NBER WORKING PAPER SERIES

REAL BUSINESS CYCLES: A NEW KEYNESIAN PERSPECTIVE

N. Gregory Mankiw

Working Paper No. 2882

NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue

Cambridge, MA 02138March 1989

Professor of Economics, Harvard University, and Research Associate, NationalBureau of Economic Research. This paper was prepared for the Journal ofEconomic Persyectives. I am grateful to Laurence Ball, Susanto Basu,Marianne Baxter, Mark Bils, Lawrence Katz, Deborah Mankiw, David Romer,Joseph Stiglitz, Lawrence Summers, Timothy Taylor, David Weil, and MichaelWoodford, for helpful discussions and comments, and to the National ScienceFoundation for financial support. This paper is part of NBER's researchprograms in Economic Fluctuations and in Financial Markets and MonetaryEconomics. Any opinions expressed are those of the author not those of theNational Bureau of Economic Research.

NBER Working Paper #2882March 1989

REAL BUSINESS CYCLES: A NEW KEYNESIAN PERSPECTIVE

ABSTRACT

This paper is a critique of the latest new classical theory of

economic fluctuations. According to this theory, the business cycle

is the natural and efficient response of the economy to exogenous changes

in the available production technology. This paper discusses several

versions of this theory and argues that this line of research is unlikely

to yield an empirically plausible explanation of observed economic

fluctuations.

N. Cregory MankiwNational Bureau of Economic Research1050 Massachusetts AvenueCambridge, MA 02138

the aftermath

the Great Dep

The debate over the source and propagati

fluctuations rages as fiercely today as it di

of Keynes's The Theory

ression. Today, there

ssical school es the

actors, the a of re

demand, and iency

nesian schoo s that

ons requires study

rium, but als ating

The cia

economic

supply and

The Key

fluctuati

thought

private

equate

markets.

economic

of general

market fai

Real

classical

on of economj

d fifty years

and in the mi

are two schoo

optimization

lative prices

of unfettered

understand in

ing the intri

the possibil

C

ago in

dst of

is of

of

to

g

cacies

ity of

as then,

emphas iz

dj ustment

the effic

1 believe

not just

0 appreci

scale.

theory is the

fluctuations.

e qu

lure

bus

view

ii ib

on a grand

mess cycle

of economic

late

It

are large

change.

rationally

The busine

efficient

product i o

My

in the rate of

these fluctuations, i

levels of labor supply

according to this theo

the ec

random fluctuations

n response to

alter their

ss cycle is,

response of

n technology.

goal in this essay

approach to the business cyc

St incarnation of the

assumes that there

technological

ndividuals

and consumption.

ry, the natural and

onomy to changes in the available

am not an advocate.

is

le

In my view,

to appraise this

I should admit

real business

not provide an

fluctuations

empir

Zoth

ically plaus

its reliance

newly revived

in advance that I

cycle theory does

of economic

ogical

ible expi

on large

1

anat ion

technol

disturbances as the primary source of economic fluctuations and

its reliance on the intertemporal substitution of leisure to

explain changes in employment are fundamental weaknesses.

Moreover, to the extent that it trivializes the social cost of

observed fluctuations, real business cycle theory is potentially

dangerous. The danger is that those who advise policy-makers

might attempt to use it to evaluate the effects of alternative

macroeconomic policies or to conclude that macroeconomic policies

are unnecessary.

Wairasian Equilibrium and The Classical Dichotomy

The typical undergraduate course in microeconomics begins

with partial equilibrium analysis of individual markets. A

market for a good is characterized by a downward sloping demand

curve and an upward sloping supply curve. The price of the good

is assumed to adjust until the quantity supplied equals the

quantity demanded.

The course then builds up to Wairasian general equilibrium.

In this Wairasian equilibrium, prices adjust to equate supply and

demand in every market simultaneously. The general equilibrium

system determines the quantities of all goods and services sold

and their relative prices. The most important theoretical

result, after the existence of such a Wairasian equilibrium, is

the "invisible hand" theorem: the equilibrium is Pareto

efficient.

Courses in microeconomics thus show how employment.

