Farm size and productivity

MISSING IMAGE

Material Information

Title:
Farm size and productivity an empirical analysis of the farm size-productivity relationship in Ecuador
Physical Description:
vii, 152 leaves : ill. ; 29 cm.
Language:
English
Creator:
Langedyk, Kimberly, 1968-
Publication Date:

Subjects

Subjects / Keywords:
Farms, Size of -- Ecuador   ( lcsh )
Agricultural productivity -- Ecuador   ( lcsh )
Food and Resource Economics thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Food and Resource Economics -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 2001.
Bibliography:
Includes bibliographical references (leaves 147-151).
Statement of Responsibility:
by Kimberly Langedyk.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002726017
oclc - 47291994
notis - ANK3742
System ID:
AA00008998:00001

Full Text











FARM SIZE AND PRODUCTIVITY: AN EMPIRICAL ANALYSIS OF
THE FARM SIZE-PRODUCTIVITY RELATIONSHIP IN ECUADOR












By

KIMBERLY LANGEDYK


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2001































Copyright 2001

By

Kimberly Langedyk














ACKNOWLEDGEMENTS


I wish to express my sincere gratitude to all those who helped me through the

process of completing this dissertation. First and foremost, I would like to thank Dr.

Donna Lee and Dr. Chris Andrew, without whose advice, support, and encouragement

this dissertation would never have been completed. I wish to extend my thanks to all of

the other members of my supervisory committee as well Drs. Thomas Spreen, Peter

Hildebrand, and Anthony Oliver-Smith. Their willingness to serve on the committee and

to provide help whenever it was needed is greatly appreciated.

I wish to thank the U.S. Fulbright Commission for the fellowship which allowed

me to conduct field research in Ecuador. Dr. Susana Cabeza de Vaca, Executive Director

of the Fulbright Commission in Ecuador, deserves a special note of thanks for all of the

assistance she provided in Ecuador. All of the employees of the Fulbright Commission in

Ecuador were extremely helpful when called upon, and my thanks go out to each of them.

My thanks also go out to all of the farmers who participated in this study.

I would also like to acknowledge the support and encouragement of my parents,

Richard and Ruth Langedyk, who never doubted my ability to complete this study, even

at those times when I did. Finally, my greatest thanks go to my husband, Ramiro

Vdsquez, who has taught me to see the beauty in every day we live.















TABLE OF CONTENTS


page

ACKNOWLEDGMENTS ..................................................................................................vi

ABSTRACT........................................................................................... .... ............vi

CHAPTERS

1 INTRODU CTION ................................................................................................. 1

2 THE QUESTION OF LAND REFORM IN LATIN AMERICA .................6....

Land R eform .......................................................................................................6*** *
The History of Land Reform in Latin America ................................................... 13
Land Reform in Ecuador.................................................................................. 16
Summary and Conclusion......................................................................... .....25

3 THE FARM SIZE-PRODUCTIVITY RELATIONSHIP ..................26

The Existence of an Inverse Farm Size-Productivity Relationship .........................26
Definitions of Productivity and Efficiency..........................................................30
Productivity.............. .......................................... ........................................30
Partial Productivity Measures.....................................................................32
Efficiency.......................................................................... ....................... 33
The Relationship Between Productivity and Efficiency........... ..............................35
Measuring Productivity ...... ................................................................ ............... 37
The Role of Land Quality .......................................................................................38
Differences in the Environmental Impact of Small and Large Farms .................39
Possible Explanations for the Inverse Farm Size-
Productivity Relationship.......................................................................... ........42
Testing for the Inverse Relationship and Its Causes in Ecuador ............................45

4 STUDY AREA AND METHODOLOGY..............................................................49

Ecuador .......................... ................................................. .. ................... .............49
Geography...................................................................................... .............49
Socioeconomic Characteristics ...................................................... .... 52
A agriculture ................................................................ ..................................... 54









B olivar .....................................................................................................................55
Survey Methodology............................................................................................58

5 THE SAMPLE STUDY GROUP ........................................................................61

Agricultural Production on the Sample Farms.....................................................61
Forms of Tenancy................................................................................................68
Socioeconomic Characteristics of Farm Operators' Families .............................70
Credit, Technical Assistance and Agrarian Reform ............................................71
Interviews with Large Farm Operators ................................................................74

6 ANALYSIS OF THE FARM SIZE-PRODUCTIVITY
RELATIONSHIP IN BOLIVAR...................................................................79

Equations and Variables ......................................................................................79
The Farm Size-Productivity Relationship............................................................83
Family Labor Versus Hired Labor.......................................................................97
Differences in Pesticide Use Between Small and Large Farms......................... 100

7 SUMMARY, CONCLUSIONS AND RECOMMENDATIONS...................... 103

Conclusions........................................................................................................ 103
Strengths and Shortcomings of this Study......................................................... 107
Recommendations for Further Research............................................................ 109

APPENDIX TSP PROGRAM USED FOR DATA ANALYSIS ............................110

REFERENCES .........................................................................................................147

BIOGRAPHICAL SKETCH ................................................................................... 152














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

FARM SIZE AND PRODUCTIVITY: AN EMPIRICAL ANALYSIS OF
THE FARM SIZE-PRODUCTIVITY RELATIONSHIP IN ECUADOR

By

Kimberly Langedyk

May, 2001


Chair: Dr. Donna Lee
Major Department: Food and Resource Economics

This study examines the relationship between farm size and productivity in the

northern highlands of Ecuador using data collected from a sample of forty-three farms in

the region. The data were collected between October, 1998 and August, 1999.

The relationships between farm size and land productivity and between farm size

and total factor productivity are determined. This study concludes that a significant

inverse relationship exists between farm size and land productivity, but that the

relationship between farm size and total factor productivity in the region is significant

and positive. This throws into doubt the argument that redistributing land from large to

small farms would increase agricultural productivity in developing countries.

Furthermore, this study finds that when land quality indicators and variables on farmer

age and education are included in the analysis of the farm size-land productivity

relationship, the inverse relationship is no longer significant.









When land productivity and total factor productivity of crop production alone are

examined, a significant inverse relationship is found to exist between farm size and land

productivity, while no significant relationship is found to exist between farm size and

total factor productivity. It is concluded that smaller farms have a higher land

productivity than larger farms because smaller farms use more inputs per unit of land.

Smaller farms are also found to use more pesticides per unit of land than larger farms, but

pesticide expenditures as a percentage of total non-labor input expenditures are not found

to vary greatly by size of farm.

The results of this study indicate the need for a reexamination of the farm size-

productivity relationship in developing countries. Future studies should examine both the

farm size-land productivity relationship and the farm size-total factor productivity

relationship, and should take into account differences in land quality and farmer age and

education between farms of different sizes.














CHAPTER 1
INTRODUCTION


A great amount of research has been done on the relationship between farm size

and farm productivity in developing countries. The vast majority of this research

supports the existence of an inverse farm size-productivity relationship, i.e., the smaller

the farm the more productive it is per unit of land. Farm productivity is typically

measured as yields per land unit or gross output value per land unit. Numerous studies of

agriculture in developing countries have noted the existence of this inverse relationship.

The relationship was first observed by Chayanov in his work on Russian agriculture in

the 1920s (Chayanov, 1966), and has since been observed in many other developing

countries, from Brazil and Colombia to India and Malaysia. The inverse farm size-

productivity relationship has been observed so often that various authors refer to its

existence as a "stylized fact" of agriculture in developing countries (Bardhan, 1973;

Bhalla and Roy, 1988).

A variety of causes have been proposed to explain this inverse farm size-

productivity relationship. These include imperfections in land, labor and capital markets,

labor dualism, the effects of uncertainty on farmer decisions, and land quality differences

among farms of different sizes. There are far fewer studies which test the validity of the

proposed explanations of the inverse farm-size productivity relationship than there are

studies which document the existence of the relationship.









The existence of an inverse relationship between farm size and productivity could

provide an economic justification for land reform. Land reform in the sense in which it is

used in this paper refers to the redistribution of land from large to small farmers. Land

reform is typically promoted for reasons of equity and social justice; it is expected to

bring about a more equitable distribution of wealth in the countryside and to improve the

situation of the rural poor. It could also be argued that land reform represents a means of

improving agricultural productivity in developing countries. If productivity declines as

farm size increases, then redistributing land to create smaller farms could lead to

increased agricultural productivity. Whether or not land redistribution of this type will

actually be productivity-improving depends upon the causes of the inverse farm size-

productivity relationship and upon the exact nature of the relationship. In many studies,

what is referred to as an inverse farm size-productivity relationship is actually an inverse

relationship between farm size and land productivity. If this inverse relationship is to be

used to promote land reform as a means of improving agricultural productivity, then it

must exist between farm size and total factor productivity, not just farm size and land

productivity.

This study seeks to examine the relationship between farm size and productivity

in the South American country of Ecuador. Land distribution in Ecuador is extremely

skewed. In the Ecuadorian highlands, the 1.6 percent of farms which are larger than one-

hundred hectares comprise 42.9 percent of total farmland, while 80.2 percent of all farms

are smaller than five hectares and comprise only 14.1 percent of total farmland (World

Bank, 1996). Ecuador has enacted a series of land reform laws since the early 1960s, but

the overall redistributive effect of those laws has been minimal.









A recent World Bank report on poverty in Ecuador (1996) found that small farms

in that country have higher yields per unit of land than large farms. The authors of the

report suggested that the Ecuadorian government should encourage the transfer of land

from large to small farms because this would lead to increased agricultural production in

the country.

The World Bank study raises several questions about the relationship between farm

size and farm productivity in Ecuador. For example, why do small farms attain higher

yields per unit of land than large farms? Do they have a more intensive input use per unit of

land? Could it be due to land quality differences between small and large farms? The likely

impact of land redistribution on agricultural production and on farmer welfare depends on

the answers to these questions. Furthermore, are small farms more efficient than large

farms? Higher physical yields per land or labor unit do not necessarily imply greater

efficiency, since yields are only a partial measure of productivity, and do not take into

account multiple output prices, differences in crop mix, or differences in input use. Finally,

do small and large farms differ in other important ways, such as in the impact of their

farming practices on the natural environment? If they do, then the transfer of land from

large to small farms may affect environmental quality as well as agricultural production.

This study seeks to answer the above questions. The three main objectives of this

study are

1. to determine the cause of the inverse relationship between farm size and yields

per hectare in Ecuador,

2. to determine whether an inverse relationship exists between farm size and total

factor productivity (as opposed to simply physical yields), and









3. to determine whether the use of pesticides in agricultural production varies by

size of farm.

In order to answer these questions, a group of sixty farm operators in the Parish of

Bolivar in northern Ecuador was studied over a period often months. Twelve farms from

each of five different size groups were selected for the study. The operators of each of the

sixty sample farms were interviewed approximately every eight weeks from October, 1998

through August, 1999. Data on production for each farm were gathered, as well as

socioeconomic data on the family of each farm operator. Seventeen of the sixty farms were

eliminated during data collection due to problems collecting reliable data from the farm

operators, thereby reducing the sample size to forty-three farms. Data from these forty-three

farms are used in this study to determine the type of relationship between farm size and

productivity in the region, as well as to explore the possible causes of this relationship.

In addition, interviews were conducted with twenty large landowners in the Province

of Carchi in northern Ecuador. These landowners all owned farms often hectares or more.

The interviews were conducted for the purpose of determining the owner's relationship with

their farms and the productive use of those farms.

Chapter 2 of this paper addresses the issue of land reform in Latin America in

general and in Ecuador in particular. The definition of land reform is discussed, and a very

brief history of land reform in Latin America is presented. A more detailed history of land

reform in Ecuador and its impact on land distribution in that country follows.

Chapter 3 provides theoretical and empirical background on the inverse farm size-

productivity relationship. Studies documenting the inverse farm size-productivity

relationship are cited, and the possible explanations for such a relationship are discussed.









The equations which will be used to test for an inverse farm size-productivity relationship

using the sample data are laid out in this chapter.

Chapter 4 provides background information on Ecuador and the study site of

Bolivar. Geographic, agricultural and socioeconomic background is given. Chapter 4 also

gives more information on the data collection method used for this study.

Chapter 5 presents information on the sample farms, including information on the

types of crops produced, prices received for crops, forms of tenancy and socioeconomic

characteristics of the farm operators' families. Summary statistics from the sample group

are presented in this chapter, and the results of the interviews with the twenty large

landowners are discussed.

Chapter 6 presents the analysis of the farm size-productivity relationship in the study

region. The results of several regressions to determine the relationship between farm size

and land productivity and between farm size and total factor productivity are presented. An

analysis of the cause of the relationship is given, as well as an analysis of the relationship

between farm size and pesticide use. Chapter 7 provides a summary of the information

presented in the paper and gives recommendations for future studies.















CHAPTER 2
THE QUESTION OF LAND REFORM IN LATIN AMERICA


Land reform has been a subject of great debate in Latin America for the past fifty

years. Much of the debate has been highly emotional, as land reform is often carried out

for political or social reasons. Many countries in Latin America have undertaken land

reforms, with varying degrees of success. In recent years some countries, such as

Mexico, have passed laws intended to roll back previous reforms.

There is substantial disagreement regarding the value of land reform, due in part

to the multiple objectives of reform. The term itself has many different definitions, and

different writers and politicians often mean different things when they speak of land

reform. While the primary goal of a particular land reform may not be economic, land

reform always has economic consequences, and a great deal of debate also exists about

the economic consequences of land reform.

This chapter examines the theory behind land reform, particularly the economic

justification for land reform. A brief history of land reform in Latin America is

presented, along with a more detailed history of land reform in Ecuador.


Land Reform

Land reform has been defined in many different ways. Some authors use the term

interchangeably with agrarian reform, while others distinguish between the two terms.

