SUPPLY RESPONSEIN ARGENTINA: __ i  
AGGREGATE PLANTED AREA IN CROP PRODUCTION

Man 8 paper for the Degree of MS -
MICHIGAN STATE UNIVERSITY ~'
CHRISTINE A; MARTIN ‘
1998

rmMP‘Yfi
__.

 

 

l..iBRARY
Michigan State
University

 

 

M

MICH. STATE UNIV.

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 o'JCIRC/DatoDuopss-sz

SUPPLY RESPONSE IN ARGENTINA:
AGGREGATE PLAN TED AREA IN‘CROP
PRODUCTION

by
Christine A. Martin
A Plan B paper submitted inpartial fialfillment of the

requirements for "the degree of

Master of Science

31.10% M f0 A53 if» adj wacfl é..w‘mt9~rrw/LJ
Michiganl State University

December 1998

Michigan State University
Abstract

SUPPLY RESPONSE IN ARGENTINA: AGGREGATE PLANTED AREA
IN CROP PRODUCTION

by Christine A. Martin

Research Supervisor: Dr. Scott Swinton

Major Professor: Dr. Christopher Peterson

This study attempts to explain the variation in aggregate crop acreage in Argentina in the
period 1960-97. It improves on, previous studies by incorporating an expected yield
variable and input price variables, extending the period of analysis, and attempting an
aggregate crop acreage supply response of six crops. A single equation model is
developed and the naive, moving average and simple weighted average price expectation
mechanisms'are examined. Results show that a moving average specification of price

expectation accounts better for the data.

ACKNOWLEDGEMENTS

I thank Dr. Chris Peterson for his willingness to accept a non-traditional ag-econ student
and his guidance throughout the master’s program. I am also grateful to Dr. Scott
Swinton for his direction in the completion of this paper and to Dr. Roy Black for his

guidance.

I am thankful to my classmates and colleagues in their support and assistance

throughout this program.

Last, but not least, I am indebted to Santiago Liboreiro for the loving and patient
assistance and support he gave, without whom this paper could not have been

completed.

I dedicate this paper to my children, Megan and Nicolas. Now they can play on the

computer!

CHAPTER 1

Introduction

Supply Response in Argentina: Aggregate Planted Area in Grain Production

Argentina is well known for its beefi which is raised on extensive natural pastures.
Argentina also produces a substantial amount of grain and oilseeds, which include
wheat, corn, sorghum, soybeans and sunflower- Sixty-five percent of Argentina’s 279
million hectares is suitable for either agriculture or livestock. Of the 183 million acres
of arable land, in 1992, approximately 80 percent was in natural pasture, the remaining
land in annual or perennial crops. Annual crop acreage in 1992 was approximately
19.75 million hectares, and the cattle stocks was approximately 53 million head. In
1996, apnua] crop acreage increased by 14.2 percent to 26.4 million hectares, while
cattle stock decreased 3 percent to approximately 51 million head. In recent years,
analystshave'predicted that With grain prices rising faster than cattle prices, most of

the pasture land would. shifi into grain production even further.

There have been periods in Argentina’s agricultural history when government has

played a major role in affecting the uses of arable land. Today, the Argentine farmer is
fi‘ee of any government regulation, and the use of the land is based on his/her decision.
This deciSiOn regarding the use of land is primarily based on income. Consequently, as

the prices pf grains, and yields, have risen, farmers have shifted to crop production.

Studies conducted on the supply response of crops to prices in Argentina are not very

extensive.

Reca (1980) examines pricing policy in Argentina from 1950 to 1975 for seven major
agricultural products: wheat, corn, grain sorghum, beef cattle, wool, rice, and cotton.
Reca presents three estimates of the Argentine producer’s price responsiveness:
aggregate crop production, crop production‘in the Pampas, and outside the Pampas.l
Using a Nerlovian model, Reca analyzed aggregate output as a filnction of prices
lagged one year, credit availability, technology and weather. These last. two variables
were entered as dummy variables. "Reca concludes that “the model discussed strongly
supports the contention that product prices were one of the key determining variables

of the behavior of agricultural production'in Argentinain the last twenty-five years.”

To analyze the supply responsiveness in the Pampas region, Reca’s linear regression.
model described area planted to crops and Oilseeds as a filnction of ‘expected’ product
prices, prices of substitutes, level, of technology, availability of credit, and weather.
These irt'clu‘ded product price lagged one year, yield, and credit availability, beef cattle
slaughter, and stock of beef cattle as independent. variables. The yield variable was
detemrined as a dummy variableiinstead of the actual yieId. Reca concludes that the
influence of cattle production appears strong, and that cattle and crops compete for

the use of land.

A second study conducted by Wainio (1983) looked at the effects of government
intervention on the responsiveness of grain area to price movements during the late
1940’s through the late 1 970’s. His linear model consisted of four equations, one for
each of the following crops: wheat, corn, flaxseed, and sunflowerseed. He includes in
his model the price of each cropi'lagged one year, price of cattle lagged one year, and
acreage planted to the crop lagged one year (partial adjustment). His decision to lag
prices one year was based on the results ofReca’s study. Input costs and crop yields
were not included in the regression model- Wainio’s study also did not include
soybeans since they were not very important during the period of the analysis. Wainio
determines that farmer’s price expectations seemed to be based on prices received for
the previous'harvest, however, these same expectationshad a delayed effect on

subsequent years’ decisions as well.

In a third study, Sturzenegger describes the effects of intervention on output between
1961 and 1985. He assumes the" producer maximizes profit as a function of price of
wheat, corn, sorghum, soybean, sunflower, beef, variable inputs and fixed inputs. No
variable for crop yield is'in'clu'ded in the estimation medel. All variable inputs are
combined into one input variable due to lack of information The fixed inputs
represent soil conditions, rainfall and sunlight. Sturzenegger includes a coefficient of

adjustment which is extrapolated from two previous empirical studies.

 

1 The bulk of the grain and oilseed production in Argentina comes from the Pampas. Grain crops and
oilseeds oonpete-withbeef cattle fertheuseof IandinthePampas.

7

In the US. and other countries, studies on the response of crops to prices are
extensive. Askari and Cummings summarize most of the work performed until 1976.
Until then, most of the studies mentioned were directly or indirectly based on
Nerlove’s supply response model, published in 1958. Studies on the measurement and
analysis of price responsiveness 0f agricultural supply has continued since then. More
recently, many studies have incorporated the rational expectations hypothesis and risk
in their modeling to analyze and measure supply response. (Gardner (1976), Fisher

(1932), Eckstein (198.4), Holt and Johnson (1987), Holt and Ardubyala (1992))

With the implementationof ArgentinePresident CarlosMenem’s Convertibility Plan
(CP) in 1991, the Argentine economy has improved drastically and this may have
affected the Argentine agricultural prOducef’ s'behavior. The stabilization of the
economy has allowed the Argentine farmer to change. his farming practices in several.
ways, piimai'ily by being able to purchase imported goods less expensively, purchasing
and adopting new technology, and by not having to keep cattle as a capital investment

due to rising inflation.

Menem’s CP also reduced government intervention in Argentine agriculture- The
government has reduced the trade taxes drastically, and it does not intervene in the
determination of grain prices. The ports and elevators have been privatized and. the
National Grain Board has been dismantled. In signing Mercosur, the trade agreement
with Brasil, Paraguay and Uruguay, Argentina has removed all taxes on trade with

these countries.

Reca and Wainio’s analysis did not include soybean since it was not a relevant crop
during their period of study. The first commercial scale soybean production did not
occur until 1975 when about 0.5 million hectares were planted. By 1997, there were
7.2 million hectares of soybean seeded acres, becoming one of the nation’s most

important crOp.

Cattle is still an important aspect of Argentine agricultural production, primarily since
domestic consumption of beef is still very high. However, with the improved economy
and high world grain prices, and that most of the grain production is exported, the
Argentine producer has increased his grain production at the expense of cattle pasture.
Several Argentine analysts, through personal interviews, predict that more acreage will

be shified .from pasture tograinproduction.2

This study works towards explainingthe variations in aggregate grain acreage in the
line of Reca and Wainio. This study improves ontheirs in that the period of analysis is
extended to 1997, itincludes soybeans. (an increasingly important crop), and it
improves on their model specifications by incorporating an expected yield variable and
input price variables- Other price expectation behaviors, aside from naive price
expectations where the previous,period’s price is used as the expect price for this

period, are also analyzed.

 

2 Sparks America del Sur and Secretaria dc Agricultura, Ganaderia y Pesca.
9 .

Aside fi'om an attempt at analyzing the aggregate crop acreage, the results obtained
from this paper may benefit those in the industry who might be able to use this

information in their organizations’ strategic plans.

The purpose of this study is to develop a single equation econometric model that
explains the variation in aggregate crop acreage in Argentina. It is a simple
comparative advantage model between cattle and crop. It will examine several price
expectation mechanisms, adjustment processes, and the explanatory variables needed

to explain the aggregate acreage put into crop production.

It is expected that the:

0 Argentine producers have naive price expectation behavior,

. Argentine producers experience a significant lag between economic shocks and
crop production, and

o Cattle and grain prices, input costs and technology afl’ect grain production

acreage.

The chapters to follow include a background on Argentina and its agriculture, a
literature review on supply response studies in Argentina and on other supply response
studies, a conceptual and empirical model, results and discussion of the econometric

model, and conclusions.

10

CHAPTER 2
Argentina

“Until the Great Depression of the 1930’s, agriculture was the staple sector of the
Argentine economy.”3 Between 1860 and 1930, the exploitation of the rich land of the
Pampas strongly pushed economic growth. During this period, Argentina grew more
rapidly than the United States, Canada, Australia, or Brazil, countries similarly
endowed with rich land, which. also accommodated large inflows of capital and
European imnrigration. During the first three decades of this century, Argentina
outgrew the other four countries in population, total income, and. per capita income as

shown in the table below.

T_able 1: Comparative Growth in Income and Population 1900-1984

 

 

 

 

 

 

 

 

 

 

Pieriod/Item
1900-04-to 1925-29» . Argentina Australia --Brazil Canada United States
Population 2:8 ’ 1.8 2.1 2.2 - 1.)
Income " 4.6 7.2.6 3:3 3.4 2.9
Per cgita Income . . .18 ~ 0.8 1.2 1.2 . 1.3
1925-29 to 1980-84
Population " " '“1i8 ‘ ~17 2.5 1.5 ‘13
:Income - 2 8 3 9 5 5 3 9 3.1
Per capita income 10 2 2 3 0 2. 4 1.8

 

 

 

 

 

 

 

(Avera armual rates in percentages, Cavatlo and'Mundlak, 1989)

However, beginning in the 1930’s, the Argentine economic vigor deteriorated rapidly.

This loss ip vitality was especially‘dramatic in agriculture.

11

2.1 Policies

The Great Depression in the 1930’s reduced Argentine agricultural exports. The
Argentine govemment’s response to this was to start an import-substitution program
for manufactured goods, a multiple exchange rate scheme, and a price support
program to stimulate farm preduction. Two central government agencies were
created, the National Grain Board (IN G) for grain and oilseed production and
National Meat Board (IN C) for meat and its derivative products. Support prices were
fixed, which exceeded market levels for wheat, corn and flaxseed during 1933-35, and
support’ed by ‘JN G purChases. The price support program together with favorable
exchange rates for agricultural exports, began alarge expansion of grain and. oilseed.
area in Argentina. Strong export” demand during 'thelast half of the thirties stimulated

further giowth.

World War H brought difficult times for agricultural producers and traders Disrupted
market and shipping activitiesproduced the ingest grain surpluses in Argentine
history- However, storage was inadequate and losses were great. Afier WWII,
agricultural prices rose substantially, and'had it not been for government intervention,
this would have been a good opportunity for Argentina to increase its exports. The

government of Juan Peron implemented a strong protection for domestic

 

3.Mundlak, ~Yair and Domingo Cavallo, ‘Agn‘culume and Eoenemis Growth in Argentina, 1913-
1984”, p. 12. ,

12

manufacturers and high taxation on agricultural exports. As manufacturing and urban
services were more labor intensive than export oriented agriculture, the policies
implemented reflected Peron’s main concern for income redistribution to labor.
Specific agricultural objectives of the Peron government were to shift economic and
human resources fi'om rural to urban sectors, redistribute resources within the
agricultural sector from landed class to the landless tenants and workers, and generate
revenue and foreign reserves for industrial development through agricultural exports.
The government nationalized mass transportation, warehousing, port operations, and
commupication services. Intervention extended to electrical generation and
distribution, to controls in banking. and foreigntrade which limited imports mainly to
raw materials‘and semi-manufactured goods for useby beginning domestic industries.
The Argentine Institute for the Promotion of Trade (IAPI) controlled agricultural
trade. It controlled all exports and, as sole’buyer, made all domestic purchases of
grains and oilseeds. IAPI also controlled imports of food and agricultural inputs. The
governrr’ent 'fixed producer prices and subsidized retail food prices through 1946-55,
with substantial subsidization of domestic wheat consumption. Farmgate prices were
announded before harvest the. .Thegovernment ,nationéliZed the railroads in 1946-47
when trains carried the bulk of the internal grain shipments. The futures market closed
and wasn’t reopened until the late 1980’s. In the late forties and early fifties (in 1951,
Argentina experienced a severe drought causing an agricultural crisis), some price and
marketing regulations were eased in an attempt to stimulate production and exports of

Argentina’s major agricultural exports, however, not in time to impede the steady

decline in grain production due to a full decade of inadequate incentives.
l

13

In 1955, Peron’s government was overthrown by the military, beginning a period of
political instability which lasted until 1973. The Argentine economy slowly became
more market oriented through this periOd. The NGB took over most of the duties of
IAPI (which was closed). and refrained fiom. direct market intervention except when it
imported grains to offset prOduction shortfalls. International grain and oilseed trade
gradually returned to the private sector- Market determined prices set by daily
quotations replaced administered piices on the commodity exchanges. Export duties
were generally low except for a 25% grain export duty imposed in 1958. Atthe end
of the 1950’s, the National Institute of Agricultural TechnOlogy (INT A) was created

and was responsible for technological research and extension.

