This PDF is a selection from an out-of-print volume from the National Bureau
of Economic Research
Volume Title: Business Cycles: Theory, History, Indicators, and Forecasting
Volume Author/Editor: Victor Zarnowitz
Volume Publisher: University of Chicago Press
Volume ISBN: 0-226-97890-7
Volume URL: http://www.nber.org/books/zarn92-1
Conference Date: n/a
Publication Date: January 1992
Chapter Title: Recent Work on Business Cycles in Historical Perspective
Chapter Author: Victor Zarnowitz
Chapter URL: http://www.nber.org/chapters/c10373
Chapter pages in book: (p. 20 - 76)
Recent Work on Business
Cycles in Historical Perspective
Interest in business cycles is itself subject to a wavelike movement, waxing
during and after periods of turbulence and depression, waning in periods of
substantial stability and continuing growth. I At time, confidence in government institutions and actions persuaded many that cyclical instability had
ceased to be a serious problem. Thus in the early heyday of the Federal Reserve System, 1922-29, monetary policies were expected to help maintain
prosperity. In the 1960s, the late heyday of Keynesian economics, fiscal finetuning evoked similar hopes.
The present is another time of disillusionment-now extending to both
types of stabilization policy. The sequence of serious worldwide recessions in
the last decade soon refuted the perennially attractive idea that business cycles
had become obsolete. Beyond that, the credibility of both Keynesian and monetarist explanations has diminished. Once again, the apparent failure of old
solutions prompts the profession to pay more attention to the continued existence of business cycles.
Reprinted from the Journal ofEconomic Literature 23 (June 1985): 523-80.
The author thanks Alan Blinder, William Fellner, Stanley Fischer, Milton Friedman, Robert
Gordon, Robert Hall, Bert Hickman, Wilhelm Krelle, David Laidler, Ronald McKinnon, John
Taylor, and two anonymous referees for many helpful suggestions and criticisms of the first outline and earlier drafts of this paper. Also gratefully acknowledged is the financial and intellectual
support received from the National Bureau of Economic Research and the Graduate School of
Business of the University of Chicago. Naturally, responsibility for the expressed opinions and
any remaining errors is exclusively the author's.
1. This is well illustrated in the early literature, which focused on the episodes of commercial
crises, but it is also reflected in the timing of later, classical studies of the nature and causes of
business cycles at large. There is little doubt about the impetus provided in this context by the
major depressions of the late 1830s, 1870s, 1890s, 1907-8, 1920-21, and most strikingly the
Recent Work on Business Cycles in Historical Perspective
The rediscovery of an important subject is always welcome, even if long
overdue. However, much of the recent work has neglected the long history of
both the phenomena of major economic fluctuations and their interpretations,
concentrating instead on contemporary theoretical and policy controversies,
mainly in the United States. An overview of selected literature will attempt to
demonstrate that this myopia is costly and needs to be corrected.
The study of business cycles is almost coextensive with short-term macrodynamics and it has a large interface with the economics of growth, money,
inflation, and expectations. The literature is huge; its level of difficulty is in
general high. This survey attempts to provide a historical background and
outline the evolution of thought leading to the recent developments in theory
and related evidence. The coverage is extensive, yet of necessity much is left
out, including theories that are largely concerned with unemployment and
inflation, much less with business cycles directly. 2
In particular, no attempt can be made here to discuss in any detail the statistical and historical work on the observed regularities and idiosyncrasies of
business cycles and their possible long-term changes. This empirical literature
is rich and important: it deserves a separate review. After all, it is the "stylized
facts" which it provides that ought to be explained by the theory. Those main
facts are summarized in section 2.2, but with a minimum of references and
Section 2.3 discusses the main elements of older theories, before and after
the Great Depression, and proceeds to more recent models driven by changes
in investment, credit, and price-cost-profit relations. Most of these theories
and models are primarily endogenous and deterministic. Exogenous factors
and stochastic elements are introduced early in section 2.4.
