Ebook Macroeconomics (6th edition): Part 2

•

u S l n ess
C

s an
•

acroeco n o m l c
•

o IC


CHAPTER
•

•

•

u S l n ess

c es

ince the Industrial Revolution, the economies of the United States and many
other countries have grown tremendously. That growth has transformed
economies and greatly improved living standards. Yet even in prosperous countries,
economic expansion has been periodically interrupted by episodes of declining
production and income and rising unemployment. Sometimes fortunately, not
very often these episodes have been severe and prolonged. But whether brief or

more extended, declines in economic activity have been followed almost invariably
by a resumption of economic growth.
This repeated sequence of economic expansion giving way to temporary
decline followed by recovery, is known as the business cycle. The business cycle is a
central concern in macroeconomics because business cycle fluctuations the ups
and downs in overall economic activity are felt throughout the economy. When
the economy is growing strongly, prosperity is shared by most of the nation's
industries and their workers and owners of capital. When the economy weakens,
many sectors of the economy experience declining sales and production, and
workers are laid off or forced to work only part-time. Because the effects of busi­
ness cycles are so widespread, and because economic downturns can cause great
hardship, economists have tried to find the causes of these episodes and to deter­
mine what, if anything, can be done to counteract them. The two basic questions of
(1) what causes business cycles and (2) how policymakers should respond to cyclical
fluctuations are the main concern of Part 3 of this book.
The answers to these two questions remain highly controversial. Much of this
controversy involves the proponents of the classical and Keynesian approaches to
macroeconomics, introduced in Chapter 1 . In brief, classical economists view busi­
ness cycles as generally representing the economy's best response to disturbances
in production or spending. Thus classical economists do not see much, if any, need
for government action to counteract these fluctuations. In contrast, Keynesian econ­
omists argue that, because wages and prices adjust slowly, disturbances in pro­
duction or spending may drive the economy away from its most desirable level of
output and employment for long periods of time. According to the Keynesian
view, government should intervene to smooth business cycle fluctuations.
We explore the debate between classicals and Keynesians, and the implications
of that debate for economic analysis and macroeconomic policy, in Chapters 9-11. In
this chapter we provide essential background for that discussion by presenting the
basic features of the business cycle. We begin with a definition and a brief history of
282

8.1

What Is a Business Cycle?

283

the business cycle in the United States. We then turn to a more detailed discussion of
business cycle characteristics, or "business cycle facts." We conclude the chapter
with a brief preview of the alternative approaches to the analysis of business cycles.

8.1

What Is a B us i ness Cycle?
Countries have experienced ups and downs in overall economic activity since they
began to industrialize. Economists have measured and studied these fluctuations
for more than a century. Marx and Engels referred to "commercial crises," an early
term for business cycles, in their Communist Manifesto in 1848. In the United
States, the National Bureau of Economic Research (NBER), a private nonprofit
organization of economists founded in 1920, pioneered business cycle research. The
NBER developed and continues to update the business cycle chronology, a
detailed history of business cycles in the United States and other countries. The
NBER has also sponsored many studies of the business cycle: One landmark study
was the 1946 book Measuring Business Cycles, by Arthur Burns (who served as Fed­
eral Reserve chairman from 1970 until 1978) and Wesley Mitchell (a principal
founder of the NBER). This work was among the first to document and analyze the
empirical facts about business cycles. It begins with the following definition:

Business cycles are a type of fluctuation found in the aggregate economic activity of

nations that organize their work mainly in business enterprises. A cycle consists of
expansions occurring at about the same time in many economic activities, followed by
similarly general recessions, contractions, and revivals which merge into the expansion
phase of the next cycle; this sequence of changes is recurrent but not periodic; in dura­
tion business cycles vary from more than one year to ten or twelve years.'
Five points in this definition should be clarified and emphasized.
1.

Aggregate economic activity. Business cycles are defined broadly as fluctua­

tions of "aggregate economic activity" rather than as fluctuations in a single, specific
economic variable such as real GDP. Although real GDP may be the single variable
that most closely measures aggregate economic activity, Burns and Mitchell also
thought it important to look at other indicators of activity, such as employment and
financial market variables.
2. Expansions and contractions. Figure 8.1 a diagram of a typical business
cycle helps explain what Burns and Mitchell meant by expansions and contractions.
The dashed line shows the average, or nonnal, growth path of aggregate economic
activity, as determined by the factors we considered in Chapter 6. The solid curve
shows the rises and falls of actual economic activity. The period of time during which
aggregate economic activity is falling is a contraction or recession. If the recession is
particularly severe, it becomes a depression. After reaching the low point of the
contraction, the trough (T), aggregate economic activity begins to increase. The
period of time during which aggregate economic activity grows is an expansion or
a boom. After reaching the high point of the expansion, the peak (P), aggregate
economic activity begins to decline again. The entire sequence of decline followed by
recovery, measured from peak to peak or trough to trough, is a business cycle.
IBurns and Mitchell, Measuring Business Cycles, New York: National Bureau of Economic Research,
1946, p. 1.

284

Chapter 8

Business Cycles

Figure 8.1
A business cycle
The solid curve graphs
the behavior of aggregate
economic activity over a
typical business cycle.
The dashed line shows
the economy's normal
growth path. During a
contraction aggregate
economic activity falls
until it reaches a trough,
T. The trough is followed
by an expansion during
whkh economic activity
increases until it reaches
a peak, P A complete
cycle is measured from
peak to peak or trough
to trough.

.�

>

.�
-

v
..
v
.�

=
o
"

o
v

'"

•
•
•

�

•

Aggregate
economIC
activity

gs"

•
•
•

•

�

."
."

..:

•
•
•
•

·
..
-"'":
•

..

..

..

..

..

..

..

....

.

..

..

•
•
•
•
•

•
•

•

•

•

·

.

.

'

.

'

:
Expansion
:
-+
..;... :-..� :
'

p
'

:
:

•

Normal
growt h
p ath

•
•
•
.
.

:

.
•

•

•

T

..

..

..

•

•

.

..

..

..

..

.

•

.

....

....

.;.
•

•

•

•

•

•

•

•
•

•

•

•

•

•

•

•

•

•

•

•

•

:

•

•

.
•

.

•

.

•

Contraction :

� :-

Expansion
.;....

-

•

T
•

:
.
�:

-

.
.

p

Time

Figure 8.1 suggests that business cycles are purely temporary deviations from
the economy's normal growth path. However, part of the output losses and gains
that occur during a business cycle may become permanent.
Peaks and troughs in the business cycle are known collectively as turning points.
One goal of business cycle research is to identify when turning points occur. Aggre­
gate economic activity isn't measured directly by any single variable, so there's no
simple formula that tells economists when a peak or trough has been reached.2 In
practice, a small group of economists who fonn the NBER's Business Cycle Dating
Committee determine that date. The committee meets only when its members believe
that a turning point may have occurred. By examining a variety of economic data, the
committee determines whether a peak or trough has been reached and, if so, the
month it happened. However, the committee's announcements usually come well
after a peak or trough occurs, so their judgments are more useful for historical analy­

sis of business cycles than as a guide to current policymaking.
3. Comovement. Business cycles do not occur in just a few sectors or in just a
few economic variables. Instead, expansions or contractions "occur at about the
same time in many economic activities." Thus, although some industries are more
sensitive to the business cycle than others, output and employment in most indus­
tries tend to fall in recessions and rise in expansions. Many other economic vari­
ables, such as prices, productivity, investment, and government purchases, also have
regular and predictable patterns of behavior over the course of the business cycle.
2A conventional definition used by the media-that a recession has occurred when there are two
consecutive quarters of negative real GDP growth-isn't widely accepted by economists. The reason
that economists tend not to like this definition is that real GDP is only one of many possible indicators
of economic a ctivity
.


8.2

The American Business Cycle: The Historical Record

285

The tendency of many economic variables to move together in a predictable way
over the business cycle is called comovement.
4. Recurrent but not periodic. The business cycle isn't periodic, in that it does not
occur at regular, predictable intervals and doesn't last for a fixed or predetermined
length of time. (Box 8.1, p. 301, discusses the seasonal cycle or economic fluctuations
over the seasons of the year which, unlike the business cycle, is periodic.) Although
the business cycle isn't periodic, it is recurrent; that is, the standard pattern of
contraction-trough--€xpansion-peak recurs again and again in industrial economies.
5. Persistence. The duration of a complete business cycle can vary greatly,

from about a year to more than a decade, and predicting it is extremely difficult.
However, once a recession begins, the economy tends to keep contracting for a
period of time, perhaps for a year or more. Similarly, an expansion, once begun,
usually lasts a while. This tendency for declines in economic activity to be fol­
lowed by further declines, and for growth in economic activity to be followed by
more growth, is called persistence. Because movements in economic activity have
some persistence, economic forecasters are always on the lookout for turning
points, which are likely to indicate a change in the direction of economic activity.