In 1973, Peron was re-elected and most of the firstPeron govemment’s policies were
reintroduced- The state once again became the only buyer of wheat, corn, sorghum
and sunflower. Fixed producer pricesbecame effective at the farmgate for wheat in
1974 and for the other three products in 1975- High export taxes and exchange rate
controls were implemented again .The , government banned vegetable oil exports

between 1973 and 1976 to insure domestic supplies and control consumer prices.

A military junta took over in‘March 1976 and gradually returned control of the
economy to the private sector. The government. removed domestic price controls and
returned domestic marketing to the private traders. In late 1979, private traders could

invest in terminal port facilities and to rent state-owned storage capacity. The

14

government removed export taxes on all major grains and oilseeds by November 1976.
The NGB continued to administer the commodity price support program for grains
and oilseeds (although market prices were higher than support prices for the entire
period), to manage government-owned storage facilities, to collect export taxes, to

issue export licenses and to set export quotas when necessary.

Taxes on eXports varied fitom 2 to 13 percent, and restrictions on imports and
exchange controls shifted from extremely high levels in 1975 to almost complete
elimination by 1980. Tarifl's on permitted imports varied in an opposite direction from
the quantitative restrictions and the exchange controls, moving from a subsidy of 6

percent in 1975 to an actual tarifl of 26 percent in 1984.

As a consequence of the'poli'cies implemented during the period 1970-84, the rate of
inflation rose to historicalhigh levels, becoming volatile while the economy stagnated,

except fpr a short-livediboom'during' the world commodity crisis in 1973-74.

A fourth economic period inArgentina began in 1989 with the election of President
Carlos Menenr, and the implementation of the “Convertibility Plan” (CP), devised by
the. Minister of Finance Domingo Cavallo, in 1991. President Menem’s political
agenda was to drastically change the nationis economic structure. This change in
structure included: the convertibility to dollars of the Argentine currency, the Central

Bank’s autonomy, fiscal adjustments,.modemization of the government, opening the

15

economy to foreign investors, privatization of public companies and the deregulation

of the commodity markets.

The CP, implemented in April of 1991, fixed the Argentine peso to the US. dollar at
1: 1 parity and obliged the CentralBank to fund this convertibility with gold in reserves
valued at market prices. The same law prohibits the Central Bank to finance the
Treasury; deficit and to give loans to the government, national, provincial or municipal.
These laws have been strictly enforced, even, during the Mexican financial crisis of

1995.

Since the implementation of the CP, inflation has been greatly’reduced, from an annual
rate of 84%‘in 1991 to 0.1% in 1996. Another consequence of the new plan,
Argentina’s gross domestic product (GDP) in 1996 increased 4.4% from 1995. The

economy is improving and inflation is non-existent.

Market deregulation and reduction in export taxes and import tariffs were also
established by the CP. The elimination of commercial obstacles has led to a dynamic
market environment. Until 1990, there was a certain. anti-commerce and anti-export
bias, manifested by the barriers .to importsand. exports. In 1991, when the CP came
into effect, these barriers were slowly reduced. Average tariffs decreased from 39
percent In 1988 to 10 percent in I995, together with the elimination of almost all

export licenses. In 1992, all export taxes were annulled, with the exception of those

16

on oilseeds and cowhides. As a consequence, Argentine total exports, which totaled

US$11,977 million in 1991, reached US$23,774 in 1996.

The CP also greatly affected the agricultural sector. Widespread privatization has
improved the network of country and terminal elevators, railroads, and waterways.
These improvements have lead to an increase in storage and loading/unloading
capabilities, and quality of service. An improved economy and non-existent inflation
has also allowed the Argentine producer to improve technology, i.e. purchase of new

farm equipment, purchase fertilizer, purchase better seed.
2. 2 Geography and Land Occupancy

Argentina encompasses distinct climactic zones, fi'om sub-tropical to arctic, which ,
allows for the production of a Wide variety of agricultural commodities. The bulk of
Argentina’s arable land is in. the Temperate Zone. Sixty five percent of Argentina’s
279 million hectares of land are suitablefor either crop orlivestock. Sixteen percent is
forested, and the remainder is unsuitable for either cultivation or forestry. In 1993, of
the arable land, 23 million hectares was used as annual or perennial crop production,
15 million hectares was under improved pasture, and 145 million hectares was natural

pasture (Mch and Schmitz).

The major geographic zones are: the Pampas, the Chaco, the Andean zone, and

Patagonia.

l7

The Pampas is that largest ecological zone and the dominant grain-producing area.
Its 45 million hectares spread out from the city of Buenos Aires and covers most
of the provinces of Buenos Aires, Cordoba, La Pampa, Santa Fe and Entre Rios.
The area has temperate climate, between'750 and 1000 mm of rainfall annually,
and contains about 50 percent of Argentina’s arable land. Two thirds of the

Pampas afea has deep fertile sOils ’similar to the Great Plains of theUnited States.

' The Chaco is the northern lowlands along the River Plate. This area has heavier

rainfall and poorer sOils, especially in thenortheast, and contains important forest
resources. The principal agricultural products are cotton, nuts, tea, rice, tobacco,
citrus, and cross-bred cattle.

The Andean zone, west and northwest of the Pampas, is hilly to mountainous,
poorly Watered, and with aIew exceptions, .can only be cultivated under irrigation.
The principal agricultural products are sugar, tobacco, citrus, grapes for wine, and
miscellanéOus cash crops, including beans.

Patagonia, about the. same size as the Pampas, receives only minimal rainfall, is
sparsely populated, and isdevoted mainly to sheepfarming and to the production

of some high quality temperate-zone fi'uits (McCarry and Schmitz).

2. 3 Agricultural Sector

According to Reca, between 1960-1974, the share of crops in total agricultural

production grew at the expense oflivestock." The increased share of crops was due to
l

a number of factors, which included the introduction of improved crop varieties and

18

technologies, which increased the relative profitability of crops, and the considerable
cyclical fluctuations in beef prices which discouraged development in this sector.
Within crops, grains and oilseeds increased their proportion of production in the
period mentioned at the expense of industrial crops (cotton, sugarcane, tobacco, wine
grapes, etc. . .). The share of fruits and vegetables remainedconstant. The
composition of livestock production showed some changes, but. not enough to alter
substantially the picture prevailing at the beginning of the 1950’s. The share of beef
production increased slightly, but remained close to two-thirds of all livestock

production.
2. 3. 1 Grain and Oilseed Production

Wheat was the first crop grownin thePampas area of Argentina, and is Argentina’s
most important crop. In 1996, 7366 thousand. hectares were planted. Wheat is sown
in the fall and harvested in late spring and early sumrirer (N ovember/December).
Wheat yields have improved due to the use of improved varieties and the use of
chemicalinpUts in the 1990’s."'1'he most significant 'eVent'in wheat production was the
introduction of semi-dwarf wheats in the mid-1970’s. An estimated “96 percent of the
area used for wheat is now sown—to.these.high-yielding varieties.”4 The greatest
advantage derived from the‘introfluCtiOn of these varieties is an earlier maturity of the

l

crop. Use; of early-maturing varieties has enabled farmers to double-crop wheat with

 

r
" Lacroix‘; Richard, Michael I'MCCmyJVIamiew Melviahomand Lowell Hill, “Argentina: Grain and
Marketing, finstitutions and Policies”, p.12.

19

0N

 

 

 

 

 

 

 

 

 

 

.._..._l...rlr_7+_r.......nl4_..+...Frrnhnhtwo

18.:

.......~

-88

woos.

.8...

-88

:82

. . . 5.8..

 

 

25:an c. 3.2 .35 “.85..“ 3 2:2“.

 

serene” puesnoul

 

soybeans. See Figure 1 for planted area per crop.

The seeded area of corn in Argentina has been declining primarily due to the increase
in wheat-soybean cropping. Over “80 percent corn grown comes from northern
Buenos Aires, southern Santa Fe and southeastern Cordoba provinces.”5 Though the
seeded area has decreased, corn yields have been increasing, as a result of several
factors, including the introduction ofh'ybrids and the use of improved agronomic
practices (such as planting at higher densities and increased use of fertilizer). Corn is

sown in Dabber/November and harveSted inMay.

Soybean is a relatively new crop in Argentina and did not become economically
significant until the mid-T970’ 5. Argentina is now one of the world’s four main
producer of soybeans as a consequence to the increase in production. Production
technology was introduced to Argentina from theUnited States, and Argentine
farmers have adopted this new crop into their rotation quite rapidly. This increase in
production was stimulated by warldwide demand for vegetable oils and oilseed meals
and the availability of agricultural technology. Area sown to soybean is concentrated
to the “Pampa Humeda”, which comprises northernBuenos Aires, southern Santa Fe
and southeastern Cordoba, however, soybeans have also spread both south and west
into drier areas. Soybeans'are sownin July/August for the first crop and

December/January for the second. crop and harvest begins in April. Expansion in

soybean seeded area is being achieved at the expense of area sown to com, sorghum

 

5 Ibid p.15.
21

and permanent pasture. Since 1980, soybean yields have been constant due to a
buildup of pests and diseases in the traditional areas and to the expansion of soybean

cropping into less favorable land.

Sunflower is a traditional crop in Argentina. Production was stimulated in the 1970’s
by the same factors that stimulated soybean production, i.e. world demand for
vegetable oil and oilseed meals. Sunflower production began in southern Buenos ,
Aires province. In the 1980’s, the area sown expanded northwards. This increase in
area sownwas in responseto improved technology, especially the introduction of .
superior hybiids, as depicted in the increasein yields. Sunflower is normally sown in

September and harvested in March.

Sorghum production in Argentina is declining due to the fall in world sorghum prices.
Sorghum is normally sown in fire drier areas 01' the Pampas towards the west.
However, sorghum is being replaced by soybean and sunflower. It is sown in
October/November and harvested inMarchIApfil. Sorghum yields have been

practically stagnant since the introduction of hybrids in the ‘60’s and ‘70’s.

F laxseeq production in Argentina has’beendeClining during the period of this study
due to decreasing world prices and at the expense of increasing wheat, acreage. .
Flaxseed is Sown in May7June andharvestedinDeceinberlJanuary. It is normally

sown in the eastern Buenos Aires province, Entre Rios, Santa Fe and Cordoba.

22

2. 3.2 Fertilizer

Argentina does not produce suficient fertilizer to satisfy domestic demand,
consequently, imports are required. According to Argentine Secretary of Agriculture,
since the CP in 1991, the apparent domestic consumption of fertilizer has increased
tremendously. The favorable outlook of grain prices, substandard soil nutrients and
the genetic yield potential with fertilizer application were the major incentives for the

increase in fertilizer consumption.

 

{igure 2: mm in Argentina, 1984-1995

 

1409'
1399 «
1000 9
800...
$99 9
.490‘
zoo -

7‘0 - r r r r T l l T 1 7 r

 

Thousand Tons

 

 

 

 

1992
1993
1994
1996

1984
1985
1986
1987
1988
1989
1999
1991

 

 

—National '~—¢—‘1mpofled +Total

 

 

 

 

 

As Figure 2 shows, fertilizer consumption during the 1980’s fluctuated between 280
thousand tons and 350 thousand tons This apparent consumption implies a “6 or 7
kilogram of fertilizer per hectare of arable land usage.”6 Despite the apparent low
consumption of fertilizer, the Argentine producer compensated for this by

incorporiitin‘g crop residuals into the soil afier harvest and by rotating crops on the

 

6SAGYP
23

land. According to Wainio (1980), about 75% of the fertilizer consumed is applied to
intensively cultivated crops such as sugar cane, wine grapes, and links and vegetables.
The remaining 25% is used on grains. The most heavily fertilized grain is wheat,

although only about 15.5% of total wheat acreage is fertilized.

According to SAGyP, the fertilizers most used by Argentine producers are urea and

diammonium phosphate.

24

CHAPTER 3
Literature Review

There have been few Argentine studies conducted on the supply response of crops to
prices. The literature on supply response is, however, extensive. What follows is a
review of the Argentine studies, and selected studies pertaining to risk, and pricing
behaviors and rational expectations.

-

3.1 Supply Response in Argentina

One of the Argentine studies looks at the shift in the cattle industry with respect to
prices and economic incentives during the mid 1930’s through the mid 1960’s
conducted by Jarvis in 1969. Jarvis believed previous studies were oversimplified for
the cattle industry in Argentina, and these studies resulted in negative short-run
responses on cattle for slaughter. According to Jarvis, these oversimplified studies
extrapolated the negative short-run response into long-run responses, not taking into
consideration that with less slaughter, what results is a growing herd through larger
calf crops, Which would lead to increased slaughter over time. Consequently, he
developed microeconomic models to illustrate his views, and from this, he built an
economic model that may be used to explain the historical. reaction of the cattle sector
to exogenous shocks such as those caused by climatic variation and by changes in the

domestic demand for beef or for other related agricultural products. According to

25

Jarvis, results show that cattle producers systematically reallocate their “portfolios”--
that producers are price responsive--and due to the interactivity shift with the
agricultural sector between grains and livestock, this also implies supply response by

field crop producers.