2

g money in t

tinkin 1956)

ative prices,

by the Ja1ras

ce level, the nomi

then determined by

since nominal van

arket is not very

suggests that, fo

The profe

class ical

librium,

either

Real variables, such as

including the real inte

jan system. Nominal

nal wage, and the nom

the equilibrium in t

The

money

nd

evel of

employment,

rest rate,

production, and relative

mention of the existence

simplest way to append in

demand function and an e

depends on the level of

prices are determined without any

of money, the medium of exchange.

oney to the model is to specify a

xogenous morey supply. Money dema

output and the price level. The 1

output is already determined in the

level, however, can adjust to equat

Walrasian system.

e supply and demand

money market.

Introduc in

dichotomy. (Pa

output, and tel

are determined

The price

in the

his way leads to the classical

as

rat

Of

the

the

mar

the

e,

cou

ec

ke t

pri

are

rse

ney m

onomy

can be ignored

var j

inal

he in

t real

lass ic

uss ion

ables

impor

r mos

do

tant

t po

ables

inte

oney

var i

al vi

s, th

not affec

This c

licy disc

such

rest

market.

ables

ew of

e money

the

equi

must

with

ssor of macroeconomic

dichotomy. Given th

money is largely irre

destroy this classical

s must in some way deal with

e assumptions of Walrasian

levant. The macroeconomjst

dichotomy or learn to live

it

Keynesian macroeconomics destroys the classical dichotomy by

abandoning the assumption that wages and prices adjust instantly

to clear markets. This approach is motivated by the observation

3

chat many nominal wages are fixed by long-term labor contracts

and many product prices remain unchanged for long periods of

time. Once the inflexibility of wages and prices is admitted

into a macroeconomic model, the classical dichotomy and the

irrelevance of money quickly di

of the early

attempted to

the fundamen

sappear

Much work in the new classical revolution of

the l970s destroy the classical dichotomy without

abandoning tal axiom of continuous market clearing.

(Lucas 1972,1973) These models were based on the assumption that

individuals have imperfect information regarding prices. These

individuals therefore confuse movements in the overall price

level (which under the classical dichotomy should not matter)

with movements in relative prices (which should matter). An

unanticipated decrease in the money supply leads individuals to

infer that the relative prices of the goods they produce are

temporarily low, which induces them to reduce the quantity

supplied. TJhile the fascination with this sort of story was

substantial in the l970s, it has attracted relatively few

adherents in the 1980s. It is hard to believe that confusion

about the price level is sufficiently great to generate the large

changes in quantities observed over the business cycle.

In contrast to both the Keynesian and the early new

classical approaches to the business cycle, real business cycle

theory embraces the classical dichotomy. It accepts the complete

irrelevance of monetary policy,

by almost all macroeconomists a

thereby denying a tenet accepted

decade ago. Nominal variables,

4

such as the money supply and the price level, are assumed to have

no role in explaining fluctuations in real variables, such as

output and employment.

Real business cycle theory thus pushes the Walrasian model

farther than it has been pushed before. In evaluating whether it

provides a successful explanation of recessions and booms, two

questions naturally arise. First, why are there such large

fluctuations in output and employment? And second, why do

movements in nominal variables, such as the money supply, appear

related to movements in real variables, such as output?

Classical and Keynesian Views of Economic Fluctuations

The only forces that can cause economic fluctuations,

according to real business cycle theory, are those forces that

change the Wairasian equilibrium. The Walrasian equilibrium is

simply the set of quantities and relative prices that

simultaneously equate supply and demand in all markets in the

economy. To understand how real business cycle theory explains

the business cycle, it is necessary to look into the fundamental

forces that change the supplies and demands for various goods and

services

Many sorts of macroeconomic disturbances can in principle

generate fluctuations in real business cycle models. For

example, changes in the level of government purchases or in the

investment tax credit alter the demand for goods and therefore

affect the WaIrasian equilibrium. Changes in the relative price

5

of oil alter the equilibrium allocation of labor among

alternative uses. Many of the macroeconomic disturbances that

receive much attention among Keynesian macroeconomists will also

have important effects in real business cycle models. There is,

however, substantial disagreement between the two schools

regarding the mechanisms through which these disturbances work.