8.2

The American B usiness Cycle: The H istorica l Record
An overview of American business cycle history is provided by the NBER's monthly
business cycle chronology? as summarized in Table 8.1. It gives the dates of the
troughs and peaks of the thirty-two complete business cycles that the U.s. economy
has experienced since 1854. Also shown are the number of months that each con­
traction and expansion lasted.
T h e P re-World Wa r I P e r i o d

The period between the Civil War (1861-1865) and World War I (1917-1918) was
one of rapid economic growth in the United States. Nevertheless, as Table 8.1
shows, recessions were a serious problem during that time. Indeed, the longest con­
traction on record is the 65-month-long decline between October 1873 and March
1879, a contraction that was worldwide in scope and is referred to by economic his­
torians as the Depression of the 1870s. Overall, during the 1854-1914 period the
economy suffered 338 months of contraction, or nearly as many as the 382 months
of expansion. In contrast, from the end of World War II in 1945 through October
2006, the number of months of expansion (627) outnumbered months of contraction
(104) by more than six to one.
T h e G re a t D e p re s s i o n a n d Wo r l d Wa r I I

The worst economic contraction in the history of the United States was the Great
Depression of the 1930s. After a prosperous decade in the 1920s, aggregate eco­
nomic activity reached a peak in August 1929, two months before the stock market
crash in October 1929. Between the 1929 peak and the 1933 trough, real GDP fell by
3For a detailed discussion of the NBER chronologies, see Geoffrey H. Moore and Victor Zarnowitz,
"The NBER's Business Cycle Chronologies," in Robert j. Gordon, ed., The American BlIsiness Cycle:
CO/llillllily a/ld Change, Chicago: University of Chicago Press, 1986. The NBER chronology is available
at the NBER's Web site, www.nber.org.


286

Chapter 8

Business Cycles

Table 8.1
NBER Business Cycle Turning Points and Durations of Post-1854 Business Cycles

Trough

Dec. 1854
Dec. 1858
june 1861
Dec. 1867
Dec. 1870
Mar. 1879
May 1885
Apr. 1888

May 1891
june 1894
june 1897
Dec. 1900
Aug. 1 904
june 1908
jan. 1912
Dec. 1914
Mar. 1919
july 1921
july 1924
Nov. 1927
Mar. 1933
june 1938
Oct. 1945
Oct. 1 949
May 1954
Apr. 1958
Feb. 1961
Nov. 1970
Mar. 1975
july 1980
Nov. 1982
Mar. 1991
Nov. 2001

Expansion
(months from
trough to peak)

Peak

30
22
46 (Civil War)
18
34
36
22
27
20
18
24
21
33
19
12
44 (WWI)
10
22
27
21
50
80 (WWII)
37
45 (Korean War)
39
24
106 (Vietnam War)
36

58
12
92
1 20

june 1857
Oct. 1 860
Apr. 1865
june 1869
Oct. 1 873
Mar. 1882
Mar. 1887
july 1 890
jan. 1893
Dec. 1895
june 1899
Sept. 1902
May 1907
jan. 1910
jan. 1 9 1 3
Aug. 1 9 1 8
jan. 1920
May 1923
Oct. 1926
Aug. 1929
May 1937
Feb. 1945
Nov. 1948
july 1953
Aug. 1957

Apr. 1960
Dec. 1969
Nov. 1973
jan. 1980
july 1981
july 1990
Mar. 2001

Contraction
(months from peak
to next trough)

18
8
32
18
65
38
13
10
17
18
18
23
13
24
23
7
18
14

13
43 (Depression)
1 3 (Depression)
8
11
10
8
10
11
16
6
16
8
8

Source: NBER Web site, www.nber. org/cycies.html.

nearly 30%. During the same period the unemployment rate rose from about 3% to
nearly 25%, with many of those lucky enough to have jobs only able to work part­
time. To appreciate how severe the Great Depression was, compare it with the two
worst post-World War II recessions of 1973-1975 and 1981-1982. In contrast to the
30% real GDP decline and 25% unemployment rate of the Great Depression, in the
1973-1975 recession real GDP fell by 3.4% and the unemployment rate rose from
about 4% to about 9%; in the 1981-1982 recession real GDP fell by 2.8% and the
unemployment rate rose from about 7% to about 11 %.
Although no sector escaped the Great Depression, some were particularly hard
hit. In the financial sector, stock prices continued to collapse after the crash. Depos­
itors withdrew their money from banks, and borrowers, unable to repay their

8.2

The American Business Cycle: The Historical Record

287

bank loans, were forced to default; as a result, thousands of banks were forced to go
out of business or merge with other banks. In agriculture, farmers were bankrupted
by low crop prices, and a prolonged drought in the Midwest turned thousands of
farm families into homeless migrants. Investment, both business and residential,
fell to extremely low levels, and a "trade war" in which countries competed in
erecting barriers to imports virtually halted international trade.
Although most people think of the Great Depression as a single episode, tech­
nically it consisted of two business cycles, as Table 8.1 shows. The contraction
phase of the first cycle lasted forty-three months, from August 1929 until March
1933, and was the most precipitous economic decline in U.s. history. After Franklin
Roosevelt took office as President in March 1933 and instituted a set of policies
known collectively as the New Deal, a strong expansion began and continued for
fifty months, from March 1933 to May 1937. By 1937 real GDP was almost back to
its 1929 level, although at 14% the unemployment rate remained high. Unemploy­
ment remained high in 1937 despite the recovery of real GDP because the number
of people of working age had grown since 1929 and because increases in produc­
tivity allowed employment to grow more slowly than output.
The second cycle of the Great Depression began in May 1937 with a contraction
phase that lasted more than a year. Despite a new recovery that began in June
1938, the unemployment rate was still more than 17% in 1939.
The Great Depression ended dramatically with the advent of World War II.
Even before the Japanese attack on Pearl Harbor brought the United States into the
war in December 1941, the economy was gearing up for increased armaments pro­
duction. After the shock of Pearl Harbor, the United States prepared for total war.

With production supervised by government boards and driven by the insatiable
demands of the military for more guns, planes, and ships, real GDP almost doubled
between 1939 and 1944. Unemployment dropped sharply, averaging less than 2%
of the labor force in 1943-1945 and bottoming out at 1.2% in 1944.
P ost-W o r l d W a r I I U . S . B usi n e s s Cyc l e s

As World War II was ending in 1945, economists and policymakers were concerned
that the economy would relapse into depression. As an expression of this concern,
Congress passed the Employment Act of 1946, which required the government to
fight recessions and depressions with any measures at its disposal. But instead of
falling into a new depression as feared, the U.s. economy began to grow strongly.
Only a few relatively brief and mild recessions interrupted the economic expansion
of the early postwar period. None of the five contractions that occurred between 1945
and 1970 lasted more than a year, whereas eighteen of the twenty-two previous cycli­
cal contractions in the NBER's monthly chronology had lasted a year or more. The
largest drop in real GDP between 1945 and 1970 was 3.3% during the 1957-1958 reces­
sion, and throughout this period unemployment never exceeded 8.1 % of the work force.
Again, there was a correlation between economic expansion and war: The 1949-1953
expansion corresponded closely to the Korean War, and the latter part of the strong
1961-1969 expansion occurred during the military buildup to fight the Vietnam War.
Because no serious recession occurred between 1945 and 1970, some economists
suggested that the business cycle had been "tamed," or even that it was "dead." This
view was especially popular during the 106-month expansion of 1961-1969, which
was widely attributed not only to high rates of military spending during the Vietnam
War but also to the macroeconomic policies of Presidents Kennedy and Johnson.


