ECONOMETRIC ANALYSIS OF THE STRUCTURAL RELATIONSHIPS OF THE U.S. COTTON ECONOMY by DOVI-AKUE K. ALIPOE, AG. ENGR., M.A. A DISSERTATION IN AGRICULTURE Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY Approved December, 1984
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
RELATIONSHIPS OF THE U.S. COTTON ECONOMY
by
A DISSERTATION
Submitted to the Graduate Faculty of Texas Tech University in
Partial Fulfillment of the Requirements for
the Degree of
DOCTOR OF PHILOSOPHY
^ ACKNOWLEDGMENTS
I would like to thank my advisor, Dr. Sujit IC. Roy, for his
patience, his continuous guidance, and his search for the
utmost
quality throughout all stages of the dissertation research. I
am
also very thankful of Dr. Don Ethridge's assistance which was
most
instrumental in completion of this dissertation. I am very
grate
ful also to the other members of my doctoral committee, Drs.
Hong
Lee, Roger Troub, and Oswald Bowl in, for their helpful
comments
and highly professional examination of the final output.
I am ^^T'j appreciative of Dr. Thomas Owens's editing of my
dissertation. His editing was offered and carried out at a
^^v)/
critical period in the completion of this work. Lastly, but
not
least, I thank my wife, Tchotchovi, and Barbi Dickensheet for
their
diligent typing of the several preliminary drafts and final
copy
of the dissertation.
Organization of Presentation 8
Domestic Production and Governmental
Domestic Mill Consumption, Exports,
III. REVIEW OF LITERATURE 25
Econometric Models of the U.S. Cotton Sector and Textile Industry
25 Simulation Analysis Using Commodity Econometric Models 51
IV. THEORETICAL CONSIDERATIONS AND
m
Summary of Major Hypotheses 70
V. METHODOLOGY 76
Identification, Linearity, and Reduced Form Equations 83
Stability Condition for Dynamic Models Applied to the Annual Model
of the U.S. Cotton Economy 86
Model Validation Procedures 89
The Single Equation Distributed Lag Model
of Domestic Cotton Mill Consumption 100
VI. RESULTS AND ANALYSIS 103
Estimated Structural Equations 103 Reduced Form Equations, Test of
Model Stability, and Evaluation of Alternative Estimators 113
Impact Multipliers 121
Trend Projections for the Baseline Simulation 130
Simulation Results to 1989-1990 135
IV
Cotton Mill Demand 150
LIST OF REFERENCES 164
1. COEFFICIENTS USED TO PROJECT DOMESTIC PRODUCTION, BASELINE
SCENARIO 174
2. COEFFICIENTS USED TO PROJECT WPOC., RCGEF., AND CEXCC^, BASELINE
SCENARIO ^ ^ 177
3. PROJECTIONS OF ACREAGE PLANTED, YIELD, AND DOMESTIC PRODUCTION,
BASELINE SCENARIO 178
4. PROJECTIONS OF ACREAGE PLANTED AND DOMESTIC PRODUCTION WITH
RATES OF GROWTH SET AT FIVE PERCENTAGE POINTS ABOVE THE BASELINE
179
5. PROJECTIONS OF ACREAGE PLANTED AND DOMESTIC PRODUCTION WITH
RATES OF GROWTH SET AT TEN PERCENTAGE POINTS BELOW THE BASELINE
180
6. DATA 181
8. MAJOR IMPORTERS OF U.S. COTTON, 1960-61 TO 1979-80 188
9. INPUT DATA OF EXOGENOUS VARIABLES USED TO PROJECT BASE LEVEL OF
ENDOGENOUS VARIABLES 189
10. SUMMARY OF SELECTED MARKET AND INDUSTRY SIMULATION STUDIES
190
ABSTRACT
cotton economy of the U.S. since 1933. After five decades,
governmental policies and economic and technological
developments
have produced a downward long-term trend in planted acreage.
Goals
of price and income stability at the farm level have not been
fully
realized because of an imperfect knowledge of the relationships
in
the sector.
The objectives of this study were to Cl) identify and
estimate
the structural relations of the U.S. cotton economy with
alter
native single equation and multiequation methodologies and
12)
simulate policies involving supply controls. Commodity Credit
Corporation loan rates, and U.S. export financing.
The alternative estimators of the structural relationships
were
validated with the use of Theil's inequality coefficients,
the
RMSE's, and the turning points. The I3SLS model provided
better
estimates of the endogenous variables than the 2SLS, 3SLS, and
SUR
models.
Results indicated that the price competition between cotton
and polyester have diminished from the 1960's to the 1970's.
Cot
ton price elasticities of mill demand and export demand
obtained
from the I3SLS were -0.47 and -5.64, respectively. In the
world
market, the main criterion of competition between Upland
cotton
vi
The elasticity of price transmission between U,S. domestic
prices and world prices is 0.57, The Almon lag model revealed
that the full impact of changing fiber prices is realized in
five
years. The main policy implications of the ex ante simulation
scenarios are:
in combination with reasonably restrictive supply controls,
would create the most distortion the first year. Subsequent
adjustments would take place in the domestic fiber economy,
and gross farm income loss would be minimized or reduced to
zero after six to seven years following the initial shock.
- Highly stringent supply control policies would result in
higher farm prices. However, they would also cause gross
farm income to fall due to the severe decline in quantities
of disappearance.
to also put upward pressure on domestic and world prices,
causing U.S. cotton to be less competitive.
v n
LIST OF TABLES
1. Domestic Production, Yield, Carryover, and Foreign Production of
Cotton, 2
2. Structural Equation for U.S. Mill Consumption of Cotton,
Equation 4.19. 104
3. Structural Equation for Domestic Inventories, Equation 4.20.
107
4. Structural Equation for U.S Cotton Export to Non-Communist
Countries, Equation 4.21. 109
5. Structural Equation for the World Price of U.S. Cotton, Equation
4.22. 110
6. Structural Equation for Domestic Farm Price, Equation 4.23.
112
7. Comparison of Alternative Models on the Basis of Root Mean
Square Errors, 1960-61 to 1979-80. 117
8. Reduced Form Equations for the Iterative Three-Stage Least
Squares Model. 118
9. Percent Errors for Endogenous Variables within the Period of
Fit: The Iterative Three-Stage Least Squares Model. 120
10. Root Mean Square Errors and Theil's Inequality Coefficients for
the Endogenous Variables. 122
11. Prediction of Turning Points for U.S. Cotton Exports: I3SLS
Model. 123
12. Point Elasticities Associated with Per Capita Domestic Mill
Consumption of Cotton, August 1960-July 1980. 126
13. Average Elasticities Derived from I3SLS Structural Equations.
128
vm
14. Projections of World Texti le Ac t iv i ty , Domestic Polyester
Prices, and Consumer Price Index, 1984-85 to 1989-90. 132
15. Projections of Regional Cotton Yields and Acreages: Base
Scenario 133
16. Projections of Domestic Cotton Production and U.S. Population.
1984-85 to 1989-90. 134
17. Projected Mill Use of Cotton, Million Bales, 1984-85 to
1989-90. 137
18. Projected U.S. Exports of Cotton, Million Bales, 1984-85 to
1989-90. 138
19. Projected Domestic Cotton Inventories, Million Bales, 1984-85
to 1989-90. 139
20. Projected Domestic Mill Price of S.L.M. 15/16", Cents Per
Pound, 1984-85 to 1989-90. 140
21. Projected Domestic Farm Price of Cotton, Cents Per Pound,
1984-85 to 1989-90. 141
22. Projected World Price of American Short and Medium Staple
Cotton, Cents Per Pound, 1984-85 to 1989-90. 142
23. Projected Domestic Gross Farm Income Originating from Sales of
Lint, Billion Dollars, 1984-85 to 1989-90. 144
24. U.S. Export Earnings Generated from Sales of Cotton Abroad,
Billion Dollars, 1984-85 to 1989-90. 145
IX
LIST OF FIGURES
1. Cotton and Manmade Fibers Shares of the U.S, Fiber Market.
4
2. States and Regions of Cotton Crop Production
in the U.S. 12
4. Flow of Ownership Documents for Merchandising U.S. Cotton.
21
5. Diagrammatic Presentation of the Structural Relationships of the
U.S. Cotton Economy. 71
6. Lagged Effects of Fiber Prices on Mill Use of Cotton:
Distributed Impacts. 152
CHAPTER I
The Federal government has intervened extensively in the
cotton
sector of the U.S. econorny over the last five decades. Before
1933,
when the first cotton legislation was enacted, the American
cotton
crop constituted more than one-half of world cotton output. In
1930,
U.S. production and other world production were, respectively,
13.9
million bales and 12.3 million bales for all growths of
cotton.
Table 1. In the 1920's and 1930's, cotton was the largest of
all
U.S. agricultural exports, and the U.S. share of the world
cotton
market was well above 50%.