Consider the case of a temporary increase in government

purchases. Almost all macroeconomists agree that such a change

causes an increase in output and employment, and the evidence,

mainly from wartime experience, supports this prediction. Yet

the explanations of this effect of government purchases differ

greatly. (Cf. Barro 1987 and Dornbusch and Fischer 1987)

Real business cycle theory emphasizes the intertemporal

substitution of goods and leisure. It begins by pointing out

that an increase in government purchases increases the demand for

goods. To achieve equilibrium in the goods market, the real

interest rate must rise, which reduces consumption and

investment. The increase in the real interest rate also causes

individuals to reallocate leisure across time. In particular, at

a higher real interest rate, working today becomes relatively

more attractive than working in the future; today's labor supply

therefore increases. This increase in labor supply causes

equilibrium employment and output to rise.

While Keynesian theory also predicts an increase in the real

interest rate in response to a temporary increase in government

purchases, the effect of the real interest rate on labor supply

6

market is charac

• Inc

theory

oyment.

Both real busi

conclude that incre

and employment

imp 1 i cation s

Wairasian mod

observational

with both the

increase in

in the amount of

Keynesian theory, the

state of excess

y thus

output

minent

emp oral

rox ima te

are consistent

The Cen

Wh

princ ip

mode

The

gen

act

bus

dir

fal

tral

ile

le cause

used on technologic

of disturbances ar

business cycle mode

is that over

sure move in

into a reces

economy goes

al disturbances.

e unlikely to

Is that resemble

the typical

opposite

sion, consumption

into a boom,

does not play a crucial role. Instead, the

employment and output is due to a reduction

labor unemployed or underutilized. In most

labor

supply

cycle

unemp I

ontrast

does not

terized as

the Wa].ras

allow for

of

ian

the

ten in a

approach of

possibility

real bus

of invol

mes s

untary

ness cycle theory and Keynesian theor

ases in government purchases increase

This example shows that some of the pro

Keynesian models also come Out of intert

Macroeconomists face a problem of app

of

els

eq

ci

uivalen

assical

Ce: many ob

and Keynes

served phenomena

ian paradigms.

many sorts of macroeconomic disturbances

economic fluctuations in real busine

can

ss c

in

ycle

ls, most attention has foc

reason is that other sorts

erate fluctuations in real

ual economic fluctuations.

An obvious but important fact

mess cycle, consumption and lei

ections. When the economy goes

Is and leisure rises. When the

7

increase in government purchases,

should fall. Many other changes

a change due to a temporary inves

cause consumption and leisure to

Real business cycle theo

recession find it rational to

they demand at the same time

they demand. The answer mu

relative to goods, the real

crucial implication of real

real wage is procyclical.1

If the production function w

were the source of fluctuations,

abor and

roduction

produce a

real, wage

and leisure

goods, such as

also should

iduals in

leisure

of goods

ure

Hence

C the

cons

is p

cons

s inc

ump t

0 tCfl

urnp t

e bo

Lon

tia

ion

th

rises and leisur

lly problematic f

and leisure woul

are normal goods.

e falls. Explaining t

or real business cycle

d often be expected to

In the example of a

his p

theo

move

t emp 0

henome non

ry:

together.

rary

both consumpt

in the demand

tment tax cred

move together.

ion

for

it,

ry must explain why mdiv

increase the quantity of

they decrease the quantity

st be that the price of leis

wage, falls in a recession.

business cycle theory is tha

a

a

generating a p

in a recession

ere unchanging and demand shocks

real business cycle theory would

rocyc I

one

have tro

input is

product

unchang i

labor wo

procyci i

uble

low

of 1

ng p

uld

cal

ical real wage. S

would expect that

ince labor

the marginal

thus the real wage should be high. With an

function, diminishing marginal returns to

countercyclical real wage, not the

necessary to explain the fluctuations in

'Alternatively, one could explain the observed patternwithout a procyclical real wage by positing that tastes forconsumption relative to leisure varyover time. Recessions arethen periods of "chronic laziness." As far as I know, no one hasseriously proposed this explanation of the business cycle.

