Académique Documents
Professionnel Documents
Culture Documents
Abstract
The debate over whether "rms smooth production relative to demand continues. In this paper we "nd that results
obtained in Blinder (Quarterly Journal of Economics 101 (3) (1986) 431}453) and Blinder and Maccini (Journal of
Economic Surveys 5 (4) (1991) 293}328), indicating that "rms do not smooth production, are greatly in#uenced by the
seasonal adjustment of the Census data for US manufacturing. Using seasonally adjusted data, we "nd signi"cant
evidence of production smoothing in 10 of 14 manufacturing industries for the years 1982}1997. Further, the prevalence
of production smoothing has increased over time, which we attribute either to fundamental changes in the collection of
the data or to the advent of improved methods of inventory management. 2000 Elsevier Science B.V. All rights
reserved.
Keywords: Inventory; Backorder; Production-smoothing; Seasonal adjustment
1. Introduction
The main goal of the production-smoothing
literature has been to discern the extent to which
"rms use inventories to smooth production levels
in the face of #uctuating demand. Whether "rms
smooth or bunch production over time is not
merely an academic question; it has broad implications for the business cycle, inventory demand and
even seasonal and frictional unemployment.
Others have shown that seasonally unadjusted
data are preferable to seasonally adjusted data
when doing dynamic production studies [1}3].
* Corresponding author.
E-mail addresses: michael.gorman@bnsf.com (M.F. Gorman),
brannon@uwosh.edu (J.I. Brannon)
0925-5273/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 5 - 5 2 7 3 ( 9 9 ) 0 0 0 4 9 - 3
174
2. Literature review
Blinder [4] and Blinder and Maccini [6] "nd
little evidence to support the production-smoothing hypothesis. Using monthly, seasonally adjusted
data for durable and nondurable manufacturing
from the Census Bureau, the two studies "nd the
ratio of the variance of production to the variance
of shipments falls between 1.03 and 1.20 for the two
sectors and the covariance of shipments and change
in inventories to be positive. For two-digit SIC
industries, the ratio ranges from 0.93 to 1.38, with
17 out of 20 industries having a ratio greater than
one. A ratio of the variances of production and
shipments less than one strongly supports the production-smoothing hypothesis.
In light of the paucity of evidence supporting
production smoothing, researchers responded by
attempting to refute the very idea of production
smoothing. Production smoothing may not exist, it
has been suggested, due to the stochastic behavior
of demand, concavity of costs and the existence of
signi"cant supply shocks in manufacturing.
Researchers have also tried to "nd new ways of
testing the production-smoothing hypothesis, focusing on the properties of the data used to test the
hypothesis. For instance, there might be a positive
correlation between inventories and sales due to
the behavior of demand shocks and the develop-
of production smoothing of seasonal and nonseasonal #uctuations for various two-digit SIC industries.
Ghali and Dimelis [11], using data for six highly
disaggregated industries from 1950 to 1960, "nd
signi"cant evidence of production smoothing either
using the variance bounds test or estimating a reduced form equation. They suggest that the failure
of previous models to detect production smoothing
is either due to the level of aggregation or the test
used.
3. Data
We have monthly, seasonally adjusted and unadjusted data on production, inventory, shipments
and new orders from 1959 to 1997 for 14 2-digit
SIC industries in both durable and nondurable
manufacturing. We also have the same aggregate
data for durable, nondurable and total manufacturing. All but the production series comes from data
"les supplied by the Census Bureau; production
numbers were created from the identity
P ,S #(F !F ),
(1)
R
R
R
R\
where P is the production in period t, S the shipR
R
ments in period t, and F the inventory at the end of
R
period t.
Because of their intertemporal nature, inventories cannot be properly de#ated by a simple index.
We calculate real values by considering both current and lagged price levels, as well as the age
composition of the goods held in inventory. To
de#ate nominal seasonally unadjusted data we calculated the implied de#ator of the BEA for seasonally adjusted inventories from the ratio of
nominal and real seasonally adjusted inventory
numbers [2]. As in West [8], inventory numbers
are converted from cost to market values by an
industry-speci"c factor in order to adjust for the
ratio of shipments to cost of goods sold.
Due to a change in inventory recording methods
by the Census bureau in 1982, the inventory num-
175
bers from before and after that time are not consistent. Thus, we separate our data analysis into
pre-1982 and post-1982 results, placing more
weight on the latter simply because the revision in
the data collection appears to have signi"cantly
improved the reliability of the data.