288

Chapter 8

Business Cycles

Some argued that policymakers should stop worrying about recessions and focus
their attention on inflation, which had been gradually increasing over the 1960s.
Unfortunately, reports of the business cycle's death proved premature. Shortly
after the Organization of Petroleum Exporting Countries (OPEC) succeeded in
quadrupling oil prices in the fall of 1973, the U.s. economy and the economies of
many other nations fell into a severe recession. In the 1973-1975 recession U.s.
real GDP fell by 3.4% and the unemployment rate reached 9% not a depression
but a serious downturn, nonetheless. Also disturbing was the fact that inflation,
which had fallen during most previous recessions, shot up to unprecedented
double-digit levels. Inflation continued to be a problem for the rest of the 1970s,
even as the economy recovered from the 1973-1975 recession.
More evidence that the business cycle wasn't dead came with the sharp
1981-1982 recession. This contraction lasted sixteen months, the same length as the
1973-1975 decline, and the unemployment rate reached 11 %, a postwar high. Many
economists claim that the Fed knowingly created this recession to reduce inflation,
a claim we discuss in Chapter 11. Inflation did drop dramatically, from about 11 %
to less than 4% per year. The recovery from this recession was strong, however.
T h e " Lo n g B o o m "

The expansion that followed the 1981-1982 recession lasted almost eight years,
until July 1990, when the economy again entered a recession. This contraction was
relatively short (the trough came in March 1991, only eight months after the peak)
and shallow (the unemployment rate peaked in mid 1992 at 7.7% not particularly
high for a recession). Moreover, after some initial sluggishness, the 1990-1991 reces­
sion was followed by another sustained expansion. Indeed, in February 2000, after
107 months without a recession, the expansion of the 1990s became the longest in
U.s. history, exceeding in length the Vietnam War-era expansion of the 1960s.

Taking the expansions of the 1980s and 1990s together, you can see that the U.s.
economy experienced a period of more than eighteen years during which only
one relatively minor recession occurred. Some observers referred to this lengthy
period of prosperity as the "long boom." The long boom ended with the business
cycle peak in March 2001, after which the U.s. economy suffered a mild recession
and sluggish growth.
H a ve A m e r i c a n B us i n e s s Cyc l es B e c o m e Less S e v e r e ?

Until recently, macroeconomists believed that, over the long sweep of history, business
cycles generally have become less severe. Obviously, no recession in the United States
since World War II can begin to rival the severity of the Great Depression. Even
putting aside the Great Depression, economists generally believed that business
downturns before 1929 were longer and deeper than those since 1945. According to
the NBER business cycle chronology (Table 8.1), for example, the average contraction
before 1929 lasted nearly twenty-one months and the average expansion lasted
slightly more than twenty-five months. Since 1945, contractions have shortened to an
average of eleven months, and expansions have lengthened to an average of fifty
months, even excluding the lengthy expansion of the 1990s. Standard measures of eco­
nomic fluctuations, such as real GDP growth and the unemployment rate, also show
considerably less volatility since 1945, relative to data available for the pre-1929 era.


8.2

The American Business Cycle: The Historical Record

289

Since World War II a major goal of economic policy has been to reduce the size
and frequency of recessions. If researchers found contrary to the generally accepted

view-that business cycles had not moderated in the postwar period, serious doubt
would be cast on the ability of economic policymakers to achieve this goal. For this
reason, although the question of whether the business cycle has moderated over time
may seem to be a matter of interest only to economic historians, this issue is of great
practical importance.
Thus Christina Romer, now at the University of California at Berkeley, sparked
a heated controversy by writing a series of articles denying the claim that the busi­
ness cycle has moderated over time 4 Romer's main point concerned the dubious
quality of the pre-1929 data. Unlike today, in earlier periods the government didn't
collect comprehensive data on economic variables such as GDP. Instead, economic
historians, using whatever fragmentary information they could find, have had to
estimate historical measures of these variables.
Romer argued that methods used for estimating historical data typically over­
stated the size of earlier cyclical fluctuations. For example, widely accepted estimates
of pre-1929 GNp5 were based on estimates of just the goods-producing sectors of the
economy, which are volatile, while ignoring less-volatile sectors such as wholesale and
retail distribution, transportation, and services. As a result, the volatility of GNP was
overstated. Measured properly, GNP varied substantially less over time than the official
statistics showed. Romer's arguments sparked additional research, though none proved
decisively whether volatility truly declined after 1929. Nonetheless, the debate served
the useful purpose of forcing a careful reexamination of the historical data.
New research shows that economic volatility declined in the mid 1980s and has
remained low since then. Beca use the quality of the data is not an issue for the
period following World War II, the decline in volatility in the mid 1980s, relative to
the preceding forty years, probably reflects a genuine change in economic volatility
rather than a change in how economic data are produced.
Other economic variables, including inflation, residential investment, output of
durable goods, and output of structures, also appear to fluctuate less in the past
twenty years than they did in the preceding forty years. Research by James Stock of
Harvard University and Mark Watson of Princeton University6 shows that the

volatility, as measured by the standard deviation of a variable, declined by 20 to 40%
for many of the twenty-one variables they examine, including a decline of 33% for
real GDP, 27% for employment, and 50% for inflation. Because the decline in
volatility of macroeconomic variables has been so widespread, economists have
dubbed this episode "the Great Moderation."?
4The articles included "Is the Stabilization of the Postwar Economy a Figment of the Data?" American
Economic Review, June 1986, pp. 314-334; "The Prewar Business Cycle Reconsidered: New Estimates of
Gross National Product, 1869-1908," journal of Political Economy, February 1989, pp. 1-37; and "The
Cyclical Behavior of Individual Production Series, 1889-1984," Quarterly joumal of Econol1lics, February
1991, pp. 1-3l.
sAs discussed in Chapter 2, unti1 1991 the U.s. national income and product accounts focused on
GNP rather than GOP. As a result, studies of business cycle behavior have often focused on GNP
rather than GOP
6"Has the Business Cycle Changed and Why?" NBER MacroecOIlOmics Al1IllIa l 2002 (Cambridge, MA:
MIT Press, 2002), pp. 159-218.
7 See Ben S. Bernanke, "The Great Moderation," Speech at the Eastern Economic Association meetings,
February 20, 2004, available at wwwfederalreserve.gov.


290

Chapter 8

Business Cycles

Somewhat surprisingly, the reduction in volatility seemed to corne from a
sudden, one-time drop rather than a gradual decline. The break seems to have
corne around 1984 for many economic variables, though for some variables the
break occurred much later.
What accounts for this reduction in the volatility of the economy? Stock and

Watson found that better monetary policy is responsible for about 20% to 30% of
the reduction in output volatility, with reduced shocks to the economy's produc­
tivity accounting for about 15% and reduced shocks to food and commodity prices
accounting for another 15%. The remainder is attributable to some unknown form
of good luck in terms of smaller shocks to the economy.s

8.3

Business Cycle Facts
Although no two business cycles are identical, all (or most) cycles have features in
common. This point has been made strongly by a leading business cycle theorist,
Nobel laureate Robert E. Lucas, Jr., of the University of Chicago:

Though there is absolutely no theoretical reason to anticipate it, one is led by the facts
to conclude that, with respect to the qualitative behavior of comovements among
series [that is, economic variables], business cycles are all alike. To theoretically inclined
economists, this conclusion should be attractive and challenging, for it suggests the
possibility of a unified explanation of business cycles, grounded in the general laws
governing market economies, rather than in political or institutional characteristics
specific to particular countries or periods?
Lucas's statement that business cycles are all alike (or more accurately, that they
have many features in common) is based on examinations of comovements among eco­
nomic variables over the business cycle. In this section, we study these comovements,
which we call business cycle facts, for the post-World War II period in the United
States. Knowing these business cycle facts is useful for interpreting economic data and
evaluating the state of the economy. In addition, they provide guidance and discipline
for developing economic theories of the business cycle. When we discuss alternative
theories of the business cycle in Chapters 10 and 11, we evaluate the theories principally
by determining how well they account for business cycle facts. To be successful, a
theory of the business cycle must explain the cyclical behavior of not just a few vari­

ables, such as output and employment, but of a wide range of key economic variables.
The Cyc l i c a l B e h a v i o r of E c o n o m i c Va r i a b l e s : D i re c t i o n a n d Ti m i ng

Two characteristics of the cyclical behavior of macroeconomic variables are impor­
tant to our discussion of the business cycle facts. The first is the direction in which
SSince the Stock and Watson paper was written, much additional research has been undertaken, with
mixed results. For example, Shaghil Ahmed, Andrew Levin, and Beth Ann Wilson ["Recent Improve­
ments in U.s. Macroeconomic Stability: Good Policy, Good Practices, or Good Luck?" Review of Eco­
Ilomics alld Statistics, vol. 86 (2004), pp. 824-8321 suggest that good luck played the biggest role, while
others find a larger role for monetary policy, including Peter M. Summers ["What Caused the Great
Moderation? Some Cross-Country Evidence," Federal Reserve Bank of Kansas City, Economic Review
(Third Quarter 2005), pp. 5-321.
'Robert E. Lucas, Jr., "Understanding Business Cycles," in K. Brunner and A. H. Meltzer, eds., Camegie­
Rochester COllferellce Series 011 Public Policy, vol. 5, Autumn 1977, p. 10.


