aggregate
der the assumption of a cleared labor situation as seen by an outside observer.
market; its slope therefore reflects labor Accordingly,we shall attempt to keep these
and product market "rigidities." two points of view separate, turning first
The structure of this model, which is to the situation faced by individual sup-
essentially that suggested in Lucas and pliers.
Leonard Rapping (1969), will be greatly Quantity supplied in each market will be
simplifiedby an additionalspecial assump- viewed as the product of a normal (or
tion: that the aggregate demand curve is secular) component common to all markets
unit elastic.2 In this case, the level of and a cyclical component which varies
nominal output can be treated as an from market to market. Letting z index
"exogenous"'variable with respect to the markets, and using ynt and yet to denote
goods market, and the entire burden of ac- the logs of these components, supply in
counting for the breakdown of nominal market z is:
income into real output and price is placed = Ynt +
( 1) yt(Z) yet(Z)
on the aggregate supply side. In the next
subsection, A, a supply model designed to The secular component, reflecting capital
serve this purpose is developed. In subsec- accumulation and population change, fol-
tion B, solutions to the full (demand and lows the trend line:
supply) model are obtained. a+? t
(2) y.t
A. AggregateSupply The cyclical component varies with per-
All formulations of the natural rate ceived, relative prices and with its own
theory postulate rational agents, whose lagged value:
decisions depend on relative prices only, y[Pt(z) -
(3) yet(z) E(Pt It(Z))]
placed in an economic setting in which
they cannot distinguish relative from gen- + NyC_t-1(z)
eral price movements. Obviously, there is where Pt(z) is the actual price in z at t and
no limit to the number of models one can E(PtI It(z)) is the mean current, general
construct where agents are placed in this price level, conditioned on information
situation of imperfect information; the available in z at t, It(z).3 Since yet is a
trick is to find tractable schemes with this deviation from trend, f I < 1.
feature. One such model is developed 3 A supply function for labor which varies with the
below. ratio of actual to expected prices is developed and veri-
We imagine suppliers as located in a fied empirically by Lucas and Rapping (1969). The
large number of scattered, competitive effect of lagged on actual employment is also shown.
In our 1972 paper, in response to Albert Rees's criti-
markets. Demand for goods in each period cism, we found that this persistence in employment
cannot be fully explained by price expectations behav-
2 An explicit derivation of the price-output relation- ior. Both these effects-an expectations and a persis-
ship from the IS-LM framework is given by Frederic tence effect-will be transmitted by firms to the goods
Raines. Of course, this framework does not imply an market. In addition, they are probably augmented by
elasticity of unity, though it is consistent with it. Since speculative behavior on the part of firms (as analyzed
the unit elasticity hypothesis is primarily a matter of for example, by Paul Taubman and Maurice Wilkinson).
convenience in the present study, I shall comment below For a general equilibrium model in which suppliers
on the probable consequences of relaxing it. behave essentially as given by (3), see my 1972 papers.
328 THE AMERICAN ECONOMIC REVIEW JUNE 1973