8

ure.

ycle

tua t

theorists

ions in the

there

rate

fore

of

assume that

technological

there

change

and leis

usiness c

tial fluc

ion, th

The

In

reduce

real bus

5 as a c

uations

and the

of emplo

Attempts

e available production

marginal product of

response to the low

consumption and inc

mess cycle theory

hanging Walrasian e

are efficient. Civ

technological p055

yment, output, and

by the government

market, such as policies

technology is relatively

labor and thus the real

return to working,

rease leisure.

describes economic

quilibrium, it implies that

en the tastes of

ibilities facing society,

consumption

Real b

are substan

In a recess

unfavorable

wage are low

individuals

Since

fluc tuat ion

these fluct

individuals

the levels

improved.

the private

best are me

"invisible h

Of all

optimalicy

It seems un

recession t

explains th

coordinatjo

ffective and

consumption

to alter the

to stabilize

by iat worst can do harm

cannot be

allocations of

employment, at

mpeding the

and

the

of e

deni

han

N

implicatconomic f

able that

in the bo

ions of real business

luctuations is perhaps

the level of welfare

om that preceded it.

cycle theory, the

the most shocking.

is lower in a

Keynesian theory

e in economic

adjust

tion in welfare by a failur

use wages and prices do not

e reduc

n: beca

taneously to

from trade go

ss cycle theo

welfare is 1

ins tan

gains

bus me

reason

equate supply and demand in all markets, s

unrealized in a recession. In contrast,

ry allows no unrealized gains from trade.

ower in a recessiqn is, according to these

9

ome

real

The

theories, that the technological capabilities of society have

declined.

Advocates

convincing ske

sudden changes

that the accum

cycle theorists

run fluctuation

Edward Pr

the importance

changes in tot

economy- - the p

inputs, where

shares. This

technological

substantial El

suggests a pot

disturbances

Figure

the percent

are for the

services

ical Di

real business

s that the ec

technology.

ion of knowle

must

5 in

escot

of te

al fac

ercent

the di

Solo w

progre

uc tuat

ential

s turbance s

cycle

on o my

It is

dge a

uaLLy

ness

short

direct evidence on

He examines

nited States

e percent change in

ed by their factor

the rate of

are

g which

The Evidence on Technolo

of

pticin

ulat

theories have trouble

is subject to such large

a more standard presumpt

nd the concurrent increas

the economy's techno

over time. Yet to m

and

ion

e in

logical opportunities take place grad

imic observed fluctuations, real busi

ub Stanti a 1mai

the

t (Ichno

tor

ntain

produ

986)

logic

produ

that there are sction function.has offered some

al disturbances.ctivity for the U

in Out

inputsI" sho

putare

uld

less th

we ight

measure

change

ffe rent

res idua

ss. Pre

ions in

ly impor

as a source of

1 presents my ca

change in output

private economy

Like Prescott, I

scott points out that there

the Solow residual, a findin

tant role for technological

business cycle fluctuations.

iculation of the Solow residual and

yearly since 1948. (Both variables

less agriculture and housing

find substantial fluctuations in

10

year.

Figure 1.

cyclical. In

productivity a

of the change

recessions are

The Solow

regarding exoge

standard explan

labor hoarding

Productivity ap

also shows

every

lso fe

fn the

perio

res id

nous

ation

and o

pears

r input

earl

that

lid measure

then

2A related explanation of the procyclical behavior of the

Solow residual has recently been proposed by Hall (1987): Hallpoints Out that if price exceeds marginal cost because ofimperfect competition, then the measured Solow residual willappear procyclical even if the true production technology isunchanging. Alternatively, the Solow residual could reflectendogenous changes in technology due to demand shocks: suchendogeneity might arise if, for example, learning-by-doing isimportant.

11

1 factor

tivity fe

measured tota

factor produc

3.4 percent.

the economy'

aggregate pr

productivity. For

11 by 3.5 percent,

example,

while in

in 1982

1984 it

total

rose by

One might interpret these numbers as showing

s ability to convert inputs into outputs--the

oduction function--varies substantially from year

that measured productivity is highly

total factor

to

year in which output fell,11. If the Solow residual is a va

available production technology,

ds of technological regress.

ual need not be interpreted as evidence

technological disturbances, however. The

of cyclical productivity is that it reflects

ther "off the production function" behavior.