8.3

Business Cycle Facts

291

a macroeconomic variable moves, relative to the direction of aggregate economic
activity. An economic variable that moves in the same direction as aggregate economic
activity (up in expansions, down in contractions) is procyclical. A variable that
moves in the opposite direction to aggregate economic activity (up in contractions,
down in expansions) is countercyclical. Variables that do not display a clear pattern
over the business cycle are acyclical.
The second characteristic is the timing of the variable's turning points (peaks and
troughs) relative to the turning points of the business cycle. An economic variable is

a leading variable if it tends to move in advance of aggregate economic activity. In
other words, the peaks and troughs in a leading variable occur before the corre­
sponding peaks and troughs in the business cycle. A coincident variable is one
whose peaks and troughs occur at about the same time as the corresponding business
cycle peaks and troughs. Finally, a lagging variable is one whose peaks and troughs
tend to occur later than the corresponding peaks and troughs in the business cycle.
The fact that some economic variables consistently lead the business cycle sug­
gests that they might be used to forecast the future course of the economy. Some
analysts have used downturns in the stock market to predict recessions, but such
an indicator is not infallible. As Paul Samuelson noted: "Wall Street indexes pre­
dicted nine out of the last five recessions." lO The idea that recessions can be forecast
also underlies the index of leading indicators, discussed in the box, "In Touch with
the Macroeconomy: Leading Indicators," on page 292.
In some cases, the cyclical timing of a variable is obvious from a graph of its
behavior over the course of several business cycles; in other cases, elaborate statis­
tical techniques are needed to determine timing. Conveniently, The Conference
Board has analyzed the timing of dozens of economic variables. This information
is published monthly in Business Cycle Indicators, along with the most recent data
for these variables. For the most part, in this chapter we rely on The Conference
Board's timing classifications.
Let's now examine the cyclical behavior of some key macroeconomic variables.
We showed the historical behavior of several of these variables in Figs. 1.1-1.4.
Those figures covered a long time period and were based on annual data. We can
get a better view of short-run cyclical behavior by looking at quarterly or monthly
data. The direction and timing of the variables considered are presented in Summary
table 10 on page 293.
Production

Because the level of production is a basic indicator of aggregate economic activity,
peaks and troughs in production tend to occur at about the same time as peaks and

troughs in aggregate economic activity. Thus production is a coincident and pro­
cyclical variable. Figure 8.2, p. 294, shows the behavior of the industrial production
index in the United States since 1947. This index is a broad measure of production
in manufacturing, mining, and utilities. The vertical lines P and T in Figs. 8.2-8.8
indicate the dates of business cycle peaks and troughs, as determined by the NBER
(see Table 8.1). The turning points in industrial production correspond closely to
the turning points of the cycle.
lONewsweek column, September 19, 1966, as quoted in John Bartlett, Bartlett's Familiar QltDtatiDIlS, Boston:

Little Brown, 2002.


Chapter 8

292

a

Business Cycles

W I T H T H E M A C RO E C O N O M Y
Lea d i n g I n d i cators

:J
o Many different economic variables are considered to
I-

lead the business cycle, but because none give an exact
indication of when a turning point may arrive, econo­
Z

- mists have spent considerable effort trying to determine
if a combination of those variables can help indicate
when a peak or trough may occur.
The first such index was originally developed in
1938 at the National Bureau of Economic Research
(NBER) by Wesley Mitchell and Arthur Burns: whose
important early work on business cycles was mentioned
earlier in this chapter. The NBER's work was made offi­
cial by the U.s. Department of Commerce, which first
published the "composite index of leading indicators"
in its publication Business Conditions Digest in November
1968. In 1995, the Commerce Department passed the
composite index back to the private sector, and it is now
produced by The Conference Board.
Although the composite index of leading indicators
was designed to predict the onset of recessions and expan­
sions, its history is spotty. When the index declines for
two or three consecutive months, it warns that a recession
is likely. However, its forecasting acumen in real time has
not been very good because of the following problems:
1.

2.
3.
4.

Data on the components of the index are often
revised when more complete data become avail­
able. Revisions change the value of the index and
may even reverse a signal of a future recession.

The index is prone to giving false signals, predict­
ing recessions that did not materialize.
The index does not provide any information on when
a recession might arrive or how severe it might be.
Changes in the structure of the economy over time
may cause some variables to become better predic­
tors of the economy and others to become worse.
For this reason, the index must be revised periodi­
cally, as the list of component indicators is changed.

Research by Francis Diebold and Glenn Rudebusch
showed that the revisions were substantia!." The
agency calculating the index (the Commerce Depart­
ment or The Conference Board) often demonstrates the
value of the index with a plot of the index over time,
showing how it turns down just before every recession.

But Diebold and Rudebusch showed that such a plot is
illusory because the index plotted was not the one used
at the time of each recession, but rather a revised index
made many years after the fact. In real time, they con­
cluded, the use of the index does not improve forecasts
of industrial production.
For example, suppose you were examining the
changes over time in the composite index of leading
indicators, and used the rule of thumb that a decline in
the index for three months in a row meant that a reces­
sion was likely in the next six months to one year. You
would have noticed in December 1969 that the index
had declined two months in a row; by January 1970 you

would have seen the third monthly decline. In fact, the
NBER declared that a recession had begun in December
1969, so the index did not give you any advance warn­
ing. Even worse, if you had been following the index in
1973 to 1974, you would have thought all was well until
September 1974, when the index declined for the second
month in a row, or October 1974, when the third monthly
decline occurred. But the NBER declared that a reces­
sion had actually begun in November 1973, so the index
was nearly a year late in calling the recession.
After missing a recession's onset so badly, the cre­
ators of the index naturally want to improve it. So, they
may revise the index with different variables, give the
variables different weights, or manipulate the statistics so
that, if the new revised index had been available, it would
have indicated that a recession were coming. For exam­
ple, the revised index published in April 1979 would
have given eight months of lead time before the recession
that began in December 1969 and six months of lead time
before the recession that began in November 1973. But of
course, that index was not available to forecasters when it
would have been useful-before the recessions began.
Because of the problems of the official composite
index of leading indicators, James Stock and Mark
Watson'" set out to create some new indexes that would
improve the value of such indexes in forecasting. They
created several experimental leading indexes, with the
hope that such indexes would prove better at helping
economists forecast turning points in the business cycle.
However, it appears that the two most recent recessions

(Continued)

'Statistical Illdicators of Cyclical Revivals (New York: National Bureau of Economic Research, 1938).
" "Forecasting Output with the Composite Leading Index: A Real-Time Analysis." JOllrnal of the Americall Statistical Associatiol1
(September 1991), pp. 603-610.
"'*""New Indexes of Coincident and Leading Economic Indicators," in Olivier J. Blanchard and Stanley Fischer, eds., NBER
Macroecol1omics Amlllal, 1989 (Cambridge, MA: MIT Press, 1989), pp. 351-394.


8.3

have been sufficiently different from earlier recessions
that the experimental leading indexes did not suggest an
appreciable probability of recession in either 1990 or 2001.
In early 1990, the experimental recession index of Stock
and Watson showed that the probability that a recession
would occur in the next six months never exceeded 10%.
Also, in late 2000 and early 2001, the index did not rise
above 10%. So, although the Stock and Watson approach

Business Cycle Facts

293

appeared promising in prospect, it did not deliver any
improvement in forecasting recessions.
The inability of leading indicators to forecast reces­
sions may simply mean that recessions are often unusual
events, caused by large, unpredictable shocks such as
disruptions in the world oil supply. If so, then the pur­

suit of the perfect index of leading indicators may prove
to be frustrating.