The information available to suppliers Averaging over markets (integrating with


in z at t comes from two sources. First, respect to the distribution of z) gives the
traders enter period t with knowledge of aggregate supply function:
the past courseof demand shifts, of normal
supply ynt, and of past deviations ye,t-i, (7) = yYnt+ G7y(Pt- Pt)
yc,t-2 . While this information does + xLyt-1 - yn,t-1I
not permit exact inference of the log of the The slope of the aggregate supply func-
current general price level, Pt, it does de- tion (7) thus varies with the fraction 0 of
termine a "prior"distribution on Pt, com- total individual price variance, a2+r2,
mon to traders in all markets. We assume which is due to relative price variation,,In
that this distribution is known to be nor- cases where r2 is relatively small, so that
mal, with mean Pt (depending in a known individual price changes are virtually cer-
way on the above history) and a constant tain to reflect general price changes, the
variance a2.
supply curve is nearly vertical. At the
Second, we suppose that the actual price other extreme when general prices are
deviates from the (geometric) economy- stable (a2 is relatively small) the slope of
wide average by an amount which is dis- the supply curve approaches the limiting
tributed independently of Pt. Specifically, value of y.4
let the percentage deviation of the price in
z from the average Pt be denoted by z (so B. Completionand Solution of the Model
that markets are indexed by their price A central assumption in the develop-
deviations from average) where z is nor- ment above is that supply behavior is
mally distributed, independent of Pt, with based on the correct distribution of the
mean zero and variance r2. Then the ob- unobserved current price level, Pt. To
served price in z, Pt(z) (in logs) is the sum proceed, then, it is necessary to determine
of independent, normal variates what this correct distribution is, a step
(4) Pt(z) = Pt + z which requires the completion of the
model by inclusion of an aggregate de-
The information It(z) relevant for estima- mand side.
tion of the unobserved (by suppliers in z As suggested earlier, this will be done by
at t) Pt, consists then of the observed postulating a demandfunction for goods of
price Pt(z) and the history summarized the form:
in P.t
To utilize this information, suppliersuse (8) yt + Pt = xt
(4) to calculate the distribution of Pt, where xt is an exogenous shift variable-
conditional on Pt(z) and Pt. This distribu- equal to the observable log of nominal
tion is (by straightforward calculation) GNP. Further, let Axt} be a sequence of
normal with mean: independent, normal variates with mean 6
r

E(Pt I(z)) = E(Pt Pt(z), iPt) ande var'lanceax 0


and variane2

(~) ~ ~~~ = (1-)Pt (z) + 07Pt 4This predicted relationship between a supply elas-
ticity and the variance of a component of the price series
where 0= i2/(u2+i2) , and variance Ooa2. is analogous to the link between the income elasticity of
Combining (1), (3), and (5) yields the consumption demand and the variances of permanent
supply function for market z: and transitory income components which Friedman
(1957) observes. As will be seen in Section II, it works
- PTt]
yt(Z) = Ynt + O@Y[Pt(z) in empirical testing in much the same way as well.
(6) This particular characterization of the "shocks" to
+ XYc,t-1(z) the economy is not central to the theory, but to discuss
VOL. 63 NO. 3 LUCAS: OUTPUT-INFLATIONTRADEOFF 329