The World Bank (1975, p. 5) defines land reform as "intervention in the







7

prevailing pattern of land-ownership, control and usage in order to change the structure of

holdings, improve land productivity and broaden the distribution of benefits." According

to the World Bank, land reform can include redistribution of public or private land,

consolidation of individual holdings to eliminate fragmentation, changes in land

ownership and tenurial rights, and changes in conditions of tenure. The World Bank

distinguishes land reform from agrarian reform, which it regards as a much more

comprehensive concept. According to the World Bank, agrarian reform involves

modification of a wide range of conditions that affect the agricultural sector, such as price

policies, agricultural extension and research, provision of irrigation and other

infrastructure improvements, access to credit, and marketing assistance. Agrarian reform

may or may not include land reform, which is to say that it may or may not involve

changes in the prevailing pattern of land ownership.

Lipton (1974) would find the World Bank's definition problematic. He describes

three errors which are commonly made in defining land reform: first, the definition may

be insufficient and allow extraneous elements to enter, second, it may be more than

sufficient and exclude elements that should be allowed to enter, and third, it can be "plain

wrong" and specify characteristics that do not really belong to land reform. Lipton would

likely classify the World Bank's definition as insufficient, in that it includes changes in

patterns of land ownership and conditions of tenure that he does not believe constitute

land reform. Lipton proposes his own definition of land reform. According to this

definition, land reform consists of "(1) compulsory take-over of land, usually (a) by the

State, (b) from the biggest landowners, and (c) with partial compensation; and (2) the

farming of that land in such a way as to spread the benefits of the man-land relationship









more widely than before the take-over" (p. 270). Lipton's definition is far more

restrictive than that of the World Bank. It also makes no distinction between land reform

and agrarian reform; Lipton uses the two terms synonymously. Depending upon one's

point of view, Lipton's definition may be guilty of the error of being more-than-

sufficient, in that it excludes elements that many would include within land reform.

Lipton states that errors of definition often are due to a desire to "define" as land reform

only those agrarian changes that seem to the definer to be desirable; his definition could

be considered erroneous for this very reason.

Thiesenhusen (1989b) also fails to make a distinction between land reform and

agrarian reform. He notes that in Spanish there are no separate terms for land and

agrarian reform; the entire concept is covered by the term reform agraria. He defines

land reform as "a fundamental reordering of a land-tenure pattern that... occurs

sometimes by revolution, sometimes by alliances between the peasants and the middle

class which pressure the government, sometimes by less straightforward coalitions

involving the Church and the military, and sometimes even through technological

change" (p. 12). According to Thiesenhusen, the general idea of land reform is not only

that land-tenure patterns should change, but that the rural poor should benefit in the

process.

Like the World Bank, Koo (1982) distinguishes between the two main areas of

reform in the agricultural sector, although he uses a different terminology. Koo groups

reform under two broad headings: land tenure reform and land operation reform. Land

tenure reform is designed to eliminate land market distortions, and includes "removing

barriers to free access to factor input and product output markets, information, and









government services and policies directed at the agricultural sector in general and at

small and tenant farmers in particular" (p. 5). Land operation reform involves changes in

the distribution of land ownership, and can refer to the breaking up of larger estates and

the redistribution of land to landless tenants or to the consolidation of smaller peasant

holdings into larger collective or cooperative farms. Koo's distinction between land

tenure reform and land operation reform is fairly analogous to the World Bank's

distinction between agrarian reform and land reform. Koo also states that land tenure

reform can be independent of land operation reform.

As is evident from the above discussion, there is no one universally accepted

definition of land reform. It is not the intent of this author to provide one. Any definition

proposed would likely be embraced by some while scorned by others. Such is the

emotional nature of the debate over land reform. However, it is necessary for the author

to specify what is meant by the term "land reform" when it appears in this study. Since

this study is concerned with an economic justification of land reform based on the

relationship between farm size and productivity, the author is concerned only with land

reforms which change the distribution of land ownership by breaking up larger farms to

create smaller ones. When the term "land reform" is used here, it will refer only to this

type of reform. This is not to imply that other changes in the agrarian structure do not

constitute land reform; they simply do not constitute the type of land reform which this

paper seeks to evaluate as potentially productivity-improving.

Land reform is advocated for a variety of reasons. It has been recommended as a

means of reducing poverty, bringing about a more equitable distribution of income,

raising agricultural productivity, and preserving political stability. Obviously, the goals










and implications of land reform vary enormously in different contexts, and the success of

any particular land reform must be judged in light of that particular reform's goals.

Typically land reform is promoted for reasons of equity and social justice; land reform is

expected to bring about a more equitable distribution of wealth in the countryside and to

improve the situation of the rural poor. Arguments advocating reform as a means of

improving agricultural productivity are much less common. In fact, it is often argued that

agricultural productivity will fall with land reform. Nicholls (1971, p. 35), discussing the

potential impact of land reform on productivity in Brazil, states that "In Brazil at least,

land reform ranks very low in the priorities for sound public policy and. indeed it

would probably do more harm than good."

However, a large body of research indicates that smaller farms may actually be

more productive, per unit of land, than larger farms. This research is summarized in

Chapter Three of this study. If smaller farms are more productive than larger farms, then

land reform could be expected to raise agricultural productivity. This would provide an

economic justification of land reform based on productivity reasons, in addition to the

equity justification.

If smaller farms are more efficient than larger farms, the question arises as to why

land rental and sales markets have not reduced the operational size of farms naturally,

without government intervention. One would expect the less efficient larger farms to sell

or rent part of their land to small farmers. However, high land prices may keep this from

happening.







11

Binswanger et al. (1995) discuss several factors that may cause the price of land

to be greater than the present discounted value of the agricultural income stream

produced from that land. If the price of land is greater than the present discounted value

of the land's agricultural profits, then it becomes very difficult for poor farmers to

purchase land. If the percentage increase in the net present value of the land when

managed as a small farm is less than the prevailing loan rate, then small farmers without

capital do not stand to gain by borrowing money to purchase land. Part of the value of

family labor must go to the loan, thus reducing the income stream available for

consumption. This reduction in consumption could only be avoided if the productivity

advantage of small farms over larger farms is great enough to offset the high price of

land, or if the buyer is able to purchase the land using savings rather than taking out a

loan. Otherwise, the poor would have to reduce their consumption below the level of

their potential earnings in the labor market in order to buy land.

The first factor behind high land prices which Binswanger et al. discuss is the

value of land as collateral. Because land is immobile, it is often a preferred form of

collateral in credit markets. This confers additional utility from landownership, beyond

the utility derived from the land's agricultural income stream. This additional utility is

incorporated into the land's price. If the buyer must mortgage the land in order to

purchase it, then the land can no longer be used as collateral for working capital, and the

buyer does not receive the credit advantage associated with owning the land. However,

he still pays for this advantage, as it is incorporated into the land's sale price. Purchasing

land through mortgage would therefore be unprofitable, and land sales will likely be

limited to buyers with savings that can be used for the land purchase.









A second factor which raises land prices above the capitalized present value of

agricultural profits is that land is sometimes used as an asset to hedge against inflation in

periods of macroeconomic instability. An inflation premium is thereby incorporated into

the real land price. This will not affect agricultural land prices if expected inflation is

fully reflected in interest rates, but if it is not then excess demand for land will increase

the price of land as a speculative asset.

Agricultural credit subsidies are also capitalized into land values. A study by

Brandao and de Rezende (1992) found that six percent of the land prices in Brazil can be

attributed to credit subsidies. This should not be an obstacle to small farmers in

purchasing land, as they would benefit from the credit subsidies once the land belongs to

them. This will only hold, however, if the credit subsidies are available for small farms

as well as large farms. In Latin America, large landowners often have easier access to

credit subsidies than small farmers (Thiesenhusen, 1989b).

Finally, agricultural land located near urban areas often has a higher price due to

growing urban populations. The price of land near urban areas is expected to appreciate

as urban demand for land increases, and some of this appreciation is capitalized into the

current land price.

If the land sales market is unlikely to bring about a redistribution of land in favor

of smaller farms, yet smaller farms are more efficient, then the rental market could be

expected to reduce the operational size of holdings through small farmers renting portions

of land from large landowners. However, legal restrictions on tenancy and the fear of

expropriation of land if it is rented out for long periods of time can make this option

unfeasible (Binswanger et al., 1995). Sharecropping arrangements are another way of







13

reducing operational farm size, and are very common in developing countries. However,

there is much debate about the efficiency of sharecropping.


The History of Land Reform in Latin America

The pattern of land ownership in Latin America since colonial times has been

dominated by the latifundio-minifundio system. Large farms (called haciendas or

latifundios) make up the majority of the agricultural land, with tiny farms (minifundios)

scattered among them. The minifundios are of a size too small to provide sustenance for

the families which farm them, so the residents of the minifundios must either work for

wages on the neighboring haciendas or migrate to the cities in search of work.

For centuries the latifundio-minifundio system has been the basis not only of

economic relations in rural Latin America, but of social relations as well. The owners of

the latifundios have long had a patron-client relationship with the minifundistas. The

latifundio owner provides employment and grants small favors to the minifundistas, who

in turn give their labor power and service. Historically, the obedience of the minifundista

to the hacienda owner was guaranteed through a variety of institutions, ranging from

outright slavery to debt peonage. These institutions were designed to guarantee the

provision of a stable, sufficient workforce for the hacienda. In colonial times, one

common practice was the granting of encomiendas, or trusts over the Indians who resided

on hacienda land. The landowner who was granted an encomienda was obligated to

watch over the welfare of his Indians and to teach them Christianity. In return, he had the

right to force his charges to work on his lands and to collect tribute from them. Over

time, the Indians came to be tied to the land. When properties were sold, both land and

Indians were sold together. This practice continued into the mid-twentieth century in










some Latin American countries (Thiesenhusen, 1989b). In others, the hacienda owner

ensured a sufficient year-round workforce by granting laborers the right to farm a small

plot for themselves in exchange for working on the main fields. In order to continue to

cultivate their plots, the laborers had to continue working for the hacienda; if they

stopped working for the hacienda, they lost their plots.

With modernization and the increasing mechanization of large farms, the need for

a year-round resident labor force to work the haciendas diminished in the latter half of the

twentieth century. The technological advances of the green revolution also reduced the

need for a large year-round labor force. Many of the Latin American land reforms of the

1960s and 1970s were intended to eliminate pre-capitalist labor relations in the

countryside, a process that was occurring naturally as the need for a permanent resident

labor force declined. Most of the labor needed by haciendas today comes from an off-

farm workforce. The workers are often minifundistas whose plots are too small to

maintain their families, so they must accept wage work on the haciendas.

The latifundio-minifundio system has been condemned as highly inegalitarian; a

small group of elites control most of the land while the poor majority own plots which are

too small to support a family. The small farmers are forced to seek wage labor to meet

the needs which their small plots cannot provide, and since wage labor is in high demand

only during planting and harvesting times, these small landowners must often migrate

seasonally. Furthermore, the existence of large numbers of underemployed peasants in

the countryside causes labor to be cheap and wages to be low. The result is widespread

rural poverty (Thiesenhusen, 1989b).









Land reform has been advocated throughout Latin America as a means of

reducing rural poverty and improving the income and resource distribution in rural areas.

Land reforms have been undertaken in the majority of Latin American countries during

the past century. These reforms have had varying impacts on rural land distribution and

on agricultural production.

The most recent wave of Latin American land reforms began in the early 1960s

and was inspired by the 1961 declaration of the Inter-American Economic and Social

Council of the Organization of American States that established the Alliance for Progress

(Dorner, 1992). The Alliance for Progress agreement was signed by nineteen Western

Hemisphere countries. It required Latin American countries to undertake land reform in

order to be eligible for foreign aid. Much of the motivation behind the Alliance for

Progress was political; the U.S. feared that further revolutions like the one in Cuba would

occur if something was not done to pacify the peasantry in Latin America. Land reform

was seen as one means of keeping the peasantry content.

The land area and number of farming families affected by land reform for a

selected group of Latin American countries are revealing (Tables 2.1 and 2.2). For

several of these countries, both the area of agricultural land affected and the number of

peasant families benefited by reform are quite small. Only a few countries (Bolivia and

Mexico, in particular) have undertaken reforms that have substantially changed the

distribution of agricultural land. The impact of reform on land distribution in the

majority of countries has been minimal. This explains why many social scientists who

have studied land reform in the region are critical of the reforms undertaken to date and

of their impact on those they are supposed to benefit (See for example, Grindle, 1986, De







16

Janvry, 1981, and Bret6n, 1997). In the words of Thiesenhusen (1989a, p. 488), "reform

programs to date in the region have been too small, too late, too underfunded, too dictated

from above, too hierarchically organized, and too infrequently responsive to pressure

from the grass roots."


TABLE 2.1
Surface area affected by land reform in selected Latin American countries

Forest and agricultural surface
(thousands of hectares)

Percentage
Country Total Affected Affected
Bolivia 3,275 2,730 83
Chile 28.759 2,940 10
Costa Rica 3,122 222 7
Dominican Republic 2,676 375 14
Ecuador 7,949 718 9
Mexico 139,868 60,724 43
Panama 2,254 493 22
Peru 23,545 9,256 39
Venezuela 26,470 5,119 19

Source: Thiesenhusen (1989b)


Land Reform in Ecuador

The first modem Ecuadorian land reform law was promulgated in 1964. Prior to

this law, land tenure relations in the Ecuadorian Sierra were dominated by a form of

service tenancy known as huasipungaje. Under this arrangement, peasants received

usufruct rights to a small plot of hacienda land, the huasipungo, in exchange for working

for the hacienda. The huasipungo plots were usually two to five hectares in size, and the

peasant families lived on their lots and cultivated subsistence products on them. The







17

male head of the household (the huasipunguero) typically spent five or six days per week

working for the hacienda (Haney and Haney 1989). In some cases the huasipungueros

were paid for their work on the hacienda; when this was the case they usually received

less-than-subsistence wages, since they were expected to complement these wages with

subsistence production on their parcels of land (Barsky et al. 1982). In other cases

huasipungeros received no wages for their work on the hacienda; usufruct rights to a

small parcel of land were their sole compensation. In addition, a form of debt peonage

existed, with landlords giving laborers credit against future wages and making them

financially responsible for any livestock that got lost or died while in their charge. These

debts were hereditary and served to bind the laborer and his family to the hacienda.