Although almost all types of production rise in expansions and fall in reces­
sions, the cyclical sensitivity of production in some sectors of the economy is
greater than in others. Industries that produce relatively durable, or long-lasting,
goods houses, consumer durables (refrigerators, cars, washing machines), or cap­
ital goods (drill presses, computers, factories) respond strongly to the business
cycle, producing at high rates during expansions and at much lower rates during
recessions. In contrast, industries that produce relatively nondurable or short-lived

S U M M A RY 1 0
The Cyclical Behavior of Key Macroeconomic Variables (The Business Cycle Facts)
Variable

Direction

Timing

Procyclical

Coincident

Production
Industrial production

Durable goods industries are more volatile than nondurable goods and services
Expenditure
Consumption

Procyclical

Coincident

Business fixed investment

Procyclical

Coincident

Residential investment

Procyclical

leading

Inventory investment

Procyclical

leading

Government purchases

Procyclical

'

-

Investment is more volatile than consumption
labor Market Variables
Employment

Procyclical

Coincident

Unemployment

Countercyclical

Unclassified b

Average labor productivity

Procyclical

leadinga

Real wage

Procyclical

-

Money supply

Procyclical

leading

Inflation

Procyclical

lagging

Stock prices

Procyclical

leading

Nominal interest rates

Procyclical

lagging

Real interest rates

Acyclical

'

Money Supply and Inflation

FinanciaL VariabLes

'

-

3Timing is not designated by The Conference Board.
bOesignated as "unclassified" by The Conference Board.

Source: Business Cycle Indicators. April 2003. Industrial production: series 47 (industrial production); consumption: series 57 (manufacturing

and trade sales. constant dollars); business fixed investment: series 86 (gross private nonresidential fixed investment); residential investment:

series 29 (new private housing units started); inventory investment: series 30 (change in business inventories. constant dollars); employment:
series 41 (employees on nonagricultural payrolls); unemployment: series 43 (civilian unemployment rate); money supply: series 106 (money
supply M2. constant dollars); inflation: series 120 ((PI for services. change over six-month span); stock prices: series 19 (index of stock prices.
500 common stocks); nominal interest rates: series 1 1 9 (Federal funds rate). series 1 14 (discount rate on new 91 -day Treasury bills). series
109 (average prime rate charged by banks).


294

Chapter 8

Business Cycles

Figure 8.2
Cyclical behavior of
the index of industrial
production
The index of industrial
production, a broad mea·

sure of production in
manufacturing, mining,
and utilities, is procycli­
cal and coincident. The
peaks and troughs of the
business cycle are shown
by the vertical lines P
and T. The shaded areas
represent recessions.
SOllrce: Federal Reserve Bank
of 51. Louis FRED database at

x N
QJ
"e o
O
C
N
.�

c
o

.�

t:

..g

140

.S

PT

PT

PT P T

PT

P T

PTPT

PT

PT

°

� 1 20

II
o "

� "
Q. - ..

.. ;. 1 00

"t:: x

- "
�" "e

C
"e .�

c -

80

60

40

I N DU ST RIA L
PR O DU CTIO N
I N D EX

researell.sflOllisfcd.orglfred2/series/
fNDPRO.

20

o �� � �
�
�
�

�
�
�
�
�
�
�
�
�
�
�
�
�
�
�
�
�
�
�
1 950 1955 1960 1 965 1970 1975 1 980 1985 1990 1 995 2000 2005
__

Year

goods (foods, paper products) or services (education, insurance) are less sensitive
to the business cycle.
Ex p e n d it u r e

For components of expenditure, as for types of production, durability is the key to
determining sensitivity to the business cycle. Figure 8.3 shows the cyclical behavior of

consumption of nondurable goods, consumption of services, consumption expendi­
tures on durable goods, and investment. Investment is made up primarily of spending
on durable goods and is strongly procyclical. In contrast, consumption of nondurable
goods and consumption of services are both much smoother. Consumption expendi­
tures on durable goods are more strongly procyclical than consumption expenditures
on nondurable goods or consumption of services, but not as procyclical as investment
expenditures. With respect to timing, consumption and investment are generally coin­
cident with the business cycle, although individual components of fixed investment
vary in their cyclical timing. ll
One component of spending that seems to follow its own rules is inventory
investment, or changes in business inventories (not shown), which often displays
large fluctuations that aren't associated with business cycle peaks and troughs. In
general, however, inventory investment is procyclical and leading. Even though
goods kept in inventory need not be durable, inventory investment is also very
volatile. Although, on average, inventory investment is a small part (about 1%) of

11Summary table 10 shows that residential investment leads the cycle.


8.3

Business Cycle Facts

295

Figure 8.3
Cyclical behavior of
consumption and
investment
Both consumption and

investment are procycli­
cal. However, investment
is more sensitive than
consumption to the busi­
ness cycle, reflecting the
fact that durable goods
are a larger part of invest­
ment spending than they
are of consumption
spending. Similarly,
expenditures on con­
sumer durables are more
sensitive to the business
cycle than is consumption
of nondurable goods or
servICes.

-

�
"
� 6000
<1/
-

E�

- 0
g) ""

> 0
" 0

:; ;:::
"

'"

c:

""
<1/

PT

PT

PT

PT P T

PT

PT

.5

-..= ..=

o

5000

'"

E
';:
= 0

4000

�
�
" C

.::!
U =
.
6
o

-

3000

2000

CONS U M PT I ON O F
S ERVI C ES
CONS U M PTI ON O F

NON DURA B LE
G OO D S

I N VE ST ME NT

1 000

PE N DIT URE ON
DURA B LE G OO D S

•

Source: Federal Reserve Bank
of St. Louis FRED database at
research.stlollisfed.orglfred2 series
PCDGCC96 (durable goods),
PCNDGC96 (nondurable
goods), PCESVC96 (services),
and GPDICl (investment).

Year

total spending, sharp declines in inventory investment represented a large part of
the total decline in spending in some recessions, most notably those of 1973-1975,
1981-1982, and 200l.
Government purchases of goods and services generally are procyclical. Rapid
military buildups, as during World War II, the Korean War, and the Vietnam War,
are usually associated with economic expansions.
E m p l oy m e n t a n d U n e m p l o y m e n t

Business cycles are strongly felt in the labor market. In a recession, employment grows
slowly or falls, many workers are laid off, and jobs become more difficult to find.
Figure 8.4 shows the number of civilians employed in the United States since
1955. Employment clearly is procyclical, as more people have jobs in booms than in
recessions, and also is coincident with the cycle.
Figure 8.5 shows the civilian unemployment rate, which is the fraction of the
civilian labor force (the number of people who are available for work and want to
work) that is unemployed. The civilian unemployment rate is strongly counter­
cyclical, rising sharply in contractions but falling more slowly in expansions.
Although The Conference Board has studied the timing of unemployment, Summary
table 10 shows that the timing of this variable is designated as "unclassified,"
owing to the absence of a clear pattern in the data. Figures 8.4 and 8.5 illustrate
a worrisome change in the patterns of recent recessions: namely in both the
1990-1991 and 2001 recessions, employment growth stagnated and unem­
ployment tended to rise for some time even after the recession's trough was


296

Chapter 8

Business Cycles

Figure 8.4
Cyclical behavior of
civilian employment
Civilian employment is
procycUcal and coinci­
dent with the business
cycle.

Source: Federal Reserve Bank
of St. Louis FRED database at

-

�

�

-

1 60

e g.9 � 1 50
""
0
e" =�
"tIS ;':::0 1 40
..
;: e
1 30

PT

>' ''

-

U
.

PT

P T

PT

PTP T

PT

PT

-

.

-

researdt.stlouisfed.orglfred2/seriesl
CE160V:

1 20

1 10

1 00

90

C IVI L IAN
EMP LOYME NT

80

70
60 L__
-L
__
__
__
�
__
__
�
__
__
L__
-L
__
__
__
�
__
__
-L
__
__
LL�
1 95 5

1 960
1 965
1 970
1 975
1 980
1 985
1 990
1 995
2000
2005

Year

Figure 8.5
Cyclical behavior of
the unemployment rate
The unemployment rate
is countercyclical and
very sensitive to the
business cycle. Its timing
pattern relative to the
cycle is unclassified,
meaning that it has no
definite tendency to lead,
be coincident, or lag.

"

�

-

C

� "
u
�
= "
"

_

12

e>, .B- l l

PT

PT

PTP T

P T

PT

PT

o

Q.