The relevant history of the economy


then consists (at most) of ynt (which fixes + - Xt-1 - (1 - X)Y.t
1 + O-Y
calendar time), the demand shifts xt,
xt1,... , and past actual real outputs 1- + +X0++ 1 + -Axt
8 AX
Yt-1, Yt-2. * Since the model is linear in I8 O
logs, it is reasonable to conjecture a price + Xyt-1 + (1 - X)ynt
solution of the form: 6
In terms of APt and yet, and letting
(9) Pt= 7rO+rlXt+7r2Xt_l+7r3Xt_2+ *
7r= y/(1+G'y), the solutions are:
+l1 Yt_1+?12Yty2+ +tOynt
(11) Yet = - irb + 7rAxt + Xye,t_1
Then 7Ptwill be the expectation of Pt, (12) APt = - d + (1 - r)AXt + 7rAXt-i
based on all information except Xt (the
current demand level) or:
Pt = 750 + 7rl(Xt-1 + 5) + 72Xt-l Let us review these solutions for internal
consistency. Evidently, Pt is normally dis-
(10) + 7r3Xt-2 + ... + ?lyt-l tributed about Pt. The conditional vari-
+ n72yt.-2 + + tOynt ance of Pt will have the constant (as
assumed) variance 1/(1+0Gy)2o-,. Thus
To solve for the unknown parameters those features of the behavior of prices
7ri, Ij and 4o we first eliminate yt between which were assumed "known"by suppliers
(7) and (8), or equate quantity demanded in subsection A are, in fact, true in this
and supplied. Then inserting the right economy.
sides of (9) and (10) in place of Pt and Pt, To review, equations (11) and (12) are
one obtains an identity in Xt}, {yt , and the equilibrium values of the inflation rate
ynt, which is then used to obtain the and real output (as a percentage deviation
parameter values. The resulting solutions from trend). They give the intersection
for price and output are:' points of an aggregate demand schedule,
shifted by changes in xt, and an aggregate
Pt= - + X- , supply schedule shifted by variables
1+GY 1 +OY (lagged prices) which determine expecta-
tions. In order to avoid the introduction of
an additional, spurious "expectations pa-
rational expectations formation at all, some explicit
stochastic description is clearly required. Independence rameter," one cannot solve for this inter-
is used here partly for simplicity, partly because it is section on a period-by-period basis; 4c-
empirically roughly accurate for most countries in the cordingly, we have adopted a method
sample. The effect of autocorrelation in the shocks
would, as can be easily traced out, be to add higher order which yields equilibrium"paths" of prices
lag terms to the solutions found below. and output. Otherwise, the interpretation
6 This solution method is adapted from Lucas (1972),
of (11) and (12) is entirely conventional.
which is in turn based on the ideas of John Muth.
7 If a demand function of the form
Yt=-Pt+xt had
Not surprisingly, the solution values of
been used, these solutions would assume the same form, inflation and the cyclical component of
with different expressions for the coefficients. If t $1, real output are indicated by (11) and (12)
however, xt is an unobserved shock, unequal in general
to observed nominal income. In this case, the model still to be distributed lags of current and past
predicts the time-series structure (moments and lagged changesin nominaloutput. A change in the
moments) of the series Yetand APt and is thus, in princi- nominal expansion rate, Axt, has an im-
ple, testable. I have found empirical experimenting
along these lines suggestive, but the series used are mediate effect on real output, and lagged
simply too short to yield results of any reliability. effects which decay geometrically. The
330 THE AMERICAN ECONOMIC REVIEW JUNE 1973

immediate effect on prices is one minus the a fitted trend line must average to zero.
real output effect, with the remainder of Accordingly, we must base tests of the
the impact coming in the succeeding pe- natural rate hypothesis (in this context)
riod. We note in particular that this lag on (13): a relationship between an ob-
pattern may well produceperiods of simul- servable variance and a slope parameter.
taneous inflation and below average real
output. Though these periods arise be- II. Test Results
cause of supply shifts, the shifts result Testing the hypothesis advanced above
from lagged perceptionof demand changes, involves two steps. First, within each
and not from autonomous changes in the country (11) and (12) should perform
cost structure of suppliers. reasonably well. In particular, under the
In addition to these features, the model presumption that demand fluctuations are
does indeed assert the existence of a nat- the major source of variation in APt and
ural rate of output: the averagerate of de- yct, the fits should be "good." The esti-
mand expansion, 5, appears in (11) with a mated values of 7rand X should be between
coefficient equal in magnitude to the co- zero and one. Finally, since (11) and (12)
efficient of the current rate, and with the involve five slope parametersbut only two
opposite sign. Thus changes in the average theoretical ones, the estimated ir and X
rate of nominal income growth will have no values obtained from fitting (11) should
effect on average real output. On the other work reasonably well in explaining varia-
hand, unanticipated demand shifts do tions in APt.
have output effects, with magnitude given The main object of this study, however,
by the parameter ir. Since this effect de- is not to "explain" output and price level
pends on "fooling" suppliers (in the sense movements within a given country, but
of subsectionA), one expects that 7rwill be rather to see whether the terms of the
larger the smaller the variance of the output-inflation "tradeoff'' vary across
demand shifts. We next develop this im- countries in the way predicted by the nat-
plication explicitly. ural rate theory. For this purpose, we shall
From the definition of ir in terms of 0 utilize the theoretical relationship (13)
and y, and the definition of 6 in terms of and the estimated values of r and a'2.
q2 and r2 we have Under the assumption that r2 and y are
relatively stable across countries, the esti-
r27 mated r values should decline as the sam-
a2 + r2(1 + y) ple variance of Axt increases.
Descriptive statistics for the eighteen
Combining with the expression for u2 ob-
tained above, this gives countries in the sample are given in Table
1., As is evident, there is no association