In the late 1970's, after nearly fifty years of constant
government intervention in the domestic cotton sector, annual
U.S.
cotton acreage fell to 12.8 million acres, down 29.6 million
acres
from the 1930 level. Although yields in the late 1970's were
more
than three times what they were in the 1930's, total production
in
the United States was up by less than one million bales. The
U.S.
share of the world cotton export trade fell from an average of
47%
in the 1930's to an average of 26% in the 1970's, Table 1. If
the
U.S. export share had fallen to only 35% in 1981, U.S. gross
farm
income of cotton producers would have been $412.8 million
greater
• c o
o c o
A
•t— + J o Z3
-a o s- a. u .^ + J CO 0 ) E o a
. • — t
3 • ! - • -a (A ZD O - M S . =1
Q . O
C/) T 3
o «a:
• C C/)
- O 4 J O u. c
Q . - r -
Q - - M </) (U
• r - «0 > - 3 =
-o Qi Q)
4-> c n CO to 0 ) (U > $- &• u ro < C
O l
> * o > o i OJ 3 ocK
0 )
CVJ CO P ^ Lf>
CVJ CO t-H o
r- CD O CVJ CVI
VO CO VO CO
^ pv. CO r ^ t - H
C>sJ cr» CO 0 0 CVJ
CO O 1—1 CO
o CO
• r ^ i n 0 0 l o IT ) 0 0 CO o t-H i-H CVJ CVJ
^ ^ i n 0 0 CVJ
CVJ r ^ CO t-H ^ C\J CVJ CO
<T> CO
o i n CTt 1 CO CO CO o O^ 0 > CT> CO t—1 1—1 t—1 CT>
1—1
CO CVJ c n ^a-
<y\ CO CO i n o
CO t n CO r > . i-H r*^ ^ CO
CVJ f-H p>. o r ^
f - 4 CO CO CVJ
t - H CO I-H CO ^
o t-H i n 0 0 r - l I-H
CVJ CO o »—» o
CT>
• CVJ <:!• CVJ i n i n i n 0 0 VO CVJ CVJ CVJ CVJ
t—1 O ^ O « ^ VO
CO r>« r ^ 1—1 CVJ I-H CVJ CVJ
<y\ •sf
o i n CT> 1
^ ^ ^ o cr> o^ o^ ^ 1—1 1—1 1—1 O l
I - H
CO CVJ O *-H CT>
0 0 0 0 CVJ i n » - l CVJ CO CVJ
LT) VO CO CO
i n r ^ CVJ t n CO r-H «d- CO
CVJ CO CO ^ O l
"d- CVJ P>* ^
VO 1-H 0 0 0 0 1-H
i n o r ^ VO i n
O «d- «;:*• CO
«d-
• O l r ^ t-H CVJ CO T - l CO CO CVJ ^ ^ CO
CVJ 0 0 O l t-H r-^
r > . CO i n 0 0
O l i n
o i n O l 1
in in in o O l O l O l i n t-H t-H t-H O l
t—I
CVJ 0 0 t-H VO CO CO ^ " CO
cn CO t-H O CO
0 0 r>» CO ^ CO I-H t-H CVJ
o CO CJl 0 0 C J
CO CVJ CVJ ^ j -
r ^ ^ CO o t-H I-H
r-» CO (71 O CO
«;d- ^d- O CVJ
0 0
• VO CO * : f r ^ ' d - c \ j CO r-* ' d - i n ^ ^
CO CO CO t-H r > .
i n CO t-H CVJ
O l VO
o i n O l 1 CO CO VO o O l O l O l CO t-H 1-H t-H O l
I - H
CO VO O l O <:t O l
VO t-H t-H t-H ^ 1 - i n i n i n
t - H O VO O l o
» - i r*^ cy» VO CVJ t-H CO CVJ
VO r ^ CO CVJ o
CO CO CJl i n
O l 0 0 VO O l CVJ
L D i n CO ^
0 0 CVJ CVJ i n i n 0 0
O 0 0 « * . -H
CJl
• 0 0 CO t ^ «;J- CO i n ^ r ^ ^ «:1- i n ^
':^ C\J 0 0 CO 0 0
t-H 0 0 CVJ f-H
C31
r^ o i n O l 1
r^ p r^ o O l O l O l r~* t—t T—t t-H O l
1—I
en ts
> -o <u ro
1 I-H
c • T—
ro 4->
CO
0 0
in that year alone. Competing strength of U.S. cotton was not
only
weakened abroad but also in domestic fiber markets.
Cotton's share of the U.S. domestic fiber market dropped
steadily from 71% in 1951 to 23.5% in 1981, while manmade
fiber's
share increased from less than 22% in 1950 to more than 75% in
1981,
Figure 1. If cotton's share of domestic fiber market had been
main
tained at 30%, gross farm income loss of cotton producers as a
whole
would have been reduced by more than $409 million in 1981.
The goals of the U.S. government in its policies with regard
to the agricultural sector, in general,and the cotton industry,
in
particular, were to support and stabilize farm incomes and
prices,
and to avoid commodity surpluses. These initial objectives
were
generally supported by producers. The secondary effects of
these
policies were not, however, anticipated because of imperfect
knowl
edge about the forces and interrelationships existing in the
cotton
economy.
In practice, the Commodity Credit Corporation loan rate
served
as a floor below which farm level prices could not fall.
Actual
prices, were, however, often characterized by wide seasonal
fluc
tuations and broad yearly changes. For example, in the decade
of
the 1970"s, average annual farm prices for Upland cotton
fluctuated
from a low of 22.81 cents per pound in 1970 to a high of 63.8
cents
per pound in 1976. Annual changes of farm level prices were
highest
from the 1975 to the 1976 crop year t+12.7 cents per pound) and
Towest
75-
70-
65-
60-
55-
: 50-
I • ' ' •
1960
Figure 1. Cotton and Manmade Fibers Shares of U.S. Fiber
Market.
from 1976 to 1977 C-11.7 cents per pound). These wide
fluctuations
in prices caused large variations in income to cotton
producers.
Market share losses in the export and domestic fiber markets and
the
wide farm level price fluctuations can be attributed, at
least
partly, to an Imperfect knowledge about the basic structural
relation
ships of the cotton econorny. Knowledge of such relationships
is
important in order to formulate policies that would stabilize
farm
price and income and maintain the competitive position of U.S.
cot
ton both abroad and in the domestic fiber market.
Objectives of the Study
The overall objective of this research was to analyze the
effects of selected government policies on the U.S. cotton
industry.
Specific objectives were to:
alternative single equation and multi-equation methodologies.
(2) Evaluate the relative performance of alternative models.
(3) Derive elasticities and assess the probable impact of
changes in prices and income on cotton consumption and
other variables.
export financing.
(5) Examine the effects of stated policies on mill
consumption
of cotton, mill price, inventories, exports, world price,
domestic farm price, gross farm Income, and gross export
earnings generated from cotton sales.
Justification of the Study
The 1981 U.S. cotton crop of 15.6 million bales generated
more
than 4 billion dollars In total revenue for U.S. cotton
producers.
Exports represented more than 55% of U.S. cotton disappearance
during
the 1981 crop year and contributed greatly to the national
balance
of trade. Cotton farmers are not the only beneficiaries of a
healthy cotton econorny. Cotton is processed or marketed by
gins,
cotton shippers and merchants, cottonseed oil mills,
warehouses,
transportation facilities, textile mills, garment makers and
other
final producers, and retail outlets. Output of cotton
broadwoven
fabrics represented 33% of the total 7.6 billion dollar
broadwoven
industry in 1981. Collins, Evans, and Barry (1979) reported that
20
million people derived all or part of their earnings from
industries
directly or Indirectly associated with cotton production and
mar-
keti ng.
Knowledge of the basic structural relationships of the cotton
econorny, as well as projections for future years based on
certain
market or policy assumptions, should be of Interest to producers
and
others associated with the cotton Industry and other related
indus
tries.
Costs of government programs related to cotton have Increased
through the years. Expenditures by the Commodity Credit
Corporation
CC.C.C.) between 1934 and 1980 for Upland and extra long
staple
guaranteed loans totaled more than $11 billion dollars (C.C.C,
1980).
Costs were $941 million for the 1977 crop year alone CC.C.C,
1980).
In the early 1980's, political concerns regarding the cost of
government farm programs, including cotton programs, were
increasing
because of a general public awareness of government budget
deficits
and their probable effects. The availability of recent models
of
the U.S. agriculture and the cotton sector,in particular, would
be
useful In devising policies that would reduce or minimize the
cost
of governmental stabilization programs.
national markets are most Important, because the fiber markets
of
many foreign countries Cespeclally developing countries') are
not
mature yet, and thus, manmade fiber consumption Is still
relatively
low. Thigpen (1978) reported growth rates of cotton mill
consump
tion of 4.1 and 3.4% per annum in developing countries and
centrally
planned economies between 1955 and 1973, as opposed to a
negative
0.2% for developed western countries. In addition, analysts
pre
dicted increased political instability in many exporting third
world
countries during the 1980's. An increased need for food
production
due to rising populations was also expected in those countries.
The
U.S.S.R. and the People's Republic of China were expected to
have
8
greater domestic requirements for cotton. The Ejidal system
in
Mexico had caused production to trend down since 1974, and
such
trends are not expected to be reversed substantially in the
1980's.
The cumulative Impact of these events could result in a
slower
average annual growth of foreign cotton exports, and provide
U.S.
cotton producers and shippers with opportunities for increasing
their
share of the world cotton trade if International linkages are
well
understood and proper actions are implemented.
Organization of Presentation
The second chapter Includes a brief overview of the U.S.
cotton
sector. The discussion focuses on production, marketing
activities,
exports, and stocks. Geographical distribution of production
and
historical policy developments are also covered.
Chapter III is a review of past studies that are related to
this research. The first part covers econometric studies of
cotton
mill demand, stocks, exports, prices, acreage, and yield.