-

e"

=
::>

10

9

8

Source: Federal Reserve Bank
of St. Louis FRED database at

researdt.stlollisfed.orglfred2/seriesl

7

UNRATE.

UNEMPLOYMENT
RAT E

6

5

4
3 -L
1 960

�

__
__

__
�

__

1 965

-L

__

1 970

�

__
__

1 975

L-

__
__

1 980

1 985

�

__

-L

__
__

1 990

�

__
__

1 995

�
__

__

2000

2005

Year


8.3

Business Cyde Facts

297

Figure 8.6
Cyclical behavior
of average labor
productivity and
the real wage
Average labor produc­
tivity, measured as real
output per employee
hour in the nonfarm
business sector, is pro­
cyclical and leading. The
economywide average
real wage is mildly pro­
cyclical, and its growth
slowed sharply between

1973 and 1997.

�8
> '"'

.•
••

u
=
""
0
�

0

"
'" 1 40
",
'"
'"'

", �

� "
o x
,Q "

..
"

""

"
..

eo "
eo

..
�

1 50

..

� �

< ..
�

PT

P T

PT P T

PT

PT

1 30
1 20
110

""
:; 1 00

90

Source: Federal Reserve Bank

80

of St. Louis FRED database at
research st/Ollis/ed.orglfred2 series
OPHNFB (productivity) and
COMPRNFB (real wage).

70

.

PT

REA L WAGE

AVERAGE
LAB O R
PRO DU CTIVITY

60
5 0 �������--��--�������--��--�����
1 975
1 965
1 970
1 985
1 990 1 995 2000 2005
1 960
1 980
Year

reached. This pattern led observers to refer to the recovery period following
these recessions as "jobless recoveries." Fortunately, both of these recoveries even­
tually gained strength and the economy showed employment growth and a
decline in the rate of unemployment.
Ave r a g e L a b o r P r o d u ctiv ity a n d t h e R e a l W a g e

Two other significant labor market variables are average labor productivity and the
real wage. As discussed in Chapter 1, average labor productivity is output per unit of
labor input. Figure 8.6 shows average labor productivity measured as total real output
in the u.s. economy (excluding farms) divided by the total number of hours worked
to produce that output. Average labor productivity tends to be procydical: In booms
workers produce more output during each hour of work than they do in recessions.12
Although The Conference Board doesn't deSignate the timing of this variable, studies
show that average labor productivity tends to lead the business cycle.B
Recall from Chapter 3 that the real wage is the compensation received by workers
per unit of time (such as an hour or a week) measured in real, or purchasing-power,
12The Application in Chapter 3, "The Production Function of the U.s. Economy and u.s. Productivity
Growth," p. 64, made the point that total factor productivity A also tends to be procyclical.

"See Robert J . Gordon, "The 'End of Expansion' Phenomenon in Short-Run Productivity Behavior,"
Brookil1gs Papers 011 Ecol1omic Activity, 1979:2, pp. 447-461.


298

Chapter 8

Business Cycles

Figure 8.7
Cyclical behavior
of nominal money
growth and inflation
Nominal money growth,
here measured as the six­
month m ov ing average
of monthly growth rates
in M2 (expressed in
annual rates), is volatile.
However, the figure
shows tha t money
growth often falls a t or
just before a cyclical peak.
Statistical and historical
studies suggest that, gen­
erally, money growth is
procyclical and leading.
Inflation, here measured
as the six-month moving

average of monthly
growth rates of the CPI
(expressed in annual
ra tes) , is procy cli c al and
lags the business cycle.
Federal Reserve Bank
of SL Louis FRED database at
researell.sf1011isfed.orglfred2 series
M2SL and CPIAUCSL.

PT

PT

P T

PT P T

PT

PT

15

M2

GR OWT H

10

5

O f-

-------.;--:--

----

I NF LATIO N
L-L
�
�
- 5 U_
_
_
�
_
_
_
_
L_
_
_
_
_L
_
_
_
_
L_

_
�
_
1 960
1 965
1 97 0
1 97 5
1 980
1 985
1 990
1 995
2000
2005

Source:

terms. The real wage, as shown in Fig. 8.6, is an especially important variable in the
study of business cycles because it is one of the main determinants of the amount
of labor supplied by workers and demanded by firms. Most of the evidence points
to the conclusion that real wages are mildly procyclical, but there is some contro­
versy on this point.14
M o n ey G rowth a n d I nf l a t i o n

Another variable whose cyclical behavior is somewhat controversial is the money
supply. Figure 8.7 shows the behavior since 1959 of the growth in the M2 measure
of the money supply.15 Note that (nominal) money growth fluctuates a great deal
and doesn't always display an obvious cyclical pattern. However, as Fig. 8.7 shows,
money growth often falls sharply at or just before the onset of a recession.
14See, for example, Mark Mitchell, Myles Wallace, and John Warner, "Real Wages Over the Business
Cycle: Some Further Evidence," Southem Ecol1omic journal, April 1985, pp. 1162-1173; and Michael

Keane, Robert Moffitt, and David Runkle, "Real Wages Over the Business Cycle: Estimating the
Impact of Heterogeneity with Micro Data," joumal of Political Economy, December 1988, pp. 1232-1266.
Stronger procyclicality for the real wage is claimed by Gary Solon, Robert Barsky, and Jonathan Parker,
"Measuring the Cyclicality of Real Wages: How Important Is Composition Bias?" Quarterly jountal of
Economics, February 1994, pp. 1-25.
I'See Table 7.1 for a definition of M2. To reduce the effect of high month-to-month volatility in money
growth, Fig. 8.7 presents a six-month moving average of money growth rates; that is, the reported
growth rate in each month is actually the average of the growth rate in the current month and in the
previous five months


8.3

Figure

Business Cycle Facts

299

S.S

Cyclical behavior
of the nominal
interest rate
The nominal interest
rate, measured here as
the interest rate on three­
month Treasury bills, is
procyclical and lagging.

.; �

18

�

'"

'"

Q.. 1 6

>,
� �

�

-

PT

PT

PT PT

PT

PT

PTP T

PT

PT

'"

-

= =
.

.

'"

u

.. �
= Q.,I

Eo -

.•

Z

'"

14

1

2

Source: Federal Reserve Bank
of St. Louis FRED database at

research.stlollisfed.orglfred2Iseriesl
TB3MS.

10

N O MI NAL
I NT ERE ST
RATE

8

6

4

2

O L�
1 950

-L

__

-L

__

1 95 5

�

__

1 960

__

__

1 965

L-

1 970

L-

__

�

-L

-L

1 980

1 985

1 990

__

1 97 5

__

__

�

__

�

__

1 995

L-

__
__

2000

2005

Year

Moreover, many statistical and historical studies including a classic work by
Milton Friedman and AlUla J. Schwartz16 that used data back to 1867 demon­
strate that money growth is procyclical and leads the cycle.
The cyclical behavior of inflation, also shown in Fig. 8.7, presents a somewhat
clearer picture. Inflation is procyclical but with some lag. Inflation typically builds
during an economic expansion, peaks slightly after the business cycle peak, and
then falls until some time after the business cycle trough is reached. Atypically,
inflation did not increase during the long boom of the 1990s.
F i n a n c i a l Va r i a b l es

Financial variables are another class of economic variables that are sensitive to the
cycle. For example, stock prices are generally procyclical (stock prices rise in good
economic times) and leading (stock prices usually fall in advance of a recession).
Nominal interest rates are procyclical and lagging. The nominal interest rate
shown in Fig. 8.8 is the rate on three-month Treasury bills. However, other inter­
est rates, such as the prime rate (charged by banks to their best customers) and the
Federal funds rate (the interest rate on overnight loans made from one bank to
another) also are procyclical and lagging. Note that nominal interest rates have the
same general cyclical pattern as inflation; in Chapter 7 we discussed why nominal
interest rates tend to move up and down with the inflation rate.
16A MO/1etary History of tlIe Ul1ited States, 1867-1960, Princeton, N.j.: Princeton University Press for

NBER, 1963. We discuss this study further in Chapter 10.


300

Chapter 8

Business Cycles

Figure 8.9
Industrial production
indexes in six major
countries
The world wide effect of
business cycles is reflected
in the similarity of the
behavior of industrial
production in each of
the six countries shown.
But individual countries
also have fluctuations
not shared with other
countries.

"
o
.

Note: Scales differ by country

t;

"
."

-

U N IT ED STAT E S

:::

...
..