TABLE 2.2
Number of peasant families benefited by land reform in selected Latin American countries

Number of farming families

Percentage
Country Total Benefited Affected
Bolivia 516,200 384,560 74.5
Chile 412,000 38,000 9.2
Costa Rica 155,200 8,349 5.4
Dominican Republic 697,800 59,411 8.5
Ecuador 749,000 78,088 10.4
Mexico 4,629,400 1,986,000 42.9
Panama 132,800 17,703 13.3
Peru 1,419,400 431,982 30.4
Venezuela 561,800 171,861 30.6

Source: Thiesenhusen (1989b)


The primary advantage of the huasipungaje system, from the viewpoint of the

hacienda owners, was that it guaranteed a stable, year-round work force for production on









the haciendas. It also guaranteed a work force which could be paid very low wages, or

none at all. However, it created enormous social tensions in the countryside. By the

1950s, peasants were demanding increased land rights. Some landowners responded to

this social pressure by voluntarily giving their huasipungueros ownership of small plots

of hacienda land and releasing them from their obligations to the hacienda. These

landowners began introducing technological innovations in agriculture (mechanization,

use of genetically improved varieties, etc.) on their land. Modernization reduced the

need for a year-round resident workforce, so that landowners also found it in their own

best interests to relieve themselves of their huasipungueros. These modernizing

landowners played an important role, together with the peasantry, in supporting a land

reform initiative.

Nevertheless, the opposition of important sectors of Ecuadorian society, including

the more traditional landowners, to any land redistribution led to a military coup in 1963.

One of the primary objectives of the military government which came to power was the

elimination of "precarious" forms of land tenancy, such as huasipungaje. The military

government viewed such forms of tenancy as socially unstable and as an impediment to

the development of a more efficient, modern capitalist agricultural sector.

The Agrarian Reform Law decreed July 11, 1964, by the military junta was

intended to relieve social tensions and increase agricultural productivity by transferring

inefficiently-used hacienda land to the peasants. The law formally abolished the

huasipungaje system and established a maximum landholding size. It provided that every

huasipunguero who had worked his plot for ten years or more would receive title to it

with no recompense due the original landowner. Those who had worked their plots for









fewer than ten years were also entitled to receive ownership, but had to make partial

payments to the landowner based on the number of years they had worked their plots

(Blankenstein and Zuvekas, Jr., 1973). The law also called for colonization of the tropical

lowland regions. The Ecuadorian Agrarian Reform and Colonization Institute (IERAC)

was established to administer the law.

Officially the 1964 law was to be applied throughout the country, but in reality it

was implemented only in the Sierra. Between 1964 and 1971, approximately 17,000

plots were awarded to huasipungueros in the Sierra. These plots averaged approximately

3.5 hectares and were usually of lower quality than the plots the huasipungueros had

worked prior to the reform (Zevallos, 1989). Land redistribution was restricted primarily

to haciendas belonging to the government's social security agency (Asistencia Social)

and to haciendas owned by the Catholic Church. By 1970, the Agrarian Reform and

Colonization Institute had expropriated only fourteen privately-owned haciendas and sold

land to peasants on an additional thirty-six haciendas (Blankenstein and Zuvekas, 1973).

According to Zevallos (1989), the most important impact of the 1964 reform was

its contribution to the spread of impersonal wage relations in highland agriculture, a

process which had already been initiated before the reform. Land redistribution was

minimal, since it was confined to huasipungueros, who made up less than eight percent of

the Sierra's rural population (Forster, 1989). The Ecuadorian Agrarian Reform and

Colonization Institute emphasized colonization over reform; over seventy-five percent of

the land distributed by the Institute to smallholders was land for colonization.

In 1970 a second land reform was instituted by special decree (Decreto 1001).

This decree eliminated "precarious tenure" relations in the rice-producing areas of the







20

coastal lowlands. Under these tenure relations, precaristas were allowed to cultivate rice

on hacienda land in exchange for sharing their harvest with the landowner. Usually,

precaristas were assigned seven hectares of land and had to give the landowner from

three to six hundredweight of milled rice for every hectare (Zevallos, 1989). Pressure for

land reform by precaristas demanding land of their own grew throughout the 1960s.

Decreto 1001 provided that all rice land cultivated under precarismo was subject to

appropriation by the state and redistribution to the precaristas. Precaristas were not given

individual title to the land; title was given instead to cooperatives of precaristas.

Like the 1964 reform, Decreto 1001 did not lead to a significant redistribution of

land (Zevallos, 1989). Precaristas who were already farming land simply obtained title to

it through cooperatives. The Decree did contribute to the elimination of pre-capitalist

labor relations in the rice-producing areas of the Coast and to the spread of wage relations

in coastal agriculture.

A third Agrarian Reform Law was promulgated in 1973. Unlike the 1964 law,

this new legislation did not establish a maximum farm size; instead it targeted for

redistribution farms which were not being "efficiently" operated. Article 25 of the law

stated that a farm was considered inefficient if it failed to meet any one of the following

three conditions: 1) by January 1, 1976 at least eighty percent of the farm's agricultural

land had to be efficiently utilized, 2) the farm had to meet production levels at least equal

to those fixed by the Ministry of Agriculture for the area where it was located, and 3) the

farm had to be equipped with the physical infrastructure necessary for its utilization.

Farms in areas of high "demographic pressure" and farms which violated labor laws were

also subject to expropriation, as were farms not administered directly by their owners or







21

relatives of their owners (or administrators if the farm was owned by a company or other

legal entity) (Bret6n, 1997).

Large landowners were strongly opposed to article 25 of the 1973 law and

campaigned against it. Opposition to the article was strong enough (and the

government's support of it weak enough) that it was never applied. Expropriations of

land on the basis of other provisions stipulated in the 1973 law were few, and generally

occurred after peasants had invaded land that had been abandoned by the owner

(Zevallos, 1989).

Land redistribution which occurred as a result of the 1964 to 1973 land reform

laws was minimal (see Table 2.3). From 1964 to 1985, about eight percent of all

agricultural land in the country (744,400 hectares) was adjudicated through land reform.

Thirty percent of adjudicated land involved peasants acquiring legal rights to land they

were already working, and should not be considered redistributed land. Subtracting this

thirty percent from adjudicated land provides an estimated amount of land redistributed

of 520,000 hectares, or 5.4 percent of total agricultural land (Zevallos, 1989).

A much greater impact of the 1964 to 1973 land reform laws occurred in the area

of colonization. Large landowners supported colonization of new lands as a means of

relieving population pressure without redistributing land. While the state never formally

accepted colonization as a substitute for redistribution, most of the Ecuadorian Land

Reform and Colonization Institute's work took place in the area of colonization. The

total area of colonized land from 1964 to 1985, and the percentage colonized land

composed of the reform sector (land awarded through redistribution or colonization) is









given in Table 2.4. Over seventy-seven percent of all land in the reform sector

(2,580,100 hectares) is land awarded through colonization.


TABLE 2.3
Land adjudicated on the basis of the 1964 1973 reform laws

Percentage of total
Period Hectares 1985 agricultural land
1964-1971 174,000 1.8
1972-1979 349,800 3.6
1980 1985 220,600 2.3
1964 -1985 744,400 7.7

Source: Zevallos (1989)


While the direct impact of the agrarian reform laws on land redistribution was

minimal, the laws did indirectly lead to a change in the distribution of land. The threat of

expropriation caused many landowners to divide and sell their farms. The oil boom of

the 1970s helped make urban investments in construction, import-substitution, and real-

estate businesses more attractive, and many large landowners, worried about the

possibility of expropriation, sold their lands and invested the proceeds in urban areas.

Land distribution in Ecuador did become more equitable from the 1950s to the 1980s,

although much of this change must be attributed to the impact of colonization (see Table

2.5).

The land reform laws also brought about a change in the productive structure of

agriculture in the highlands. Farmers who were concerned that noncompliance with labor

laws could lead to expropriation of their lands reduced their labor needs by increasing

mechanization and by transforming labor-intensive crop areas into pastures (Zevallos,

1989). Landowners who faced the possibility of having their land expropriated because







23

less than 80 percent of it was being cultivated converted uncultivated land into pastures.

The result was a shift away from traditional crops and towards livestock and dairy

production. Increased demand for animal protein due to rising incomes among the urban

middle class during the oil boom helped to encourage this shift.


TABLE 2.4
Land awarded through colonization, 1964 1985

Hectares Total hectares in Colonized land as
Period colonized the reform sector percentage of reform sector
1964-1971 518,100 692,100 74.9
1972 1979 1,012,900 1,362,700 74.3
1980 1985 1,049,100 1,269,700 82.6
1964 1985 2,580,100 3,324,500 77.6

Source: Zevallos (1989)


TABLE 2.5
Changes in the distribution of land, 1954 1984

Percentage of agricultural land

Farm size 1954 1974 1984
< 20 hectares 16.6 18.6 35.6
20 100 hectares 19.0 33.5 30.0
> 100 hectares 64.4 47.9 34.4
Total 100.0 100.0 100.0

Source: Breton (1997)


In 1979, a legislative shift in the character of the land reform laws occurred with

the Farming Development and Promotion Law (Ley de Fomento y Desarrollo

Agropecuario) approved by the military junta March 6, 1979. This law greatly modified

the efficiency criteria established under the 1973 law, eliminating the requirement that









eighty percent of a farm's agricultural land must be utilized in order for the farm to be

considered efficient. This was followed by the National Development Plan of 1980 -

1984, which made land reform a minor component of agrarian reform, concentrating the

state's efforts on integrated rural development projects which were not redistributive in

nature. Land reform became limited in practice to the granting of legal titles to land

already in the possession of peasants.

The 1994 Agrarian Development Law further weakened the legal basis for land

reform in Ecuador. This law eliminated demographic pressure as a cause for

expropriation of farms and left remaining as causes for expropriation only the continued

existence of "precarious land tenure relations," the use of technologies which gravely

threaten the conservation of natural resources, and the idleness of productive lands for

more than two years. Such was the level of discontent with the new law within the

peasantry and among the indigenous population that these two sectors united in an

uprising which forced the government to revise the law. Demographic pressure was re-

inserted as a justification for land expropriation, but with the condition that the farm in

question not be completing the "social function" which the state assigns it (essentially to

produce and generate surpluses, Bret6n, 1997).

While the 1994 Agrarian Development Law was revised due to pressure from the

peasantry and the indigenous population, the basic thrust of the law remained the same.

It continued the shift away from land redistribution that began in 1979. This shift was

due in large part to pressure from the large landowners, who campaigned fiercely against

all laws supporting land distribution.









Summary and Conclusion

The overall redistributive effect of land reform laws in Ecuador remains minimal.

While the distribution of land is more equitable than it was fifty years ago (Table 5), it is

still quite skewed, and further redistribution of land would undoubtedly lead to a more

equitable distribution of resources in the countryside. Those who support land reform for

equity reasons must now answer the question of how such reform would affect,

agricultural productivity. While land reform might be beneficial in terms of improving

income distribution in the countryside, it could be disastrous if it were to lead to a large

drop in agricultural productivity. Chapter 6 of this study will examine the possible

impact of land reform on agricultural productivity, based on the farm size-productivity

relationship.















CHAPTER 3
THE FARM SIZE-PRODUCTIVITY RELATIONSHIP


A great amount of research has been done on the relationship between farm size

and farm productivity in developing countries. The vast majority of this research

supports the existence of an inverse farm size-productivity relationship. Proponents of

land reform in developing nations point to the inverse relationship as support for their

cause. They argue that, given the existence of an inverse relationship between farm size

and productivity, land reform in developing nations is justified not only for equity

reasons but also for productivity reasons. If smaller farms are more productive than

larger ones, then further land reform should lead to increased agricultural productivity.

In this chapter previous studies on the farm size-productivity relationship are

reviewed, and both the results of those studies and the methodology used by them are

presented. Theoretical explanations for the inverse relationship are discussed and those

studies that have tested the validity of proposed explanations are reviewed. Finally,

the methodology to be used in this study to test for a farm size-productivity relationship

is developed in this chapter.


The Existence of an Inverse Farm Size-Productivity Relationship

The inverse relationship between farm size and productivity was first observed by

Chayanov in his work on Russian agriculture in the 1920s (Chayanov, 1966). It was

observed in China in the 1930s (Buck, 1937) and in Europe prior to the First World War









(Lenin, 1961). A substantial amount of investigation has been done on the inverse

relationship in India, beginning in the mid nineteen fifties when the Indian Farm

Management Surveys were begun. Sen (1975) reviewed several of the Indian studies.

The existence of an inverse farm size-productivity relationship has become part of the

conventional wisdom about agriculture in developing countries. It has been observed in

many developing countries characterized by very different agrarian structures, cropping

patterns, soil types, and climatic conditions.

Some of the more recent empirical studies on the inverse farm size-productivity

relationship are summarized in Table 3.1. The countries included in each study, the exact

relationship measured, and the nature of the relationship are given. This listing

demonstrates the generality of the inverse relationship among developing countries.