to fall in a recession because firms keep

unnece

labore

wi thou

A

ssary andrs begin

t a largen examina

unde

to pu

incr

tion

rutili

t Out

ease jof the

zed labor. In

greater effort;

n measured labo

data from the

a boom the hoarded

output increases

support this standard explanation of

2

y 1940s appears to

the cyclical behavior of

productivity. The increase in output associated with the World

War II build-up

Yet from 1939

average of 7.

productivity

1950.) One m

economic boom

rather than de

that the Solow

of changes in

Once the

year changes

longer any di

disturbances.

fluctuations,

disturbances.

available tech

real business

An advoca

economic model

formal evidenc

plausibility t

disturbances. R

widespread atten

however, no disc

demand-driven

1 factor produc

By contrast, th

since then is 5

inding as showin

fact driven by s

e appealing inter

good measure over

nological abilities.

rejected as a measure of year-to-

the

ump t

phenomenon.

tivity grew

e most

.2 percent

g that the

upply shoc

pretation

short hor

an

n

ks

is

izons

is most plausibly a

944 measured tota

cent per year.

rown in any year

interpret this f

he l940s was in

shocks. A mor

idual is not a

economy's tech

w residual is

e available production technol

evidence for substantial techn

to generate fluctuations that

bu mess cycle models require

ex stence of large fluctuatio

y s a crucial but unjustifie

eory.

to 1

6 per

has g

ight

of t

mand

res

the

Solo

in th

rec t

Yet

real

The

no lo

cyci

te 0

s of

e

0

ogy, there is no

ological

mimic observed

such

ns in

d ass ion of

S

1.

g i

e th

f re

ten

Yet

the as

ecessi

tion from

ussion of

al business

rely on ass

more casual

sumption of

ons are

that

no

cycle theory might respond

umptions for which there is

evidence also does not give

subs tantial technological

events; they receiveimportant

policy-maker

declines in

s and the medi

the available

If society suffered some important

12

a. Th

techno

ere is,

logy.

adverse technological shock,

we would be aware of it. My own reading of the newspaper,

Whether changes in energy prices affect the Solow residualcomputed from GNP depends on a variety of issues involving theconstruction of index numbers like CNP. See Bruno and Sachs(1985, P. 43) for a discussion.

4Hamilton (1983) finds oil price changes are also associatedwith the pre-OPEC recessions. Yet these prices changes are muchtoo small to explain plausibly such large declines inproductivity.

13

however, does not lead

exogenous deteriorati

The OPEC

when the econo

easily identif

that the econo

1974 and 1979,

recessions in

Solow residual

changes in the

find events wi

ene

my

i ab

my

as

the

S.

ag

th

me to associate most recessions with some

on in the economy's productive capabilities.

price changes of the l970s illustrate that

riences large real shocks, these shocks are

nd much discussed. Figure 1 indeed shows

rienced large negative Solow residuals in

might have expected.3 Yet the five other

t-war period also exhibit large negative

explain these Solow residuals as adverse

r gy

expe

le a

exp e

one

pos

To

gre

the

gate production

economic signi

The apparent absence of

function, one

ficance of the

such events is

recessions cannot be easily attributed to

would need to

OPEC price

evidence thatincreases

these

4shocks

Labor Suoolv and Intertemooral

Real business cycle theori

employment are fully voluntary.

economy always finds itself on

the typical business cycle, emp

exogenous real

Substitution

sts assume that fluctuations in

In other words, they assume the

the labor supply curve. Yet over

loyment varies substantially while

pattern, rea

very willing

significanti

response to

response to

It is u

to interternp

supply typic

substitute 1

inter tempo r a

assume, then

wage should

Individuals

work hard to

individual 1

the real

(Altonj i

time.

1 business cyci

to reallocate

y reduce the qu

small temporary

small decreases

nlikely, howeve

oral relative p

ally finds that

eisure over tim

lly

in

wor

fac

day

ab or

upply - -

ightly.e model

leisure

anti ty

reduc t

in the

r, that

rices.

the w

e is s

and the real

s observed

individuals

Individuals

ey supply in

real wage or in

metric

ess of

If lei

us mes s

increa

n the future.

their real wage

future. Yet studies of

Personal experience

to judge the behavioral re

models rely. One key beh

labor supplied reacts subs

Jithout such intertemporal

and introspec

sponses on wh

avioral respo

tantially to

substitution

14

tion provide ano

ich real busines

nse is that quan

the real interes

real business

ther way

s cycle

tity of

t rate.