.-

�

-

"
."

'"

."

-

o

1il

."
"

U N ITED K I N GD O M

GERMA N Y

-

FRA NC E

Note: The scales for the indus­
trial production indexes differ
by country; for example, the
figure does not imply that the
United Kingdom's total indus­
trial production is higher than
that of Japan.

CAN ADA

Source: Illtematiollal Fillal/ciaJ
Statistics, July 2006, from Inter­
national Monetary Fund (with
scales adjusted for clarity).

1 960

1 965

1 970

1 9 75

1 980

1 98 5

1 990

1 995

2000

2005

Year

The real interest rate doesn't have an obvious cyclical pattern. For instance, the
real interest rate actually was negative during the 1973-1975 recession but was very
high during the 1981-1982 recession. (Annual values of the real interest rate are
shown in Fig. 2.5.) The acyclicality of the real interest rate doesn't necessarily mean
its movements are unimportant over the business cycle. Instead, the lack of a stable
cyclical pattern may reflect the facts that individual business cycles have different
causes and that these different sources of cycles have different effects on the real
interest rate.

I nt e r n at i o n a l Asp ects of the B u s i n e ss Cyc l e

So far we have concentrated on business cycles in the United States. However,
business cycles are by no means unique to the United States, having been regularly
observed in all industrialized market economies. In most cases the cyclical behav­
ior of key economic variables in these other economies is similar to that described
for the United States.
The business cycle is an international phenomenon in another sense: Fre­
quently, the major industrial economies undergo recessions and expansions at
about the same time, suggesting that they share a common cycle. Figure 8.9
illustrates this common cycle by showing the index of industrial production
since 1960 for each of six major industrial countries. Note in particular the effects
of worldwide recessions in about 1975, 1982, 1991, and 2001. Figure 8.9 also
shows that each economy experiences many small fluctuations not shared by the
others.


8.4

Business Cycle Anaiysis:A Preview

301

Bus iness Cycle Ana lysis: A Preview

8.4

The business cycle facts presented in this chapter would be useful even if we
took them no further. For example, being familiar with the typical cyclical pat­
terns of key macroeconomic variables may help forecasters project the course of

the economy, as we showed when discussing leading indicators. Knowing the
facts about cycles also is important for businesspeople making investment and
hiring decisions and for financial investors trying to choose portfolios that pro­
vide the desired combinations of risk and return. However, macroeconomists are
interested not only in what happens during business cycles but also in why it
1
T h e S e a s o n a l Cyc l e a n d t h e B u s i ness Cyc l e

Did you know that the United States has a large eco­
nomic boom, followed by a deep recession, every year?
The boom always occurs in the fourth quarter of the year
(October through December). During this quarter output
is 5% higher than in the third quarter (July-September)
and about 8% higher than in the following first quarter
(January-March). Fortunately, the first-quarter recession
is always a short one, with output rising by almost 4%
in the second quarter (April-June). This regular seasonal
pattern, known as the seasonal cycle, actually accounts
for more than 85% of the total fluctuation in the growth
rate of real output!
Why don't large seasonal fluctuations appear in
Figs. 8.2-8.9? Normally, macroeconomic data are sea­
sonally adjusted, meaning that regularly recurring sea­
sonal fluctuations are removed from the data. Seasonal
adjustment allows users of economic data to ignore sea­
sonal changes and focus on business cycle fluctuations
and longer-term movements in the data. However,
Robert Barsky of the University of Michigan and Jeffrey
Miron of Boston University' argue that the practice of
seasonally adjusting macroeconomic data may throw

away information that could help economists better
understand the business cycle. Using data that hadn't
been seasonally adjusted, Barsky and Miron deter­
mined that the comovements of variables over the sea­
sonal cycle are similar to their comovements over the
business cycle. Specifically, they obtained the follow­
ing results:
1.

Of the types of expenditure, expenditures on durable
goods vary most over the seasonal cycle and expen­
ditures on services vary least.

2.
3.

4.
5.

Government spending is seasonally procyclical.
Employment is seasonally procyclical, and the
unemployment rate is seasonally countercyclical.
Average labor productivity is seasonally procyclical,
and the real wage hardly varies over the seasonal
cycle.
The nominal money stock is seasonally procyclical.

Each observation appears to be true for both the
business cycle and the seasonal cycle (although, as dis­
cussed, there is some controversy about the cyclical

behavior of the real wage). However, the seasonal fluc­
tuations of inventory investment, the price level, and
the nominal interest rate are much smaller than their
fluctuations over the business cycle.
The seasonal cycle illustrates three potential sources
of aggregate economic fluctuations: (1) changes in con­
sumer demand, as at Christmastime; (2) changes in pro­
ductivity, as when construction workers become less
productive because of winter weather in the first quarter;
and (3) changes in labor supply, as when people take
summer vacations in the third quarter. Each of these
three sources of fluctuation may also contribute to the
business cycle.
As we discuss in Chapter 10, classical economists
believe that business cycles generally represent the
economy's best response to changes in the economic
environment, a response that macroeconomic policy
need not try to eliminate. Although it doesn't necessarily
confirm this view, the seasonal cycle shows that large
economic fluctuations may be desirable responses to
various factors (Christmas, the weather) and do not need
to be offset by government policy.

'''The Seasonal Cycle and the Business Cycle," Journal of Political Economy, June 1989, pp. 503-534.


302

Chapter 8

Business Cycles

happens. This desire to understand cycles isn't just idle intellectual curiosity. For
example, as we demonstrate in Chapters 9-11, the advice that macroeconomists
give to policymakers about how to respond to a recession depends on what
they think is causing the recession. Thus, with the business cycle facts as back­
ground, in the rest of Part 3 we describe the primary alternative explanations of
business cycle fluctuations, as well as policy recommendations based on these
explanations.
In general, theories of the business cycle have two main components. The first
is a description of the types of factors that have major effects on the economy wars,
new inventions, harvest failures, and changes in government policy are examples.
Economists often refer to these (typically unpredictable) forces hitting the economy
as shocks. The other component of a business cycle theory is a model of how the
economy responds to the various shocks. Think of the economy as a car moving
down a poorly maintained highway: The shocks can be thought of as the potholes
and bumps in the road; the model describes how the components of the car (its tires
and shock absorbers) act to smooth out or amplify the effects of the shocks on the
passengers.
The two principal business cycle theories that we discuss in this book are the
classical and the Keynesian theories. Fortunately, to present and discuss these two the­
ories we don't have to develop two completely different models. Instead, both can
be considered within a general framework called the aggregate demand-aggregate
supply, or AD-AS, model. To introduce some of the key differences between the clas­
sical and Keynesian approaches to business cycle analysis, in the rest of this chapter
we preview the AD-AS model and how it is used to analyze business cycles.
A g g r e g ate D e m a n d a n d Ag g re g ate S u p p ly: A B ri ef I ntro d u c t i o n

We develop and apply the AD-AS model, and a key building block of the
AD-AS model, the IS-LM model, in Chapters 9-11. Here, we simply introduce

and briefly explain the basic components of the AD-AS model. The AD-AS
model has three components, as illustrated in Fig. 8.10: (1) the aggregate demand
curve, (2) the short-run aggregate supply curve, and (3) the long-run aggregate
supply curve. Each curve represents a relationship between the aggregate price
level, P, measured on the vertical axis in Fig. 8.10, and output, Y, measured
along the horizontal axis.
The aggregate demand (AD) curve shows for any price level, P, the total quantity
of goods and services, Y, demanded by households, firms, and governments. The AD
curve slopes downward in Fig. 8.10, implying that, when the general price level is
higher, people demand fewer goods and services. We give the precise explanation
for this downward slope in Chapter 9. The intuitive explanation for the downward
slope of the AD curve that when prices are higher people can afford to buy fewer
goods is not correct. The problem with the intuitive explanation is that, although
an increase in the general price level does reflect an increase in the prices of most
goods, it also implies an increase in the incomes of the people who produce and sell
those goods. Thus to say that a higher price level reduces the quantities of goods
and services that people can afford to buy is not correct, because their incomes, as
well as prices, have gone up.
The AD curve relates the amount of output demanded to the price level, if
we hold other economic factors constant. However, for a specific price level, any


8.4

Business Cycle Analysis: A Preview

303

Figure 8. 1 0
The aggregate

demand-aggregate
supply model
The aggregate demand
(AD) curve slopes down­
ward, reflecting the fact
that the aggregate quan­
tity of goods and services
demanded, Y, falls when
the price level, P, rises.
The short-run aggregate
supply (SRAS) curve is
horizontal, reflecting the
assumption that, in the
short run, prices are
fixed and firms simply
produce whatever quan­
tity is demanded. In the
long run, firms produce
their normal levels of
output, so the long-run
aggregate supply (LRAS)
curve is vertical at the
full-employment level of
output, Y. The economy's
short-run equilibrium is
at the point where the
AD and SRAS curves
intersect, and its long­
run equiUbrium is where
the AD and LRAS curves

intersect. ln this example,
the economy is in both
short-run and long-run
equilibrium at point E.