Most
previous studies were based on Single equation models of
cotton
sector aggregates, e.g.,Lewis (1972), Lowenstein (1954). A few
of
the past cotton models, nonetheless, were multi-equation
models,
e.g., Blakley (1961).
studies on the basis of econometric models. Although most
simulation
works were non-stochastic, e.g., Behrman's (1971) study on
rubber,
some of the simulation experiments found in the literature
were
stochastic, e.g., Barnum's (1971) study on foodgrains and
Laby's
(1971) work with 1 auric oils. Other topics of Importance in
exami
ning previous simulation studies Include problems of model
validation
and solution approaches used for non-linear models.
Chapter IV consists of theoretical considerations pertaining
to
the structural relations and major variables in the cotton
sector.
Hypotheses concerning the structural relationships of the sector
are
drawn from the theoretical discussions. Forces underlying
mill
demand for cotton are examined from the point of view of the
consumer
theory on one hand, because mill demand is derived from retail
demand
for cotton products. On the other hand, behaviors of millers
are
analyzed in the light of production theories where they are
assumed
to maximize net revenue under given or slowly changing
technologies,
and input (cotton, manmade fiber, energy, and labor) costs.
Chapter IV also Includes discussions on inventory theories and
how
they could be modified to fit Inventory formation in the
cotton
economy. Other theoretical discusssions are related to
exports,
domestic acreage, and effect of domestic supply control
policies.
Chapter V includes discussions on the general methodological
topics and the specific simulation methodologies that are
related
to the present study. The chapter includes a review of the
two-stage
least squares, three-stage least squares, Zellner's seemingly
unrelated regression, and the iterative three-stage least
square
techniques. The statistical properties of different
estimators
10
errors, Theil's Inequality coefficients, absolute, and relative
root
mean square error). Policy Inputs in the simulation were the
Commodity Credit Corporation loan rate, government export
financing
for cotton, and government production controls through
acreage
management. Other inputs for the various scenarios were
polyester
prices, world textile activity, and the world price of cotton
ori
ginating from countries other than the U.S. The chapter ends
with
a discussion of the Almon Lag methodology to examine distributed
lag
effects of changing prices of cotton and polyester on cotton
mill
demand.
In Chapter VI, all estimated models are presented and
analyzed.
Derived reduced form coefficients and the relative performance
of
alternative estimators are discussed, and projections of
exogenous
variables and simulation results are analyzed. The single
equation
distributed lag model of cotton mill demand is presented and
examined. In the final chapter. Chapter VII, the major
implications
of the study are presented and future avenues of research
explored.
CHAPTER II
The cotton fiber was Introduced into the United States at
Jamestown, Virginia, in the early 17th century. Cotton persisted
as
a minor farm crop until the invention of the cotton gin in the
18th
century (Sporleder et al., 1978). In these early days, most of
the
domestic crop was exported to Europe, especially England. In
1850,
for Instance, exports accounted for more than 90% of total
disap
pearance (Sporleder et al., 1978).
The bulk of the U.S. cotton production today consists of
Upland
cotton. Upland cotton comprises all varieties of the
Gossypium
hirsutum species. Some Americana Pima (Gossypium barbadense) is
also
grown In the U.S., but concentrated on a limited number of acres in
the
western and southwestern parts of the country. Production of
American Pima in the U.S. was less than 1% of domestic production
in
1982.
Cotton production in the U.S. is concentrated in the southern
and western parts of the country because the cotton plant requires
a
long growing season and hot summer temperatures. The four
cotton
producing regions of the United States are the West, the
Southwest,
the Southeast, and the Delta (Figure 2).
The West comprises the states of California, Arizona, New
Mexico,
11
12
13
Nevada. The western states mainly grow a longer staple Upland
cotton
with higher micronaire readings. Although the percentage of
acreage
planted in the West was relatively small in the 1920's (less
than
2%), it has Increased gradually in recent years. For instance,
during
the crop year of 1982-83, acreage in the West represented more
than
17% of the national total. Virtually all of the cotton acres
in
the western states are irrigated. Cultural practices in the
West
make it the most productive region. In 1982, for example, yields
in
the West were 3.5 times higher than those in the Southwest, and
1.4
times higher than those in the Delta and Southeast regions.
Competing
crops in the West include wheat, alfalfa, and grain sorghum in
the
mid-Arizona and Imperial Valley areas, and various grain and
vege
table crops in the San Joaquin Valley area (McArthur et al.,
1980).
The Southwest region comprises the states of Texas, Oklahoma,
and Kansas. The bulk of southwestern production comes from the
state
of Texas. Acreage planted in the Southwest has traditionally
repre
sented at least 40% of the national total. In 1982, for
instance,
southwestern acreage exceeded 55% of the total U.S. planted
acreage.
Yields in the southwestern region, however, are the lowest of
all
regions. Weather and other environmental calamities make
abandonment
in the Southwest the highest of all regions, almost 10% in the
1970's
as opposed to 4.6%, 6,8%, and 5.3%, respectively, in the West,
South
east, and Delta. Competing crops in the Southwest region
include
grain sorghum, corn, and wheat in the High Plains area; sorghum,
wheat.
14
oats, and hay crops in the Texas Blackland area; and mostly
grain
sorghum in the Rolling Plains (McArthur et al., 1980).
The Southeast region Includes the states of Virginia, North
Carolina, South Carolina, Georgia, Florida, and Alabama.
Although
the first cotton crops were harvested In the Southeast region,
total
acreage planted In the region has gradually declined. In
1982,
southeastern acreage represented 5.6% of the national total, a
sub
stantial decline from the 24.4% averaged in the 1930*s. The
decline
in cotton acreages in the Southeast may be attributed to
decreasing
return/cost ratios as well as changing comparative advantage
between
cotton and the alternative crops that are available to growers.
In
the latter part of the 1970's, crops that presented serious
competition
to cotton in the region were tobacco and peanuts in the
northeastern
part of the region, and corn and soybeans in the central and
Limestone
Valleys (McArthur et al., 1980).
The Delta region Includes the states of Missouri, Arkansas,
Tennessee, Mississippi, Louisiana, Illinois, and Kentucky.
Acreages
In the Delta have fallen from an average of 27% in the 1950's to
less
than 22% in 1982. Yields in the Delta area are comparable to
yields
in the Southeast region (over 740 pounds per harvested acre
during
the 1982 crop year). Alternative crops in the Delta region are
soy
beans and rice (Evans, 1977 and McArthur et al., 1980).
Total domestic cotton production has fluctuated widely from
year to year. The biggest crop ever harvested was that of
1937-1938.
15
(18.9 million bales). Year to year fluctuations are due to
weather,
environment, prices, and effective governmental policies.
Government
programs in the cotton sector have Included various programs of
price
support, acreage allotments, acreage diversion or soil
conservation
payments, non-recourse loans by the Conmodity Credit
Corporation,
and various export programs. Previous policies, in combination
with
economic and technological developments, have produced a
negative
long-term trend in acreage planted.
The first Federal legislation which included cotton
production
was the Agricultural Adjustment Act of 1933. Under the 1933
Act,
the Secretary of Agriculture was directed to achieve a "fair"
exchange
price for cotton (parity) by (1) securing voluntary reduction
of
acreage through agreements with producers, (2) regulating
marketing
through voluntary agreements with all participants involved in
the
sector, (3) licensing processors, producers, and other
entities
involved in the marketing process to avoid unfair practices or
charges,
(4) determining the necessity for and the rate of processing
taxes,
(5) using tax proceeds and appropriated funds for cost of
adjust
ment operations, for expansion of markets, and for the removal
of
agricultural surpluses. Low cotton prices (29 cents per pound
in
1929 and 6.5 cents per pound in 1933) Insured substantial
grower
participation. Growers agreed to plow only 25% to 50% of
their
initial base acreage in return for rental payments in cash. In
add-
tlon, a non-recourse loan rate of 10 cent/pound for the 1933
crop
16
to be Increased to 12 cents/pound for 1934, was set up by the
C.C.C,
established on October 17, 1933. The effects of the Act of
1933
combined with economic conditions prevailing at the time resulted
in
a decrease in cotton acreage harvested by 30% from 1929-1930
levels
(McArthur et al., 1980).
The Soil Conservation and Domestic Allotment Act of 1936 had
the objectives of (1) promoting soil conservation and profitable
use
of agricultural resources and (2) reestablishing and maintaining
farm
income at fair levels. Farmers were offered payments to shift
from
soil depleting crops to soil conserving crops. Prices had
continued
to fall despite earlier actions, and Congress made available
$130
million for cotton adjustment payments to producers. Despite
these
measures, the general economic environment forced cotton farm
level
prices to decline to 11.09 cents/pound in 1935 and 8.41
cents/pound
in 1937.
vation program of the 1936 legislation with characteristics aimed
at
meeting drought emergencies as well as price and income problems
due
to commodity surpluses. The 1938 cotton legislation was
reinforced
in 1942 when the crop insurance was extended to cotton. The
insurance
program further protected growers from risks related to crop
failures
caused by drought, floods, or other natural disasters. The
Adjustment Act of 1938 succeeded in reducing acreage harvested
by
about 28% (McArthur et al., 1980).
17
During World War II, Commodity Credit Corporation acquired
inventories were used In the war effort. To insure that
farmers
shared in the profits of defense contracts, the C.C.C. loan rate
was
raised to 85% parity in 1941, 90% in 1942, 95% in 1944, and 100%
in
1945 (Textile Economics Bureau, 1981).
Agricultural legislation passed in 1948 and 1949 continued
the
price support for cotton at varying levels above 75% of parity.