cycle

the determinants of labor s

interest rate- -vary only sl

the real wage

To mimic thi

s require that

over time.

of 1ãboTth

ions in the

be

must

real interes

individuals

t rateare so

evidenc

individ

sure we

cycle

ses in

responsive

e on labor

uals to

re highly

theorists

their real

substitutable, as

dividuals facing e

k little today and

ing expected decre

and enjoy leisure

supply over time

Econo

ill ingn

light.

real

xp e C te

much

ases i

in th

find

change

duals

b

d

i

n

e

t

5

do

lo

wage

1985

should

lead

Ball

to only

1985)

small

Indivi

real wage changes by substantially real

hat expe

in hour

not re

cating

cted changes in

s worked.

spond to expected

leisure over

models are unable to explain how a temporary increase in

output

em pla

t cons

0 acce

laymen

ends on

at labo

ation I

i tutab i

and employment.

usible. The real

ideration when md

Pt new employment.

and students that

the price of app

r supply depends

draw from this o

lity of leisure i

Yet

imte

ivid

Jh

the

les, it is

on the

bservation

s very

government purchases increases

behavioral response does not se

rate is simply not a significan

decide to leave their jobs or t

economists can easily convince

quantity of apples demanded dep

much harder to convince them th

real interest rate. The implic

is that the intertemporal subst

such a

rest

ual s

ile

likely- far too weak to get real business cycle models

Cycle Th

bus imes s

duct ion

eories

cycle

with Mul

theories

to work

cussing so

tiole Se

I have

c tots

been dis

Real Business

The real

far treat pro

This abstractio

business cycle

Some real

technologies of

changes in techn

highlight the in

to the different

as if it takes place in a single industry.

is not characteristicn, however,

theories

business cycle the

different sectors

ology. (Long and

teractions among

sectors are inde

of all real

ze changes in

economy-wide

These model

Even if the

outputs of th

an adverse s

ones emphasi

rather than

Plosser 1983)

the sectors.

pendent, the

For example,different

to one 5

economy;

sectors move together.

ector reduces the

these individual

the

5

shocks

e

hock

for

wealth of the individuals in the

s respond by reducing their demand

15

all goods. An observer would see an aggregate business cycle,

even without a single aggregate shock.

To get these real business cycle models to work, however,

the number of independent sectoral shocks cannot be too great.

If there were many independent sectoral shocks and labor were

mobile between sectors, then the law of large numbers would

guarantee that these shocks and their effect on the aggregate

economy would average out to zero. To get an aggregate business

cycle, these models therefore require that there be only a few

sectors and that these sectors be subject to large technological

disturbances. These models are therefore similar to the single-

sector theories and suffer from the same weaknesses: the absence

of any direct evidence for such large technological disturbances

and the implausibility of strong intertemporal substitutability

of leisure.

A second type of sectoral shock theory emphasizes the costly

adjustment of labor among sectors. (Lilien 1982) These models,

which depart more from the Walrasian paradigm, assume that when a

worker moves from one sector to another, a period of unemployment

is required, perhaps for job search. In this case, independent

shocks across many sectors do not offset each other. Recessions

are, according to these theories, periods of more sectoral shocks

and thus greater intersectoral. adjustment.

This type of real business cycle theory may appear more

plausible than those relying on substantial aggregate

productivity shocks and intertemporal substitution. It is

16

Money and Prices over the Business Cycle

Before real business cycle theory

debate in the early 1980s, almost all

agree on one conclusion: money matters.

discussions of business cycles (Friedman

more formal econometric work (Barro 1977)

Reserve as an important source of macroeconomic disturbances.

While there was controversy as to whether systematic monetary

policy could stabilize the economy, it was universally accepted

that bad monetary policy could be destabilizing.