176
Table 1
Seasonality's e!ect on inventory and shipment time series percent of total variation attributable to seasonality from Appendix
G, Current Industrial Reports
Nondurable manufacturing
Rubber and Plastics
Petroleum
Chemicals
Paper
Textile
Tobacco
Food
Durable manufacturing
Instruments
Transportation equipment
Electronic equipment
Industrial machinery
Fabricated metals
Primary metals
Stone, clay and glass
Inventories
Shipments
Min
(%)
Max
(%)
Min
(%)
Max
(%)
41.27
29.22
33.31
33.31
55.40
83.61
42.01
81.65
68.71
90.71
89.28
86.07
93.51
86.19
96.93
54.86
65.39
52.72
79.24
61.16
84.25
78.06
94.33
37.48
95.33
26.84
30.08
20.88
35.73
29.77
34.99
47.87
51.02
72.39
78.46
78.38
68.54
51.60
71.31
73.25
52.44
82.37
74.91
55.96
74.20
85.13
91.08
87.09
93.15
98.98
92.44
89.95
90.60
Not reported.
5. Conclusion
Consistent with Blinder [4] and Blinder and
Maccini [5], we measure the extent of production
smoothing by examining the ratio of the variance of
production to the variance of shipments for a "rm
or industry. Table 2 provides the variance ratios for
14 2-digit SIC manufacturing industries and for
total, durable, and non-durable manufacturing, calculated for both seasonally adjusted and unadjusted data. This method is intuitively appealing; a
ratio less than one indicates that output changes
less than shipments and hence inventories are being
used to smooth production.
Neither Blinder nor Blinder and Maccini "nd
much evidence of production smoothing using seasonally adjusted data; Blinder estimated a variance
ratio of 1.20 for DUR and 1.05 for NDUR; Blinder
and Maccini obtained 1.03 for both sectors with
a slightly larger data set. We "nd similar results for
seasonally adjusted data regardless of the industry,
aggregation, or time frame, with only 1 industry
(primary metals) showing any evidence of produc-
177
Fig. 1. Durable good shipments for seasonally adjusted and seasonally unadjusted goods.
Table 2
Var(production)/Var(shipments) for seasonally adjusted (SA)
and seasonally unadjusted (NSA) data 1958}1998
Industry
Pre-1982
Post-1982
SA
SA
NSA
NSA
Total manufacturing
1.05
1.02
1.08
0.78
Nondurable manufacturing
Rubber and plastics
Petroleum
Chemicals
Paper
Textile
Tobacco
Food
1.34
1.66
1.38
1.39
1.37
1.40
4.10
1.35
1.08
1.23
1.40
0.82
1.01
1.10
14.93
1.15
1.04
1.38
1.73
1.87
1.66
1.79
0.73
1.08
0.81
1.04
1.44
0.80
1.04
0.87
0.44
0.73
Durable manufacturing
Instruments
Transportation equipment
Electronics
Industrial and mach. equip.
Fabricated metals
Primary metals
Stone, clay and glass
1.01
2.43
1.04
1.55
1.43
1.57
0.83
1.52
1.01
1.08
0.93
1.14
1.11
1.21
0.90
0.84
1.24
1.88
1.24
2.43
3.08
1.69
0.96
1.09
0.81
0.77
0.93
0.74
0.70
1.12
0.94
0.83
178
Acknowledgements
Thomas Kniesner, Benjamin Craig, John Muth
and two anonymous referees provided helpful comments on earlier drafts, and Jennifer Ribarsky
assisted with the data collection.
References
[1] R. Fair, The production smoothing model is alive and well,
Working Paper, No. 2877, March, National Bureau of
Economic Research, Cambridge, MA, 1989.
[2] J.A. Miron, S.P. Zeldes, Seasonality, cost shocks, and the
production smoothing model of inventories, Econometrica
56 (1988) 877}908.
[3] M.A. Ghali, Seasonality, aggregation and the testing of the
production smoothing hypothesis, American Economic
Review 77 (3) (1987) 464}470.
[4] A.S. Blinder, Can the production smoothing model of
inventory behavior be saved? Quarterly Journal of Economics 101 (3) (1986) 431}453.
[5] A.S. Blinder, L.J. Maccini, Taking stock: A critical assessment of recent research on inventories, Journal of Economic Perspectives 5 (1) (1991) 73}96.
[6] A.S. Blinder, L.J. Maccini, The resurgence of inventory
research: What have we learned? Journal of Economic
Surveys 5 (4) (1991) 293}328.
[7] V.A. Ramey, Nonconvex costs and the behavior of
inventories, Journal of Political Economy 99 (1991)
306}333.
[8] K.D. West, A variance bounds test of the linear quadratic
inventory model, Journal of Political Economy 94 (1986)
374}401.
[9] S. Krane, S. Braun, Production smoothing: Evidence
from physical product data, Board of Governors of
the Federal Reserve System, Washington, DC, June
1989.
[10] O.J. Blanchard, The production and inventory behavior of
the American automobile industry, Journal of Political
Economy 91 (1983) 365}400.
[11] M.A. Ghali, S.P. Dimelis, Classical and variance bounds
tests of the production smoothing hypothesis, International Journal of Production Economics 35 (1) (1994)
15}22.