The
parity formula was modified by an amendment to the 1949 Act
to
include wages paid to hired farm labor in addition to wartime
pay
ments made to producers. Allotments were not in effect between
1943,
and 1949 and acreage harvested Increased by more than 25%
(Halcrow,
1977, McArthur et al., 1980, and Textile Economics Bureau,
1981).
Legislation passed in the 1950's included the Agricultural
Acts of 1956 and 1958. The Act of 1956 established the Soil
Bank
in order to take out of production a portion of the nation's
farm
land and provide a better balance between commodity supplies
and
demand. Under the Act, cotton acreage fell by more than 7%;
however,
this acreage reduction had little effect on farm level prices
which
remained around 33 cents per pound. The 1958 legislation
provided
farmers with a choice between (1) a regular acreage allotment
with
price supports or (2) an increase of up to 40% in allotments with
a
price support 15 points lower than the percentage of parity set
under
alternative (1). Price supports were authorized at between 70%
and
18
90% of parity for 1961 and between 65% and 90% of parity after
1961
for regular allotment crops (Halcrow, 1977, McArthur et al.,
1980,
and Textile Economics Bureau, 1981).
The Act of 1964 authorized the U.S. Secretary of Agriculture
to
make subsidy payments to domestic cotton shippers and textile
mills
in order to allow American cotton to compete effectively in
foreign
markets (Textile Economics Bureau, 1981). The Food and
Agriculture
Act of 1965 provided for a price support of U.S. cotton at
levels
not higher than 90% of world prices. These lower support
prices
improved the competitive position of U.S. cotton in the world
market,
as exemplified by an Increase in U.S. cotton exports (from 3
million
bales in 1965-66 to 4.8 million bales in 1966-67).
Under the farm legislation of 1970, farmers were required to
keep out of production up to 28% of their base acreage in order
to
be eligible for a payment equal to the difference between 65%
of
parity and market price or 35 cents per pound and market
price,
whichever was higher (Halcrow, 1977).
The Agricultural and Consumer Protection Act of 1973,
contrary
to previous legislation, emphasized increasing or maintaining
pro
duction. A new concept of target price was introduced.
Payments
were to equal the difference between market price and target
price;
however, such payments were not to exceed the difference between
the
target price and the C.C.C loan rate. Disaster payments were
authorized by the Act of 1973 for eligible cotton growers.
These
19
payments were to be available when natural disaster prevented
a
particular producer from harvesting 2/3 of normal acreage
(Halcrow,
1977, McArthur et al., 1980, and Textile Economics Bureau,
1981).
More recent policies are the Food and Agriculture Act of
1977,
its amendment for 1978, and the Payment-in-Kind program
initiated
during the 1982-1983 crop year. Under the Act of 1977, the
target
price for cotton was set at 70.87 cents per pound for
1981-1982.
Growers who planted less than their 1980 acreage were eligible
for
full deficiency payments and Commodity Credit Corporation loans
on
crops. The Payment-In-Kind program was particularly attractive
to
producers, because, unlike the 1981 program, deficiency
payments
were not limited to $50,000 per farm. In addition, payments
were
made In kind with previously acquired C.C.C. inventories. An
earlier
survey of growers' intentions taken in February 1983 showed that
as
a result of the program, acreage planted could drop by 19%
from
preceding crop year levels (Collins, 1983). Effects on prices
were
not known at the time.
Marketing Practices in the Sector
Marketing practices in the sector may be examined from two
perspectives: the physical flows of merchandise and the flows
of
ownership documents CSporleder et al., 1978 and Glade and
Ghetti,
1979). Documents of ownership, mostly used in the sector, are
receipts and bills of lading. The itineraries of cotton and
sales
documents through the marketing channel are presented in Figures 3
and 4.
20
Figure 3. Physical Flow of U.S. Cotton.
Source: E.H. Glade, Jr. and J.L. Ghetti. Marketing U.S. Cotton to
Domestic and Foreigri Outlets in 1977-78: Practices and Costs.
USDA, ESCS, No. 79, May 1979, p. 5.
21
Figure 4. Flow of Ownership Documents-for Merchandising U.S.
Cotton. 1
1 Source: Thomas Sporleder, James Haskel,*Don Ethridge, and Robert
Firch. Who Will Mar'ket Your Cotton, Producer Alternatives-. Texas
Agricultural Extension Service, D-1054, March 1978, p. 5.
22
The marketing process begins when the producer delivers seed
cotton to the gin where seed and lint are separated. The lint
is
then sent to warehouses for storage and/or further
compression.
After compression to universal density (28 pounds per cubit
foot),
the lint in the form of bales weighing approximately 480 pounds
is
sent to domestic textile mills, domestic ports, or Canadian
mills.
Southern textile mills do not require universal density bales.
How
ever, most new cotton gins built today install universal
density
presses in order to avoid the need for additional
compression.
Documents of ownership may be directly transferred from the
producer to the shipper. They may also be transmitted via
mill
buyers, ginners, local buyers, brokers, conmlssion firms, the
Commodity Credit Corporation, or cooperatives. The services
per
formed and marketing bills increase with the number of
intermediaries
Involved.
A number of other services, besides ginning and
reconcentration,
are performed between the farm and mills or ports. These
include
transportation, other warehouse services (resampling,
reweighing,
patching, etc.), insurance, hedging, financing, selling, etc.
Trans
portation was the largest marketing cost item during the
1977-1978
crop year. It constituted 28% and 42% of total marketing bills
for
domestic and foreign shipments, respectively (Glade and Ghetti,
1979).
23
Domestic Mill Consumption, Exports, and Inventories
In the earlier days, the bulk of American textile mills were
located In the northeastern part of the country (Sporleder et
al.,
1978). Today, more than 90% of U.S. textile mills producing
100%
cotton or mixed fiber fabrics are located in cotton growing
states.
In 1980, for Instance, 97.29% of cotton mill consumption was
attri
butable to textile mills located In the cotton growing states of
the
Southeast, Delta, and Southwest. Mill consumption in the state
of
North Carolina was by far the largest of all states, with
2,082,000
bales or 35.47% of the national total.
Aggregate domestic mill consumption of cotton was highest in
the 1940's (9.667 million bales), and lowest in the 1930's
(6.202_
million bales). Although year to year fluctuations in
aggregate
mill consumption are small, per capita mill use has dropped
steadily
through the years. This loss of market share has been attributed
to
several factors, including unstable cotton prices and
quantities
(Shafer, 1978), high cotton prices (Waugh, 1964), and certain
characteristics present in manmade fibers but not currently
attainable
with cotton (Ward, 1968).
Aggregate U.S. cotton exports averaged 7.426 million bales in
the 1920's. They fell to 2.721 million bales in the 1940's,
because
of greater domestic mill use destined to sustain the war
effort.
Exports are very sensitive to domestic price supports,
international
trade policies, and the world price of cotton originating in
other
24
countries. Per capita export has decreased substantially, as
did
the U.S. share of the world cotton trade (47% in the 1930's and
26%
In the 1970's). Since the 1950's, the consistent major importers
of
U.S. cotton have been Japan, South Korea, and the Western
European
countries. In recent years, the People's Republic of China has
also
emerged as a major importer. During the 1980-81 crop year,
for
Instance, 23% of American exports were destined for the
People's
Republic of China, 22.06% for South Korea, 19.29% for Japan,
and
5.95% for Taiwan.
during the years of intensive governmental price support.
Carryover
increased from an average of 3.1 million bales in the 1920's to
an
average of 8.4 million bales in the 1930's. Stocks averaged
more
than 8 million bales in the 1940's, 1950's, and 1960's.
Sectorwide
inventories have been lower in recent years, averaging only
3.6
million bales during the years of the 1970's.
CHAPTER III
Econometric Models of the U.S. Cotton Sector and Textile
Industry
Some pioneer studies pertaining to cotton acreage response In
the United States include those by Moore (1917), Smith (1928),
and
Manny (1933). The Moore study was one of the first to
demonstrate
that there Is a direct relationship between the change in
cotton
acreage for any given year and changes in the price of cotton
lint
in the preceding year. The Smith study showed similar
results.
The Manny study revealed further that farmers' intentions to
plant
are depended on preceding years' prices and provided the
behavioral
underpinings of the relationships observed earlier by Moore
and
Smith. No attempt was made in any of the three studies to
measure
the elasticity of acreage response to price.
Walsh (1944) provided one of the first formal regression
models
of cotton acreage in the U.S. The Walsh model was used to
derive
an elasticity of farmers' response to lagged price in making
production plans. The depended variable—acreage of cotton in
cultivation on July 1—was explained by the price of cotton at
the
farm level adjusted for changes in the index of prices paid
by
farmers for all inputs purchased. Statistical results
revealed
that the acreage-price response operated at two distinct levels
in
the periods 1910-1924 and 1925-1933. The change in the
acreage-
25
26
price relationship between 1924 and 1925 was attributed to
(1)
control of the boll weevil and (2) expansion of cotton into
new
areas in the Mississippi region and the Southwest. The price
of
cottonseed changed from being an insignificant to a
significant
Independent variable from the first period to the second.
Price
elasticities at the means of the sample data, in the period
1925-
1933, implied that a 1% increase in cotton prices was followed
by
a 0.24% Increase in acreage planted. Response of production
of
cottonseed to the price of cottonseed yielded elasticities of
supply near zero.
Two of the early formal studies of factors affecting cotton
demand and prices were undertaken by Cox (1926) and Smith
(1928).
The Cox study was mainly descriptive while that of Smith
included
a regression equation specified in a double log functional
form.