The inverse relationship has been documented for Ecuador as well. A recent

World Bank (1996) report on Ecuador found an inverse relationship between farm size

and yields per hectare in that country. The authors of the World Bank report list this

finding as one of the most important findings of their entire study. They go so far as to

suggest that the Ecuadorian government should encourage the transfer of land from large

to small farms, stating that this would lead to increased agricultural production in

Ecuador.

While most studies on the farm size-productivity relationship in developing

countries have documented an inverse relationship, some have found a different

relationship. Two different studies on agriculture in Africa demonstrate both

theoretically and empirically how imperfect markets can lead to a positive relationship

between farm size and productivity. Carter and Wiebe (1990), noting that agriculture in








28

Kenya exhibits a U-shaped farm size-productivity relationship, demonstrate formally that

this relationship can be explained by capital constraints on small farmers. Kevane (1996)

notes that theoretical work on the inverse relationship tends to omit household wealth

from the analysis, and shows that insurance and finance constraints combined with

imperfect land rental markets can lead to a positive relationship between wealth and

yields. He uses village-level data from western Sudan to support the existence of this

type of positive relationship.


Table 3.1. Summary of Some Recent Studies on the Inverse Farm Size-Productivity Relationship

Study Countries Included Relationship Investigated Results
Significant inverse
Bardhan, 1973 India Farm size output per acre relationship

Brazil, Colombia, India, Significant inverse
Berry and Cline, Malaysia, Pakistan, Farm size value added per relationship for all
1979 Malaysia, Pakistan, unit of land
Philippines countries
Significant inverse
Carter, 1984 India Farm size output per hectare relationship

Bangladesh, Barbados, Significant inverse
Burma, Ethiopia, India, relationship for all
Korea, Mexico, Nepal, Farm size gross output per countries except
Comia, 1985 Nigeria, Peru, Sudan, hectare Bangladesh, Peru and
Syria, Tanzania, Thailand, Thailand, for which no
Uganda relationship was found
Significant inverse
relationship for some
Farm size output per net relationship for some
Ghose, 1979 India Farm size districts and time
sown acre periods, no relationship
for others
Significant inverse
Farm size yields per acre of relationship at
Taslim, 1989 Bangladesh Far eaaggregate level, no
operational area relationship at district
level









Existing studies on the inverse farm size-productivity relationship are subject to

several weaknesses. The first and most important is with the definition of productivity.

The phrase "farm size-productivity relationship" is misleading. Most existing studies

document the inverse relationship via estimation of an equation such as:

lnPO/X= a+ plnX + e (1)

or

InGO/X= a + plnX + e (2)

where PO represents physical output, GO represents gross output value and Xrepresents

farm size. When J3 is shown to be negative and significant, then it is often stated that an

inverse farm size-productivity relationship exists. In some cases, an inverse relationship of

this type is taken as evidence that smaller farms are more efficient than larger farms. The

following quote is a good example of the implied equivalence between land productivity

and efficiency that is often found in the literature on the inverse relationship:

All in all, in traditional labor surplus agriculture, small farms characterized
by acute land and financial scarcity may have, on the whole... a total
resource use per unit of land substantially higher than that of large farms;
consequently, output per unit of land is expected to be higher. Small
farming would therefore appear to be the most efficient (Cornia, 1985, p.
516).

Unfortunately, the terms "productivity" and "efficiency" are often used very loosely in the

literature on the inverse relationship. As will be shown below, both terms have very precise

definitions, and it would seem appropriate to be very clear about just what type of farm size-

productivity relationship exists before drawing policy implications from such a relationship.









Definitions of Productivity and Efficiency


Productivity

The definition of productivity is simple; it is the ratio of output to input. If the

production process yields a single output using a single input, then this ratio is easy to

compute and can be expressed in physical or economic units. However, in the case of

multiple inputs or outputs, the inputs or outputs must be aggregated in a meaningful

manner. This is typically done by weighting the inputs or outputs by their prices. In this

case, the result is an economic rather than a physical ratio. As Norsworthy and Lang

state, "productivity is an economic concept" (1992, p. 8).

More precisely, Norsworthy and Lang define total factor productivity, or TFP, as

the weighted average productivity of all purchased inputs, where the weights are the

shares in the total cost of production. Thus

TFP = Ejwjyj/2ivixi (1)

Where yj is the physical quantity of output j

xi is the physical quantity of input i

wj is the share of output j in total revenue

vi is the share of input i in total cost

and

wj = qjyj/Yjqjyj

vi= Pixii/iPiXi

where qj = the price of output j j = 1,...,J

pi = the price of input i i = 1,...,I









yj is the physical quantity of output j

xi is the physical quantity of input i.

However, not all economists agree with the interpretation of productivity as an

economic concept. Silver, for example, contends that "our concern is with physical

relationships in productivity measurement" (1984, p. 44). His interpretation of

productivity is based on production function analysis. Take for example a Cobb-Douglas

production function:

Y=ALaKP (2)

where Y = output, L = labor, and K = capital. For Silver, the value of A in this

expression is an indicator of the total factor productivity (TFP) of the production process.

That is,

TFP = A = Y/La KP. (3)

For two production units employing the same inputs of capital and labor, all variables

being measured in the same units, the one with the higher value of A will generate, for a

given amount of L and K, a higher value of Y. In other words, its total factor

productivity will be greater.

Fleisher and Liu (1992) use this definition of total factor productivity in their

analysis of the effects of plot size on productivity in Chinese agriculture. Their

production function takes the general form

Q = A La KTYX'8 (4)

where Q = aggregated output, L = labor services, K = capital services, T = land services,

and X = other purchased inputs. Like Silver they define total factor productivity as the









ratio of outputs to inputs, with the inputs being weighted by the parameters of the

estimated production function:

TFP = Q/ La K TYX'= A. (5)

While the interpretation of total factor productivity as a physical concept derived

from the production function may be valid for the case of multiple inputs (which can be

aggregated using the parameters from the estimated production function), complications

arise in the case of multiple outputs. There exists no meaningful way of aggregating

multiple outputs without resorting to economic weights (i.e. prices). In the case of

multiple outputs, the definition of total factor productivity developed by Norsworthy and

Lang is more useful.


Partial Productivity Measures

Total factor productivity is a measure of the average productivity of all purchased

inputs. It is also possible to measure the productivity of any particular input. Partial

productivity is the ratio of output to a particular input. For example, in assessing the

performance of agriculture, researchers often look at land productivity, or the ratio of

output to units of land. This is precisely what is most commonly done in the literature on

the inverse relationship. By using output per unit of land or yields per unit of land as

their productivity measurement, researchers are using a partial productivity measurement

- land productivity.

Partial productivity measures can be quite useful. Often we are interested in

assessing productivity in terms of the scarcest or most limiting input. However, when

using partial productivity measures it is important to keep in mind that they are just that -

partial. A farm which has relatively high land productivity will not necessarily be doing









better than its competitors; the higher land productivity may be offset by lower

productivity of other factors.


Efficiency

Like productivity, efficiency can be expressed as a ratio. It is the ratio of an

observed value to its optimal value. For example, we may measure productive efficiency

by comparing the observed profit of the production unit to optimal profit or by comparing

observed output to maximum potential output obtainable from the inputs used (Lovell,

1993). The first ratio is an indicator of economic efficiency; the second is an indicator of

technical efficiency.

In production, there are three main types of efficiency: economic, technical and

allocative. The broadest of these is economic efficiency, and it encompasses the other

two. A firm which is economically efficient will also be technically and allocatively

efficient.

The definition of economic efficiency depends upon the economic objective the

production unit is assumed to pursue. If the objective of the production unit is to

maximize profits, then the production unit is economically efficient if its observed profit

equals its maximum potential profit. A measure of economic efficiency is therefore the

ratio of observed profit to maximum profit. If the production unit's objective is assumed

to be revenue maximization, then the measure of economic efficiency is the ratio of

observed revenue to maximum revenue.

Economic inefficiency can be caused by technical inefficiency, allocative

inefficiency, or both. A production unit is technically efficient if it produces the

maximum possible output given its input usage or, equivalently, if it uses the minimum







34

possible inputs given its output. A more formal definition of technical efficiency is given

by Koopmans (1951). According to Koopmans, a producer is technically efficient if an

increase in any output requires a reduction in at least one other output or an increase in at

least one input, and if a reduction in any input requires an increase in at least one other

input or a reduction in at least one output. Thus, a technically efficient producer

produces on his production possibility frontier. A widely used measure of technical

efficiency is the Debreu-Farrell measure. This measure is defined as one minus the

maximum equiproportionate reduction in all inputs that still allows continued production

of given outputs (Lovell, 1993). A ratio of unity indicates that no equiproportionate input

reduction is feasible and production is technically efficient. A score of less than unity

indicates the severity of technical inefficiency.

Allocative efficiency measures the degree of correctness in the adaptation of

factor and product proportions to input and product prices. For an allocatively efficient

firm, the marginal rate of substitution between any two inputs equals the price ratio of

those inputs, and the marginal rate of transformation between any two outputs equals the

price ratio of those outputs. The measure of allocative efficiency is usually obtained

residually as the ratio of the measure of economic efficiency to the measure of technical

efficiency.

Jamison and Lau propose an additional type of efficiency, which they call

"market efficiency." They define the market efficiency of a firm as the firm's "capacity

to get a good price for (its) inputs and outputs" (1982, p. 64). Different firms often pay

different prices for identical inputs and receive different prices for identical or nearly

identical outputs. It is commonly suggested that small farms in developing countries pay







35

a higher price for purchased inputs and receive a lower price for their products than large

farms. If this is true, then small farms have a lower market efficiency than large farms,

using the terminology of Jamison and Lau.


The Relationship between Productivity and Efficiency

Given the definitions of productivity and efficiency presented above, the question

arises as to the nature of the relationship between the two concepts. Before this

relationship can be fully explained, it is necessary to introduce the concept of

technological level. With the exception of market efficiency, the efficiency concepts

introduced above relate the performance of an individual firm at a given time to its

optimal performance at that same point in time. Technological level, on the other hand,

relates the optimal performance of a specific firm to the optimal performance of other

firms or to the optimal performance of that same firm at a different point in time.

Specifically, firm a is at a higher technological level than firm b if firm b's production

possibility set is a proper subset of firm a's production possibility set. In other words,

firm a has available to it all of the production choices available to firm b and then some.

This relationship is illustrated in Figure 3.1 for a one-input, one-output production

process. Firm a's production function is given by Fa(X) and firm b's production function

is given by Fb(X). Firm a has a higher technological level than firm b, since its

production frontier lies everywhere above that of firm b.

The relationship between productivity and efficiency can now be defined.

Productivity encompasses both efficiency and technological level. Productivity

differences between firms may be attributed to the firms having different production

frontiers (being at different technological levels), to differences in efficiency among the









firms, or to a combination of both. Similarly, productivity growth may be caused by

improved technology or by more efficient production using the same technology. Thus,

in a world where inefficiency exists, productivity growth is not necessarily equivalent to

technological advancement.


Fa(X)


Fb(X)


Input (X)


Figure 3.1. Technological Level


When we are concerned with how productive performance changes over time we

typically use total factor productivity as our measurement of comparison. Thus, our

measure of productivity change includes both shifts in the production frontier (changes in

technological level) and changes in how far production occurs from that frontier (changes

in efficiency). It is possible to decompose productivity changes into these two

components. Total factor productivity can also be used to compare the performance of









different firms at the same point in time, and such a comparison will incorporate

differences in efficiency as well as differences in technological level between firms.

Total factor productivity will be used in this study to determine the relationship between

farm size and productivity.


Measuring Productivity

Our concern is with measuring total factor productivity, or TFP, rather than with

partial productivity measures, since total factor productivity will be the main productivity

measurement used in this study. The techniques for measuring total factor productivity

can easily be translated into techniques for measuring partial productivity, with fewer

difficulties as one does not have to deal with the problem of how to aggregate inputs

when measuring partial productivity.

There are two main problems in measuring total factor productivity. The first is

to determine which inputs and outputs to include in the measurement, and the second is to

develop an appropriate means of weighting those inputs and outputs. Knight addressed

these problems in 1933, noting that if all inputs and outputs are included, then according

to the first law of thermodynamics (which states that neither matter nor energy can be

created or destroyed), total factor productivity will always equal unity. Knight proposed

to redefine productivity as the ratio of useful output to useful input. Representing

usefulness with weights incorporating market prices gives a measure of total factor

productivity like that of Norsworthy and Lang described previously. Alternatively,

representing the usefulness of inputs with weights taken from the parameters of the

production function provides a measure of total factor productivity like that proposed by

Silver. Both types of TFP measurements are found throughout the productivity literature.









For this study, inputs will be weighted using market prices, since our concern is

with economic productivity. Therefore, total factor productivity is defined, following

Norsworthy and Lang, as:

TFP = jwjyj/Eivixi (6)

where yj is the physical quantity of output j

xi is the physical quantity of input I

wj is the share of output j in total revenue

vi is the share of input i in total cost

and

wj= qjyj/jqjyj

Vi = pixiI/iPiXi

where qj = the price of output j j = 1,...,J

pi = the price of input ii = 1,...,I


The Role of Land Quality

A second weakness of most of the existing studies on the farm size-productivity

relationship is that they do not take into account differences in land quality between small

and large farms. If productivity is measured per unit area, then adjustments must be

made for land quality differences between farms, unless it can be shown that there is no

correlation between farm size and land quality. Several authors suggest that the observed

inverse relationship between farm size and productivity may be due simply to differences in

land quality between small and large farms, with small farms having better quality land

(Bhalla, 1988, Bhalla and Roy, 1988, Newell et al., 1997, Sampath, 1992). Theoretically,

areas with higher quality land might attract a higher population density, leading to pressures









to subdivide farms and resulting in smaller-sized landholdings than in areas with lower

quality land. Using Indian data, researchers have found that controlling for village-specific

effects reduces the observed effect of farm size on productivity, i.e. while villages with

smaller farms tend to have higher output per unit of land, there is no inverse relationship

within villages (Bhalla and Roy, 1988, Newell et al., 1997). Furthermore, when farm-level

land quality variables are included in the analysis, the inverse relationship significantly

weakens and in some cases disappears (Bhalla and Roy, 1988, Sampath, 1992).