The study performed by Reca in 1980 examines pricing policy in Argentina fi'om 1950-
1975 for seven major agricultural products: Wheat, corn, grain sorghum, beef cattle,
wool, rice, and cotton As part of this, Reca explores the relationship between prices
and agricultural output, Since it'has important economic and political consequences;
Reca presents three estimates of the Argentine producer’s price responsiveness: inthe
aggregate, crop production‘in' the Pampas, and outSide the Pampas (the latter not
discussed inthis study).7 Using a‘Nerloviantype linear model, Reca analyzed
aggregate output as a function of prices lagged one year, credit, technology and
weather. The agricultural WPI (WholeSalePrice Index) relative to non-agricultural
products lagged onyear was used. Technology and weather were entered as dummy
variables. Reca defined two difl‘erent technological levels, one covering the period of
1950-64, and the other the rest of the period- Reca admitted that this was a very
crude approximation to a difficult problem Difi‘erent alternatives were analyzed.
Reca concludes that “the. coeflicients show a remarkable stability.[...]Keeping in mind
the conceptual and statistical limitations of the analysis carried out, the model

discussed strongly supports the contention that product prices were one of the key

 

» 7 The bulk of the grain and oilswdproductioninArgentina comes from the Pampas Grain crops
and oilseedsreompete with beef- cattle-for thcuse of India the-Pampas.

26

determining variables of the behavior of agricultural production in Argentina in the last

twenty-five years.”8

To analyze supply response in the Pampas region, Reca’s regression model described
area planted to crops and oilseeds as a function of ‘expected’ product prices, prices of
substitutes, level of technology, availability of credit, and weather. Price was
determined as a weighted sum of the previous year prices of the five crops. The
structure of weights used to construct the price series changed according to the
relative importance of each crop in aggregate production value in three different
periods: 1950-59, 1960-68, 1969-74. Yield was determined as a dummy variable,
taking on a value of zero until 1965, one thereafier. Credit, beef cattle slaughter, and
stock of beef cattle were the other independent variables. Reca’s equation does not
include the price of cattle (price of a substitute) as an independent variable, nor does it
include input prices. Reca concludes from this regression study that, in terms of
policy, “ an artificially depressed price may not cut down substantially production from
one year to the next...0ne crop may very well expand in area, but this will not hold for
all the crops taken together. The effects of artificially low prices for crops will become
evident some time later through the disinvestrnent process in agriculture which they
will inevitably trigger.” One of Reca”s hypotheses was that even though there is some
flexibility in the animal unit/land. ratio, a larger stock of cattle requires more land which

is then excluded from crop use. Based on his regression coefficient, which was only

 

8 Rear, Lucio 6., “Argentina: Country Case Study of Agricultural Prices and Subsidies”, World Bank
Staff Working Paper No. 386, April 1980. ‘ ‘

27

marginally significant, he concluded that his hypothesis was not correct. It must be
noted that by including the stock of cattle asan independent variable, one might be
causing multicollinearity since there might be a relationship between independent
variables, for example, if .price of grain increases, stock of cattle might decrease,
requiring less pasture acreage. Reca also concludes that the influence of cattle

production also appears strong, andtbat cattle and crops compete for the use of land.

A study by Wainio in 1983 looked at the efi‘ects of government intervention on the
responsiveness of grainareatopricemovements duringthe late 1940’s through the
late 1970’s. In this study, Wainio calculated short and long-run price elasticities,
together with cross price elasticities for wheat, corn,flaxseed, and sunflowerseed (the
four important grains competing for land during the length of the study), under
government interventionand nongovemment intervention price programs. His linear
regression model consisted of four separate equations (one for each crop) with the
following explanatory variables:«ircrreegeplantedto.thernlh crop lagged one year;
deflated price of each croplagged one year; and deflated price of beef lagged one year.
The decision to lag prices only one year was based in part on the results of Reca’s
study which yielded better results withasimple one year lag than with a more
sophisticated geometric lag of a series of past annual prices. Beef prices were included
to capture the competitiverelationship between grains and livestock. Wainio did not
include input costs in the regression model- A partial adjustment process is
hypothesized for each of the four. commodities. These four equations were estimated

simultaneously, using Seemingly Unrelated Regression (SUR). In order to test

28

whether govermnent intervention influenced producer decisions, zero-one dummy
variables representing market-managed periods were included. Price indices for the
four crops were a three month average based on the three months after the crop was
harvested. These prices were deflated by the cost of living index. Acreage data was
aggregated at the national level. From this, Wainio learns that farmers react
differently, but not uniformly, for different crops and in difl‘erent price determining
environments. Wainio also determines that farmer’s price expectations seemed to be
based on prices received for the previous harvest, however, these same expectations
had a delayed effect on subsequent years’ decisions as well. For wheat, approximately
65 percent of the total wheat area adjustment took place in the year following the
harvest. For com the first year adjustment averaged approximately 20 percent; for flax
about 25 percent; and for sunflowers, about 50 percent- Wainio’s study did not

include soybeans since they werenot very important during the period of the analysis.

A fourth study by Sturzenegger describes the effects of intervention on output,
between 1961 and 1985, where .he assumes the producer maximizes profit as a
function of price of wheat, corn, sorghum, soybean, sunflower, beef, variable inputs ,
and fixed inputs. Due to lack -ofinformation, all variable inputs are combined into one
input variable. The fixed inputs represent the fixed agronomical conditions for
agricultural production, such as soil.conditions,rainfall, and sunlight. No variable for
crop yield is included in the. estimation equation. Specific one month averages for ,
each crop were used for the calculation for producer prices (January values for wheat,

May values for corn, sorghum, soybean and sunflower) and an annual average for beef

29

cattle. Sturzenegger deducts transportation and distribution costs fiom the quoted
market price to derive a farmgate price in domestic currency per ton. A coefficient of
adjustment is also included in Sturzenegger’s study, however, the size of the
coefficient is extrapolated fi'om two previous empirical studies. Sturzenegger’s results

are then compared to Mundlak, Cavallo and Domenech’s and Fulginiti’s study9

In Mundlak, Cavallo and Domenech’s, “Agriculture and Economic Growth in
Argentina, 1913-84”, the authors examine the relationship between agriculture and
overall economic growth, particularly, the influences of economic policies of three
sectors; agriculture, nonagriculture excluding government, and government; with a _
special emphasis on examining the important role ofthe real rate of exchange. It is a

' “comprehensive and formal analysis of the causes behind the poor performance of the
Argentine-economy” during mostofthe 20m.century. The main conclusion derived by
the authors is that incorrect economic policies led Argentina to lag behind the trend

grth of countries with similar potential.

In a study by the USDAanleTA ofsupply and demand for selected agricultural
products fi'om 1946-1965, supply projections for grains and oilseeds (the only
agricultural products discussed'in this review) are derived. Area of grain and oilseed.
production was estimated as a firnction of real price of product, real price of

competing or complementary.product,real price ofmajor inputs used in production,

 

9 Fulginitipetfonned a snrdy-on ‘rrhe Structureef Agricultural Technology: The Case of Argentina”
in 1986. 'Fulginiti’s results are published'in'Sturzenegger’sstudy.

30

yield and climatic conditions. Price variables were lagged for different periods of time
in weighted and non weighted form. Each product (crop) was estimated using a linear
equation model for each regionand for-thecountryas a whole. 1° The best modefwas
chosen (based on estimated signs of the coefficients, statistical test of significance and
magnitude of the partial regression coeflicients) and used to make the supply
projections. Since the products estimated were numerous, a review of the wheat and
sunflower acreage estimation is detailed here. For the northern Santa Fe region
(which produces about 4% of total wheat production) after more than 35 alternative
linear equations obtained by combining the various variables (price of wheat, price of
com, price of flaxseed, price. ol'beef cattle, input prices, yields, weather, and seeded
acreage to wheat lagged one year), the authors obtained an equation with what they
deemed a decent R-squa'red of 0.61. Howevenit entailed removing two variables that
had unexpected signs (input prices and yield lagged one year) and using the support
price of wheat rather than the market price of wheat. For the La Pampa and western
Buenos Aires province (which produces about one-third of Argentina’s wheat) afier
the 35 linear equation attempts, no equation Wasdeemed satisfactory due to the
magnitude of the statistical coeflicients, therefore the projection of the. wheat. seeded
area for this region was made usinga lineartrend method. In the case of sunflower
acreage, afier several multiple linear. regressions with results that did not allow the
choice of an equation suitable forprojectionpurposes, the authors derived a linear

trend equation for the area seeded to srmflower.

 

'0 The country is divided into specificagricultural‘prodnction‘regions for each crop.
3 l

3. 2 Review of Other Supply Response Literature

Askari and Cummings (1976) detailed at least 500 studies in supply response analysis,
directly or indirectly related to the basic framework proposed by Nerlove in 1958.
Studies on the measurement and analysis of price responsiveness of agricultural supply
has continued since then. Below is abriefsummary of the most representative articles

of what has been accomplished since Nerlove.

Supply response analysis has evolved to include risk. It is understood that output
response in the case of arisk averse producer would respond negatively to related
price variability. Just (1974) introduced the concept of price uncertainty into supply
response. Just proposed including the variance (or standard deviation) of past prices
as an explanatory variable in the supplyresponsemodel. Aradhulla and Holt (1989)
obtained good results in such a framework for the broiler industry. In Chem and
Just’s (1978) supply responseand demand model for processing tomatoes in
California, they formulate supply and demand equations in which they include. the ,
producer’s expected price and expected yieldin their Nerlovian adaptive expectation
model- While making price and yield predictions, the producer is subject to making
errors in prediction. The fact that the grower has to make decisions in an unsure
environinent is called uncertainty or risk- The widely adopted mean-variance criterion
is used to measure riskand uncertainty by incorporating the variance (or standard
deviation) of past yields and prices in addition. to the expected yield and price as

another explanatory variable in the supply response model.

1

32

It was not until the early 1980’s that price rationality was introduced in agricultural
supply response as a price expectation mechanism for econometric modeling and
estimation based on the work of Muth (1961). Irwin and Thraen (1991) evaluate that
literature in detail. They concluded that results about rational expectation in supply
response are mixed. Although results in Goodwin and Shefl'rin (1982) give support to
the hypothesis, in the broiler chicken industry, while studies in the soybean sector are

not that conclusive.

Some studies, Gardner (1976), used fiitures prices as a proxy for rational expectations
price behavior held by producers. However, several doubts can be put in this type of
approach, since futures prices are “endogeneity determined”. In his article, “Futures
Prices in Supply Analysis”, Gardner hypothesizes that the price of a fiitures contract

for next year’s crop reflects the market’s estimate of next year’s cash price.

3.3 Summary

Previous studies on supply response in Argentina have basically three limitations. The
three limitations are: the use naive price expectation behavior, the omission of a yield
variable and/or inputs price variables, and they do not include soybeans in their
analysis which has become an important crop in Argentina. These three limitations are

addressed in the model analyzed.

33

CHAPTER 4
Econometric Modeling: Conceptual and Empirical
4.1 Background and Conceptual Model
It is assumed that Argentine producers want to maximize their profits given their
production constraints, that of two outputs (crops and cattle) and various (inputs. In

Argentina, cattle and crops compete for the use of land. Let the production firnction

be

yt = f(xtt, X21...xnt) (i)

y2 = g(X12, Xzzmxnz) (11)
where y; and y; are vectors denoting the two outputs, f and g the production
fimctions for y1 and y2, and the first subscript on each x denotes the input, the second
subscript. denotes the product to whichitis applied.

The total amount of XI and x2 used in the production of y; and y; are

X1 = X11+X21+.-.+ an

X2 = X12fX22+...+ Xn2

34

Total revenue R from the sale of y] and y: is

R = Pryl+ P2Y2 (iii)

= Prf(X11, X21.-.an)+ P280112, X22---Xn2) (iV)

where p; is the price of y, and p2 isthe price of y2.

The total cost is the sum of the quantities of inputs multiplied by their respective prices

C = v1x1+ V2X2+...+ v..xn (v)

= V1(er+ X12)+V2(X21+ X22)+---+Vn(an+ an) (Vi)

Profit is revenue minus cost:

11 = R-C (vii)
= P1Y1+ WY? va1- V2X2‘~--‘ an“ (viii)

= ptflxtt, X21...xnt)+ ng(xt2, X22...xa2)- vt(xtt+ xt2)- V2(X2t+ X22)-...- vn(xal+ xn2) (ix)

Maximizing profits entails taking the first derivative of the profit function with respect

to each input used, in the production of the output andsetting it equal to zero:

6W6x11=p18t76x1pv1=0 (X)
6W6xzr=p16fl OX21-V2=0 (X1)

35

6W6xn1=p16t7 aXu-Vn=0 (xii)
8W6x12=p28g/6x12-v1=0 (Xlll)
6W6x22=p26gléx22-v2=0 (XlV)

arr/aitfpzag/axnflfio (xv)

From the system of equations x through xv we can solve to obtain the input demand

equations.

X11: f11(1)» P2, V1, vzr'": Va) (XVi)
x21= f2}(p1:p2, v1, V2,..., v“) (xvii)
an= flrlpr, P2, V1, V2,.--, Vn) (XViii)
x12: f12(p1, p;, v;, v;,..., V“) (xix)

X22: f22(1’1, P2, V1, Vzama Vn) (XX)

. . Xn2= £12031, 122, V1, V2,---, Va) (705) .
Substituting xvi through xxi into i and ii, the outputfirnctions are obtarned:

Yr: MRI, P2, V1, VZr") Va) (xxii)

YZ= £2031, 132, V1, V2,-.., VII) (XXiii)
In the case of crop production, output is determined by acreage allocation and weather
conditions afl‘ecting crop growth, acreage being the single most important input whose

usage level the producer has to choose.

36

From the input demand equation above obtained, a functional relationship of crop

acreage to product prices and other input prices can be stated.
A: f(p1, p2, V1, V2,..., Vn) (XXlX)

Total agricultural output is a function of planted area and weather. Supply response
models in agriculture are typically modeled as planted area and not total output due to
the difficulty in modeling weather.