The Smith equation sought to explain domestic cotton prices
with
supply and the overall domestic price level. Cotton was the
dominant fiber in the domestic fiber market, so no
consideration
was given to synthetics. The signs in the Smith equation were
consistent with a priori expectations. However, appropriate
statistical tests were not performed on the regression
coefficients
Furthermore, the specified relationship, although
incorporating
elements directly related to prices, was neither a demand nor
a
supply relationship. The period of fit covered 1905 through
1924.
27
A study by Lowenstein (1954) was made in a period when rayon
consumption had become an Important determinant of cotton
mill
demand. The objective of Lowenstein's study was to ascertain
factors that affected cotton mill consumption as well as
determine
their Individual effects. The period of study was 1921-1950,
excluding the war years and 1946. The functional form of the
equation was logarithmic. The Independent variables were: the
index of industrial production per capita, rayon consumption
per
capita, and the average price of middling 7/8-inch cotton at
the
10 spot markets. Results indicated that a change of 1% in
Industrial
production was associated with a change in cotton consumption
of
0.84%, ±0.12%, In the same direction. A 1% Increase in rayon
con
sumption produced a 0.12% decrease In cotton consumption,
±0.03%,
while a change of 1% in the price of cotton caused cotton
con
sumption to change by 0.30% in the opposite direction. The
price
inelasticity of cotton demand at the mill level was explained
by
its "derived" nature. Raw cotton demand by mills was directly
dependent on mill output of grey goods and cotton yarn, which
reflected ultimate consumers' and industrial users' demand.
Cromarty (1959) integrated agricultural and non-agricultural
sectors of the United State econorny. Econometric models of
several commodities were developed to measure agricultural
effects
of non-agricultural shocks. The agricultural sector was
disaggre
gated into twelve product categories, including cotton. The
28
cotton component of the study was composed of four stochastic
equations and one identity. The stochastic equations of the
model
sought to explain cotton production, mill consumption, farm
price,
cotton Inventories (excluding C C C stocks) and government
demand,
Cromarty's equation on domestic production included policy
variables such as allotments and price supports. A notable
result
was the effect of acreage in the West on total domestic
production.
A 1% Increase in the proportion of Far West acreage increased
domestic production by more than 71,000 bales. Domestic mill
consumption in the model was determined by farm gate prices,
pri
vate inventories, disposable Income, and the general price
level.
Inventory demand was dependent on farm price, available
domestic
supply, and foreign supply.
Estimates of the model's parameters were obtained with
annual data from 1929 to 1953. The model was evaluated on the
basis of historical forecasts (1954 and 1955) and a comparison
of
the model's point elasticities with elasticities from other
studies. The author concluded that the model overestimated
prices
and underestimated the quantity variables. Prediction errors
were
attributed to an overestimation of price level variables in
the
master Klein-Golberger model of the United States economy.
The
price elasticity of cotton supply was +0.361. Price
elasticities
of demand were -0.30 for mill demand, -0.211 for inventory
demand.
29
+1.252 for government demand. The price elasticity of mill
demand
derived by Cromarty was close to Lowenstein's (1954). Direct
policy
analyses were not undertaken with Cromarty's model.
Blakley's model (1961), unlike Cromarty's, was solely
addressed
to the U.S. cotton economy. The postulated structural model
con
tained the following endogenous variables: per capita
domestic
mill consumption, the 10-market spot price of middling 15/16",
per
capita commercial exports of cotton from the United States,
per
capita domestic ending Inventories, the world cotton price,
and
per capita foreign mill consumption of cotton. The model had
14
exogenous variables. Including disposable income, domestic
produc
tion of manmade fibers, foreign supply and inventories, the
domestic loan rate, and transfer cost from the United States
to
foreign countries. The model was initially designed to
capture
short-run as well as long-run responses in the sector. Using
Nerlove's methodology, a partial adjustment mechanism was
assumed
for the domestic and foreign mill demand equations. The
partial
adjustment model was: Y. - Y._, = Y ( Y % - Y^_J where Y. is
the
actual current quantity variable, Y._. is the quantity
variable
lagged one period, Y* is the desired quantity, and y is the
coefficient of partial adjustment. For a generalized demand
equation of the form Y*. = aX. + U., with X. the current
observed
price, the partial adjustment model became Y. = ayX. + (a
-Y)Y^_,
+ yU.. The existence of a long-run relationship distinct from
the
30
short-run one was tested by the statistical significance of
the
estimate of the quantity (a - y ) .
The structural parameters were estimated using ordinary least
squares, the limited information single-equation maximum
likelihood
(L. I .S.E. ) , and the Theil-Basmann methods. The data used
extended
from 1921 to 1956 or from 1931 to 1956. In both cases, the
war
years and 1946 were omitted, providing periods of f i t twenty
years
or thirty years long. Estimates of the parameter pertaining
to
lagged domestic mill consumption were not statistically
significant,
and the hypothesis that short-run and long-run elasticities of
mill
demand were equal was substantiated.
The various estimates of the parameters of the system also
represented elasticities because of the double-log
specification.
Price elasticity of mill consumption and Inventory demand
were
-0.85 and -0.83, respectively, higher than Cromarty's, -0.30
and
-0.211. Short-run and long-run price elasticity of foreign
mill
demand were, respectively, -0.13 and -0.66 for the L.I.S.E.
pro
cedure .
equations of total domestic production and acreage were
estimated
and used for conditional projections. Total domestic
production
was dependent on acreage planted, abandonment rates, yield
reduc
tion from weather and pest infestations, and a trend. Planted
31
lagged acreage or allotments, and a trend.
Results of specific disturbances indicated that if domestic
production controls had been Imposed at 8 million bales during
any
of the years of the periods of fit, domestic prices would
have
risen to 100 percent of parity level. The foreign market of
U.S.
cotton, however, would have disappeared and gross farm income
from
cotton would have fallen substantially. A two-price plan with
the
domestic price set at 100% of parity and a world price set
suffi
ciently low was found to result in the highest gross farm
income
from cotton.
agriculture sought to (1) develop a comprehensive description
of
the relations between prices of farm products and the
quantities
that can be disposed of through commercial market channels at
those
prices and (2) explore market possibilities for increasing
farm
prices and farmers' cash receipts through supply control. The
aggregate agricultural model Included components of commodity
demands at the retail, export, industrial, and derived farm
levels.
Double-logarithmic and semi-logarithmic functional forms were
used in specifying the export demand, industrial demand, and
farm
level demand equations. The domestic price at the farm level
was
used in all equations.
year price-quantity situation as well as the expected 1965
price-
quantity situation. The industrial demand equation was
expected
to hold for 1955-1957 and 1965. The computed total demand
equation satisfied both the 1955-1957 situation as well as
that
of 1965. Export demand, industrial demand, and farm level
demand
price elasticities of American cotton were -3.66, -0.40, and
-0.127,
respectively. The magnitude of Brandow's price elasticity of
industrial demand was close to Cromarty's and Lowenstein's
(-0.30).
Furthermore, results indicated that i f total production were
cleared from the market during the 1955-1957 period, farm
level
prices would have been 20% lower on the average. Restrictions
on the size of the cotton crop would have reduced the total
value
of l int and seed in the long-run. Total cotton production
costs
would have declined; however, no analysis was made to see
whether
the decline In costs would offset the loss of gross revenue.
Waugh (1964) estimated long-run demand relationships for
cotton fiber using a somewhat unique procedure. The Waugh
methodology, unlike Nerlove's, Koyck's, and Cagan's, did not
assume
any form of distributed lag for long-run effects to take place. I
t
was postulated that a rise in the price of cotton had only a
small
direct and immediate effect on cotton consumption. Over a
period
of years, however, such rises would increase the production
and
consumption of rayon and non-eellulosic fibers, further
decreasing
33
cotton consumption. It was assumed that the ratio of mill
con
sumption of cotton and mill consumption of rayon and acetate
was
dependent on past years' price ratios of cotton to rayon and
acetate.
The data utilized in the analysis was annual and covered 1933
through 1947. The estimated equations were:
Q^ = 11.70 - 4.28 P^_3 - 2.08 P^_g - 0.23 P^_g (3.1)
(0.70) (0.77) (0.52)
Q^ = 11.32 + 0.73 P^ - 4.79 P^_3 - 2.21 P^_g (3.2)
(0.63) (0.69) (0.48)
sumption ratio of cotton to rayon and
acetate;
P. = current three-year moving average of price
ratio of cotton to rayon and acetate.
P._3, Pf.g* and P. g are price ratios centered 3, 6, and 9
years
before the current year. The R for the equations (3.1) and
(3.2)
were 0.95 and 0.97, respectively.
The regression results were converted to reflect impacts on
an
annual basis by dividing each coefficient by 3. The
coefficients
were then plotted as a smooth curve to obtain weights by
individual
years as well as cumulative weights. Cumulative weights
represented
long-run impacts. A long-run (after 9 years) price elasticity
of
mill consumption of -1.84 was derived. This long-run
elasticity
was found to be distributed among the years as follows: -0.29
for
34
for one year, -0.65 for two, -1.00 for three, -1.54 for five,
and
-1.84 for eight and 9 years. The one-year cotton price
elasticity
of mill consumption derived by Waugh was close to the
elasticity
of -0.30 derived by Lowenstein (1954) and Cromarty (1959). In
the
long-run, mill consumption of cotton was found to be highly
elas
tic. A major policy implication of the Waugh analysis was that
if
cotton programs could be designed to keep cotton prices low,
long-
run benefits such as increased mill consumption would follow;
however, the total impact would not be felt before 9 years.