This study will control for differences in land quality among farms of different sizes

by incorporating land quality variables into the analysis. The regression of farm

productivity will include a vector of land quality variables as independent variables in the

analysis.


Differences in the Environmental Impact of Small and Large Farms

A third weakness of the current literature on the farm size-productivity relationship

is that it ignores other potential differences between small and large farms, focusing

exclusively on differences in productivity. That economists should focus on productivity is

natural and understandable. Unfortunately, however, policy recommendations are often

made based exclusively on their productivity impact, and unintended consequences can arise

when these recommendations are acted upon. If there are other important differences

between small and large farms, then changing the farm size distribution may have other,

unanticipated consequences.

This study seeks to address whether activity on small farms has a lesser impact on

the natural environment than activity on large farms, specifically with respect to

contamination from pesticides. There are many different ways in which farming impacts the









surrounding environment, and this study focuses on just one of them pesticide

contamination. Pesticide contamination is a serious problem in Ecuador, with farm worker

illnesses caused by pesticide exposure occurring often (Crissman et al., 1994). The theory

behind why small farms might utilize pesticides differently from large farms will be

discussed below, and actual differences in pesticide use between farms of different sizes in

Ecuador will be discussed in Chapter 6. This study measures only differences in quantities

of pesticide used, not in type of application or actual measured environmental damage.

In a perfectly competitive economy with no factor market imperfections and with

homogeneous land quality we would expect all farms to use inputs in the same proportion.

Farmers would apply pesticides to the point where their marginal product equals their price.

However, factor markets are not always perfect. Imperfections in factor markets are often

assumed to be the driving force behind the inverse farm size-productivity relationship, and

quite likely give rise to differences in input use between small and large farms.

Factor market imperfections may cause small farms to confront different factor

prices than large farms. Specifically, small farms likely face a lower price for labor and a

higher price for land, capital and other purchased inputs than large farms. There are several

reasons for this. Labor supervision costs may cause the effective price of labor on small

farms, which use mainly family labor, to be lower than the price of labor for large farms,

which must hire in a substantial portion of the labor they use. Higher levels of available

capital and increased access to credit for large farms may make purchased inputs relatively

cheaper for large farms. Furthermore, large farms may be able to exercise market power

and political connections to obtain a lower price for purchased inputs than that faced by

small farms. The result is that large farms utilize capital and purchased inputs (including









pesticides) more intensively than small farms, while small farms utilize labor more

intensively.

Formally, both small and large farms equate their rate of technical substitution

between pesticides and labor with the price ratio of these two inputs:

RTSs = pm /p (7)

RTS'=p,,l/pI (8)

where RTS is the rate of technical substitution, pm is the price of pesticides, pi is the price of

labor, and the superscripts s and I refer to the small and large farm cases, respectively.

Small and large farms may face different prices for labor and pesticides, such that:

pt < P1 (9)

and

PmS > Pm. (10)

Therefore,

RTSs > RTS'. (11)

The small farm will produce at a point where its rate of technical substitution between

pesticides and labor is greater than that of the large farm, and will use more labor and fewer

pesticides than the large farm.

This point can be illustrated with an isoquant diagram as in Figure 3.2. In this

diagram all farms are assumed to face the same technical substitution possibilities between

labor and pesticides as indicated by the isoquant Q. Large farms confront relative factor

prices (isocost line /) such that labor is expensive and pesticides are cheap. Small farms

confront relative factor prices (isocost line s) such that labor is cheap and pesticides are

expensive. Optimum resource use for large farms occurs at point A, while optimum







42
resource use for small farms occurs at point B. Large farms will use relatively less labor and

relatively more pesticides than small farms.

This result will not necessarily hold if the owners of large farms are using their land

less intensively than small farmers, as the literature on the inverse relationship suggests. In

this case large farms will have a lower total output than small farms, producing Q' rather

than Q, with optimum resource use occurring at point C. In this case, large farms will use

fewer pesticides than small farms. The issue of differential use of pesticides by small and

large farms is one that must be decided empirically.


s Q





Q









Pesticides

Figure 3.2. Optimum Use of Labor and Pesticides on Small and Large Farms


Possible Explanations for the Inverse Farm Size-Productivity Relationship
Several hypotheses exist about the cause or causes of the inverse relationship.

The theory that it may be due to land quality differences between farms of different sizes









has already been discussed. The other, most important proposed explanations for the

inverse relationship are labor dualism, labor supervision costs combined with market

imperfections, decreasing returns to scale, and the effects of uncertainty on farmer

decisions. A number of theoretical and empirical papers have been written in support of

these alternative explanations.

The seminal paper for the labor dualism explanation is that of Sen (1966), although

it is preceded by the work of Chayanov (1966). In his paper, Sen models the economic

equilibrium of a peasant family and the dual equilibrium of a partly peasant, partly capitalist

agriculture. He suggests that there may be a significant difference between the opportunity

cost of family labor and the wage of hired labor, which leads to a greater per-unit application

of labor on small farms than on large farms, and to the inverse farm size-productivity

relationship. Sen notes that data from India tend to support this explanation, with smaller

farms having both a higher value of output per acre and a higher amount of labor applied per

acre than larger farms.

While the labor dualism explanation is commonly employed to explain the inverse

farm size-productivity relationship, it is rarely tested for directly. Rather, as Bhalla and Roy

(1988) point out, many authors use the existence of the inverse relationship as evidence to

support the hypothesis of labor dualism. (Sen himself did this.) Those studies that do test

for labor dualism do so by testing for nonseparation of labor supply and demand decisions.

Theoretically, with perfect labor markets and no labor dualism, the labor supply and labor

demand decisions of the household should be separable. Benjamin (1992) and Pitt and

Rosenzweig (1986) find evidence of separability using data from Java.









Another hypothesized explanation of the inverse relationship is that it is caused by

labor supervision costs combined with imperfections in the labor, capital, and/or land

markets in developing countries. The most common formulation assumes that principal-

agent problems and the related labor supervision costs cause effective labor costs to be

higher on large farms which must hire most of their labor than on small peasant farms where

most of the labor is provided by the family (Bardhan, 1973, Eswaran and Kotwal, 1986,

Feder, 1985). As in the labor dualism explanation, this leads to small farms using labor

more intensively and achieving a higher output per unit of land (and a lower output per unit

of labor). As Feder notes, land or capital market failures are also necessary for this model to

hold, since with perfect markets each family would lease in or lease out as much land as

would be required to maintain an optimal operational holding proportionate to the size of the

family. Market imperfections which would prevent this from happening were discussed in

Chapter 2.

The next proposed explanation of the inverse relationship is that agriculture exhibits

decreasing returns to scale. This explanation has been almost universally rejected in

empirical research on the inverse relationship. The majority of empirical studies of

developing country agricultural production functions find nearly constant returns to scale

(Bardhan, 1973, Berry and Cline, 1979, Comia, 1985, Johnson and Ruttan, 1994, and

references therein). Furthermore, while decreasing returns to scale would produce an

inverse farm size-productivity relationship, decreasing returns should also favor a natural

subdivision of land into smaller holdings to achieve a scale efficient equilibrium, and this is

not observed in developing countries (Barrett, 1996).









Finally, both Srinivasan (1972) and Barrett (1996) have developed models which

show that the inverse relationship can arise out of the effects of uncertainty on farmers'

decision making. Srinivasan demonstrates formally that yield uncertainty caused by

"vagaries of weather" is sufficient in and of itself to generate the inverse relationship. He

shows that even in the absence of imperfections in input markets and of differences in

quality of land it may still be optimal for small farmers to use more inputs per hectare (and

thus obtain higher expected yields) than large farmers, provided that all farmers have the

same utility function for income that exhibits non-increasing absolute risk aversion and non-

decreasing relative risk aversion as income increases. Barrett develops an analytical model

which shows that price risk combined with a non-degenerate land distribution can also lead

to an inverse relationship between farm size and productivity. The basic intuition of

Barrett's model is that, where land or credit market failures constrain small farmers'

capacity to outbid larger farmers for land, food security stress created by food price risk

induces small, net buyer farms to utilize extraordinary amounts of labor in production, even

beyond their shadow valuation of labor. In contrast, large, net seller farms react to food

price risk by reducing their use of costly inputs. Thus, objectively identical risk exposure

stimulates different reactions conditional on endowments, thereby generating the inverse

farm size-land productivity relationship.


Testing for the Inverse Relationship and Its Causes in Ecuador

For this study, the following equation will be used to test for an inverse relationship

between farm size and productivity:

InTFP = a + 3lnOP + ylnHW+ pQ + XZ + s (12)







46
where TFP is total factor productivity, OP is land operated, i.e. land owned plus land leased

in minus land leased out, HWis the number of household workers, Q is a vector of land

quality variables (soil type, slope, etc.), and Z is a vector of operator human capital variables

such as age and education. If an inverse relationship exists between farm size and total

factor productivity, then P should be negative and significant. The expected sign of the

remaining coefficients are y (+) and W (+), as is discussed below.

Equation (12) is also estimated with gross output value per hectare replacing the

TFP variable, as follows:

lnGO= a + plnOP+ylnHW+ Q + XZ+ c (13)

where GO is gross output value and the other variables are as in equation (12). This

equation is estimated for two purposes. The first is to maintain consistency with previous

studies on the inverse relationship and to verify whether or not the type of inverse

relationship typically found in other developing countries exists in Ecuador. The second

is that the inverse relationship may not show up strongly in an analysis of total factor

productivity if larger farms are underutilizing the productive potential of their land

relative to smaller farms. In this case, large farms will be using fewer factor inputs than

small farms, and an inverse farm size-land productivity relationship may exist while no

inverse farm size-total factor productivity relationship exists

The first possible explanation of the inverse farm size-land productivity relationship

to be addressed in this study is that of differences in land quality between large and small

farms. If land quality differences between small and large farms are the major cause of the

inverse relationship, then no inverse relationship will appear when the data are analyzed,

since a vector of land quality variables is included in the specification.









The hypothesis that imperfections in the labor market (labor dualism, supervision

costs, etc.) are part of the cause of the inverse relationship can be tested with equations (12)

and (13). If no labor market imperfections exist, then farm output should be independent of

the number of farm household workers. For this reason number of household workers is

included as an independent variable in equations (12) and (13). If y, the coefficient on the

household workers variable, is found to be positive and significant for these equations, then

labor market imperfections are part of the cause of the inverse relationship in Ecuador.

Whether or not family and hired labor are perfect substitutes can also be tested

through use of the production function, as per Frisvold (1994). Effective labor input can be

expressed as a function of labor force composition, as follows:

E=L[(F + 1)IL] (14)

where E represents effective labor input, L equals the sum of family and hired labor used on

the farm, and F equals family labor used on the farm. If family and hired labor are perfect

substitutes, then 8 will be zero. The sample data is used to estimate a production function of

the form:

InGO = a + PilnOP + 21X+ 31nE + 4Q + 5Z + (15)

where GO is gross output value, OP is size of land operated, Xis the value of purchased

non-labor inputs, Q is a vector of land quality variables, Z is a vector of operator human

capital variables such as experience and education, and E is as defined above. This can be

rewritten as

InGO= a + pilnOP+ P321nX+ P31nL + O1n[(F+ 1)/L] + 4Q + PsZ + (16)

where 0 = P38. If 9 is significantly different from zero, then family and hired labor are

imperfect substitutes (assuming that 03 is also significantly different from zero).







48
The estimated equations are presented in Chapter 6 with data collected from the

study area. The next chapter provides background information on Ecuador, where data for

this study were collected. The methodology used for data collection is also described.














CHAPTER 4
STUDY AREA AND METHODOLOGY


Data for this study were collected in the area of Bolivar, a small town in the

Andean region of Ecuador. Bolivar is located in Carchi, the northernmost province of

Ecuador, approximately eighty kilometers from the Colombian border. The town is

situated along the Pan-American highway, the main road through the Ecuadorian Andes,

and is approximately 180 kilometers north of Quito, the country's capital. The primary

economic activity in the area is agricultural production.

This chapter provides background information on the study area and information

on the methodology used in this study. The chapter begins with general information on

Ecuador, then gives specific information on Bolivar. The survey methods used to collect

data for the study are then described.


Ecuador

Geography

Ecuador is located in western South America, bordering the Pacific Ocean at the

equator, between Colombia and Peru. It has a total area of 283,560 square kilometers,

which makes it slightly smaller than the State of Nevada. Land elevation in Ecuador

ranges from 0 meters at the country's lowest point (along the Pacific Ocean) to 6,267

meters at its highest point (the peak of Mount Chimborazo).









50
Ecuador is divided into four geographic regions. Three of these are continental -

the Costa, the Sierra, and the Oriente. The fourth region consists of the Galipagos

Islands, a group of islands of varied size located approximately 1,000 kilometers west of

the Ecuadorian coast in the Pacific Ocean. Climate varies greatly from one region to

another, and different agricultural products are produced in different regions.

The Costa, or coastal region, is located in the western part of Ecuador, between

the Pacific Ocean and the Andes Mountains. It consists of coastal lowlands, coastal

mountains, and river valleys separated by rolling hills. The coastal lowlands are 200

meters or less in elevation, and the coastal mountains have an elevation of no more than

1,000 meters. The climate in the Costa is tropical, with temperatures fairly similar along

the entire region and with minor seasonal variations in temperature. Average

temperatures range from 230 C in the South to 260 C in the north. The hottest

temperatures occur during the winter rainy season, especially from February to April.