(13) *,
r =-
(1
8 The raw data on real and nominal GNP are from
- )2a2 + r2(1 + y)
Yearbookof National Accounts Statistics, where series
from many countries are collected and put on a uniform
For fixed r2 and y, then, ir takes the value basis. The choice of countries is by no means random:
y/(1 +y) at ax= 0 and tends monotonically the eighteen used are all the countries from which con-
tinuous series are available. The sample could thus be
to zero as o2 tends to infinity. broadened considerably by use of sources from indi-
The prediction that the average devia- vidual countries. To obtain the variables used in the
tion of output from trend, E(y,,), is in- tests, the logs of real and nominal output, Ytand xt, are
logs of the series in the source. The log of the price level,
variant under demand policies is not, of
Pt, is the difference xt-yt; yt is the residual from the
course, subject to test: the deviations from trend line yt= a+bt, fit by least squares from the sample
VOL. 63 NO. 3 LUCAS: OUTPUT-INFLATIONTRADEOFF 331

TABLE1-DESCRIPTIVESTATISTICS,
1952-67

Mean Mean Variance. Variance Variance


Cyt Apt Yet Axt

Argentina .026 .220 .00096 .01998 .01555


Austria .048 .038 .00104 .00113 .00124
Belgium .034 .021 .00075 .00033 .00072
Canada .043 .024 .00109 .00018 .00139
Denmark .039 .041 .00082 .00038 .00084
West Germany .056 .026 .00147 .00026 .00073
Guatemala .046 .004 .00111 .00079 .00096
Honduras .044 .012 .00042 .00084 .00109
Ireland .025 .038 .00139 .00060 .00111
Italy .053 .032 .00022 .00044 .00040
Netherlands .047 .036 .00055 .00043 .00101
Norway .038 .034 .00092 .00033 .00098
Paraguay .054 .157 .00488 .03192 .03450
Puerto Rico .058 .024 .00205 .00021 .00077
Sweden .039 .036 .00030 .00043 .00041
United Kingdom .028 .034 .00022 .00037 .00014
United States .036 .019 .00105 .00007 .00064
Venezuela .060 .016 .00175 .00068 .00127

between average real growth rates and and Puerto Rico, so do the estimated X
average rates of inflation: this fact seems values. The R's indicate that for many, or
to be consistent with both the conventional perhaps most countries, important out-
and natural rate views of the tradeoff. put-determiningvariables have been omit-
Since our interest is in comparingreal out- ted from the model. The R s for the infla-
put and price behavior underdifferenttime tion rate equation, (12), are given in
patterns of nominal income, these statistics column (4) of Table 2. In general, these
are somewhat disappointing. Essentially tend to be lower than for equation (11),
two types of nominal income behavior are and not surprisingly the estimated co-
observed:the highly volatile and expansive efficients from (12) (which are not shown)
policies of Argentina and Paraguay, and tend to behave erratically. Column (5) of
the relatively smooth and moderately ex- Table 2 gives the fraction of the variance
pansive policies of the remaining sixteen of AP' explained by (12) when the co-
countries. But if the sample provides only efficient estimates from (11) are imposed.
two "points," they are indeed widely (A "-" indicates a negative value.)9
separated: the estimated variance of de- With respect to its performance as an
mand in the high inflation countries is on intracountry model of income and price
the order of 10 times that in the stable determination, then, the system (11)-(12)
price countries. passes the formal tests of significance. On
The first three columns of Table 2 sum- the other hand, the goodness-of-fit statis-
marize the performance of equation (11)
in accounting for movements in yet. The 9 The loss of explanatory power when these coeffi-
estimated values for ir all lie between zero cients are imposed on (12) can be assessed formally by
an approximate Chi-square test. By this measure, the
and one; with the exceptions of Argentina loss is significant at the .05 level for Paraguay only. As
Table 2 shows, however, this test is somewhat decep-
period. The moments given in Table 1 are maximum tive: for several countries the least squares estimates of
likelihood estimates based on these series. The estimates (12) are so poor that there is little explanatory power to
reported in Table 2 are by ordinary least squares. lose, and the test is "passed" vacuously.
332 THE AMERICAN ECONOMIC REVIEW JUNE 1973