Most agricultural commodities take time to produce. Characteristically, decisions
about resource allocation are taken well before output is generated, due to the very
nature of the biological processes involved. Consequently, the realized output and
actual resulting income may differ from the ones prevailing at the time the decisions
are taken. Typically, resource allocation decisions are tied to the expectation of the

final outcome and price.
As one of many possible functional forms, Nerlove proposes a model to measure the
responsiveness of acreage to prices: desired acreage is a linear function of expected

prices, as follows:

A*t=ao+atP*t+a2Zt +Ut (1)

37

where A*. is the expected acreage in. period t, P*t is the expected price in period t, Z.
is the orher factors that affect supply/production in period t, and U. is the error term.
However, the desired acreage is not directly observable, therefore, Nerlove proposes

the following hypothesis of desired acreage‘behavior:
At-At-1=Y(A*t-At-1) (2)

where A.-A..1 measures the actual change in acreage between two periods, A*.-A..1 is
the desired change in acreage between two periods, and y is known’as the coeficient
of adjustment, such that 0<y<=.l. The actual change inacreage in any given period t is

some fiaction y of thedesired change. Equation 2 can be written as:
At=yA*t+(1-Y)Aet (3)

And, substituting equation 1 into 3, we derive an equation for acreage with observable

acreage vgriables, using Nerlove’spartial adjustment hypothesis:
At=7ao+7atP*t+ra2Z. +1Ut+( 1-7)At.t (4)

Once we estimate equation 4, we can derive the coeflicient of adjustment, and in so
doing, also derive the constant term, at), and the coefficients, 8,, for the, variables.

Equation 4 can be written as:

38

A¢=Bo+B 1P*r+B2Zt +B3At-l+l-1t (5)

where ya...1 is substituted by B“ and 1--Y=B3 on the lagged acreage variable, forming a
partial adjustment acreage response model. However, equation 5 still contains one
unobservable variable, P*,. Several different price behavior hypotheses can be
proposed to derive an observable equation for acreage responsiveness to prices. The
first price expectation formation behavior hypothesis is naive behavior, in which the
producer looks at last period’s price to form his next period’s price expectation, as

follows:
P*t= Pp] (6)

Substituting equation 6 into 5, we derive an acreage response model using naive price

expectation: ..
At=Bo+BrPer+Bzzt +B3Akl+l~lt (7)

An alternative hypothesis is the moving average price expectation behavior, in which

the producers form their price expectatiOn for period t as an average of past prices for

several periods:

P*t=(Pt-.|+Pt.2+ . . . Pug/n (8)

39

Substitutin equation 8 into 5, we derive an acreage response model using moving

averages price expectation:

At:B0+Bl(Pt-1+Pt-2+- . ~+Pt-n)/n+B4Zt +BsAt-r+l~lt (9)
A third approach to solve for the unobservable price variable in equation 5 is to
propose a weighted average price formation hypothesis in which the producer takes a
weighted average of past prices for this period’s price expectation, as follows:

P *t=CP t-l+(1-C)P [.2 ( 10)

Re-writing equation 10 so as to calculate weight c, assuming that P*t= Pt so as to be

able to use available data:
P t=CP t-1+P r-2'CPt-2
= 6(1) t-r-P t-2)+P t-2
Pt‘ PM: C(P t-t'P 1-2) (103)

If we assume:

P r- P 92:31

P r-t‘P r-zzbr

40

we can rewrite equation 10a as:
a. = c*b, ( 10b)

Equation 10b will be used to estimate weight c. Once the weight is derived,
substituting equation 10 into 5, we derive an acreage response model using weighted

average price expectation:
At=Bo+BrCPer+Bz(1-C)Pt.2+l54zt +BSAt-l+u-t - (1 1)

In order to analyze for different hypotheses of producer’s price expectation behavior,

equations 7, 9 and 11 will be the models used in this study.

Based on the Argentine supply response studies described in Chapter 3 and these
studies’ results, it is assumed that a linear functional form is a acceptable

representation of Argentine agricultural supply response.

The models presented above assumed constant input quality, input technology, and
knowledge. The model also does not address all the dynamics in the crops versus

livestock issue.

In order to maximize his/her profits, it is assumed that the Argentine producer decides

on acreage based on the expected gross income (his/her expectation of prices

41

multiplied by expected yield per hectare) less input (production) costs. These input
costs are represented by variable Z (the other factors that affect supply (production) in
period t) in equation 11. Theoretically the input costs should include: prices of
fertilizer, pesticides, fuel, seed, and labor for the product(s) (i.e. crops and livestock)
being produced. Other factors that affect supply represented by the Z variable are

technological changes, government policy, and weather.

Cattle production competes with crop production for land use. However, the decisiOn
to increase or decrease the herd is not instantaneous. Ifthe Argentine producer were
to decide to decrease the herd because of lower cattle prices, and consequently reduce
the amount of land on pasture, the producer would liquidate the herd and till the land
for crop production. This process would not happen in one period; there would be a
lag between the time the producer made the decision to decrease cattle production and
the time acreage in crop production increased. This is the concept of lags: an
economic cause, for example, a price change, produces its effect only after some lag in
time. A distributed lag implies that the effect produced by an economic cause is

noticed over more that. one period.

Incorporation of distributed lags with respectto price changes into a supply response

equation would result in an equation as follows:
At=ao+cr1P r-t+(1-2Pr-2+Q3P r-3+ . - . +013 ..,.+azZ.+rt. ( 12)

42

where A. represented acreage, PM represented price of the crop being studied in the
previous period, n represented the number of periods needed for the full effect of the

economic cause to produce its effect, and Z represents other factors that affect supply.

Equation 12 is distinct from equation 9 in that the partial adjustment is not included.
The distinction between partial adjustment and distributed lags is that with the partial
adjustment there is no explanation why the partial adjustment occurs since it is not
based on optimization behavior (i.e. do not know what causes the partial adjustment to
occur, whether it is prices, technology, capital constraints, etc. ..). The partial
adjustment coefficient captures the process of the producers’ slow adjustment to

economic causes (shocks) as proposed by Nerlove and demonstrated in equation 2.

In a supply response model of crop acreage, in order to incorporate the competing
cattle production, cattle prices should be included as an explanatory variable- If one
were to use the distributed lag equation 12 to analyze acreage response, one would
have to include crop prices and cattle prices distributed n periods, which would
increase degrees of fi'eedom and reduce the measurement power of the equation in
econometric analysis. Consequently, the decision is whether to estimate the aggregate
supply response using distributed lags, or whether to use partial adjustment. In this
supply response analysis, the partial adjustment hypothesis will be used to represent

the adjustment cost of switching acreage fi'om cattle to crop production.

43

4. 2 Econometric Estimation Issues

A supply response equation includes several explanatory variables. Two consequences
of this are muticollinearity (where a linear relationship between explanatory variables
exits) and degrees of freedom can decrease as more explanatory variables are included
which reduces adjusted R squared”. One way around including so many explanatory
variables is to work with expected income, calculated as expected price multiplied by

the expected yield.

Inc‘.= P‘. *Y‘. (13)
where Incl represents the expected income in period t, PX represents the expected
price in period t, and Y'. represents the expected yield in period t. The expected yield
can be calculated by estimatingalinear trend based on historical data:

Y’t =Co+C1Trend (14)

where the trend variable represents the year. Trend yield rather than lagged yield is

used since it would be more representative of the producer’s expectations-

 

” Degrees of freedom are defined as the number of observations minus the number of coefficients
estimated (including the constant) or n-K- I", where n is the sample size and K is the number of

explanatory variables.
44

A useful method to detect multicollinearity is to check if the R squared is high with no
significant individual t-tests, and if the simple correlation coefficients between the
explanatory variables are high. Determination of how high is high for the R squared
and simple correlation coefficients is based. on the fact that previous supply response
studies on Argentina had R squared results of 0.80 or higher. It is assumed that a R

squared of 0.80 is high and that a simple correlation higher than 60% is deemed high.

Heteroskedasticity occurs when the variance of the error term is not constant. The
consequences of heteroskedasticity are: it increases the varianceof the estimated B’s
distribution; it causes OLS to tend to underestimate the variances (and standard
errors) of the coeflicients; and it does bias the estimated standard errors, although the
estimated B’s are not biased. A test for heteroskedasticity is the White test, which
regresses the squared residuals on the original explanatory variables. The decision rule
of the White test is: if nR2 (the test statistic) is larger than the critical chi-squared
value, reject the null hypothesis of no heteroskedasticity; if nR2 is less than the critical ‘
chi-square value, the null hypothesis cannot be rejected. Ifthe supply response model
has been specified correctly, the White test should detect no heteroskedasticity.
Heteroskedasticity can be caused by a specification error, such as an omitted variable.
An omitted variable can cause a heteroskedastic error term because the portion of the
omitted effect not represented by one of the included explanatory variables must be

absorbed by the error term. Ifthis efi‘ecthas a “heteroskedastic component, the error

45

term of the misspecified equation might be heteroskedastic even if the error term of

the true equation is not”?

In time series data, a large external shock to an economy in one period may linger on
for several time periods, implying serial correlation where the error terms are
correlated with each other. In Argentina, there have many economical and political
changes that might have caused such shocks. Serial correlation may also be present in
an incorrectly specified equation since the error term includes a portion of the effect of
the difl‘erence between the proper fiinctional form and the equation being estimated
The consequence if there is serial correlation is increased variances of the estimated B
coefficients. If this is due to improper specification, the coefficient estimates may be
biased. Therefore, a test to detect if there is serial correlation in the three price
expectation hypotheses regressions should be performed. However, due to the partial
adjustment, a lagged dependant variable, as an independent variable, the Durbin-
Watson d test is potentially invalid and a Durbin h test should be performed. The

equation for Durbin h statiStic is:
h=(r-o.s*d)[n/(r-n*s.2)]*
where d is the Durbin-Watson statistic, n is the sample size, and $12 is the square of the

estimated standard error of BAH, the estimated coeflicient of the lagged dependant

variable. Durbin h is normally distributed, therefore to determine the critical z-value, a

 

'2 Studenmund, AH. Using Econometrics, p. 372.
46

two-tailed test at a determined significance level is calculated. The decision rule of the
Durbin h test is: if the absolute value of h is greater than the critical z-value, reject the
null hypothesis of no serial correlation; if the absolute value of h is less than critical 2-

value, do not reject the null hypothesis of no serial correlation.

4. 3 Empirical Model

The Pampas region of Argentina was chosen as the area of focus since it is the largest
agricultural crop producing region in Argentina. Six of the major crops grown in this
region, wheat, corn, sorghum, soybean, sunflower, and flaxseed, are the focus for
aggregate grain production. Planted area data from 1960 to 1997, on a crop year of
June through May, for each of the six crops was collected fi'om the Bolsa de Cereales
(Grain Exchange) annual statistics, and totaled to derive a total planted area series, in
thousands of hectares. Due to double cropping, actual planted area may exceed actual

land in crop production.

47

 

Figure 3: Total Crop Planted Ana-1n Argentina

 

Thousand Hectares

 

 

 

 

 

Annual yields in tons per hectare for each of the six. crops were collected from the
Bolsa de Cereales. Using each crop’s yield, a trend yield was calculated to determine
expected yields. It is assumed that yield per hectare is a good proxy for technological

improvement.

Price data were also collected fi'om the Bolsa de Cereales annual statistics fi'om 1960
to 1996. Monthly average pfices for each of the six crops, in pesos per ton, were used
'to calculate a price index- In Argentina, almost of all the crops are marketed during
the first six months of the season (January through June). The first crop seeded is
wheat in May or June. Beforethis seedingtime, the Argentine producer needs tomake
a decision about what crops to produce. It is assumed that the prices the producer
looks at are the prices during the period of January to June. Due to this assumption,
monthly prices fi'om the first six months of every year were used as the best indicators

for the producers seeding decisions. These prices were converted to US dollars per

48

 

 

 

Figure 4: Crop Yield in Argentina

 

“ 1.6/966l
:.96/l766L
_ 96/3661.
L6/066L
69/996L
: £9/996l .
“ 99/1796L .
: 89/2961.
: L9/096l
: 61/8261
: LUQLGL
: 9Ul7£6l
: QUZLGL
: LUOL61
: 69/996L
: L9/996l
: 99/796l
E 99/396lv

l l l l T l

 

 

 

7000

 

- 19/096L

   

000 -

r
O
O
O
N

3000 -

O O
O O
O O
(D ID V .
aretoaH/suol arnaw

 

 

Soybean

- Sorghum —o— Flaxseed

 

49

—-I— Corn

 

 

+Sunflower - - - - -

—-a—Wheat

 

 

..____. _

 

 

 

 

 

Figure 5: Real Crop Prices (January-June) 1
g 900--r~—-~w -—--—--—--—- » ‘ --—-——-——— -------- ~w*
I- 800 -
0
b 700» t
g 600 « / f
E soc-r A j
g 400 « / ‘ . i
8 3%} ’— J . - \\: . Al\v._\ . ,
m ’ .4 , ‘ I ' ' ' - .A /A O h "‘
a 100 l -’ 7.. .. .’ .. " -- -- - . I‘M. -— .. A. n
a 0 l
g I I I I IIIIII .I I I I n I I T f I I I I If I I I I I T T T I T—I
O O N V (D co 0 N V ‘0 co 0 N V co
sgéésaaaasasaaaasafis
—-—wheat +com —soybean
sunflower ------ sorghum —o—flaxseed

 

 

 

 

 

 

 

metric ton using ofiicial exchange rate figures, due to the fact that export prices are
quoted in US dollars per metric ton, and since most of the grains and oilseeds are
exported and thus directly related. These prices in US dollars per metric ton were then
deflated by the US Producer Pfice Index ('PPI). The real price series for each of the

six crops is then used in the gross expected income calculation.