Rainfall in the Costa decreases from north to south, due to the effects of the Humboldt

Current, a cold ocean current that flows north from Chile and turns westward near the

Gulf of Guayaquil in Ecuador. The cold water and air currents associated with the

Humboldt Current inhibit rainfall along the southern coast of Ecuador. Coastal

vegetation changes from tropical rainforest in the north to tropical savannah and desert in

the south. Average precipitation varies from 300 centimeters in the north to 30

centimeters in the south (Hanratty, 1989).

Coastal climate is strongly influenced by El Niflo, which occurs periodically,

approximately every six years. During El Nifto occurrences a change in atmospheric

pressure shifts ocean currents, displacing the cold waters of the Humboldt Current and










allowing warm waters to come closer to shore. An increase in air and water

temperatures, sea levels and relative humidity occurs, producing unusually heavy rainfall

and much flooding. El Nifio typically causes great damage to crops and infrastructure in

the coastal region.

The Sierra, or highlands, consists of two major chains of the Andes Mountains,

the Cordillera Oriental and Cordillera Occidental (Eastern and Western Chains) and the

plateau between the two chains. The Sierra has at least twenty-two peaks over 4,200

meters in height, and the average altitude of the intermontane plateau is 2,650 meters

(Hanratty, 1989).

In the Sierra, climate is divided into levels based on altitude. The tropical level is

located between 400 and 1,800 meters and experiences temperatures ranging from 200 C

to 250 C with heavy precipitation. The subtropical level occurs between 1,800 and 2,500

meters, with temperatures from 150 C to 200 C and moderate precipitation. Between

2,500 and 3,200 meters is the temperate level, with a temperature range of 100 C to 150 C

and an annual rainfall of 100 centimeters. The temperate level has a rainy season lasting

from January through June (winter) and a dry season from July through December

(summer). The cold level extends from 3,200 meters to 4,650 meters, with average

temperatures from 3 C to 90 C and precipitation appearing in the form of rain, hail, and

thick fog. The frozen level occurs above 4,650 meters, with snow- and ice-capped peaks

year-round and temperatures below 30 C.

The Oriente region consists of the Andean foothills east of the Cordillera Oriental

and the Eastern lowlands. Climate in this region is tropical, with abundant rainfall and








52

temperatures surpassing 280 C. Rainfall in the Andean foothills sometimes exceeds 500

centimeters per year (Hanratty, 1989).

The Galipagos Islands are located in the Pacific Ocean approximately 1,000

kilometers west of the Ecuadorian coast. The Islands' climate is strongly influenced by

the Humboldt Current, without which the Islands would have an equatorial climate. At

sea level temperatures average 21 C. There is no precipitation during the eight summer

months, and some fog and drizzle occurs during the winter months of January through

April. Above sea level, the islands have a mixture of tropical, subtropical, temperate, and

cold climates (Hanratty, 1989).


Socioeconomic Characteristics

As of July, 1999, the estimated population of Ecuador was just over 12.5 million

people. Of that population, 35% were between 0 and 14 years of age, 60% were between

15 and 64 years of age, and 5% were age 65 years and older. The estimated population

growth rate for 1999 is 1.78% (CIA, 1999).

The population of Ecuador is composed of various ethnic groups. According to

the 1999 World Factbook of the CIA, 55% of the Ecuadorian population is mestizo

(mixed Amerindian and Spanish), 25% is Amerindian, 10% is Spanish, and 10% is black.

The official language is Spanish, and several Amerindian languages are also spoken

within the country (especially Quechua). Ninety-five percent of the population is Roman

Catholic.

Poverty is widespread in Ecuador, and has been increasing in recent years.

According to the United Nations Development Program, in 1995 there were four million

people living in poverty in Ecuador; today there are eight million. The country's infant










mortality rate is 30.69 deaths per 1,000 live births (CIA), and per capital GDP in 1999

was $1,100 (Cornell University, N.D.).

In 1999 Ecuador entered into its most severe economic crisis in decades. Since

1995, various factors had hurt the Ecuadorian economy, including a border conflict with

Peru, damage caused by the 1997-1998 occurrence of El Niflo, a fall in the prices of some

of the country's principal exports, and a financial crisis brought about by the insolvency

of several of the country's largest banks. The real growth rate of GDP fell from 3.4% in

1997 to 0.4% in 1998 to -7.3% in 1999, and urban unemployment rose from 9.3% in

1997 to 15.1% in 1999 (Cornell University, N.D.). Underemployment, defined as the

proportion of the population working less than 40 hours per week or earning less than

minimum wage, rose from 42.7% in 1997 to 57.0% in 1999 (Cornell University, N.D.).

Inflation in 1999 reached 60.7%, up from 43.4% in 1998. In March of 1999, due

to an impending banking crisis, the government froze most deposits for a period of one

year. In June of 1999 it undertook an audit of the entire banking system, after which it

liquidated one bank, merged three others, and ordered four others to recapitalize. Lack of

confidence in the country's banking system led to capital flight which, combined with the

general air of uncertainty, brought about a depreciation of the sucre relative to the dollar

of 196.6% (Cornell University, N.D.).

An additional factor contributing to the economic crisis in Ecuador is the

country's high level of foreign debt. In 1999, Ecuador's total foreign debt reached

$16,282 million, or 119% of the country's GDP. Of this, $2,530 million was private debt

and $13,752 millions was public sector debt (Cornell University, N.D.).










Agriculture

Agriculture in Ecuador varies widely both in terms of technologies used and of

crops and livestock produced. Due to its enormous variety of climatic regions, the

country is able to produce a wide range of agricultural products, from tropical fruits such

as bananas and coffee to temperate-zone grains and tubers such as wheat and potatoes.

The techniques used to produce these crops range from slash and bum agriculture in the

newly colonized portions of the Oriente to modem agriculture utilizing high-yielding

seed varieties, fertilizers, pesticides, and agricultural machinery. It is not uncommon to

see a horse or oxen plowing in one field and a tractor or combine working in the

neighboring field.

Sierra crops and livestock have traditionally been utilized almost exclusively for

internal consumption. They include tubers, legumes, soft corn, small grains, vegetables

and fruits (Whitaker and Alzamora, 1990). Livestock in the Sierra is comprised primarily

of dairy cattle. One important exception to Sierra agricultural production being destined

primarily for internal consumption is the cut flower industry, which has been growing in

importance since the early 1990s. A substantial portion of the flowers grown in the

Sierra are exported. Most crop production in the Sierra is on smaller farms, ranging in

size up to ten hectares.

The Costa produces major export crops of bananas, coffee, cacao and shrimp. It

also produces rice, sugar, plantains, vegetable oils and beef, destined primarily for

internal consumption. Costa crops are typically produced on larger farms of 50 to 100

hectares or more (Whitaker and Alzamora, 1990). Agricultural products in the Oriente

are similar to those of the Costa, with the addition of oilpalm, produced primarily for








55

export, and yuca (cassava), produced primarily for internal consumption. Agriculture in

the Galapagos Islands is insignificant in terms of total production.

A significant proportion of total farm output is destined for family subsistence,

especially in the Sierra and Oriente. In the Sierra, only 39 percent of total agricultural

output value is sold on the market. In the Oriente, this figure is 36 percent. In the Costa,

70 percent of total output value is sold on the market (World Bank, 1994). The percent

of output sold on the market increases with farm size, as shown in Table 4-1.


Table 4-1.
Average Percent of Output Sold and Per Capita Landholding

Land Size Percent of Output Sold
Class
(per cavita) Sierra Costa Oriente Ecuador

< 1 ha. 37 71 29 49
1 2.5 ha. 45 71 46 55
2.5 5 ha. 60 68 35 61
5 30 ha. 50 68 44 59
>30 ha. 84 59 56 65

All 39 70 36 50

Source: World Bank (1994).


Bolivar

Data for this study were collected from farmers in the parish (parroquia) of

Bolivar, in the northern Sierra. The parish of Bolivar is located in the county of Bolivar

in the province of Carchi, the northernmost province of Ecuador within the Sierra region.

Bolivar is located approximately eighty kilometers south of the Colombian border and










180 kilometers north of Quito, the capital of Ecuador. It is situated along the Pan-

American highway, which is the main road through the Sierra.

According to the last census, the population of the county of Bolivar was 15,157

persons in 1990. The primary economic activity in the region is agricultural production.

In 1990, 79 percent of the economically active population was employed in the

agricultural sector. The remaining 21 percent held diverse occupations, such as school

teacher, merchant, public servant, carpenter, etc. (Instituto Nacional de Estadistica y

Censos, 1991).

According to the Ministry of Agriculture, there are approximately 4,650 farms in

the county, of which 44 percent are less than one hectare in size, 17 percent are between

one and five hectares, 33 percent are between five and ten hectares, and 6 percent are

larger than ten hectares (Ministerio de Agricultura, 1996). Only 22 percent of the

cultivated land in the county has access to irrigation water.

The parish of Bolivar consists of the town of Bolivar and the surrounding rural

areas. Its area is approximately 4,086 hectares and its total population in 1990 was 4,348

people, of which 1,938 inhabited the town itself and 2,410 lived in the rural periphery

(Instituto Nacional de Estadistica y Censos, 1991). The parish is situated between 2,500

and 2,600 meters above sea level and experiences a temperate climate. Average monthly

temperature fluctuates between 10.50 C and 13.50 C. A variety of grains, legumes, fruits,

vegetables and tubers are grown in the parish. The most important of these are beans,

onions, maize, peas, wheat, barley, avocado, and potato (Municipio del Cant6n Bolivar,

1998). There is a wholesale market twice a week just outside of town where farmers can










sell their production to wholesalers who transport it to Colombia, Quito, or other

principal cities in Ecuador for resale.

Of the 4,086 hectares of land in Bolivar, 1,298 hectares are considered inapt for

agriculture due to their steep slopes. Another 530 hectares are made up of cangagua, an

extremely hard-packed soil composed primarily of calcium carbonate and clay.

Cangagua land is quite fertile in terms of mineral content, but is so hard-packed that it

cannot be farmed without a recovery process which takes several years; the lands made

up of cangagua in Bolivar are not cultivated. The total land area utilized for agricultural

production in Bolivar is approximately 2,258 hectares (Municipalidad de Bolivar,

A.M.E., 1997).

The farms in the parish of Bolivar are fairly homogeneous. The land is of similar

quality throughout the parish, and nearly all farms have access to irrigation through a

system of canals that run throughout the parish. This is in contrast to the rest of the

county, for which only 22 percent of the cultivated land area is irrigated. The size range

of farms located within the parish is also fairly narrow. There are several farms of less

than one hectare located within the parish, and only a few larger farms of twenty hectares

or more. A farm of five hectares or more is considered to be a large farm within the

parish.

Most farmers and landowners in Bolivar were born in the parish or county and

attended one of the county's public or private elementary schools. There are two private

elementary schools and one public elementary school located within the parish. There is

also one public high school in the parish. Most parish residents of forty years of age or

more have an education level of six years or less, while approximately half of the










younger residents possess a high school level education. Very few of the residents

possess university degrees. The homogeneity of land quality and the similar levels of

education of local farmers should lead to similar production levels among the farms of

the parish, and any relationship between size of farm and productivity is unlikely to be

due to land quality differences or differences in farmer education level among the farms.

Most farmers purchase materials such as fertilizers, seed and pesticides from one

of two agricultural supply stores located in the town of Bolivar. These are typically

purchased on a cash basis; the local stores rarely extend credit to farmers. Most farmers

in the parish conduct all transactions on a cash basis, and many of them expressed

reluctance to use credit for agricultural production on their farms. This is due to the

difficulty in obtaining credit (most banks will only loan money to farmers if they

mortgage their land as collateral for the loan, and then will only loan a fraction of the

value of the land to the farmer) and to the high interest rates for bank loans (typically

greater than 70 percent annually). Several farmers mentioned cases of people they knew

who had taken out loans in the past and lost their land when they were unable to repay the

loans.


Survey Methodology

Data for this study were collected from a sample of sixty farms located within the

parish of Bolivar. The sample farms were randomly selected from a list of parish farms

provided by the employees of the county's property records office. The property records

office furnished a list of all farms located within the parish of Bolivar, and these farms

were then divided into five groups based on size. The first group consisted of farms of

less than two hectares, the second of farms from two to three hectares, the third of farms










between three and five hectares, the fourth of farms from five to seven and one-half

hectares, and the last of farms over seven and one-half hectares. Twelve farms were

selected from each group to form part of the sample.

An initial interview was then conducted with the principal operator of each of the

sixty sample farms. The term operator refers to the person most involved in the day-to-

day operation of the farm. In some cases this is the farm owner, in others it is the tenant

leasing the farm or the sharecropper working the farm. At the initial interview data on

the size of the farm were confirmed and information was obtained about crops planted.

This interview was conducted in October, 1998.

Sixty sample farm operators were interviewed approximately every eight weeks

from October, 1998 through August, 1999. Through these interviews data on production

for each farm were gathered, including input amounts and prices, crop yields, product

price, etc. Socioeconomic data on the family of each farm operator were also collected.

Repeated visits to each of the sample farms allowed for the collection of very detailed

data, which is the reason this type of survey methodology was chosen. Originally, a

different survey methodology had been designed, involving a single interview with each

farmer in a much larger sample. However, this design was deemed impractical after

discussions with scientists at INIAP, the National Agricultural Research Institute, in the

province of Carchi. These individuals pointed out that farmers in the region rarely keep

written records of farm activities and were unlikely to provide reliable data on inputs

during a single post-harvest interview. It was therefore decided to work with a smaller

sample size and conduct repeated interviews with each farmer in the sample throughout

the growing season.