The sections that follow deal, first, with the monetarist interpretation of
business cycles, then with the newer equilibrium models with price misperceptions and intertemporal substitution. The route leads generally from
"adaptive" to "rational" expectations. The approach is generally monetarist in
the sense of relying on monetary shocks, but the emphasis shifts from nominal
demand changes and lagged price adjustments to informational lags and supply reactions. Various problems and complications arise, revealed in large part
by intensive testing and criticism. This leads to new attempts to explain the
persistence of cyclical movements, the role of uncertainty and financial instability, real shocks, gradual price adjustments, etc. Conclusions are drawn in
the last section, which stresses the need for a realistic synthesis.
2. Two sets of writings should be mentioned in this context, namely, the theories of the new
radical economists and those embodied in some of the recent "disequilibrium" models. For summaries or surveys, see Sherman 1976; Malinvaud 1977; and Drazen 1980.
Also, the early mathematical models and the more recent theories of the "political business
cycle". receive little attention in the present paper; monographs surveying this literature are Rau
1974; Gapinski 1982; and Mullineux 1984.
The Overall Aspects and Varying Dimensions of Business Cycles
The term "business cycle," is a misnomer insofar as no unique periodicities
are involved, but its wide acceptance reflects the recognition of important
regularities of long standing. The observed fluctuations vary greatly in amplitude and scope as well as duration, yet they also have much in common. First,
they are national, indeed often international, in scope, showing up in a multitude of processes, not just in total output, employment, and unemployment.
Second, they are persistent-lasting, as a rule several years, that is, long
enough to permit the development of cumulative movements in the downward
as well as upward direction. This is well established by the historical chronologies of business cycles in the United States, Great Britain, France, and Germany, a product of a long series of studies by the National Bureau of Economic Research (Bums and Mitchell 1946; Moore 1961, 1983; Moore and
Zarnowitz 1986; see also chapter 5). For all their differences, business expansions and contractions consist of patterns of recurrent, serially correlated and
cross-correlated movements in many economic (and even other) activities.
Seasonal movements, which are periodic but often variable in size and pattern, may obscure the cyclical developments from an observer of current
changes in individual time series. The same applies to short, erratic movements, which are similarly ubiquitous. But, looking back across monthly or
quarterly data representing many different variables, business cycles can be
clearly distinguished from the other fluctuations in that they are as a rule
larger, longer, and more widely diffused. They dominate changes in the economy over spans of several years, in contrast to the seasonal and other variations which spend themselves over spans of a year or less. They reflect, and
interact with, long growth trends which dominate developments across decades.
Peacetime expansions in the United States averaged about 3 years in the last
half-century, 2 years in the earlier periods containing 10 cycles each (table
2.1). Each of the wartime expansions was much longer. Contractions have
averaged close to 1 year since 1933, almost twice as long in the earlier periods. This suggests a strong shift toward longer and more variable expansions
and shorter and more uniform contractions since the Great Contraction of the
early 1930s. However, the NBER dates for the early business cycles are based
on limited information and may overstate the length of some of the recessions
(see note to the table). The mean duration of full peacetime cycles has remained approximately stable at 4 years.
The individual phase and cycle durations show considerable variability over
time, as shown by the standard deviations in table 2.1. However, when the
relatively rare outliers are discounted, fairly clear central tendencies emerge.
Thus the ranges of 1.5-3 years, 1-2 years, and 2.5-5 years account for three
Recent Work on Business Cycles in Historical Perspective
Average Duration of Business Cycles in the United States, 1854-1982
Average Measures of Phase and Cycle Durations
(years, T to T)
1933-82, excl. wars
1854-1982, excl. wars
(T to T)
Note: All means and standard deviations (S.D.) are rounded to full months. Expansions are measured
from troughs (T) to peaks (P), contractions from P to T, the full cycles from T to T. Figures in line 4
exclude the expansions during World War II, the Korean War, and the Vietnam War and the immediately
following contractions. Figures in line 6 exclude also the expansions during the Civil War and World
War I and the immediately followng contractions. For references and underlying detail see Moore and
It should be noted that the data available for identifying and measuring the historical business cycles
are fragmentary and often weak. Much of the evidence relates to cyclically sensitive sectors and processes. Hence some of the early fluctuations may have involved only slowdowns rather than absolute
declines in total output and employment. If so, the averages in lines 1 and 2 should be somewhat larger
for expansions and smaller for contractions. This would only moderately reduce the contrast with
the entries in line 3 and would not significantly alter the overall conclusions drawn in the text (see chapter 7).