Cattle price data were gathered from the National Meat Board (Junta Nacional de
Came), Bolsa de Cereales, and the. Secretaria de Agricultura, Ganaderia y Pesca
(SAGyP). Monthly prices from the first six months of every year, in pesos per kilo
from 1960 to 1997, converted to US dollars per kilo (using official exchange rates),
deflated by the US PPI, were used as the cattle price series. The first six months were

used since the same was used for the grainprice series.

50

 

Figure 6: Reel Cattle Price (January-June)

 

2.5
2.0 ~
1.5 ~

1.0 A

Real US Dollarleilo

0.5 ~

 

 

 

o. o r I I I I I I I I I I I I I I T I I r I I I I I I T T I I I I I I I I j

sfldefideef §§§$§§§§$

 

 

 

Cattle weight data, in kilos per head, were gathered fi'om FIEL (Latin American
Foundation for Economic Research). This is the annual average weight per head of
the cattle sold to market. Using the annual weight, a linear trend weight was calculated
to determine expected weight (cattle yield) for gross expected income calculation. '3 It
is also assumed the cattle weight trend is a good proxy for technological improvement

in cattle raising.

 

'3 Average cattle weight increased during 1955-82, and then-began decreasing A quadratic function
might be a better trend line, but it does not appear reasonable that average cattle weight will continue
decreasing The three price behavior expectation models were estimated with the expected eattle
gross income calculated using a quadratic trend line and a linear trend line. Results were not
signifieantly difl‘erent. Further work is required to determine the reason for decreasing average cattle

weight.
5 l

 

 

Kilos/Head

Figure 7: Average Cattle Weightin Argentina

47o -
460 —
450 ~
440 q
430 -
420 -
410'}
400 1

390

 

 

 

1 955

IIIIIIIIIIIIII

i

I I I I I I T I I I I I I r I T I I I I I I I
8 8 N N l‘ co co Q C) 05
O) O) O) 03 O) O) 0’ O a: G

v- 1- v- ‘- 1- 1- ‘- F P v- 1- P ‘-

 

From various personal conversations with sources in Argentina, it was determined that

the major variable input costs for Argentine grain production were diesel firel (gasoil),

labor wages, and urea (a nitrogenous fertilizer). Farm labor wage data were not

available, and it was determined that an index of industrial wages (the only wage series

obtained by this author) was not an adequate substitute due to it being from a different

sector and government policy influences in that sector. Monthly average diesel prices,

 

Real US Dollars/Liter

 

Figure 8: Real Diesel Price. (Januaryalune)

 

 

 

 

 

I I I I I T I I I I I I I f1 T I I fI r I I I I I I I I r I I I IV;
C m a) N II) n v- V’ N O ('0
a) a) a) a) a Q m a} a)

 

 

 

Inn's.‘ .

in pesos per liter, were obtained fiom F IEL. These prices were converted to US
dollars per ton using official exchange rate data and then deflated by US PPI. The

diesel price series was composed of the first six months of each year.

Urea price data were gathered from the SAGyP for the period 1970-1998- Appendix
A provides a detailed description of the urea price series. As with diesel, grain and

cattle prices, the urea price dataused was composed of January to June prices.

 

 

 

 

 

 

. Figure 9: Real Urea Price (JanuaryaJune)
600 -
8
'—
.§
é
a
In
:3
8
K
o ff VT IIIIIII I rTj r I rI
O (‘0 CD 03 N In no 1- N O (O
eeasaaaeéassi

 

 

 

Weather was assumed not to be an important factor afl‘ecting total seeded area, if

weather prevented seeding of one crop, another crop might have been sown.

4.3.1 Model Specification

 

Equation 7 can be rewritten to incorporate the specific above mentioned variables.

53

Ar=Bo+B1GPr-1+I32CPr-1+I33GY.r+B4CY.r+l3sUfeaPr+BsGasPr+BvAr-1+l—lr (7a)

where A represents total seeded area in thousand hectares in the period t, GP
represents the grain price series in US$ per ton in the period t-l, CP is cattle price
series in US$ per kilo in the period t-l, GY' is expected grain yield per hectare in
period t, CY ' is expected cattle yield (weight) in kilos per head in period t, UreaP is

urea price in the period t, and GasP is diesel price in the period t.

The price variables and the expected yield variables can be combined into a gross

income variable:
GInct=Zaikl(GPi¢.1*GYTir) (15)
CInct=(CP,.1*CYTt) (16)
where,
GY‘.=GYT;.
CY‘.=CYTa

where GInc is expected grain gross income per hectare in period t, CInc is expected
cattle gross income per kilo in period t, CP,.I is the cattle price in the period t-l (naive
expectations), GP,“ is the grain price in the period H for crop .i, GYTi. is trend grain

54

yield of crop i in period t, CYT, is trend cattle yield (weight) in period t, and am is the
proportion of crop i in period t-l, in annual in aggregate planted area. The trend
yields on grain and cattle were calculated using a simple trend regression for the entire

sample of historical data gathered, as stated in equation 14.

Since it was determined earlier that urea and diesel are important inputs in grain
production, these two inputs will be used as explanatory variables in the three
aggregate crop planted area equations. With the omission of a labor price variable, a
possible relevant independent variable, estimated coefiicient results may be biased.
According to Studenmund (1997), an omitted variable may be detected if an estimated
coefficient is significant and has an unexpected sign, or if the model has a surprisingly

poor fit.

Incorporating the expected gross income variables into equation 7, the aggregate crop

planted area response is:
Ar=I30+BIGInCr+I32CInCr +BsArr+BsUreaPr+B1GasPr+iu (7b)

Equation 7b will be used to analyzethefirst price behavior hypothesis: naive

expectations.

55

To analyze the second price behavior hypothesis, moving average expectation,
equation 8 is substituted into equation 15 and the expected gross income is calculated

as follows:
MAGInq=ZaM([(GP;..1+GP;..2)/2.] *GYTi.) ( 1 5a)
MACInc‘=([(CPt.1+CP.-2)/2]*CYTt) (16a)

where _MACInc is an average of the lasttwo periodsfor cattle prices, (151.; is the
proportion of crop i in the previous period, and MAGInc is an average of the last two
periods for grain prices. These new movingaverage expected gross incomes are then
substituted into equation 9 to derive an aggregate crop planted area response with

moving average price behavior expectation;
A=BO+B1MAGInq+BzMACInq+B3A..1+B4UreaP.+BsGasPt+p. (9a)

Two period moving average is used for the above stated equation based on personal
interviews which stated that it appeared that Argentine producers do not ‘remember’

or take into consideration prices. fiom before two periods prior.

The third price behavior hypothesis to be analyzed is the simple weighted average of
past prices. Substituting equation 10 into equation 15, we derive the simple weighted

average expected gross income as follows:

56

.' 5W..-flh:o .

" . oihuflnfl~ - .

SWGIncF-Zou.-1((ciGGPi,-1+( 1 -ciG)GP;.-2)*GYTt) . (1 5b)

SWCInc,=((ccCPH+(1-cc)CP..2)*CYT.) (16b)

where SWGInc is the simple weighted average of grain prices, a,“ is the proportion
of crop i in the previous period, on is the calculated price weight for crop i, SWCInc is
the simple weighted average of cattle prices, and cc is the calculated price weight for

cattle. The weight, c, is estimatedusingequation lOaand 10b as follows:

GPit-GPir-2=CIG(GPir-i-GPir-z) (100)
such that:
GP it'GPit-z': air
GPit-I‘GPit-2= bit
air: CiGbir (10d)
and:

CPg'CP92=Cc(CP¢.1'CPg.2) ( 1 0e)

Equations 10d and 10e estimated individual crop and cattle weight results are used to

calculate the simple weighted average expected gross crop and cattle income in

57

equations 15b and 16b- Substituting the simple weighted expected gross margins into
equation'l 1, we derive an aggregate crop planted area response with simple weighted

price behavior expectation:

Arr-[304131SWGInCfl'BzSWCInCH'B3M1+B4UICaPt+BsGaSPt+L1¢ ( 1 I a)

Equations 7b, 9a, and 11a are used to estimate Argentina’s aggregate crop acreage
supply response under three difi‘erent price behavior hypotheses. Determination of the

best structural equation (model) is based on four criteria:

1. The signs of the estimated coeflicients,
2. Statistical significance of estimated coeficients
3. Adjusted R-square is high.

4. The reasonableness of the results.

Ordinary least squares, usingE-Vrews software Version 1.0, is used to estimate the

above equations.

58

CHAPTER 5
Results

This chapter presents the regressions results for the three price expectation behavior
hypothesis models used to estimate Argentina’s aggregate crop planted area supply

response. These are:

Naive:
AFBo‘l'B lGIncr+BZCInct +BsA+r+36UfeaPr+B7GaSPr+llr (73)

Moving Average:

A.=BO+BIMAGInct+BzMACInq+BgAH+BaUreaP.+BsGasP¢+u. (9a)
Simple Weighted Average:
A¢=Bo+B1SWGIDCfi'BzSWCIIlCH’B3A4.1+B4UTCaPI+BsGaSPg+l1¢ ( I 1 a)
5.1 Results

Below is a table of empirical regression results for the three linear price expectation
behavior models: naive, moving average, and simple weighted average. In Appendix
B, the results of the three price expectationbehavior: models estimated. usinga log-log
firnctional form are presented as a comparison to the linear form presented below. In

summary, the log-log model results are not much difi'erent than those presented below.

59

Igble 2: Results of aggregate crop planted area using three specificgtions of price
expectation behgvior. Argenting 1960-1997

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Model Naive Moving Average Simple Weighted
Average
B t B t B t
Statistic Statistic Statistic
Bo -934.11 -0.37 -103 7.04 -0.47 -802.86 -0.32
Area” 1.04 7. 77 0.99 8.29 1.03 7-67
~UreaP. 4.04 1.37 4.69 1.74 3.62 1.23
GasP. -8132.09 '-2.34' -8376.59 -2.55 -8794.66 -2.53
Ginc 0.0070 2.18
Cinc . -2.56 -I.57
MAGinc ' 0.0115 3.51
MACinc -4.71 -2.60
SWGinc .0070 2. 15
SWCinc -1 .95 -1 .38
'n 37 37 37
Adj. R2 0.82 0.85 0.81
Model Naive Moving Average Simple Weighted
Average
White Test 10.68 13.43 1 1.19
Durbin h Test 1.64 0.85 -l.09

 

 

 

 

It was stated in Chapter 4 that the criteria to determine the best structural model are:

signs of the estimated coefficients, statistically significant estimated coefficients,

adjusted R2, and reasonableness of results. As the table shows, adjusted R2 for all

three models are above 0-80. The moving average price. expectation. model has the

highest adjusted R2 of the three. In comparison, Reca’s adjusted R2 for his grain

subsectpr model were between 0.82 to 0.88, and for his aggregate model were

between 0.82 to 0.87. Wainio’s average R2 for his four single equations was 0.90.

The signs on the estimated coeflicients are not all as expected. The constant

coefficient in all three models has a negative sign, however they are not significantly

60

 

 

different from zero, even at the 20% significance level. According to Studenmund
(1997), one cannot rely on the estimates of the constant term for inference for two
reasons. The first is that the “constant term acts as a garbage collector, with an
unknown amount of the mean effect ...of the omission of marginal independent
variables...being dumped into it” and second, “the origin ofien lies outside the range of

sample observations”.

' The urea price variable, in all three models,the sign is positive and, it is statistically
significant in the naive model at a 18% significance level, in the moving average model
at a 9% level, and in the simple weighted average model at a 23% level. The expected
sign of the estimated coefficient depended on whether the input is a complement or
substitute to planted area. The ‘a priori’ expected signwas negative, however,
according to Helmberger and Chavas (1996), a possible reason for the positive sign on
the urea price variable is that fertilizer and planted area are substitute inputs. An
increase in the price of fertilizer increases the producer’s demand for land. Another
possible explanation of the estimated positive sign is thatthere is a correlation of urea
price with soybean planted area (soybean area has expanded considerably at the
expense of flaxseed and sorghum, andalso as a result of double cropping with wheat).
A last possible explanation for the positive sign is as a result of an omitted variable
problem, although the estimated coeflicientis only significant ata9%. level (using .
Studenmund’s (1997) suggestion of detection for an omitted variable). Further study

is required to determine the reason for a positive sign for the urea price variable.

61

In the naive price expectation model, the estimated coefficients are statistically
significant at a 5% significance level, except for the estimated coefficient of the cattle
gross income variable which is statistically significant only at a 13% level and the urea
variable mentioned above. In the moving average price expectation model, the only
estimated coefficient which is not statistically significant at a 5% significance level is
the urea price variable- In the simple weighted average price behavior model, the
estimated coefficient of the diesel variable is statistically significant at a 2% level, while
the simple weighted grain gross income estimatedcoeficientis significantat an 4%
level and the simple weighted cattle gross income is significant at a 18% significance

level.
5.1. 1 Diagnostics for Multicollinearity, Heteroskedasticity, and Serial Correlation

Using the simple correlation coefficients between the explanatory variables in the three
price behavioral models, noneof the models appeared to show serious
multicollinearity. In the naive model, the simple correlation between diesel price and
grain gross income at 0.75,. and diesel price and cattle gross income at.0.64 are higher
than the 0.60 specified as the criteria in Chapter 4, together with the simple correlation
between grain gross income and cattle gross income at0-70. However, these two
naive model variables are statistically significant at a 4% significance level and 13%
level respectively- The simple correlationsbetween ureaprice and the other
explanatory variables are low, however, urea price is only statistically significant at a

18% significance level, therefore it may be concluded that multicollinearity is not

62

evident. In the moving average model, as in the naive model, the simple correlation of
the diesel price variable and moving average grain gross income variable at 0.75 and
diesel price and the moving average cattle gross income variable at 0.71 are high,
together with the simple correlation of and moving average cattle gross income
variables at 0.72. All of these variables, except for the urea price variable, are
statistically significant at a 2% significance level or lower, therefore it may be
concluded that multicollinearity in this model is not a concern. In the simple weighted
average model, the simple correlation of dieseland simple weighted grain income
variables at 0.75 and simple weighted grain gross income and simple weighted cattle
gross income at 0.67 are high, The dieseland simple weighted grain gross income
variables are statistically significant at a 4% level or lower, while the simple weighted
cattle gross income is significant at a 18% level. Using the criteria that R2 is above
0.80 and that the diesel variable is statistically significantat a2% level,

multicollinearity does not appear to be a concern in this model.