60

Of the original sixty farms chosen to be part of the sample, seventeen were

eliminated during the data collection process due to problems collecting reliable data

from the farm operators. The sample size was thereby reduced to forty-three farms.

Analysis of the data collected from these forty-three farms is presented in the next two

chapters.
















CHAPTER 5
THE SAMPLE STUDY GROUP


The sample for this study consisted of forty-three farms ranging in size from 0.25

to 10 hectares. All of the farms were located within a radius of three kilometers from the

center of the town of Bolivar. A variety of different crops were grown on the sample

farms, with most farms producing more than one type of crop. Several of the farms also

raised dairy cows.

This chapter will provide information on the sample farms, including information

on the types of crops produced, prices received for crops, forms of tenancy, and

socioeconomic characteristics of the farm operators' families. Summary statistics for the

sample group will be presented in this chapter. The question of the type of relationship

existing between farm size and productivity in the region will be addressed in Chapter 6.


Agricultural Production on the Sample Farms

The most commonly produced crops on the sample farms were onions, peas,

beans, and corn. Each of these crops were grown on farms of all different sizes; no crop

was singled out for production on only farms of a certain size group. The number of

farms in each size group and the types of crops grown on those farms is shown in Table

5-1. Crops are listed in order of predominance for each size group. Onions were the

most commonly produced crop among farms from all of the different size categories, and

beans or peas were the second most commonly produced crop among farms of all










size classes, with the exception of farms in the 5 to 7.5 hectare category, for which corn

was the second most commonly produced crop. The third most commonly produced crop

was either beans or peas for all of the farm size classes.


Table 5-1.
Crops Grown by Size of Farm


Number of
Farm Size Farms i Sample Crops Grown
Farms in Sample

< 2 ha. 7 Onions, Peas, Beans
2 3 ha. 12 Onions, Beans, Potatoes, Peas, Corn
3.25 4.75 ha. 6 Onions, Beans, Peas
5 7.5 ha. 10 Onions, Corn, Beans, Peas, Wheat
> 7.5 ha. 8 Onions, Peas, Beans, Corn
All Farms 43 Onions. Beans. Peas, Corn. Potatoes, Wheat

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


The predominance of onion crops is a relatively new phenomenon in the region.

Traditionally, beans, peas, and corn were the main crops in the Parish of Bolivar, with

only minor production of onions taking place. The occurrence of El Niflo in 1997-1998

caused major flooding in the coastal region of Ecuador and coastal onion crops were

devastated, leading to extremely high onion prices. Onion farmers in the Sierra saw their

profits skyrocket, and there was a mad rush among farmers in Bolivar and other regions

of the Sierra to convert from traditional crops to onions. This drove profits back down to

normal levels, but many farmers continued producing onions in the hopes that another

bad year on the Coast would raise profits once again. This has not happened, and most

onion farmers in the sample realized only normal profits. Some even experienced losses,









63

since above average rainfall in the northern Sierra led to the loss of onion crops for many

farmers.

The unexpectedly heavy rains in the region also led to losses for many farmers

who planted peas. Peas are a very delicate crop, and an overabundance of rain when peas

are still young can severely damage the crop. Of the thirteen farms in the sample where

peas were planted, four had no pea harvest at all and two were able to harvest only a few

quintales of peas.

It should be noted that a loss of one crop does not normally imply a total loss for

the farm operator. Most farms produce more than one crop at a time, and among the

sample farms it was common to have as many as three different crops planted in different

sections of the farm at a time. Also, since Ecuador is located along the Equator and does

not experience the strong seasonal climatic variances of countries in the more northern or

southern latitudes, crops can be planted and harvested year-round. If a crop is lost, the

farmer does not have to wait six months or more to replant; he or she can replant

immediately, provided he or she has the resources to do so. The one exception to this is

during winters with unusually heavy rains, as occurred during the time these data were

collected. Many of the farmers in the sample complained that they were unable to plant

when they wanted because the unusually heavy winter rains turned the fields to mud.

Nearly all of the production of the sample farms was sold; very little was

maintained for household consumption. The vast majority of farm operators sold their

crops at the wholesale market located just outside of the town of Bolivar. When a crop

was harvested, the farm operator typically brought it to the market and negotiated with

the buyers there for a favorable price. Buyers came to the wholesale market twice a










week, on Mondays and Thursdays, and farmers typically scheduled harvests to coincide

with these days. Two farmers in the survey transported their products to Ibarra, a city

approximately sixty kilometers to the south of Bolivar, in search of better prices. Dried

beans and peas were not sold at the wholesale market; they were sold to buyers from the

cities of Tulcain and Ibarra who were known in Bolivar.

The fact that nearly all farm production was sold and little was kept for family

consumption separates the region from other parts of the Ecuadorian Sierra. As stated in

Chapter 4, only thirty-nine percent of total agricultural output value is sold in the

Ecuadorian Sierra; the majority is used for family consumption. Among the sample

farms, over ninety percent of total agricultural output was sold.

Price uncertainty was a problem for all of the farmers in the sample. Product

prices varied widely, often from one week to the next, and it was difficult to know at

planting time what price the crop would fetch come harvest time. In fact, when asked

how they knew what price to expect for their crops at planting time, twenty-five of the

forty-three farm operators surveyed said either that they did not know or that they would

sell at whatever price the market offered at harvest time. The responses most commonly

given by farm operators to the question of how they predicted what price they would

receive for their crops are listed in Table 5-2.

The high, low, mean and median prices received by the farm operators for each of

the crops produced on the sample farms are given in Table 5-3. The wide variance in

prices is evident in the disparity between the high and low price for all of the crops listed.

The fact that the average and median prices are relatively close to each other for all of the










crops indicates that the wide variance between high and low prices cannot simply be

attributed to a few farmers in the sample who received abnormally high or low prices.


Table 5-2.
Answers to the Survey Question "How do you know what price to expect for
the crop at planting time?"


Number of Farms Percent of Farms
with this Answer with this Answer Answer

17 40 "I don't know," "No way to tell"

8 19 Q"It depends on the market at harvest time," "At
whatever price it comes out at"
4 9 "It's determined by supply and demand, by buyers
____ and sellers"
3 7 "It depends on the weather"

1 2 "If it's high one season, it will be low the next"

1 2 "Prices are higher during holy week"

1 2 "It's a secret"

8 19 No answer

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Neither smaller nor larger farms were able to receive better prices for their

products; the distribution of prices received for onions and beans, the two most common

crops, follows no notable pattern by farm size (see Tables 5.4 and 5.5). This would

indicate that larger farms in the region are not able to demand a higher price for their

products than smaller farms can demand.










Table 5-3.
Prices Received by Sample Farms for Principal Crops


Crop High Price Low Price Average Price Median Price
Onions 120,000 22,500 47,000 40,000
Peas 200,000 70,000 132,500 145,000
Beans 500,000 37,500 136,600 85,000
Dry Beans 500,000 200,000 275,000 250,000
Corn 400,000 42,000 216,400 250,000

Note: All prices are in sucres; exchange rate = 7,400 sucres/1 U.S. dollar (February, 1999)

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Table 5-4.
Price Received per Costal of Onions by Farm Size*


Farm Size Price per Farm Siz Price per
F Se Costal Costal
(hectares) s (hectares) s
sucress) sucress)
0.25 38000 4 50000
1.5 25000 4 60000
1.5 56000 4.5 30000
1.75 50000 5 40000
2 28000 5 35000
2 30000 5 30000
2 50000 5 60000
2 65000 6 40000
2 120000 6 60000
2.5 40000 7 50000
3 30000 7 45000
3 60000 8 35000
3 90000 8 22500
3 60000 10 35000
4 30000 10 55000
4 40000

Exchange Rate = 7,400 sucres/1 U.S. dollar (February, 1999)
*A coastal is a bag with approximately six cubic feet of volume

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.










Several of the farms in the survey also raised dairy cows. Milk from these cows

was sold to buyers from different pasteurizing plants located in the Province of Carchi.

Eleven of the forty-three sample farms were involved in dairy production. All but one of

these farms possessed six cows or less. The larger the size of the farm, the more likely it

was to have cows. The percentage of farms with dairy cows, the average number of dairy

cows per hectare and the average milk production per hectare increase steadily as farm

size increases (see Table 5-6). The average number of hectares of pasture, average

number of cows per hectare of pasture, and average weekly milk production per hectare

of pasture by size of farm are listed in Table 5-7.


Table 5-5.
Price Received per Quintal of Beans by Farm Size*

Price per
Farm Size Q a
Quintal
(hectares) u
sucress)
0.25 85,000
1.5 37,500
2 95,000
3 150,000
4 245,000
5 45,000
5 500,000
6 50,000
8 150,000
10 60,000
10 85,000

Exchange rate = 7,400 sucres/1 U.S. dollar (February, 1999)
*One quintal equals one hundred pounds

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.










Forms of Tenancy

Three main types of tenancy arrangements existed among the sample farms. The

most common form of tenancy was that of the "owner-operator," in which the landowner

or one of his employees oversaw all aspects of production on the farm. Under this

arrangement all profits or losses experienced by the farm accrued to the landowner. Just

over sixty-five percent of the sample farms were operated under this type of tenancy

arrangement. Fourteen percent of the sample farms were operated under some type of

sharecropping arrangement, through which the landowner shared both expenses and

revenues with a sharecropper who oversaw the operation of the farm. Slightly over

twenty percent of the sample farms were operated under mixed tenancy, with some crops

planted with a sharecropper and others planted solely by the owner. None of the sample

farms were rented or leased.


Table 5-6.
Cows per Hectare of Total Farmland and Milk Production
on Farm by Farm Size

Percentage of Cows per Weekly Milk
Farm Size Farms with Hectare Production per
Cows Hectare (liters)
< 2 ha. 0.0 0.00 0.00
2 3 ha. 8.3 0.08 4.48
3.25 4.75 ha. 16.7 0.12 3.54
5 7.5 ha. 30.0 0.34 4.23
> 7.5 ha. 75.0 0.58 20.51
All Farms 25.6 0.23 9.74

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


A breakdown of type of tenancy by farm size demonstrates that the most common

form of tenancy for every size group was the owner-operator form (Table 5-8). Mixed-










tenancy, in which some portions of the farm are sharecropped and others are planted

solely by the landowner, was the second most common form of tenancy for all but the

largest size group. Sharecropping was the third most common form of tenancy for all

groups, with the exception of farms over 7.5 hectares, for which sharecropping was the

second most common form of tenancy.


Table 5-7.
Hectares of Pasture, Number of Cows per Hectare of Pasture and
Weekly Milk Production per Hectare of Pasture by Size of Farm


Average Number Average Cows Weekly Milk
Farm Size of Hectares of per Hectare of Production per Hectare
Pasture Pasture of Pasture (liters)

< 2 ha. 0.00 0.00 0.00
2 3 ha. 0.25 0.33 18.67
3.25 4.75 ha. 0.25 2.00 56.00
5 7.5 ha. 0.90 0.89 27.22
> 7.5 ha. 3.81 1.38 48.43
All Farms 1.02 1.23 42.32

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Table 5-8.
Type of Tenancy by Farm Size


Percentage of Percentage of Percentage Mixed-
Owner-Operators Sharecroppers Tenancy
< 2 ha. 71.4 14.3 14.3
2 3 ha. 58.3 16.7 25.0
3.25 4.75 ha. 66.7 0.0 33.3
5 7.5 ha. 70.0 10.0 20.0
> 7.5 ha. 62.5 25.0 12.5
All Farms 65.1 14.0 20.9

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.










Sharecropping arrangements varied among the sample farms. Revenue from the

sale of the crop was divided equally between the landowner and the sharecropper in all

cases, but costs were divided in different manners. One common arrangement was for

the sharecropper to provide the seed and for all other costs to be shared equally between

the landowner and the sharecropper. Under another common arrangement, the

sharecropper and the landowner divided all costs, including seed, equally. In some cases

the sharecropper was paid daily wages for his work on the farm; in others he provided his

labor at no charge. The only universal standard for sharecropping arrangements among

the sample farms was that revenues were always split equally between the landowner and

the sharecropper.


Socioeconomic Characteristics of Farm Operators' Families

No specific relationship exists between the average age or number of years of

education of the farm operator and farm size (Table 5-9). The average age of the farm

operators in the sample group was 52 years, and the average number of years of

education of the farm operators was slightly over six and one-half years. All but two of

the forty-three farm operators were male.

Eighty-one percent of the farm operators in the sample group (35 of 43) described

their primary occupation as farmer or agricultural worker (agricultor). Four listed their

primary occupation as businessman, one as policeman, one as truck driver, one as public

worker and one as head of personnel. The average number of days per week the farm

operators spent working off their farms was 2.3 days. Farm operators of the smallest two

size groups (less than 2 hectares and 2 to 3 hectares) spent the most number of days

working off of their farms. They spent an average of 3 days and 2.8 days respectively










working off-farm. Operators of farms greater than three hectares spent between 1.8 and

2.1 days on average working off-farm (Table 5-10).

The average family size for the sample group was four, and the average number

of household workers was 1.8 (Table 5-11). The number of household workers in a farm

operator's family was calculated by weighting different types of family members. Males

over the age of 16 living in the household were counted as one household worker.

Females over the age of 16 were counted as .5 household workers, since females typically

had to divide their time between working in the fields and carrying out household chores

such as cooking, cleaning and childcare. Males between the ages of eight and sixteen

were counted as .5 household workers and females between the ages of eight and sixteen

were counted as .3 household workers.