fourths or more of the peacetime expansions, contractions, and full cycles in
the United States, respectively.
The amplitudes of cyclical expansions vary as much as their durations, with
which they tend to be well correlated. The rates of change (velocities) and
diffusion show less variability across the cycles. Table 2.2 provides some evidence in support of these generalizations.
In the 20 years between the two world wars three major depressions occurred, including the uniquely deep one of 1929-33. Since then no general
declines of comparable magnitude have occurred, notwithstanding the gravity
of recent conditions of rising and high unemployment in some countries, such
as the United Kingdom. On the whole recessions have become not only much
shorter but also shallower and less diffused. Table 2.3, using a sampling of
measures for the U.S. business contractions of 1920-82, illustrates the contrasting dimensions of major depressions versus other declines and the much
smaller but consistent differences between the "severe" and "mild" recessions.
Main Features of Cyclical Behavior
Most industries and sectors of the economy participate in the general business cycles with substantial regularity (i.e., they exhibit high conformity or
Selected Characteristics of Seven Expansions, United States, 1949-82
Total increases (%)
Rate of increase (% per year)
Total decline (% points)
Percent of industries expanding
Source: Moore and Zamowitz 1986, table 6.
Note: The entries in col. 1 refer to the expansion of 2/1961-12/1969 (lines 1-3) and 10/1948-7/
1953 (lines 4 and 5). The entries in col. 2 refer to the expansion of 7/1980-7/1981. The entries
in col. 3 and 4 cover all seven expansions. Line 5 shows the maximum percentage of nonagricultural industries with rising employment, based on changes over 6-month spans.
Average Duration, Depth, and Diffusion of Thirteen Contractions,
United States, 1920-82
Average duration (months)
Total increase (% points)
Percent of industries contracting
Two Major Six Severe Four Mild
Depression Depressions Recessions Recessions
Source: Moore and Zamowitz 1986, table 7.
Note: The contraction of 8/1929-3/1933 is referred to as the Great Depression; the contractions
of 1/1920-7/1921 and 5/1937-611938 are the major depressions. The dates of the six severe
recessions are 5/1923-711924, 11/1948-10/1949, 711953-5/1954, 8/1957-4/1958, 11/1973-3/
1975, and 7/1981-11/1982. The dates of the four mild recessions are 1011926-11/1927,4119602/1961, 12/1969-11/1970; and 111980-7/1980.
coherence), but some do not (e.g., agriculture, which depends heavily on the
weather, and production of naturally scarce resources). 3 Durable producer and
consumer goods tend to have high conformity and large amplitudes of cyclical
movements in production, employment, and inventories. The amplitudes are
3. This section is based primarily on studies of U. S. economic history, but many of the qualitative features of cyclical behavior summarized here are found as well in the data for other major
industrialized countries with private enterprise and free markets: Mitchell 1913, 1927; Schumpe-
Recent Work on Business Cycles in Historical Perspective
much smaller for nondurable goods, and still smaller for most of the (nonstorable) services. Manufacturers' sales move with greater amplitudes than
wholesalers' sales, and the latter with greater amplitudes than retailers' sales.
In many industries, particularly manufacturing of durables, production is, in
large measure, guided by advance orders, which show large fluctuations followed, with variable lags, by much smaller fluctuations in outputs and shipments. The resulting changes in backlogs of unfilled orders and average delivery lags are themselves procyclical.