The hypothesis testing which follows is conductedat a5% significance level for the

following reasons:

0 It is believed. that the data, except for a few assumptions in the urea price series, is
accurate, the variance of the data is high and that the sample size is reasonable at
37 years.

0 It is assumed that the models beingtestedarewellspecified.

63

o The ‘costs’ of making a Type I error (rejecting the hypothesis when it is true) are
large to those who might be interested in an Argentine aggregate crop supply

response.

Testing the hypothesis that there is no heteroskedasticity on the three price
expectation behavior models, where:
Ho: Homoskedasticity
HA: Heteroskedasticity
.It was determined, with degrees of freedom equal to 10, the test statistic- at 5 percent
level of significance, the critical chi-squared value is 18.31, that:
1. In the naive model, with a nR2 of 10.68 as shown in the above table, the null
hypothesis of homoskedasticity” cannot be rejected.
2. In the moving average model, with a nR2 of 13.43, the null hypothesis of
homoskedasticity cannot be rejected.
3. In the simple weighted average model, with a nR2 of 11.19, the null hypothesis of

homoskedasticity cannot be rejected.

Using the Durbin h test for serial correlation, and the hypothesis that there is no serial
correlation in the three price expectation behavior models, where:
Ho: No serial correlation

HA: Serial correlation

 

‘4 Ifan error term is homoskedastic, the variance of the distribution of the error term is constant.

64

It is determined, with a two-tailed test at a 5 percent significance level implying a

critical z-value of 1.96, that:

1. In the naive model, since the Durbin h test is 1-64, the null hypothesis of no serial
correlation cannot be rejected.

2. In the moving average model, with a Durbin h test of 0.85, the null hypothesis
cannot be rejected.

3. In the simple weightedaverage model, with aDurbin h test of —l .09, the null
hypothesis cannot be rejected.

Before determining which is the best structuralmodel, the results of the weight

estimation for the simple weighted average price behavior model is discussed below.

5.1.2 Weight Estimation for Simple Werghfid’ Average P1133 Model

 

 

Estimated individual crop price weights used in the calculation for the aggregate grain

 

 

 

 

 

 

 

 

 

 

 

 

 

simple weighted average model15 is presented below:
Crop Wgeight

Wheat 0.95

Corn - 0.94

Soybean l 1.04

Sunflower 0.89

-Sorghum . 1.02

Hmeed 0:86

Cattle 1.21

'5 Equation 10c and 10e in Chapter 4.
65

Therefore, this implies that the weights for the second lagged period, where the weight

is defined as l-c, are:

 

 

 

 

 

 

 

 

Crop 2"" Lagged

Period Weight
Wheat 0.05
Corn 0.06
Soybean -0.04
Sunflower - 0.1 l -
Soi‘ghum . .002
Raxseed 0.1'4'
‘ "Cattle .0121

 

 

 

 

The price weights for soybean; sorghum and cattle for the first lagged period are
greater than one, which imply. anega'tive weight for the second lagged period. This
would imply that the Argentine producers consider the previous period’s price and not
the second previous period. Thisisimpliesa naive expectation price behavior for
these three variables. It must be mentioned that the above calculated price weights

were not statistically significant at a20%_level.

The calculated weight results do not appear reasonable. Interpretation of these results
imply that a combination of naive and simple weighted price behavior mechanisms are
being analyzed. Further work is required to determine these results, and consequently,

this model is not considered apotential structural model.

5. 2 Best Structural Model

Given the above mentioned results for the three price behavior models, it appears that

a moving average behavior of expected price is a better representation of the sample

data than a naive expectation.

The moving average price expectation behavior model was chosen as the best

structural model for the following reasons:

It had the highest adjusted R2, at 0.35, of the three estimated model, implying that
this model explains 85% of the variation around the average area.

The estimated coefficients which had the correct sign also had the highest t-tests of
the three models considered, implying that the probability that these estimated
coefficients equal to zero are smaller.

Calculated weights in the simple weighted average price behavior model do not
appear reasonable and theoretically consistent.

The rate of adjustment for the naive and simple weighted average price behavior
models. are negative, -0.04fihousand'hectaresfor the naive model and —0.03
thousand hectares for the simple weighted average model, and statistically
significant at a 0% level. ltdoes notmake sense that a producer would have a
negative rate of adjustment (it would imply that the producer ‘go backwards’) due

to an economic shock.

67

5. 2. 1 Moving Average Price Ermctation Behavior Model Interpretations

As stated above, the adjusted R squared is 0.85 which implies that this equation
explains 85% of the variability in area, adjusted for the degrees of freedom. The sign
of the coefficient for expected grain gross income is positive as theory suggests; as
grain prices increase, seeded area increases. Since this is a moving average of
expected grain income, this implies that as the moving average of two previous
periods’ grain income increases, Argentine seeded area increases.

The coefficient of adjustment (the lagged area variable) is statistically significant at a
zero significance level implying that the rate of adjustment the Argentine producer
responds (adjusts) to changes (shocks) in prices is 0.01 thousand hectares, which

implies that Argentine producersare slowin responding to changes in prices.16

The elasticities of thismodeLare”:

Table 3:“ Short and Lon Run Elasticities

 

 

 

Elasticity - Short-Run Mpg Run
base ~ -- ~ - .032.~ . -1.2
MAGInc 0.21 2.1
MACinc -0.10 -1.0

 

 

 

 

 

 

‘5 Conceptually, the coeflicient forthepartial‘adjustment, Bu, represents (l-y), where y is themeasure
of adjustment Where 0<y< I.“1fy=1,’then theactual planted area is equal to the desired planted area
lfy=0, thenthe planted areainthispericdis equal totheplantcdarea in the previouspcfiodandtlre
desired planted area will neverbe reached '

‘7 Short run elasticities calculated asfollews: SW;where x=(x1 +...+x..-)/ttadd
Y=(Y,+..%=tY.)/n. Long run elastiCitieS’Calc‘ulated asfollows: meme/(remit The coeflicient
for the partial adjustment is used in calculating the long run elasticity since it estimatcsthciongterm
adjustments to prices.

68

Urea elasticity was not calculated since the estimated coefficient was insignificant at a

5% level.

A one percent increase in a two period (year) moving average grain income will
increase seeded acreage by 0.21 percent in the short run, while a one percent increase
in a two period moving average cattle income will decrease acreage by 0.10 percent in
the short run. A one percent increase in diesel prices will decrease seeded acreage by

0.12 percent in the short run.

Long run elasticities give the efl‘ect of a permanent change in expected price. A one
percent increase in Argentina’s two period moving average grain income would imply
a 2.1 percent increase in seeded acreage. A one percent increase in Argentina’s two

period moving average cattle income Would imply a 1.0 percent decrease in seeded

acreage.

 

Figure WM Forecaster! Planted Area,

 

 

 

 

 

 

 

 

[Wicca-97

O

2

5

0

0

:I:

'0

c

3

3

o

S

P 0. llIIIIIIIWITIIIIIIIIIflIFIIIIIIIIIII
1- rs oven—or «gm F N
sissssssgégég
§§§§§§§§e§e§e

[— Actual Total ------ i Forecasted Area]

 

 

69

Comparison of actual planted area with forecasted planted area shows that the moving
average price expectation behavior is a good representation of the data. The forecast

error of the moving average model is 4%.

70

CHAPTER 6
Conclusions

Previous studies of planted area supply response in Argentina assume that the
producer wants to maximize his profit, therefore theory implies that the Argentine
farmer decides on acreage based on the expected income (his/her expectation of prices
multiplied by expected yield per hectare) less input (production) costs. The input costs
should include: price of fertilizer, pesticides, fire], seed, and labor for the product(s)
(i.e. crops and livestock) being produced. Other factors that affect supply are
technological changes, government policy, and weather. Reca’s, Wainio’s and
Sturzenegger’s studies incorporate the above mentionedtheory with some difficulty
due to lack of data or omission. This study works towards explaining the variations in
aggregate crop acreage while improving the specification of a profit maximizing
producer by: a) incorporating input variables and yield variables, b) incorporating a
longer period of analysis to include recent data, andc). including soybeans in. the

analysis due to the increasing importance of its production to the nation.

The primary objective of this study was to develop asingle equation model to predict
Argentine aggregate planted area supply response to various variables. Secondary
objectives were: 1) determine how Argentine producers form their price expectations,
' 2) determine the period of adjustment between economic shocks and its effect on

planted area, and 3) determine the explanatory variables needed to explain the area in

71

crop production. The model is a simple comparative advantage model between cattle
and crops. Using a simplified Nerlove linear model specification of aggregate planted
area of six crops dependant on expected grain gross income (grain price multiplied by
grain trend yield), competing expected cattle gross income (cattle price multiplied by
cattle trend weight), diesel price, urea price and a partial adjustment variable, three

difi‘erent price expectation behaviors were analyzed-

The three price expectation behaviors analyzed were naive, moving average and simple
weighted average. Results show thatthe moving average model is a better
representation of the sample data. This implies that Argentine farmers respond to a
moving average of the previous two years’ prices. Previous studies have- found that
Argentine producers respond to the previous period’s price, i.e. naive expectations.
This moving average price expectation behavior. appears reasonable if one were to take
into consideration that it takes more than one year to adjust fiom cattle production to
grain production given that the cattle stock needs to be liquidated and the land
prepared for grain production. The implication that Argentine producers take at least
two periods to adjusttheir crop acreageisalso corroborated by the partial adjustment
coefficient which also shows that the Argentine producer is reasonably slow in
adjusting crop acreageto economic shocks, althoughthe actual length of adjustment is

not determined.

An attempt at a well specified supply response model appears to be successfirl, as

implied by the adjusted R2 of 0.85 (which is comparable to the other Argentine supply

72

response studies by Reca, Wainio, and Sturzenegger). The following explanatory
variables were included in the models: a yield variable, product price variable,
competing product price variable, and input price variables. These explanatory
variables’ statistical significance at a, 2% significance. level or lower is in agreement
with profit maximizing theory. Results show that, in making his/her crop production
decision, the producer is influenced by expected cattle gross income, expected. grain
gross income, and input prices. Estimated elasticities of expected grain gross income
are 0.21 in the short run and2-1 in the long run whichcompares to Wainio’s wheat
short run elasticity of 0.48 and long run of 0.67. Estimated elasticities of expected
cattle gross income is —0. 10 in the short runand-1.0 in the longrun, which compares

to Wainio’s beef price elasticity on wheat acreage of 0. 10 in the short run.

This analysis improves on previous eflmm by using more recent data, Studies on
supply response in Argentina were last performed with data through 1985 by
Sturzenegger. In an effort to update the analysis, this study uses datathrough 1997,
which incorporates an entirely new economic period in Argentina, that of President
Carlos Menem’s Convertibility Plan, which has removed governmentintervention in

agriculture, and has stabilized the economy.

This study also attempts an aggregate crop planted area supply response, instead of .
individual crop planted area response, to determine the variations in total crop planted
area in Argentina. In addition, soybeans were includedin this study, where in other

studies mentioned soybeans were not included. Soybeans only became an important

73

crop in the late 70’s when planted area was 0.5 million hectares. By 1996, soybean
planted area was 7.2 million hectares, becoming one of the nation’s most important

crops.

The explanatory variables in this study included input prices. These inputs were
limited to crop production. Inclusion of inputs related to cattle production would

improve the aggregate crop acreage supply response model-

The inclusion of urea and diesel price variables appear to explain the variation in
planted crop area However, further work is neededto determine whether the positive
sign on the estimated coeflicient of the urea price variable is a result of an omitted
variable. or if in fact it demonstratesthat urea priceis a. substitute input with land, as

suggested by Helmberger and Chavas (1996).

In order to improve the present aggregate supply response model, firrther work could
include risk, government intervention in the form of subsidies, and cattle as a capital
good A producer, by the very nature of crop production, has to make production ,
decisions based on potential output and price scenarios since the actual scenario is not
known at decision time. There are many uncontrollable factors that affect these
potential scenarios, such that the expected output and price scenario may not be
realized. Incorporation of Just’s concept of price uncertainty into supply response by
including the variance (0r standard deviation) of past prices over the period of analysis

as an explanatory variable might be one possible method of improving this supply

74

response model. The producer is also faced with yield uncertainties when making his
production decision, therefore inclusion of yield risk into the model in the same

manner as in price uncertainties should also be investigated.

Although this study incorporated government intervention in price determination (for
1973-76) by using domestic prices converted into US dollars (the domestic price in
domestic currency is the final outcome of government pricing policy, and converting
the domestic price into US dollars takes into consideration government exchange rate
policy), this does not address all of the ways in which government intervened in

C.

Argentine agriculture. For example, inclusion of the availability of credit in the

O

aggregate supply response may improve the model.

