Question

Explain why investment is usually more volatile then consumption across the business cycle using and interpreting the coefficient of variation

Answer 1

IT IS A well-known fact that inventory disinvestment can account for
much of the movement in output during recessions. Almost one-half of
the shortfall in output, averaged over the five interwar business cycles,
can be accounted for by inventory disinvestment, and the proportion
has been even larger for postwar recessions.' A lesser-known fact is that
corporate profits, and therefore internal-finance flows, are also ex-
tremely procyclical and tend to lead the cycle. Wesley Mitchell finds
that the percentage change in corporate income over the business cycle
is several times greater than that in any other macroeconomic series in
his study.2 Robert Lucas lists the high conformity and large variation of
corporate income as one of the seven main qualitative features of the
business cycle.3 The volatility of internal finance, which is also com-
We thank Lee Benham, Robert Chirinko, Mark Gertler, Simon Gilchrist, Edward
Greenberg, Charles Himmelberg, Anil Kashyap, Louis Maccini, Dorothy Petersen, and
Toni Whited for helpful comments, and Andrew Meyer, Bruce Rayton, and James Mom-
tazee for excellent research assistance. Robert Parks and Daniel Levy provided technical
assistance. We acknowledge financial support from the Jerome Levy Economics Institute,
the University Research Committee of Emory University, and Washington University.
1. See Abramovitz (1950, p. 5) and Blinder and Maccini (199la).
2. Mitchell (1951, p. 286). These series include the rate of bankruptcy, employment,
and pig iron production, among others.
3. Lucas (1977, p. 9).

This paper links these two stylized facts by examining whether fluc-
tuations of internal finance are an important cause of changes in inven-
tory investment. Our exploration is motivated by a rapidly growing body
of theoretical research arguing that changes in either internal finance or
net worth will affect firm behavior if the markets for external finance are
imperfect. Although most previous empirical work in this area has fo-
cused on fixed investment, the dramatic cyclical fluctuations in both in-
ventory investment and internal finance suggest that efforts to examine
their possible link are overdue.
When capital markets are imperfect, fluctuations of internal finance
should affect all components of investment. We argue, however, that in-
ventories should be especially sensitive to such imperfections. In re-
sponse to a negative shock to internal finance, financially constrained
firms will reduce their accumulation of all assets, with the effect on each
asset determined by its relative liquidation and adjustment costs. Be-
cause inventory investment has low adjustment costs, its share of a de-
cline in total investment caused by the contraction of internal finance
will be disproportionately large relative to fixed investment or other uses
of funds (research and development, for example).
While the modern literature on inventories typically excludes finan-
cial effects, the connection between internal finance and inventory in-
vestment may help resolve an empirical puzzle about inventory behav-
ior. Numerous studies have found that production varies more than
sales and that inventory investment is positively correlated with con-
temporaneous sales shocks.4 Both results are inconsistent with the
production-smoothing model that predicts inventories will buffer pro-
duction from sales shocks. These findings may arise from an omitted-
variable bias. The presence of financing constraints induces a positive
correlation between inventory investment and internal-finance flows.
When internal-finance variables are excluded from inventory invest-
ment regressions, the coefficient on contemporaneous sales may reflect
the impact of financing constraints, overwhelming any negative correla-
tion caused by buffer-stock effects alone.
We test for a linkage between inventory investment and internal fi

nance by estimating a standard inventory investment model augmented
by measures of internal finance. The data are taken from Compustat's
quarterly "full coverage" files for manufacturing firms. The sample pe-
riod from 1981 to 1992 contains pronounced swings in inventory invest-
ment as well as large fluctuations in internal finance, with troughs in
1982, 1986, and 1991. To our knowledge, our study is the first work on
the microfoundations of cyclical firm behavior to employ a data set with
three key features: (i) firm-level panel data, (ii) high-frequency data
(quarterly), and (iii) data covering a major fraction of the aggregate econ-
omy. The structure of these data provides several important methodo-
logical advantages.
First, with firm-level panel data, we include both fixed firm effects
and highly disaggregated industry time dummies. The fixed firm effects
control for the many possible time-invariant determinants of inventory
investment that differ across firms. The disaggregated time dummies
control for a wide range of alternative hypotheses about inventory
movements that would be observationally equivalent in tests based on
aggregate time-series data.5 For example, an alternative explanation to
our hypothesis is that cost or technological shocks at the aggregate or
industry level drive both internal finance and inventory investment. By
including industry time dummies to control for these shocks, however,
the influence of cost shocks can be disentangled from other variables.
Indeed, because cost shocks at industry or higher levels of aggregation
are often invoked to explain cyclical phenomena, we believe the empiri-
cal approach pursued here is applicable to a wide class of macroeco-
nomic issues.
A second feature of our method, critical to our study and new in the
literature, is the use of quiacrter-ly firm data. This innovation is especially
important for a high-frequency phenomenon such as inventory invest-
ment; one could miss important cyclical variations in annual data. Per-
haps more important, quarterly data increase the number of time-series
observations. We can therefore run regressions, in the time dimension
of the panel, for very short calendar periods (such as two or three years).
The cross-sectional breadth of the data in combination with its high fre-
quency allows standard panel data techniques to be used to examine in

Most of our regressions include time dummies for each four-digit SIC (standard in-
dustry classification) industry. In contrast, previous panel studies in the financing-con-
straint literature have included only aggregate time dummies.