Private investment expenditures, although much smaller in the aggregate
than consumer spending, have much larger cycles in percentage terms. Aggregate production typically fluctuates more widely than aggregate sales, which
implies a procyclical behavior of inventory investment. Business profits show
very high conformity and much greater amplitude of cyclical movements than
wages and salaries, dividends, net interest, and rental income.
The level of industrial prices tends to have wider fluctuations than the levels
of retail prices and wages. Virtually all U. S. business contractions before
World War II were associated with declines in wholesale prices. 4 However,
the last recession to be accompanied by a significant deflation was that of
1948-49. Since then the price level never fell cyclically, but each of the seven
U. S. recessions of 1953-82 resulted in a temporary reduction of the rate at
which prices rose, that is, in some disinflation. But, in contrast to the general
price indexes for consumer and producer goods, prices of industrial commodities and raw materials traded in organized auction markets continued to show
high sensitivity to business cycles, often turning down early in slowdowns as
well as contractions.
Narrowly and broadly defined monetary aggregates usually experience only
reduced growth rates, not absolute declines, in connection with ordinary
recessions. Only in cycles with severe contractions do substantial downward
movements interrupt the pronounced upward trends in these series. The income velocity of money (i.e., ratio of income to the stock of currency and
commercial bank deposits held by the public) tends to move procyclically (up
in expansions and down in contractions), allowing for its long trends (downward before World War II, then upward for some time).
Short-term interest rates display high positive conformity and generally
large amplitudes of movements relative to their average level in each cycle.
However, when measured in basis points, cyclical changes in these series are
typically small when the interest-rate levels are low. Long-term rates usually
ter 1939; Frickey 1942; Burns and Mitchell 1946; Abramovitz 1950; Mitchell 1951; Gayer, Rostow, and Schwartz 1953; Matthews, 1959; Moore 1961,1983; R. A. Gordon 1961; M. Friedman
and Schwartz 1963a; Hultgren 1965; Zamowitz 1972a, 1973; Zamowitz and Moore 1984; P. Klein
and Moore 1985.
4. This is true both for the periods of long-term inflationary trends (1843-64, 1896-1920) and
for those of long-term deflationary trends (1864-96, 1920-32).
lag behind the short-term rates and have much lower conformity and much
smaller amplitudes. The relative movements in both short-term market rates
and bond yields increased significantly in the recent past compared with their
historical averages. Near cyclical peaks, short rates tend to come close to or
exceed the long rates; near cyclical troughs, they tend to be much lower.
Along with these conformity and amplitude characteristics, the recurring
features of business cycles include an array of timing sequences. Months before total employment, output, and real income turn down, activities marking
the early stages of investment process begin to decline. These include the
formation of new business enterprises, corporate appropriations for capital
expenditure, contracts for commercial and industrial construction, new orders
of machinery and equipment, and new bond and equity issues. Investment
realizations-construction put in place, deliveries and installations of equipment-keep increasing long after the decline in these investment commitments as work continues on the backlog of orders accumulated during the
busiest stages of expansion. Indeed, business expenditures for new plant and
equipment often peak when the overall economic contraction is already well
under way. At business cycle troughs, with lower levels of capacity utilization, the delivery lags are generally shorter, but investment commitments still
tend to lead and expenditures coincide or lag.
Long before the downturn in total sales, profits per unit of sales decline.
Total profits (a product of margins times sales) also lead, but by shorter intervals. Stock prices move early as well, reflecting expected changes in corporate
earnings. Bond prices tend to tum earlier yet (bond yields are generally lagging).
Labor productivity (output per hour) fluctuates procyclically around a secularly rising trend, generally with leads. Money wages often rise less than
prices in recoveries and more than prices in late expansion stages. This combines with the marked and persistent productivity changes to induce a procyclical and lagging movement in labor costs per unit of output.
Net changes in consumer installment credit and in mortgage credit outstanding have similar ...
Purchase answer to see full attachment