As mentioned previously in this Study, cattle and crops compete for area in Argentina.
Historically, cattle has been an important product in Argentina because it has allowed
the producer'to ‘hedge’ against economic instability. As a part of his study of supply
response in cattle industry in Argentina, Jarvis addresses the concept of cattle as
capital goods. Jarvis maintainst‘natproducers considered cattle production as an
investment opportunity. Inclusion of such in this aggregate crop planted area supply

response might also improve the model.

Moving land from cattle- to creps requires liquidating the cattle stock and preparing the
field for crop production. This process takes time. In this study, a two year period is

analyzed (a moving average price expectation of two periods). It is possible that a

75

cattle producer needs to see the price change last for longer than two years before
he/she makes the decision to shift fiom cattle production to crop production due to the
capital investment involved. Incorporation of the more complex dynamics involved

would improve the present aggregate. planted area supply response.

Further work might include the role of budgeting in resource allocation. An Argentine
producer, if one assumes profit maximization, is goingto look at his potential income,
costs and determine his/her allocation based on his budget. His budget may be
imposed by capital, resources, or personal requirements (i.e. time ofl‘ during the

winter).

Further work on this aggregate crop planted area supply response might test for I
different price expectation behaviors, such as adaptive expectation and rational
expectations. Although this study shows thatit is likely, given the data used, that
Argentine producers form a moving average price expectation, analysis of other price
behavior expectations might show difl‘erentresults. Analysis of rational expectation
price behavior, where the producer forms his price expectation by looking towards the
firture and analyzing potential supply/demand scenarios,would be interesting given
that access to information in Argentina is improving, the Argentine producer is
becoming amore sophisticatedandeducatedmanager, andtbat Argentina is an

important agricultural world producer.

76

In conclusion, a single equation aggregate crop planted area supply response model for
Argentina was specified and estimated, resulting in a respectable model. This model
may be used by the those in the industry which could benefit fiom being able to

forecast aggregate crop planted area.

77

APPENDICES

78

APPENDIX A
Urea Price Series Construction
Summary

An Argentine domestic urea price series was located for the period 1970-199 7,
however, the period of stuajz is 1960-199 7, therefore an attempt was made to estimate
domestic urea prices for the missing data. Taking a World Bank publication ’s
example calculation of importing urea into Argentina, a domestic urea price series
was estimated’ 8. Yhere were many assumptions that had to be made due to lack of
information, but it is believed that the estimated Argentine. domestic urea price series
is a reasonable representation of the actual price series. Consequently, it is believed
that for the period 1960-19 70, the estimated urea price series can be used as an index

to complete the price series located.

Domestic urea price series from 1970 to date was obtained from the Argentine
Secretaria de Agricultura, Ganado y Pesca (SAGyP). Any price before 1970 is
unavailable- Since the period of study is 1960 to 1997, an attempt to estimate

domestic prices for the missingdata is described below.

 

‘8 World Bank, Argentig Economic Memorandum, p. 156.
79

To estimate Argentine domestic urea prices, Free On Board (FOB) Europe urea prices
fiom International Monetary Fund (IMF) were obtained, together with dry bulk
voyage charter index from United Nations Committee on Trade and Development ’
(UNCTAD), Argentine import tax from various sources”, and Value Added Tax
(VAT) from various sources”. In a World Bank Country Study, Argentina, Economic
Memorandum (1983), an example of how onewould calculate the costs of importing
urea for June 1983 into Argentina was obtained, which is the calculation used, and is

presented below.

Table 4: Calculation of we Fertilizer Imports
Calculation of~fertilizer imports June-83
FOB, Western Europe bagged 120.0
Insurance and freight charges 26.8
CIF Buenos Aires 1463
Import tax{25% CIFrprice) 36.7
Consular, custom chargesyfreight i 20.8

and other costs --

Bagging 15.0
SubtotaI-~ 219.3
Financing costs (4.5%) 9.9
Cost to importer 229.2
Importers margin (10%) ‘ 22.9
Wholesale Price 252.1
Farmer's price (18% VAT) 297.5
US$/ton

A historical series for dry bulk fieight fi'om Europe to Argentina is not available. In
order to arrive at some approximation of what fi'eight might be, the dry bulk voyage
charter index series (an annualaverage ofdry bulk freight) was used to calculated a

dry bulk Europe to Argentina fi'eight series, using the freight for June 1983 fi'om the

 

‘9 McCarry, Michael J. and Andrew Schmitz, The World Grain Trgae: Grain Marketing, Institutions,
and Policies. 1992.

80

World Bank study. The June 1983 Europe to Argentina fi'eight rate divided by the dry
bulk voyage charter index for 1983 was multiplied to the index to arrive at a simulated
freight rate series. In order to check that my simulated fieight series was
representative of the actual fieight market, the United States to Rotterdam dry bulk
freight rates21 was plotted against the estimated freight series. See Table 5 for the
calculation of the simulated Europe to Argentina dry bulk fi'eight series and below a
graph with the US to Rotterdam versus the simulated Europe to Argentina series. It
appears that the simulated series is a good representation of the actual freight market

for the period 1960-1980.

 

{ Figure 11: Freight Rate Comparison

 

  
 

USSIMetrlc Ton
a a 8

 

 

332313 ifi'fiiiiifi

[—O—Europe-Argentina dry bulk + US-Rotterdam dry bulk]

1%?
1970
1973

 

 

 

 

 

Below is a description of the calculation of the Argentine farmer’s price for imported

urea (see Table 4 for calculation worksheet).

 

2° World Bank, Argentig From Insolveng to growth, 1992 and from World Bank, Argentina,
Economic Memorandum, 1983.
2' Series from UNCT AD, St. Lawrence to Rotterdam.

81

. The Cost Insurance and Freight (CIF) price is calculated by summing the FOB
Europe Urea price and the freight (the simulateddry bulk freight rate series was
used).

. The import tax is then calculated, using airnport tax. series compiled from various
sources as mentioned above, by multiplying the CIF price by the import tax.

. Due to lack of information for other costs associatedwith importing fertilizer, it is
assumed that the US$35.00 used in the World Bank’s calculation is a good
approximation, andis usedthroughout theseries. These other costs are then
added to arrive at a subtotal of CIF*Irnport tax+Costs.

. Financing costs are then.calculated,using theWorldBank’ s interest. rate for 1983
of 4.5% (and assumed for the remainder to the series), from the subtotal.

. An importer’s marginof 10% (assumedfor the entire series) is then calculated to
arrive at a wholesale price.

. A VAT is then computed, using aseries compiledas mentioned above, and

summed to arrive at the farmer’s price.

82

Figure 12: Estimated Urea Price
800 _-__,__.

700
600‘
500
400
360
200
160

 

US$/ton

 

 

 

 

—a—Estimated Farm er's Price
+SAGyP
—°—Estimated SAGyP's Price

 

 

 

 

 

 

Comparison of the calculated farmer’s price for imported urea to the series provided
by SAGyP demonstrates that, although the assumptions are many, the estimated urea
price series resembles SAGyP’s series, except for the period 1974 to 1979 wherethe ‘
trend is followed but the difference is much larger. The two data sets have a
correlation of 0.554. (If one were to calculate the correlation of the two data sets

omitting 1974—1979, the correlation is 0.886;)22

Given the above, data on domestic consumption for urea in Argentina was analyzed to
determine how important urea was during 1960-1970. Data was taken fi'om FAO and
Argentine Secretary of Agriculture, and although it is not complete, it appears that
urea consumption between. 1960-70 wasminimal compared to other nitrogenous

fertilizer or other fertilizers. In 1968/69, urea consumption was estimated at 10,459

 

22 Note: The Argentine government-during 1974 through 1976 set grain prices at a fixed rate to curb
inflation. Once this was abandoned, domestic grain prices sky rocketed. There is no mention that the
government fixed prices for fertilizers, but it may be assumed that some policy was implemented that

83

 

' Figure 13: Fertilizer Consum ptlon versus Planted Area 1

 

35
30 «-
25 q
20 —
15 e
lo —

KiloslHectare

 

 

 

 

 

 

E—o—Urea ——o—Nitrogenous Fertilizer-+Total Fertilizeq

 

 

metric tons, nitrogenous fertilizer at 30,661 metric tons, and total fertilizer at 67,666 ‘ -

metric tons.

To put this fertilizer consumption into perspective, aratio of urea, nitrogenous and
total fertilizer consumption versus seeded acreage for the six crops being analyzed was
derived. In 1968/69, usage of urea was 0.7 kilos/hectare, versus nitrogenous fertilizer
at 2.2 kilos/hectare and total fertilizer at 5.0 kilos/hectare. It appears that urea

consumption was minimal during 1960-1970-

Given the above mentioned analysis, there appears to be two methods of handling the

missing domestic urea price series for 1960-1970:

 

affected domestic fertilizer prices. Also notethat FOB European fertilizer prices increased drastically
in 1974. " "

84

o The first method would be to use a constant price for the missing data period,
since urea consumption appears to be minimal, using the last known price, i.e.
1970 price provided by SAGyP.

o The second method would be to use the estimated urea price series for the missing
data period as an index to calculate a series based on the 1970 price provided by
SAGyP. Urea consumption appears to be minimal, the estimated urea price series
appears to resemble SAGyP’s price series and has a high correlation when data for

1974-1979 is omitted.

The second method, using theestimated urea prices for the period 1960-1970 as an

index, was to complete the urea price series provided by SAGyP.

85

APPENDIX B

Log-log Aggregate Crop Planted Area Supply Response Model

The table below lists the results of the three price behavior expectation models in a

log-log functional form. These are:

Naive:

lnAFBo‘l'B llnGIflCt+lenCIflCt +lenA+l+BoanfeaPt+BrlnGaSPt+l~h

Moving Average:

lnA.=Bo+B1lnMAGM.+lenMACM.+BglnA..1+BranreaP.+lenGasP.+u.

Simple Weighted Average:

lnAt=Bo+BlIDSWGIDQ+leflSWCINCt+leflAH+NEUfeapt+lenGaSPt+llt

Ethic 6: Results of aggregate crop planted area using three specifications of price

 

 

expectation behavior in logg'thms, Argentina. 1960-1997

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Model Naive Moving Average Simple Weighted
Average
9 -t p t p t
Statistic Statistic Statistic
Q. -1.26 -.97 -1.78 -1.50 -1.24 -0.95
InAreaH 0.97 7.98 0.93 8.63 0.99 8.01
anreaP. 0.06 1.06 0.08 1.61 0.06 1.04
lnGasP. -0.10 -2.13 -1.11 -2.57 -0.11 -2.17
lnGine 0.11 1.82
lnCinc -0.06 -1.35
InMAGinc 0.20 3.17
lnMACinc -0. 12 " -2.67
lnSWGinc 0.09 l .49
lnSWCinc -0.04 -1.28
n" 37 37 37
Adj. R’- 0.83 0.86 0.83

86

 

In a log-log specification, the three price expectation behavior estimation results are

not markedly different than the linear specification. Adjusted R2 on all three equations
are approximately the same. The signs arethesame as in the linear specification; once
again, the urea price is positive and the constant is negative . The level of significance

for the estimated coefficientsis also. comparableto the linear specification.

87

STATISTICAL TABLES

88

Table 5: Calculation of Freight Cost Series

1955 . 173 136 21.47
1956 212 167 26.32
1957 153 120 18.99
1958 91 72 11.30
1959 97 76 12.04
1960 100 79 12.41
1961 ‘ 107 84 13.28
1962 89 70 11.05
1963 109 86 13.53
1964 . 112 88 13.90
1965 > 127 100 15.76
1966 , 114 90 14.15
1967 121 95 15.02
1968 . 124 98 15.39
1969 91 14.35
1970 * 110 17.34
1971 70 11.04
1972 75 11.82
1973 161 25.38
1974 218 34.37
1975 142 22.39
1976 - 134 21.12
1977 ~ 133 20.97
1978 “140 ' 22.07
1979 179 28.22
1980 213 33.58
1981 195 30.74
1982 159 25.07
1983 170 26.80
1984 173 27.27
1985 ‘167 26.33
1986 '158 24.91
1987 . 174 27.43
1988 195 30.74
1989 ' ~ 204 32.16
1990 198 31.21
1991 205 32.32
1992 . 195 30.74
1993 ' 196 30.90
1994 201 31.69
1995 225 35.47
1996 193 30.43
1997 196 30.90
1998

UNCTAD, Dry Cargo Tramp Voyage Charter,
except for '55-'68 wlbase 1960=100,USDlmt

89

Igble 7: Planned Area in Argenting

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Wheat Corn} Soybean ! Sunflower Scighum Flaxseedi Total
1959/60 4792 3062’ 0.9, 1250 730 12281 11063
1960/61 4275 3222 1 l 1122 937 1 1ng 10686
1961 I62 4952' 3300 10] 1351 1075 1307:, 1 1995
1962/63 4847; 3420. 21 j 983 1072 1503 11846
1963/64 6276? 3778; 14? 873 1218 1409 13568
1964/65 6497 3693? 18} 1173 1246 1172 13799
1965/66 ’ 5724, 3921 i 171 1181 1346 12941 13483
1966/67 6291 4157T 18, 1362 1454 924; 14206
1967/68 6613! 4473 23 1194 1841 711 14855
1968/69 6680} 4595 31 1354 2151 879 15690
1969/70 6239; 4666 30 1472 2568 952 15927
1970/71 4468: 4993 38 1614 3122 973 15208
1971/72 4986’ 4439 80 1533 2759 539 14335
1972/73 5627 4251 169 1652 2974 509 15183
1973/74 4252, 4134 377 1342 3114 415 13633
1974/75 5183? 3871 370 1 196 2602 520 13741
1975/76. 5753.3 3696 443 1411 2358 471 14132
1976/77 7192; 2980 710 1460 2780 722 15844
1977/78 4600 3100 1200 2200 2650 950 14700
1978/79 5230 3300 1640 1766 2540 893 15369
1979/80 5000 3310 2100 2000 1884 1070 15364
1980/81 ' 6196 ‘4000 ”1925 1390 . 2400 780 16691
1981/82 6566. " 3695 2040 1733 2712 851 17597
1982/83 74101 3440.. 2382 1930 2657 * 910 18709
1983/84 7200; 3484 2920 2131 2550 810 19095
1984/85 6000? 3620 3300 2380 2040 620 17960
1985/86 57007 3820 3340 3140 1400 750 18150
1986/87 5000 3650 3700 1891 1127 758 16126
1987/88.. 4850 . 2825 4413 . 2117 1075 671 15951
1988/89 4750 2685 4670 2313 830 574 1582
1989/90 5500 2070 5100 2800 800 600 16870
1990/91 6178 2160 4967 2372 752 590 17019
1991/92 4751 2686 5004 2724 823 431 16420
1992/93 4548i 2963 5320 2187 810 215 16043
1993/94 4910[ 2781 5817 2206 670 148 16533
1994/95 5308’ 2958 6011 3010 622 156 18065
1995/96 5088 3415 6002 3411 671 196 18781
1996/97 7367 4153 6670 3120 804 94 2207
1997/98 5919 3752 7176 3511 920 116 21394
Bolsa de Cereales 90, in thousand hectares, SAGyA for data 1970-96