Ward (1968) concentrated on demand for cotton within an
inter-
fiber competition framework. He estimated demand
relationships
for consumer goods (e.g., men's apparel, women's apparel),
for
semi-manufactured products (e.g., yarn, broadwoven fabrics),
along
with mill demand equations for raw fibers (cotton, cellulosic
fibers, and non-eellulosic fibers). Static and dynamic
statistical
models were used to obtain end-product demand relationships.
Results indicated that retail demand for textitle products,
including cotton and manmade fibers, were income elastic.
Using
annual data, the short-run and long-run effects of price and
income
on retail demand were found to be similar. Demand
relationships
for semi-manufactures used in the domestic production of each
category of consumer products were obtained by assuming
constant
inventories and known quantities of end-products entering
trade.
35
Equations were derived for total mill use and for end use
categories to show the price relationships between competing
fibers
and to determine the role of technological change in fiber
com
petition. Conclusions indicated that prices of manmade fibers
were
important determinants of cotton mill demand. However, 50% of
«
cotton-manmade fiber substitution was due to non-price
competition,
e.g., easy care characteristics and abrasion resistance present
in
most manmade fibers. The two major determinants for growth in
cotton's share of the fiber market were research and
promotion.
Wallace, Naylor, and Sasser (1968) presented an econometric
model of the United States textile industry. The basic
objective
of the study was to provide a tool for a better understanding
of
relations in the textile economy. The final model was a
recursive
system with nine endogenous variables, namely apparel demand,
apparel output, textile demand, textile output, employment of
pro
duction workers, production worker earnings, textile
products'
prices, textile mills' profits, and investment in the textile
industry. Ordinary least squares was used to estimate the
equations of the system. An intermediate version of the
original
conceptual model was modified to remove autocorrelation of
the
error terms, following Theil and Nagar (1961).
The Theil and Nagar method of ridding of first-order corre
lation uses differences of the observed values of variables
adjusted
36
"t = P(Vl) * h C3.3)
^t-1 "" ^9ged disturbance term;
with constant variance;
1 - Durbin-Watson coefficient/2.
becomes after transformation
+ b^{x2^ - pxg^.j} + {U^ - pU^.j} (3.5)
The ordinary least squares procedure was appropriate because of
the
recursive nature of the model. The data utilized to obtain
the
textile model covered a period from 1951 through 1962 for a
total
of 144 monthly observations. The performance of the model was
evaluated by the same authors (1967) using spectral analysis
and
turning point .criteria on historical simulation data. The
model
was accepted as a valid representation of the textile
econorny
(1951-1962). No direct policy implications were drawn from
the
model.
37
A market share approach was employed by Sirhan and Johnson
(1971) to examine the demand for U.S. cotton in Great Britain
and
Germany. The estimated market share relationships was based
on
Nerlove's partial adjustment mechanism, and provided
coefficients
of short-run and long-run price effects on the U.S. cotton
market
share in Europe. Three alternative specifications were used
for
the British market. The first alternative consisted of
regressing
the U.S. share in Great Britain (Mb.) against two independent
variables: (1) the ratio of the Liverpool price of U.S.
cotton
and average of Liverpool prices of cotton originating from
Mexico,
Nicaragua, Syria, and Iran (P.) and (2) the lagged market share
of
U.S. cotton In Great Britain (Mb._J. In the second
specification,
a trend variable (T) was added to the two exogenous variables
dis
cussed earlier. The final alternative consisted of regressing
Mb.
against the logarithm of P. and the logarithm of Mb._..
Models
specified for the West German market were similar to those of
the
British market. The price variable used in the German market
was
the ratio of the Bremen price of U.S. cotton and the average
Bremen prices of Mexican, Nicaraguan, Syrian, and Iranian
cotton.
Short-run price elasticities for the British market was as
high as -8.67 and as low as -2.70. Long-run price elasticities
of
export demand for Great Britain were -20.16, -9.67, and
-14.0.
Examples of short-run and long-run elasticities derived for
the
West German market were -7.62 and -11.04. It was concluded
that
38
a high degree of price competition existed between American
and
other countries' cotton. For same qualities of cotton, price
was the major determinant in consuming foreign
establishments.
Although the methodology utilized was appropriate, some of
the statistical results, e.g. , standard error of regression
coeffi
cients were not indicative of a good statistical f i t . On
the
basis of obtained standard errors, the Nerlovian partial
adjustment
coefficients were not statistically different from one (for
the
West German market). Short-run and long-run price effects on
the
market share of U.S. cotton in the West German market were
identical,
Results regarding interfiber competition indicated by Ward
were likewise obtained by Smith and Dardis (1972) using a
Markov
chain approach. A first-order Markov process was assumed. In
such
a process, the probability P.. (transition probability) of
moving
from one state(S.)to a second state(S.) depends only on the
prior
state(S.). Although other procedures were available for
estimating
the transition probabilities, the Smith and Dardis study
utilized
minimum absolute deviations in the attainment of the init ial
objec
tives (to examine interfiber competition and make projections
of
cotton's share in alternative end uses).
Market share of cotton was projected to decline for all end
use categories. At equilibrium, i t was projected to reach zero
for
men's ut i l i ty clothing, men's shirts, men's underwear, women's
and
children's dresses, bedspreads and quilts, retail
piece-goods.
39
rugs and carpets, sewing thread, narrow fabrics, women's and
children's playwear, men's separate slacks and hosiery, rope
and
cordage. Cotton's share of all these end uses was on the
average
58.08% In 1967. Smith and Dardis Indicated, however, that the
equilibrium is not expected to be reached for many years
after
1999, meaning that a considerable amount of time was available
for
planning and Implementing cotton programs designed to improve
its
competitive position. The approach employed could not
indicate
what type of programs would be effective in lessening the loss
of
market share.
Lewis (1972), as did Blakley (1961) and Sirhan and Johnson
(1971), assumed Nerlove's partial adjustment in estimating
demand
relationships for seven textile fibers in the United States.
The
Lewis model was obtained by regressing the poundage of cotton
fiber
purchased or shipped to mills (DC.) against the domestic level
of
real per capita income (Y.), the price paid by mills for
cotton
(P-.), an Index of prices of synthetic fibers lagged (P2t.i^»
^^^
rate of change in real income (DY^), a dummy variable for the
war
years (WD.), a dummy variable for the depression years (DD^),
and
DC. T. The ordinary least squares estimate of the cotton model was:
w— J.
DC^ = 17.88 + 2.13 DC^_j + 0.57 Y^ - 1.045 P^^
(10.0) (2.6) (3.8) (3.5)
(5.9) (3.4) (6.9) (5.4)
Price and Income elasticities computed at the means of the
data series were -0.37 and +0.17 for cotton and -1.45 and
+0.46
for rayon and acetate staple. The partial coefficient of
adjust
ment and adjustment path for cotton were estimated at +0.79
and
+1.69. Reaction speeds were fastest for cotton and wool.
Observed
rapid adjustments were attributed to non-durability and
well-known
cotton qualities by consumers. The main policy implication
drawn
in the Lewis study was:
The significant cross-relationship between synthetic fiber and
cotton and wool Indicated that policy decisions affecting the price
of one fiber may have some impact on the other markets in the
industry. If such is the case, government pricing and stockpile
policies for natural fibers may affect or even be offset or
confounded by changes in the manmade fiber sector.
Principal factors affecting total fiber consumption, textile
mill use of cotton and apparel wool, were likewise examined
by
Evans in 1972. Ordinary least squares were applied to data
extending from 1955 to 1976. Three alternative specifications
were put forth. In all cases, domestic mill consumption of
cotton
was determined by domestic fiber consumption and prices of
cotton
and polyester.
The Evans models differ from those by Cromarty (1959),
Blakley
(1961), Lowenstein (1954), Brandow (1961), and Lewis (1972)
in
that the own price and cross price variables were lagged one
year.
Results indicated a mill demand price elasticity of -0.32.
Cotton
mill demand curves were estimated for three levels of total
fiber
41
use (50, 60, and 70 pounds per capita). The t-statistics
pertaining
to the estimates were indicative of good statistical fit. The
rationale for using lagged cotton and polyester prices was
that
processed Inventories were acquired prior to use. Mill use
was
planned, and plans were made according to current observed
prices.
Factors affecting the wholesale price of cotton-broadwoven
fabrics were analyzed by Evans in 1977. Production of all
cotton-
broadwoven fabric accounted for about 60% of mill consumption
of
cotton in 1971-1976. Three alternative models were estimated.
In
the first model, the wholesale price of cotton-broadwoven
fabrics
(WPIC.) was dependent on the lagged price of cotton at group
201
mill points (PCT._,) and the lagged average hourly earnings of
pro
duction workers in domestic mills (W^_J. In the second model,
one additional independent variable (WPIC._.) was added to
(PCT.)
and (Wa^_,). The third specification included an excess
demand/
supply variable in addition to the earlier two. Ordinary
least
squares was applied to semi-annual data extending from 1966 to
1975.
Results Indicated that wages Impacted more than mill prices on
the
wholesale price of cotton-broadwoven fabrics. The third model
indicated that a 1% Increase in wages or cotton prices led to
0.48%
and 0.18% increases in (WPIC) in the short-run. Effects of
imports
were examined through the excess demand/supply variable. It
was
estimated that a net import balance equal to 5% to 10% of
domestic
42
output would eventually lead to declines of 2.9% to 5.5% in
the
index of cotton-broadwoven cloth price, other things being
equal.