�

•
-

>

"
"

-

"
u

LRAS

.

-

0..

-

�------ SRAS

'AD
y
Output, Y

change in the economy that increases the aggregate quantity of goods and ser­
vices demanded will shift the AD curve to the right (and any change that
decreases the quantity of goods and services demanded will shift the AD curve
to the left). For example, a sharp rise in the stock market, by making consumers
wealthier, would likely increase households' demand for goods and services,
shifting the AD curve to the right. Similarly, the development of more efficient
capital goods would increase firms' demand for new capital goods, again shift­
ing the AD curve to the right. Government policies also can affect the AD curve.
For example, a decline in government spending on military hardware reduces
the aggregate quantity of goods and services demanded and shifts the AD curve
to the left.
An aggregate supply curve indicates the amount of output producers are will­
ing to supply at any particular price level. Two aggregate supply curves are shown
in Fig. 8.10 one that holds in the short run and one that holds in the long run. The
short-run aggregate supply (SRAS) curve, shown in Fig. 8.10, is a horizontal line. The
horizontal SRAS curve captures the ideas that in the short run the price level is
fixed and that firms are willing to supply any amount of output at that price. If the
short run is a very short period of time, such as a day, this assumption is realistic.
For instance, an ice cream store posts the price of ice cream in the morning and sells
as much ice cream as is demanded at that price (up to its capacity to produce ice
cream). During a single day, the owner typically won't raise the price of ice cream
if the quantity demanded is unusually high; nor does the owner lower the price of
ice cream if the quantity demanded is unusually low. The tendency of a producer
to set a price for some time and then supply whatever is demanded at that price is

represented by a horizontal SRAS curve.


304

Chapter 8

Business Cycles

However, suppose that the quantity of ice cream demanded remains high day
after day, to the point that the owner is straining to produce enough ice cream to
meet demand. In this case, the owner may raise her price to reduce the quantity of
ice cream demanded to a more manageable level. The owner will keep raising the
price of ice cream as long as the quantity demanded exceeds normal production
capacity. In the long run, the price of ice cream will be whatever it has to be to
equate the quantity demanded to the owner's normal level of output. Similarly, in
the long run, all other firms in the economy will adjust their prices as necessary so
as to be able to produce their normal level of output. As discussed in Chapter 3, the
normal level of production for the economy as a whole is called the full-employ­
ment level of output, denoted Y. In the long run, then, when prices fully adjust, the
aggregate quantity of output supplied will simply equal the full-employment level
of output, Y. Thus the long-run aggregate supply (LRAS) curve is vertical, as shown in
Fig. 8.10, at the point that output supplied, Y, equals Y.
Figure 8.10 represents an economy that is simultaneously in short-run and
long-run equilibrium. The short-run equilibrium is represented by the intersection
of the AD and SRAS curves, shown as point E. The long-run equilibrium is repre­
sented by the intersection of the AD and LRAS curves, also shown as point E.
However, when some change occurs in the economy, the short-run equilibrium can
differ from the long-run equilibrium.
Recall that a theory of business cycles has to

include a description of the shocks hitting the economy. The AD-AS framework
identifies shocks by their initial effects on aggregate demand or aggregate supply.
An aggregate demand shock is a change in the economy that shifts the AD curve. For
example, a negative aggregate demand shock would occur if consumers became
more pessimistic about the future and thus reduced their current consumption
spending, shifting the AD curve to the left.
To analyze the effect of an aggregate demand shock, let's suppose that the econ­
omy initially is in both short-run and long-run equilibriwn at point E in Fig. 8.11. We
assume that, because conswners become more pessimistic, the aggregate demand
curve shifts down and to the left from AD I to AD 2 In this case, the new short-run
equilibrium (the intersection of AD2 and SRAS) is at point F, where output has fallen
to Y2 and the price level remains unchanged at PI ' Thus the decline in household con­
sumption demand causes a recession, with output falling below its normal level.
However, the economy will not stay at point F forever, because firms won't be con­
tent to keep producing below their normal capacity. Eventually firms will respond to
lower demand by adjusting their prices in this case downward until the economy
reaches its new long-run equilibrium at point H, the intersection of AD2 and LRAS. At
point H, output is at its original level, Y, but the price level has fallen to P2'
Our analysis shows that an adverse aggregate demand shock, which shifts the
AD curve down, will cause output to fall in the short run but not in the long run.
How long does it take for the economy to reach the long run? This question is crucial
to economic analysis and is one to which classical economists and Keynesian econo­
mists have very different answers. Their answers help explain why classicals and
Keynesians have different views about the appropriate role of government policy in
fighting recessions.
The classical answer is that prices adjust quite rapidly to imbalances in quantities
supplied and demanded so that the economy gets to its long-run equilibrium
A g g reg ate D e m a n d S h ocks.

8.4

Business Cycle Analysis: A Preview

305

Figure 8. 1 1
An adverse aggregate
demand shock
An adverse aggregate
demand shock red uces
the aggregate quantity
of goods and services
demanded at a given
price level; an example is
that consumers become
more pessimistic and
thus reduce their spend­
ing. This shock is repre­
sented by a shift to the
left of the aggregate
demand curve from AD'
to ADO. In the short run,
the economy moves to
point F At this short-run
equilibrium, output has
fallen to Y, and the price
level is unchanged. Even­
tually, price adjustment
causes the economy to

move to the new long­
run equilibrium at point
H, where output returns
to its full-employment
level, Y, and the price
level falls to P,. In the
strict classical view, the
economy moves almost
immediately to point H,
so the adverse aggregate
demand shock essentially
has no effect on output in
both the short run and
the long run. Keynesians
argue that the adjustment
process takes longer, so
that the adverse aggre­
gate demand shock may
lead to a sustained
decline in output.

Consumers
become
more thrifty

-

"
u

...

P,

LRAS

"

1-----.....;

F
.�

E
--..,; �------ SRAS

--

•
•
·
•

•

H

•

�

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

•

. . . . . . . . ;. . . . . . . .
•

•
•
•
•
•
•
•
•
•

AD '

•
•
•
•
•
•
•
•

Y

Output, Y

quickly in a few months or less. Thus a recession caused by a downward shift of the
AD curve is likely to end rather quickly, as the price level falls and the economy
reaches the original level of output, Y. In the strictest versions of the classical model, the
economy is assumed to reach its long-run equilibrium essentially immediately, imply­
ing that the short-run aggregate supply curve is irrelevant and that the economy
always operates on the long-run aggregate supply (LRAS) curve. Because the adjust­
ment takes place quickly, classical economists argue that little is gained by the govern­
ment actively trying to fight recessions. Note that this conclusion is consistent with the
"invisible hand" argument described in Chapter 1, according to which the free market
and unconstrained price adjustments are sufficient to achieve good economic results.
In contrast to the classical view, Keynesian economists argue that prices (and
wages, which are the price of labor) do not necessarily adjust quickly in response to
shocks. Hence the return of the economy to its long-run equilibrium may be slow,
taking perhaps years rather than months. In other words, although Keynesians
agree with classicals that the economy's level of output will eventually return from
its recessionary level (represented by Y2 in Fig. 8.11) to its full-employment level, Y,
they believe that this process may be slow. Because they lack confidence in the
self-correcting powers of the economy, Keynesians tend to see an important role for
the government in fighting recessions. For example, Keynes himself originally argued
that government could fight recessions by increasing spending. In terms of Fig. 8.11,
an increase in government spending could in principle shift the AD curve up and to
the right, from AD2 back to ADJ, restoring the economy to full employment.
Agg reg ate S u p p l y Shocks.

Because classical economists believe that aggre­
gate demand shocks don't cause sustained fluctuations in output, they generally
view aggregate supply shocks as the major force behind changes in output and