90

 

1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
'1994
1995
1996
1997
1998

Wheat
124.97
150.05
142.29
152.64
185.32
139.06
143.71
142.25
131.72
137.86
128.28
132.30

152.84 '

270.75

264.24 '

211.41

61.48
126.01
156.20
154.67
194.08
190.31
149.36

92.51

80.17
82.72
73.19
102.71
128.52

69.40
100.54
109.1 3

98.71
107.96
192.22
117.28

90.95

I_able 8: Argentine Real Crop Price in US$/Metric Ton

Corn Soybean Sunflower Sorghum

122.69
148.52
145.81
175.13
157.72
178.19
153.68
127.87
118.24
126.53
128.38
125.94
146.44
251.16
242.31
169.68
45.89“
114.30
131.29
120.18
154.21
128.60
101.76
85.03
97.89
84.69
64.95
59.95
78.50
116.52
64.39
75.24
79.81
82.73
103.58
81.73
130.59
85.01
77.87

200.67
226.37
237.25
232.96
254.85
347.47
625.81
608.16
409.17

' 227.69

373.81
254.24
286.53
218.73
214.92
204.51
163.29
192.56
144.38
124.86
154.42
212.85
237.35
113.89
138.69
160.48
172.97
192.59
162.22
206.45
229.58
175.76

Source: Bolsa de Cereales and SAGyP

91

231.70
305.06
245.37
258.62
348.22
311.70
235.58
189.84
176.45
207.22
228.35
287.84
434.93
515.62
534.88
353.18
171.45
352.14
265.48
326.08
203.80
247.15
212.85
148.51
243.45
166.31
114.31
136.20
167.25

181.73

109.09
126.05
139.71

"'171.33

190.51
164.42
172.95

" 167.57

205.37

96.10
103.09
114.59
125.85
103.52
129.50
118.79
100.77

96.41

96.18

94.60
103.05
120.26
190.11
226.77
148.15

42.85

90.19

95.89

92.59
130.55
111.24

80.19

73.48

75.08

63.64

52.75

49.00

63.18

82.20

54.35

54.84

65.95

63.21

74.12

56.73
105.79

66.40

63.57

F laxseed

238.16
306.97
273.09
268.05
281 .02
249.34
238.78
200.12
232.69
244.59
215.86
183.98
331 .20
513.03
808.72
372.01
307.73
323.80
243.68
346.64
277.08
258.50
244.95
150.66
187.88
162.92
155.69
130.35
163.53
262.07
178.59
134.70
140.43
184.82
171 .98
217.29
154.68
226.68
21 1 .13

1960/61
1961/62
1962/63
1963/64
1964165
1965/66
1966/67
1967/68
1968/69
1969/70
1970/71 .
1971/72
1972/73
1973/74
1974/75 ;
1975/76
1976/77
1977/78
1978/79'
1979/80
1980/81
1981/82 -
1962/83 j
1983/84-
1984785
1985/86 -
1986/87 '
1987/88 ~
1988/89
1989/90 -
1990/91 .
1991/92
1992/93
199394
1994795 ;
1995/96
1996/97
1997/98

VNheat
1160
1295
1522
1575
1835
1321

1198'

1260

983 ’

1352

1329 ‘

‘1267
1591
1657
1410
1626

1711 ‘

1355
1729
1692

1549'

1400
2049
1837
2305
1617
1778
1879

1836"

1892
1896
2173
2320
2022
2166
1936
2241

- 2585

_T_able 9: Argentine Crop Yield

Com Soybean Smflower Sorghtm Flaxmd

1767
1894
1648
1801

1678'

2150
2466
1942
1929
2330
2442
1862
2721
2840
‘2508
2117
3278
3647
3107
2570
3801
3028
3030
3140
3563
3745
’3189
3774
2910
3461

6078

977
1163

972
1146
1035
1147
1188
1089
1124
1032
1624
1143

1732 '

1440
1363
1603
2121
2174
2313
1724

2005'

2090-
1754

1?“!5'

1988
2141
1897
2264
1653
2156
2275
2291

2158 '

2039
2045
2105
1721
2760

651
718
611
628
746
765
902
891
737

632

-890
'948
‘868
984
1184
1262
1106
1441
1346
1268
1435
1444
1451
1752
1413

1435‘

1902
1963
1718
1812
1661

1265
2159
1660
1751
1458
2524
1805

1752 "

1908
2040
2085
1663
2328
2539
2493
2758
2776
3194
3033
2314
3595
3187
3214
2911
3155
3125
3067
3347
2531
2812
3331
3621
3952
3506
3459

3876 '

3685
4800

Source: Bolsa de Cereales and SAGyP, in tons per hectare

92

3889982689

587
698
638
634
752
568
721
625

630
899
816
700
748
762

761

8

915
916

742

S

797

8998698

TOE“
1068
1321
1175
1256
1251
1413
1380
1260
1219
1402
1488
1213
1630
1676
1544
1635
1922
2014
1972

'1655
‘ 2123

1937

2037
2214
2107

2253
1853
2103

2474
2512
2417

2411
2471

1 959/60
1 960/61

1961/62.

1962/63
1963/64-
1964765
1965766
1966/67
1967/68
1968/69
1969/70
1970/.71.,
1971772
1972773
1973774
1974775
1975/76
1976/77
1977778
1978779:
1979780
1980781
1981782
1982783 ’
1983/84
1934/95,
1985786
1986787
1987/88

1988/89'“
1989/90 _, ‘
1990791

1991792
1992793
1993194
1994795
1995/96
1996/97'
1997/98

Priceei'hnuarywfie‘ " ’

 

and Real Urea Price

Real Cattle

Pfice
0.5151
0.6135
0.5389
0.5281
0.5675
0.9473
0.9726
018585
0.5940

0.5952

0.5401

0.6599 ..
1.0634 ,
1.4020

2.0820
1.6547
1.1654

0-3987. w _
95095
0.5313; .
0.9698

1.1509
0.8682

0.6550 ‘

. 0.5905
0.6224

0. _~ ...
0.4930

0.

0.5226
0.5948;
0.4305
0.5248
0.7652

0.6576

06082

0.6296

aw.-......
0.8631

Table 10: Real Agentine Cattle Price. Cattle Weight. Real Diesel Price

Cattle Real Diesel Real Urea

Weight
437
441
428
426
431
444
443
440
437
445 .
452

.413..-

E

55.

93

£588§§§§§§§§8§§8358E5855

Pfice
0.1899
0.1909
0.1703
0.1746
0.1755
0.1945
0.1958
0.1541
0.1342

. 0129? . _
0.128
04997., -
1 0.2444, x
0.3139
0.4933

0.3044
0.1200

f 0.1360 f
02482

0-2822
0.2934
0.3691
0.2288
0.1576

0.1883.

0.1790
0.2178

0.2018?
0.2129}
0.2273, 4 f
0.2014“ '

0.2820
02641

I . 02057.! x
0.2241

0.1846
0.2173

"‘0-‘23397: ..
0.0000

Pflce
472

§

470

., 175.
257

BIBLIOFRAPHY

94

BIBLIOGRAPHY

Aradhulla, Satheesh V. and Matthew T. Holt, “Risk Behavior and Rational
Expectations in the US. Broiler Market”, American Agricultural Economics
Association, Volume 71, 1989, pp. 892-901.

Askari, Hossein and John Thomas Cummings, Agricultural Supply Response: A
Survey of the Econometric Evidence. Praeger Publishers, New York, 1976.

Bolsa de Cereales, Numero Estadistico 1990. Bolsa de Cereales, Buenos Aires, 1991.

Chem, Wen S. and Richard E. Just, “Econometric Analysis of Supply Response and .
Demand for Processing Tomatoes in California”, Giannini Foundation, Number 37,
Berkeley, CA, September 1978.

Dillon, John L. and Jock Anderson, The was of Resmse in Crop and—Livestoc 'i ~

Production, Pergamon Press, New York, 1990.

Ferris, John N., Agg'cultural Prices ancLCommodity' Market Analysis,-WCB/McGraw—
Hill, Dubuque, Iowa, 1997.

Fundacion de Investigaciones EconomicosLatinn Americanos, Statistical Database,
Buenos Aires, Argentina, 1998.

Gardner, B.L., “Futures Prices in Supply Analysis”, American Journal ' ultural
’ Economics, Volume 58, 1976, pp. 81-84.

Goodwin, T. H. and SM Shefl‘rin, “Testingthe Rational Expectations Hypothesis in
an Agricultural Market”, Review of Economics and Statistics, 64, 1982, pp. 658-
667.

Gujarati, Damodar N., Basic Econometrics, McGraw-Hill, New Yorlg 1992.

Helmberger, Peter G. and J can-Paul Chavas, The Economics of Agricultural Prices.
Prentice-Hall, Upper Saddle River, N.J., 1996.

- Institute Nacional de Tecnologia Agropecuaria, Argentina: Projections of Supply of
and Demand For Selected Agriwltural Products through 1980, Keter Publishing,
Jerusalem, 1972.

Irwin, Scott H. and Cameron S- Thraen,“Rational Expectations in Agriculture? A
Review of the Issues and the Evidence”, Review of Agricultural Economics. 16,

133-158, 1994.

95

Jarvis, Lovell 5., “Supply Response in the Cattle Industry: The Argentine Case”,
Giannini Foundation, August, Berkeley, CA, 1986.

Just, Richard E., “An Investigation of the Importance of Risk in Farmers’ Decisions”,
American Journal 'cultural Economics, 56, 14-25, 1974.

Junta Nacional de Carne, Ammo Esgtdistico. Junta Nacional de Carne, Buenos
Aires, 1976.

LaFrance, Jeffrey T. and Oscar R- Burt, “Modified Partial Adjustment Model-of
Aggregate U.S. Agricultural Suppl ”,» Western Journal of Agricultural Economics.
8(1), 1983, pp. 1-12.

Lau, Lawrence J ., “Functional Forms in Econometric Model Building”, Handbook of
Econometrics, Volume III,Elsevier Science Publishers BV, 1986, pp. 1515-1566.

McCarry, Michael J. and Andrew Schmme World Grain Trade-Grain Marketing,
Institutions. and Policies, Westview Press, Boulder, CO, 1992.

Miranda, Mario J ., Frank Novak, andMelIerohl,“Acreage Response Under-Canada’s
Westerjn Grains Stabilization Program”, American Journal of Agricultural
Economics Volume 76, May 1994,. palm-2.76.

 

Mundlak, Yair and Domingo Cavallo and Roberto Domenech, “Agriculture and
Economic Grth in Argentina, 1913-84” International Food Policy Research
Institute, Research Report”76,Washington D.C., November 1989.

Nerlove, Marc, “Distributed LagsandEstimationof Long-run Supply and Demand
Elasticities: Theoretical Considerations”,~ Journal of Farm Economics, Volume XL,
May 1958, pp. 31 1.313.

Nerlove, Marc, “The Dynamics of Supply: Retrospect and Prospect”, American
Jou;n_al_ of Agg'cultura‘l Economics, Volume , December 1979, pp. 874-88.

Pablo, Juan Carlos de, Politica Economica Argentina, Editoras Macchi, Cordoba,
March 1984.

Pindyck, Robert S. and Daniel L. Rubinfeld, Econometric Models and Economic
Forecasts, Chapter 9,’McGraw-Hill, New York, 1991, pp.203-247.

Reca, Lucio 6., “Argentina: Country Case Study of Agricultural Prices and

Subsidies”, World Bank Staff Working Paper No. 386, Washington DC, April
1980.

96

Sadoulet, Elisabeth and Alain de Janvry, ngptitative Development Policy Analysis,
John Hopkins University Press, Baltimore,‘MD, 1995.

Studenmund, A.H., Using Econometrics. Addison Wesley, Massachussetts, 1997.

Sturzenegger, Adolfo, “Trade, Exchange Rate, and Agricultural Pricing Policies in
Argentina”, World Bank Comparative Studies, Washington DC, 1990.

Tomek, William G. and Kenneth]; Robinson, Agficulturfiiroduct Prices. Chapter 4,
Cornell University Press, Ithaca, NY, 1990.

Wainio, J GMT, “The Effect of Government Intervention on Grain Acreage Supply
ResponSe: The Argentine Case, 1946/47-1979/80”, Master of Science Thesis,
Texas A&M University, May 1983.

World Bank, “Argentina, Economic Memorandum”, World Bank Country Study,
Volume 1', Washington'D.C., 1983.

97

(164432-, JK 77 /1..1

 

3 1293 02428 8429 ‘