Farm level competitive relationships between cotton and other
crops were also examined by Evans (1977). Cotton breakeven
prices
were computed for the Delta (Louisiana, Mississippi,
Missouri,
Tennessee), the Southeast (Georgia, Alabama, the Carolinas),
the
Southwest (Texas and Oklahoma), and the West (California,
Arizona,
New Mexico). The breakeven formula utilized was:
BEPCT = (P)(Y) - VC + VCCT/YCT (3.7)
where BEPCT = breakeven price of cotton, cents per
pound;
4 months of calendar year);
Y = average competing crop yields in last
3 seasons;
crop, $/acre;
costs, $/acre;
During 1976-1977, crops that were competive with cotton were:
soybeans and rice in the Delta; soybeans and corn in the
Southeast;
grain sorghum in the Southwest; and barley, alfalfa, and wheat
in
the West. Ordinary least squares analysis of Upland cotton
acreage
43
planted based on 1959-1976 annual data revealed that a 0.01
$/pound
Increase in the breakeven price of cotton resulted in a decrease
of
299,000 acres in planted area.
In most aggregate models of crops (field crops as well as
tree crops), e.g., Blakley (1961), total production is
generally
composed of acreage response and yield components. Acreage
planted
has usually been explained by market and policy variables,
e.g.,
price received by producers, allotments, and deficiency
payments.
Aggregate time-series yield variations are ordinarily
attributed
to environmental and technological factors. Houck and
Gallagher
(1976), along with Guise (1969), Krishna (1964), and Hee
(1958),
incorporated market condition variables into aggregate yield
equations.
aggregate yield modeling was reviewed by Houck and Gallagher
(1976).
Using a neoclassical production function where total
production
was determined by fertilizer and land factors, a supply
function
was derived. Total supply was determined by input costs,
output
prices, and land use. Aggregate yield by definition was equal
to
total supply divided by acreage harvested. Following the
above
theoretical development, aggregate yield was ascertained by
supply
factors, e.g., production costs, returns, and acreage
harvested.
Increased returns and/or acreage harvested were expected to
decrease
yield if production occurred in the rational stage of
production
44
(stage II), e.g., where the marginal product of land was less
than its average (yield). The significance of such reasoning
is
that if government acreage limitation policies result in
taking
out of production marginal land areas, the effect is an
increased
per acre yield. Supply reductions are, thus, less than the
relative
acreage reductions and a fully successful x% acreage
reduction
policy results in less than x% decrease in total production.
Such
hypotheses were tested and confirmed by Houck and Gallagher
(1976)
with data from the U.S. corn sector.
Following Houck and Gallagher, Evans and Bell (1978) incor
porated market and economic variables into their regional
yield
equations (Southeast and West). For the Southeast, yield per
har
vested acre (YSe.) was determined by the average price
received
by farmers in the region (PCT), the index of production costs
in
the region (INC), acreage harvested (HA), a variable
representing
departure of rainfall from normal during the growing season
(SUMRAIN), and trends in Southeast yields for 1951-1965 and
1966-
1974, respectively. The estimated equation showed a definite
effect of market variables on Southeast yields. A one-point
change
in the ratio of price received to production cost resulted in
a
change of 6.2 pounds per acre in the same direction. The model
put
forth for the western region, following a slightly different
specification, showed similar results. A one-point increase in
the
45
ratio of (PCT) to (INC) increased yield by more than eight
pounds
of lint in the West.
Price elasticities of Upland cotton supply were computed from
the models: 0.87 for the Delta; 1.95 for the Southeast; 0.22
for
the Southwest; and 1.22 for the West. Elasticities of supply
with
respect to acreage harvested were: 0.75 for the Delta; 0.94
for
the Southeast; 0.66 for the Southwest; and 1.00 for the West.
As
expected, negative yield elasticities of acreage were derived
for
all regions except for the West, where the regression
coefficient
was not statistically significant. To achieve a 10% reduction
in
production, acreage cuts ranging from 10% in the West to 15%
in
the Southeast would be required. On the average, a nationwide
13%
cut in acreage would give a 10% cut in production.
Evans, Remnele, and Bell (1978) used a 35-equation model of
the U.S. cotton economy to derive impact multipliers, analyze
policy alternatives, and evaluate the effects of
technological
changes in the production sector and on the industry as a
whole.
The structural system of stochastic equations and identities
repre
sented a simultaneous equation system.
The equations of the model were grouped into four main
sectors: (1) the farm level or cotton supply sector, (2) the
domestic mill sector, (3) the export sector, and (4) the retail
or
final demand sector. Retail fiber demand was broken down for
men's
and women's apparel, household furnishings, and industrial
products.
46
presented in a functional form, such as:
1t " ^^'^^^iV PC ND t' ^^^V ^^^t^ ^^'^^
where Q . = quantity of fiber demanded in the 1th end
use;
end use;
POP^ = U.S. population.
QCOT^ = f(ZQ.^, PCOT^.i. PPOLY^-l^ C3.9)
where QCOT = pounds of cotton consumed by U.S. mills;
2Q-. . = total fibers consumed in the four end
uses;
1-1/16", lagged one year;
for cotton blending, lagged one year.
The export demand for U.S. cotton was estimated through
equation (3.10):
47
PCOTUS^ = price of U.S. cotton C.I.F. in
Liverpool, England SM 1-1/16";
Liverpool, England;
communist countries;
foreign countries relative to the U.S.
dollar, based on special drawing rights
from the International Monetary Fund.
The model was not validated with formal parametric tests,
e.g.,
Theil's U^, or root mean square error. Results indicated that
a five cents per pound increase in the mill price of cotton
would
lower cotton mill use by 350,000 bales. The impact of shocks
caused by two events occurring in 1974 were traced through
1978-
1979. Event (A) was a 1.2 mi 11 ion bales increase in U.S.
cotton
export in 1974, and event (B) was a $0.50 per bushel annual
increase
in the farm level price of soybeans. Event (A) caused
domestic
mill consumption to decrease in 1974 and 1975, increase in
1976
and 1977, and decrease in 1978. Event (B) resulted in
decreasing
acreage planted in all years of the forecast except 1977.
Shafer (1978) demonstrated that the loss in market share by
cotton Is due more to highly variable prices and quantities of
the
48
different qualities used by mills than to high cotton prices.
It
was shown that, although cotton prices were higher than those
of
non-eellulosics prior to 1966, the loss of market share was
minimal
before 1965 when compared to 1966-1974. Between 1933 and
1965,
^ government programs resulted in little price variation and
high
domestic carryovers. The Shafer hypothesis came from Logan's
theoretical model (1969):
Risk reduction undoubtedly plays an important part in decisions to
Integrate. Further, by reducing impact of prices, quantity, and
quality uncertainty through integration, the firm effectively cuts
its probability of going broke if severe fluctuations in any of the
functional relationships (including procurement) could be fatal
financially. This type of return cannot be put into the usual rate
of return on investment figures.
The Shafer analysis relied mainly on Interviews with the
management of domestic mills. It concluded that the
oligopolistic
structure of the manmade fiber industry has helped it in
maintaining
stable prices of non-eellulosic fibers. In addition, desired
quantities and domestic production capacities were
predictable.
Questions regarding U.S. cotton exports. Including the
effects
of exchange rates and domestic price fluctuations, were
examined
by Collins (1979, 1980). Results indicated that, although
theo
retically Important, exchange rate fluctuations of currencies
of
various countries might be offsetting, and have a net zero
effect
on U.S. cotton export demand. Domestic cotton prices,
however,
had a definite impact on U.S. exports in markets with little
49
markets were, nonetheless, very limited in cases where
importing
countries' policies regulated textile activities.
In formulating an expression for the price elasticity of
export demand, several authors—Tweeten, 1967; Johnson, 1977;
Bredahl, Meyers, and Collins, 1979—have recognized the
importance
of the elasticity of price transmission mechanism. The
elasticity
of price transmission represents the response of country i's
prices
to changes in U.S. prices. Originated by Tweeten, the
expression
of the elasticity of export demand (E x) Is:
E^^ = (Ed.)(Ep.)^^)- (Es.)(Ep.)(^j (3.11)
^ef ef
country 1;
country 1;
Qd. = domestic use;
Qs. = domestic supply;
^ef ~ '•'"P "'' quantities.
Obviously, E X depends on values of or assumptions made
regarding Ed., Es., and Ep .. The controversy in the
literature
revolved around the proper value to use for Ep.. Johnson
(1977)
assumed the elasticity of price transmission to be one
(perfect
50
tively, with the perfect price transmission assumption. Such
coefficients indicate highly elastic export demand for U.S.
agri
cultural products.
In summary, factors that have been deemed important for the
cotton economy are both economic and physical in nature. At
the
mill level, competition between cotton and manmade fibers is
due
to price relationships and also to certain physical
characteristics
present in synthetic fibers as well as their availability.