17

perhaps easier to imagine that recessions are characterized b

unusually great need for intersectoral reallocation than by s

sort of major technological regress that makes leisure unusua

attractive. Yet the available evidence appears not to suppor

this intersectoral story. If workers were unemployed volunta

y an

ome

1 ly

t

r i ly

in recess

sectors

with high

opposite

levels of

Moreover,

measured

procyclic

theory is

economic

ions because

we would expe

job vacancy.

pattern: high

help wanted

in contrast

mobility of w

al. (Murphy

also unable

fluctuations

they were moving to new jobs in other

Ct to find high unemployment coinciding

Yet observed fluctuations have just the

unemployment rates coincide with low

advertising. (Abraham and Katz 1986)

to the prediction of this theory, the

orkers between sectors is strongly

and Topel 1987) This real business cycle

to be plausibly reconciled with observed

nomic

d to

entered the macroeco

macroeconomists seeme

Both historical

and Schwartz 1963)

pointed to the Fe

and

deral

it is ironic that real business cycle theory arose in the

wake of Paul Voicker's disinflation. Many economists view this

recent experience as clear confirmation of -the potency of

monetary policy. Voicker announced he was going to slow the rate

of money growth to achieve a lower rate of inflation; the rate of

money growth in fact slowed down; and one of the deepest post-war

recessions followed, as did an eventual reduction in the rate of

inflation. This set of events is easy to explain within the

context of Keynesian theory with its emphasis on the gradual

adjustment of wages and prices. It is less easy to explain

within the context of real business cycle theory.5

Robert King and Charles Plosset (1984) explain the

historical association between money and output by arguing that

the money supply endogenously responds to fluctuations in output.

Standard measures of the money supply such as Ml are mostly

inside money, that is, money created by the banking system. King

and Plosser suggest that the transactions services of inside

money should be viewed as simply the "output" of one sector of

the economy, the banking sector. Just as one should expect the

outputs of different sectors to move together within a multi-

sector real business cycle model, one should expect the output of

the banking sector to move with the outputs of other sectors. An

increase in productivity in any sector will tend to increase the

The recent disinflation is not unusual. Romer and Romer(1989) show that output typically falls after the Fed makes anexplicit decision to reduce inflation, which they interpret asevidence against real business cycle theory.

18

While the story of K

procyclical behavior of money,

behavior of prices. It is a

absence of identifiable real

changes, inflation tends to r

This famous Phillips curve co

the macroeconomic debate of t

empirical motivation for the

1970s. (Friedman 1968, Lucas

and Plosser generates procycli

transactions services, these f

associated with fluctuations i

fluctuations in prices. The

been left without an explanat

theorists6

it cannot explain the pro

documented fact that,

s such as the OPEC oil

n booms and fall in re

tion played a central

60s, and it was the pr

new classica

Yet since

money through the demand

luctuations in money will be

n real balances not with

short-run Phillips curve

ion by real business cycle

6Indeed, as King and Plosser point out, their model makes

the counterfactual prediction that the price level should becountercyclical: since the demand for real outside money probablyrises in a boom, and it is the outside money stock that pins downthe price level, equilibrium in the market for outside moneyrequires that the price level fall in a boom.

19

demand for transactions services; the banking system responds by

creating more inside money. Hence, the procyclical behavior of

standard monetary aggregates cannot necessarily be interpreted as

evidence that changes in outside money caused by the monetary

authority have real effects.

ing and Plosser can explain the

wel

sho

ise

rre

he

ea

1

1

I-

La

19

ny972

cal

cyclical

in the

price

ce S S ions

role in

ima ry

1 theories

the model o

in the

f King

for

has thus

The Tradeoff between Internal, and External Consistency

A good theory has two characteristics: internal consistency

and external consistency. An internally consistent theory is one

that is parsimonious; it invokes no ad hoc or peculiar axioms.

An externally consistent theory is one that fits the facts; it

makes empirically refutable predictions that are not refuted.

All scientists, including economists, strive for theories that

are both internally and externally consistent. Yet like all

optimizing agents, scientists face tradeoffs. One theory may be

more "beautiful," while another may be easier to reconcile with

observation.

The choice between alternative theories of the business

cycle--in particular, between real business cycle theory and new

Keynesian theory- - is partly a choice between internal and

external consistency. Real business cycle theory extends the

Wairasian paradigm, the most widely understood and taught model

in economics, and provides a unified explanation for economic

growth and economic fluctuations. New Keynesian theory, in its

attempt to mimic the world more accurately, relies on nominal

rigidities that are observed but only little understood. Indeed,

new Keynesians sometimes suggest that to understand the business

cycle, it may be necessary to reject the axiom of rational,

optimizing individuals, an act which for economists would be the

ultimate abandonment of internal consistency.