TABLE 2-SUMMARY STATISTICS BY COUNTRY, 1953-67

Country 7r R R2 R2

Argentina .011 -.126 .018 .929 .914


(.070) (.258)
Austria .319 .703 .507 .518 -
(.179) (.209)
Belgium .502 .741 .875 .772 .661
(.100) (.093)
Canada .759 .736 .936 .418 -
(.064) (.075)
Denmark .571 .679 .812 .498 .282
(.118) (.110)
West Germany .820 .784 .881 .130 -
(.136) ( .110)
Guatemala .674 .695 .356 .016
(.301) (.274)
Honduras .287 .414 .274 .521 .358
(.152) (.250)
Ireland .430 .858 .847 .499 .192
(.121) (.111)
Italy .622 .042 .746 .934 .914
(.134) (.183)
Netherlands .531 .571 .711 .627 .580
(.111) (.149)
Norway .530 .841 .893 .633 .427
(.088) (.096)
Paraguay .022 .742 .568 .941 .751
(.079) (.201)
Puerto Rico .689 1.029 .939 .419
(.121) (.072)
Sweden .287 .584 .525 .648 .405
(.166) (.186)
United Kingdom .665 .178 .394 .266 .115
(.290) (.209)
United States .910 .887 .945 .571 .464
(.086) (.070)
Venezuela .514 .937 .755 .425
(.183) (.148)

tics are generally considerably poorer than tina. For the United States, the fitted ver-
we have come to expect from annual time- sions of (11) and (12) are:
series models.
yct = - .049 + (.910),Axt + (.887)yc,t-l
In contrast to these somewhat mixed re-
sults, the behavior of the estimated 7r APt = - .028 + (.119) Axt + (.758) Axt_
values across countries is in striking con- -
(.637) Aycet_1
formity with the natural rate hypothesis. The comparable results for Argentina are:
For the sixteen stable price countries, *
ranges from .287 to .910; for the two yct = - .006 + (.011)Axt - (.126)y,.,t_j
volatile price countries, this estimate is APt = - .047 + (1.140)Axt - (.083) Axt-
smaller by a factor of 10! To illustrate this
+ (.102)Ayc,t-1
order-of-magnitude effect more sharply,
let us examine the complete results for two In a stable price country like the United
countries: the United States and Argen- States, then, policies which increase nomi-
VOL. 63 NO. 3 LUCAS: OUTPUT-INFLATION TRADEOFF 333