Vari
ables representing consumer purchasing power are often present
in
mill demand equations, because the latter is a derivation of
retail
demand for cotton products. Domestic Industrial production
vari
ables have been incorporated in cotton mill demand equations,
because mill use is also derived from industrial demand for
cotton
(automobile manufacturing, cordage, ropage, etc.). The problem
of
choice between lagged prices or current prices in mill
relation
ships remains. Although earlier studies (Cromarty, 1960;
Lowenstein,
1954; Blakley, 1961) used current prices to estimate cotton
mill
demand equations, later studies (Evans, 1977; Evans, Remmelle,
and
Bell, 1978) estimated mill consumption with lagged cotton and
syn
thetic fiber prices. In most models reviewed, export demand
for
U.S. cotton was determined by U.S. cotton and foreign cotton
prices
in world markets, exchange rate variables, and foreign
countries'
51
purchasing power. Most of the models reviewed were single
equation
estimated by ordinary least squares. Few were multi-equation
models (Cromarty, 1960; Blakley, 1961; and Evans, Remmelle,
and
Bell, 1978). Cotton production, acreage, and yield were taken
as
exogenous in the above cited simultaneous equation models.
Nation
wide policy variables of importance were acreage allotments,
deficiency of diversion payments, and the Commodity Credit
Corporation loan rate on cotton. Although price elasticities
of
cotton mill demand were as high as -0.86, most studies
reported
coefficients around 0.30. None of the models discussed
presented
an empirical estimate of the elasticity of price transmission
between U.S. prices and foreign prices.
Simulation Analysis Using Commodity Econometric Models
Econometric models of commodity sectors are particularly
suited for policy and market scenario analysis, because
alternative
outcomes may be evaluated without resulting in distortions to
the
real system. Previous commodity simulation experiments vary
from
historical or ex post experiments to ex ante policy or market
scenario forecasts. Examples of the former Include the works
of
Burrows (1971) with tungsten, Epps (1970) with coffee,
Agarwala
(1971) with eggs, and Kofi (1972) with cocoa. Historical
experi
ments Involve simulating alternative policies and evaluating
what
might have happened to the industry if such policies had been
applied in the past. Studies by Labys (1970, 1971) and
Behrman
52
(1971), among others, illustrate how the techniques of
simulation
can be used for ex ante forecasting of future dynamic paths of
the
endogenous variables of the system.
Potential problems in using econometric models for simulation
Include convergence (or model stability), validation,
linearity,
and stochastic variation. The problem of convergence concerns
a
determination as to whether the pattern is oscillating or
cyclical.
The cyclical pattern will eventually result in attaining
equilibrium
levels in the endogenous arguments, while equilibrium is
never
reached with oscillating patterns (Reutlinger, 1966; Labys,
1973;
Judge et al., 1982). The problem of validation relates to
whether
the structure identified is really a "true" representation of
the
timely behavior of the variables in the industry. In that
respect,
good statistical fit is not enough to ensure that the
equations
found were best. Validation techniques have included
parametric
tests, e.g., Theil's U^, the root mean square error, the
percent
error, and non-parametric tests such as spectral analysis.
The
problem of linearity arises in cases where the structural
system
includes non-linear endogenous variables, e.g., ratios,
products,,
or other non-linear combinations (Klein, 1953; Roy and
Johnson,
1974). In such cases, the reduced form equations cannot be
obtained
directly and one has to use alternative solution techniques,
e.g.,
Jacoblan or Gauss-Seidel. The problem of stochastic variation
relates to whether the random error terms are suppressed or
53
utilized in the simulation process. Authors who have allowed
for
stochastic variation in simulation analysis include
Duesenberry,
Eckstein, and Fromm (1959); Desai (1966); and Agarwala
(1971).
Simulation experiments may be treated as partial, total, or
final (Labys, 1973). In the partial approach, data on all
current
and lagged endogenous variables as well as the exogenous
variables
are fed to the model to obtain solutions for the dependent
variables
of the system. The total approach, in contrast, does not
require
data on the endogenous variables. The final form relies on
the
ability of the model to generate data on the lagged
endogenous
variables. In the final form, only data on exogenous policy
variables are required.
One of the earliest simulation studies with econometric model
of commodity was that of Cohen (1959) With shoes, leather, and
hide.
The dynamics of the industry were explored with a combination
of
static and dynamic relations. All simulations were
non-stochastic
in that the statistical errors were not considered in the
process.
Several authors used the technique of econometric simulation
for
policy analysis. These Include Labys (1971); Agarwala (1971);
Barnum (1971); Vernon, Rives, and Naylor (1969); Kofi (1972);
and
Raulerson and Langham (1970). Labys' experiments were
concerned
with determining norm policies that would stabilize 1 auric
oil
prices. The study by Raulerson and Langham evaluated effects
of
diverse supply control policies on the frozen concentrated
orange
54
earnings of cocoa producers and exporters' net of expenses
involved
in administering the commodity agreement. A brief summary of
selected econometric simulation studies is presented in a
tabular
form in Appendix 10.
Discussions in this chapter are related to selected theo
retical considerations pertaining to the structural relations
of
the cotton sector. General concepts regarding price
determination
in a competitive framework were used to arrive at a
hypothetical
systems model of the cotton sector. The major quantity
aggregates
of the sector are domestic mill use, domestic inventories,
and
exports of cotton. Pricing of the product was examined at
three
levels in the marketing channel: farm, textile mill, and
world
markets.
Aggregate mill demand for cotton in the United States is
derived from the final consumer demand for cotton products and
the
domestic use of cotton-made intermediate products. Individual
demand schedules for final cotton products is determined through
a
process of utility maximization, in which prices of cotton
products
as well as prices of all other related goods and services and
the
consumer's income are taken into account.
Each consumer is assumed to have a utility function which is
dependent on the quantities consumed of all accessible goods
and
services (as symbolized by equation (4.1)).
U = f(X.) (4.1)
X. = quantities of available goods and services con
sumed, including cotton-based products, with
1 = 1 , 2, ... n.
Most, if not all, of the accessible goods and services have
a price attached to them (P.). Furthermore, the consumer is
subject to a budget constraint reflecting a limited personal
dis
posable income. These limitations are introduced into the
objective
function by using a Lagrangian multiplier denoted L:
maximize Z = f(X^.) + L(P^.X. - I) (4.2)
subject to
Equation (4.3) expresses the consumer's budget constraint
that the sum of all expenditures (P X .) equals personal
disposable
income (I). The maximization of equation (4.2) is dependent
on
the fulfillment of a set of first-order and second-order
conditions.
dZ/dX. = df(X.)/dX. - LP. = 0 (4.4) subject to
Z P.X. = I with 1 ' '
dh/dl} < 0
The solution of the system of equations represented by (4.4)
yields n equilibrium values of X. (quantities purchased) and
L.
57
It can be shown using the implicit function theorem that each
individual X . is a function of all prices and income (Phlips,
1974
and Intrlligator, 1971). These are the Individual demand
functions
for all goods and services available to the consumer.
A consumer demand function for cotton products is derived
from
system (4.4) and written as follows:
" = nPC^. Pit P„t' It' ^t^ ^^-5'
The relationship described by equation (4.5) postulates that
the quantity of cotton products demanded (C.) is dependent on
the
real retail price of cotton products (PC^), perceived real
retail
prices of other fibers, prices of all other goods and
services
purchased (P.. . . .P-J.)5 and Income (I4.). The demand function
of
(4.5) is static in nature, i.e., it does not consider dynamic
adjustments which could take place in response to a one time
change
In prices and Income.
equilibrium levels of the quantity of cotton products purchased,
if
prices of competing alternatives and Income remain unchanged.
Increasing real prices of competing alternatives will cause
the
consumer to purchase higher quantities of cotton products and
lesser quantities of competing alternatives. Increasing real
per
sonal disposable income will cause quantities of cotton
products
consumed to Increase when all prices are held constant. These
are
58
the expected direction of the various effects resulting from
changes
in single variables, ceteris paribus.
It is practically impossible to measure empirically the
relationship described by equation (4.5) for two reasons.
First,
a data series on individual consumer purchases of cotton
products
is not available and obtaining these data would be
inordinately
expensive. Secondly, many consumer and industrial products
con
taining cotton fiber also contain a percentage of manmade
fibers.
Domestic mill demand is assumed to be derived directly from
the final consumer demand for cotton products and, hence, is
deter
mined by the same forces underlying the latter. Mill demand
for
cotton can also be analyzed from a production viewpoint on
the
assumption that miller maximize returns in the face of
changing
input and output prices and general economic conditions. Given
the
price of cotton at mill points, the price of manmade fibers,
the
price of broadwoven products, and manufacturing costs, each
Individual mill is assumed to maximize profits as per (4.6).
Max P = (ZPb.)(ZB.l - (Pc)(ZCot.) - (P )(2MMF.) - TOPC
+ X[(Pc)(2:Cot.) + (Pp)(ZMMF.) + TOPC - TBO] (4.6)
In equation (4.6), Pb. represents the price at the mill gate
of any of the n types of broadwoven products manufactured by
the
mill, B. represents the quantities manufactured, x is a
Lagrange
multiplier that is introduced in the objective function of the
firm
59
to reflect the limited character of the firm's budget outlays
and
credit, in the short-run. The remaining variables of equation
(4.6)
are defined as follows:
Pc = price of cotton input delivered to the mill;
Cot . = quantity of cotton utilized by the mill in the pro
duction of the ith broadwoven fabric;
Pp = price of manmade fiber(s) utilized in fabric production;
TOPC = other costs incurred by the mill in the production
pro
cess;
mills in producing the ith broadwoven product;
TBO = total budget outlays and/or credit that are available
to the individual mill.
win yield optimal quantities of cotton and other fibers given
input prices and other costs, e.g., labor and energy.
The economic relationship representing mill demand for input
factors (in this case, raw cotton and manmade fibers) will
vary
depending on whether one assumes that the quantity of manmade
fibers
used in the production process are fixed