The tension between these two goals of science will

undoubtedly continue. Each school of macroeconomic thought will

20

highlight its strengths while trying to improve on its

weaknesses. My own forecast is that real business cycle

advocates will not manage to produce convincing evidence that

there are substantial shocks to technology and that leisure is

highly substitutable over time. Without such evidence, their

theories will be judged as not persuasive. New Keynesians,

however, have made substantial progress in recent years toward

providing rigorous microeconomic foundations, the absence of

which was the fatal flaw of the Keynesian consensus of the l960s.

Jhile real business cycle theory has served the important

function of stimulating and provoking the scientific debate, it

will, I predict, ultimately be discarded as an explanation of

observed fluctuations.

21

References

Abraham, Katharine G

Unemployment: Sec

Journal of Politi

Altonji, Joseph G.,

Supply: Evidence

Economy, June 198

Ball, Laurence,

Labor Supply

115.

• and Lawrence F. Katz, "Cyclical

toral Shifts or Aggregate Disturbances?"

cal Economy, June 1986, jj, 507-522.

"Intertemporal Substitution in Labor

from Micro Data," Journal of Political

6, Part 2, 2. Sl76-S2l5.

Substitution and Constraints on

Panel Data," manuscript, M.I.T.,

Barro, Robert J

Bruno, Michael,

Staf].ption,

Dornbusch, Rudiger,

ley, 1987.

f Worldwide

"Inter temporal

Evidence from

1985

Barro

the

Robert J

United S

• ,"Unanticipated Mon

tates," American Ec

ey Growth and UnempL

pnomic Review. 1977,

oyme rtt in101-

Macroeconomics, New York: Wi

and Jeffrey Sachs, Economics o

Cambridge, MA: Harvard Unive

and Stanley Fischer, Mac

rsity Press,

roeconomics,

1985.

New

York: McGraw-Hill, 1987.

Friedman, Milton, "The Role of Monetary Policy." American

Economic Review, 1968, , 1-17.

Friedman, Milton, and Anna Schwartz, A Monetary History of the

United States, Princeton, NJ: Princeton University Press,

1963.

Hall, Robert E. , "Market Structure and Macroeconomic

Fluctuations," Byookjngs Papers on Economic Activity,

1987:1, 285-322.

22

Hamilton, James D., "Oil and the Macroeconomy since World

War II," Journal of Political Economy, April 1983, 2.1,

228-248.

King, Robert G., and Charles I. Plosser, "Money, Credit, and

Prices in a Real Business Cycle," American Economic Review,

June 1984, j, 363-380.

Lilien, David M., "Sectoral Shifts and Cyclical Unemployment,"

Journal of Political Economy, August 1982, Q., 777-793.

Long, John B. Jr., and Charles I. Plosser, "Real Business

Cycles," Journal of Political Economy, February 1983, jJ,,

39-69.

Lucas, Robert E. Jr., "Expectations and the Neutrality of Money,

Journal of Economjc Theory, 1972, , 103-124.

Lucas, Robert E. Jr., "International Evidence on Output-Inflatjc

Tradeoffs," American Economic Review, 1973, 326-334.

Murphy, Kevin M. , and Robert H. Topel, "The Evolution of

Unemployment in the United States: 1968-1985,"

NBER Macroeconomics Annual, 1987.

Patinkin, Don, Money. Interest, and Prices: An Interation of

Monetary and Value Theory, Evanston, I1l.:Row, Peterson, 1956

Prescott, Edward, "Theory Ahead of Business Cycle Measurement,"

CarneEie-Rochester Conference on Public Policy, Autumn 1986,

21, 11-44.

Rotner, Christina, and David Romer, "Does Monetary Policy

Matter: A New Test in the Spirit of Friedman and Schwartz,"

NBER Macroeconomics Annual, 1989, forthcoming.

23

</ref_section>

PE

RC

EN

T

PE

R

YE

AR

0

P1

I I

0 r%)

- - co

o

Lfl

c-f) 0

I- 0

xl P

1 (1

cr,

(0 N

) (0

0

0 -1

-D

-1 C

-) xl

0 -1