nal income tend to have a large initial first, that changes in average inflation
effect on real output, together with a small, rates will not increase average output, and
positive initial effect on the rate of infla- secondly, that the higher the variance in
tion. Thus the apparent short-term trade- average prices, the less "favorable" will be
off is favorable, as long as it remains un- the observed tradeoff.
used. In contrast, in a volatile price The most natural cross-national com-
country like Argentina, nominal income parison of these propositions would seem
changes are associated with equal, con- to be a direct examination of the associa-
temporaneous price movements with no tion of average inflation rates and average
discernible effect on real output. These output, relative to "normal" or "full em-
results are, of course, inconsistent with the ployment." Unfortunately, there seems to
existence of even moderately stable Phillips be no satisfactory way to measure normal
curves. On the other hand, they follow output. The deviation-from-fitted-trend
directly from the view that inflation method I have used defines normal output
stimulates real output if, and only if, it to be average output. The use of unem-
succeeds in "fooling" suppliers of labor ployment series suffers from the same diffi-
and goods into thinking relative prices are culty, since one must somehow select the
moving in their favor. (obviously positive) rate to be denoted full
employment.
III. Concluding Remarks Thus although the issue revolves around
The basic idea underlying the tests re- the relation between means of inflation
ported above is extremely simple, yet I am and output rates, it cannot be resolved by
afraid it may have become obscured by the examination of sample averages. Fortu-
rather special model in which it is em- nately, the existence of a stable tradeoff
bodied. In this section, I shall try to re- also implies a relationship between vari-
state this idea in a way which, though not ances of inflation and output rates, as
quite accurate enough to form the basis for illustrated in Figure 1. With a stable
econometric work, conveys its essential tradeoff, policies which lead to wide varia-
feature more directly. tion in prices must also induce comparable
The propositions to be compared empiri- variation in real output.' If these sample
cally refer to the effects of aggregate de- variances do not tend to move together
mand policies which tend to move infla- (and, as Table 1 shows, they do not) one
tion rates and output (relative to trend) in
the same direction, or alternatively, un-
employment and inflation in opposite APt'
directions. The conventional Phillips curve Var(yct) 'a2 Varf(APt) 0 0
account of this observed co-movement says
that the terms of the tradeoff arise from 00
relatively stable structural features of the 0
economy, and are thus independent of the
nature of the aggregate demand policy
pursued. The alternative explanation of
0 Stable Price Country
the same observed tradeoff is that the 0 Volatile Price Country
positive association of price changes and
output arises because suppliers misinter-
pret general price movements for relative Yct
price changes. It follows from this view, FIGURE 1
334 THE AMERICAN ECONOMIC REVIEW JUNE 1973

can only conclude that the tradeoff tends trality of Money," J. Econ. Theor., Apr.
to fade away the more frequently it is 1972, 4, 103-24.
used, or abused. , "EconometricTesting of the Natural
This simple argument leads to a formal Rate Hypothesis," Conferenceon the Econ-
test if the output-inflation association is ometrics of Price Determination, Washing-
ton 1972, 50-59.
entirely contemporaneous. In fact, how-
and L. A. Rapping, "Real Wages,
ever, it involves lagged effects which make Employment and the Price Level," J. Polit.
a direct comparison of variances, as just Econ., Sept./Oct. 1969, 77, 721-54.
suggested, difficult in short time-series. _ and , "Unemployment in the
Accordingly, it has been necessary to im- Great Depression: Is There a Full Expla-
pose a specific, simple structure on the nation?",J. Polit. Econ., Jan./Feb. 1972, 80,
data. As we have seen, this structure ac- 186-91.
counts for output and inflation rate move- J. F. Muth, "Rational Expectations and the
ments only moderately well, but well Theory of Price Movements," Economet-
enough to capture the main phenomenon rica, July 1961, 29, 315-35.
predicted by the natural rate theory: the E. S. Phelps, introductory chapter in E. S.
higher the variance of demand, the more Phelps et al., Micro-economics of Inflation
unfavorable are the terms of the Phillips and Employment Theory, New York 1969.
F. Raines, "MacroeconomicDemand and Sup-
tradeoff.
ply: an Integrative Approach," Washing-
ton Univ. working paper, Apr. 1971.
REFERENCES A. Rees, "On Equilibrium in Labor Markets,"
M. Friedman, A Theory of the Consumption J. Polit. Econ., Mar./Apr. 1970, 78, 306-10.
Function, Princeton 1957. P. Taubman and M. Wilkinson, "User Cost,
, "The Role of Monetary Policy," Capital Utilization and Investment
Amer. Econ. Rev., Mar. 1968, 58, 1-17. Theory," Int. Econ. Rev., June 1970, 11,
D. F. Gordon and A. Hynes, "On the Theory 209-15.
of Price Dynamics," in E. S. Phelps et al., United Nations, Department of Economic
Micro-economics of Inflation and Employ- and Social Affairs, United Nations Statisti-
ment Theory, New York 1969. cal Office, Yearbook of National Accounts
R. E. Lucas, Jr., "Expectations and the Neu- Statistics, 66 and 68, New York 1958.