Table 5-9.
Age and Education of Farm Operator by Farm Size

Average Number of
Farm Size Average Age of Years of Education of
Farm Operatorerator
Farm Operator
< 2 ha. 62 5.0
2 3 ha. 49 7.0
3.25 4.75 ha. 56 6.5
5 7.5 ha. 48 7.3
> 7.5 ha. 52 6.5
All Farms 52 6.6

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Credit, Technical Assistance and Agrarian Reform

Technical assistance from outside agencies was rarely received by farmers in the

sample. Only one of the forty-three farm operators stated that he had received technical










assistance from an outside agency within the past two years. He was the owner of a ten

hectare farm and had received assistance from the Ministry of Agriculture. None of the

other farm operators stated that they had received technical assistance.


Table 5-10.
Days per Week Worked Off Farm


Average Number of Days per
Farm Size Week Operator Spends
Working Off-Farm
< 2 ha. 3
2 3 ha. 2.8
3.25 4.75 ha. 1.8
5 7.5 ha. 2.1
> 7.5 ha. 1.8
All Farms 2.3

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Table 5-11.
Family Size and Number of Household Workers by Farm Size


Farm Size Average Family Average Number of
Size Household Workers
< 2 ha. 4.3 1.8
2 3 ha. 3.6 1.5
3.25 4.75 ha. 4 1.7
5 7.5 ha. 3.9 1.8
> 7.5 ha. 4.6 2.0
All Farms 4 1.8

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


The use of credit for crop production was almost as rare as the receipt of technical

assistance. Only six of the forty-three farm operators in the sample stated that they used










credit for crop production. Of these six, one was the same operator who had received

technical assistance from the Ministry of Agriculture, and he also received credit from the

Ministry. The other five operators received credit either through banks or credit unions.

The use of credit was not limited to farms of a certain size; the smallest farm which used

credit for crop production was one and one-half hectares in size, while the largest was ten

hectares.

There was very little concern among the farm operators in the sample with the

country's agrarian reform laws. When questioned, only three of the sample farm

operators answered that they were familiar with the laws, and not one of the sample farm

operators replied that he/she had been affected by any agrarian reform law. The three

operators who claimed to be familiar with the agrarian reform laws operated farms of

seven, ten, and one and one-half hectares.

When questioned as to what they considered to be the greatest problem facing

farmers in the region, the sample farm operators gave a variety of answers (Table 5-12).

The most common answer was the high price of pesticides and fertilizers. The

predominance of this answer is related to the national bank crisis and subsequent

depreciation of the Ecuadorian sucre in late 1998 and early 1999. Since Ecuador imports

a large portion of its agricultural pesticides and fertilizers, the depreciation of the sucre

led to substantial increases in the prices of these inputs. The second most common

answer to the question of what was the greatest problem facing the region's farmers was

bad weather/bad winter, which is related to the heavy rains which fell on the region at the

time of the survey. The answers to this question were unrelated to farm size, with the

exception of three of the four least common answers (lack of workers, lack of resources









74

and high interest rates), which were all given by operators of small farms (1.5 hectares or

less).


Table 5-12.
Answers to the Question: "In your opinion, what is the greatest problem facing
farmers in the region today?"


Number of Percentage of Average Size of
Answer Farms with Farms with Farms with this
this Answer this Answer Answer
(hectares)
The high price of pesticides/fertilizers 16 37.2 4.8
The bad weather/winter 8 18.6 4.9
The low prices received for our products 8 18.6 4.9
Crop diseases 3 7.0 5.4
Loss of crops 3 7.0 3.8
High price of labor 3 7.0 4.8
Lack of workers 2 4.7 1.25
Lack of resources 1 2.4 0.25
High interest rates 1 2.4 1.5
Lack of technical assistance 1 2.4 4.0
Note: Some farm operators gave more than one answer

Source: Langedyk survey, Bolivar Parish, Carchi, Ecuador, 1999.


Interviews with Large Farm Operators

In addition to the detailed surveys of the forty-three farms in the Parish of

Bolivar, interviews were conducted with twenty large landowners in the Province of

Carchi, within which Bolivar is located. The landowners interviewed all owned farms of

ten hectares or more. The interviews were conducted for the purpose of determining the

owners' relationship with their farms and the productive use of these farms. It is

sometimes suggested in the literature on the farm size-productivity relationship that large

landowners buy land for reasons other than its use in agricultural production (such as

protection against inflation), and are less interested in the productive use of that land than










are smaller farmers. One of the main arguments in favor of land reform is that large

landowners leave large portions of productive land idle or underproductive, which would

be better used if broken up and distributed to small farmers.

The average size of the farms interviewed in this survey was 44.3 hectares, of

which an average of 9.75 hectares were used for crop production, 28.65 hectares were

used as pasture for dairy cows, 1.5 hectares were used as pasture for beef cattle, and 4.4

hectares were left idle (Table 5-13). In other words, among these farms, 22 percent of the

land was used for crop production, 68 percent was used as pasture for either dairy cows

or beef cattle, and 10 percent was left idle.

At first glance, these large landowners appear to be making productive use of

nearly all of their land; only ten percent of the land was left idle. However, the large

landowners dedicate a substantially greater portion of their land to pasture than the

smaller farms do. Among the forty-three farms of one-quarter to ten hectares surveyed in

Bolivar, only 23 percent of the land was used as pasture, versus 66 percent of the land of

the twenty large landowners. The question arises as to how intensively the hectares

dedicated to pasture are utilized. Average number of dairy cows per hectare of pasture

declines substantially as farm size increases, indicating that larger farms are using their

pasture less intensively (Table 5-14). Furthermore, the average number of dairy cows per

hectare of pasture for these large farms is only 0.43 compared with 1.23 average dairy

cows per hectare of pasture for the forty-three farms often hectares or less in the Bolivar

sample (See Tables 5-7 and 5-14). Larger farms evidently use their pasture land less

intensively than smaller farms.









76
Of the twenty large landowners interviewed, only seven lived on their farms. The

remaining thirteen lived off farm, in the cities of Tulcan, el Angel and Quito. The seven

who lived on their farms operated the farms themselves, as did six who lived off farm.

Seven of the thirteen who lived off farm operated the farm with a sharecropper.


Table 5-13.
Use of land by Size of Farm Among Large Farms

Farm Hectares Used Hectares Used as Hectares Used as
(Hectares) in Crop Pasture for Dairy Pasture for Beef Left Idle
Production Cows Cattle
10 4 6 0 0
10 0 10 0 0
11 4 0 7 0
12 5 7 0 0
14 9 5 0 0
15 5 10 0 0
15 3 12 0 0
16 4 12 0 0
18 7 7 0 4
18 12 6 0 0
20 10 10 0 0
27 4 0 23 0
30 20 6 0 4
34 7 27 0 0
40 0 40 0 0
40 0 40 0 0
50 16 34 0 0
76 15 61 0 0
130 50 80 0 0
300 20 200 0 80
Average 44.3 9.75 28.65 1.5 4.4

Source: Langedyk survey, Carchi, Ecuador, 1999.










Table 5-14.
Average Number of Dairy Cows per Hectare of Pasture and Average
Weekly Milk Production per Hectare of Pasture by Size of Farm (Large Farms)


Average Dairy Weekly Milk
Farm Size Cows per Hectare Production per Hectare
of Pasture of Pasture (liters)
10 20 ha. 0.68 4.49
21 40 ha. 0.46 4.34
41 + ha. 0.36 4.16
All Farms 0.43 4.25

Source: Langedyk survey, Carchi, Ecuador, 1999.


The large landowners were more familiar with the Agrarian Reform laws than the

smaller farmers in the Parish of Bolivar. Twelve of the twenty landowners stated that

they knew about the Agrarian Reform laws. None of them said that they had been

affected by the laws.

The twenty large landowners were all asked what their primary reason for owning

their land was (Table 5-15). Some gave two or three answers. The most common answer

was that the landowner owned the lands primarily to make money through crop and dairy

production. The second most common answer was that he/she owned the lands because

they were passed down through the family and he/she did not want to sell them. Only

one owner answered that he held his lands mainly to be able to get away for vacations

and weekends.

The next chapter will address the issue of the farm-size productivity relationship

among the farms in the Parish of Bolivar. The results of several regressions completed to

determine what type of farm-size productivity relationship exists in the region will be

presented, along with explanations of the results.












Table 5-15.
Answers to the Survey Question "What is your main reason for
owning your lands?" (Large Landowners)


Source: Langedyk survey, Carchi, Ecuador, 1999.


Number of Farms Percent of Farms Answer
with this Answer with this Answer
"To make money with crop and livestock
13 65 production"
"Because they were passed down through the
family and I don't want to sell them"

5 25 "As an investment"

1 5 "As a place for vacationing"
















CHAPTER 6
ANALYSIS OF THE FARM SIZE-PRODUCTIVITY RELATIONSHIP
IN BOLIVAR


In this chapter the relationship between farm size and productivity in the Parish of

Bolivar is determined. Six equations are estimated, each with a productivity

measurement as its dependent variable. Three of the estimated equations have only farm

size as the explanatory variable, while the other three utilize farm size, number of

household workers, land quality variables, farmer age and farmer education as

explanatory variables. The parameter of most interest is the coefficient of farm size; a

negative coefficient indicates an inverse relationship between farm size and productivity.

Three different productivity measurements are used in this study. The first is

gross output value per hectare. As discussed in Chapter 3, using gross output value per

hectare as the productivity measurement gives the relationship between farm size and

land productivity. This is a partial productivity measurement, and it is the measurement

most commonly used in studies on the farm size-productivity relationship. The other two

productivity measurements used are two distinct measures of total factor productivity. In

contrast with gross output value per hectare, these are complete productivity

measurements.


-7'










Equations and Variables

The following equations are estimated in order to determine the relationship

between farm size and productivity in the study region:

1. InGO= Pi + 2lnOP + ,

2. InGO = Pi + P21nOP + P3HW + p4TENPART + r5TENMIX + P6ED5 +

p7ED7 + psLESS20 + P9LESS60 + o10PLUS60 + I 1ARENOSA +

312BARROSA + p13CANGAHUA + P14LOC1 + P15LOC2 + p16PEND +

P317MIXSLOPE + 1piDAYSOFF + E,

3. InTFPI = PI + 21lnOP + s,

4. InTFPl = Pi + 21lnOP + 13HW + P4TENPART + p5TENMIX + 36ED5 +

P7ED7 + 38LESS20 + P9LESS60 + P1oPLUS60 + PI1ARENOSA +

P12BARROSA + 313CANGAHUA + P14LOC1 + P15LOC2 + p16PEND +

P17MIXSLOPE + p3sDAYSOFF + e,

5. InTFP2 = P1 + p2lnOP + s,

6. InTFP2 = p1 + p21nOP + 03HW + P4TENPART + isTENMIX + p6ED5 +

p7ED7 + psLESS20 + 39LESS60 + PioPLUS60 + IhuARENOSA +

312BARROSA + i13CANGAHUA + 314LOC1 + P15LOC2 + 1i6PEND +

P17MIXSLOPE + P1iDAYSOFF + s.

Each equation is estimated independently using ordinary least squares, not as a system of

equations. The formal definitions of the variables used in the estimated equations are

given in Table 6-1.




































































W
02
3
a
C


SI
- o











Most farm size-productivity studies involve the estimation of an equation similar

to one of these six equations. The most commonly estimated equation is a variation of

equation number one, with either gross output value per hectare or yields per hectare as

the dependent variable and with only one explanatory variable farm size. Equation

two is a more complete version of equation one; gross output value remains the

dependent variable, but variables on number of household workers, farm operator age

and education, form of tenancy, and land quality are added as explanatory variables.

One critique of previous studies of the relationship between farm size and productivity

is precisely that they may be subject to mis-specification problems by leaving out these

explanatory variables, particularly variables on land quality (Bhalla and Roy, 1988). In

both equations one and two, the relationship between farm size and land productivity is

the subject of analysis.

The third, fourth, fifth and sixth equations are similar to equations one and two.

They have the same explanatory variables as equations one and two, but use different

productivity measurements as the dependent variable. Equations three and four use one

measurement of total factor productivity (TFP1) as the dependent variable, while

equations five and six use a different measurement of total factor productivity (TFP2) as

the dependent variable.

Two different measurements of total factor productivity are used as a response

to the debate over the proper measurement of total factor productivity. There is a great

deal of discussion over how best to weight inputs in total factor productivity

calculations using market prices or using social opportunity costs. As Binswanger, et

al. (1995) point out, using social opportunity costs eliminates the impact of distortions











caused by market imperfections and allows us to measure differences in social

efficiency. Using market prices measures differences in private efficiency. This

distinction is raised most often in discussions of how to value family labor used in farm

production. It is sometimes argued that the wage rate paid to farm laborers is artificially

high due to government regulations, and that the social opportunity cost of family labor

is much lower than this wage rate. The argument has been made that the opportunity

cost of family labor may be close to zero in areas where work opportunities do not exist

and where family workers would not be able to find work outside of the family farm.

The first measurement of total factor productivity, TFP1, values family labor at

the average wage rate paid to hired labor on the farm. The second measurement of total

factor productivity, TFP2, does not include family labor as a measured input; it

implicitly values family labor at zero. One result of interest from the analysis is

whether the relationship between farm size and TFP 1 differs from the relationship

between farm size and TFP2.


The Farm Size-Productivity Relationship

Average gross output value per hectare, TFP1 and TFP2 by size of farm are

listed in Table 6-2. This table indicates a clear decrease in average gross output value

per hectare as size of farm increases. While the smallest farms have an average gross

output value of 5,389,009 sucres per hectare, the largest farms have an average gross

output value of only 3,294,366 sucres per hectare. This supports the existence of an

inverse relationship between size of farm and land productivity. However, the

relationship between size of farm and total factor productivity is exactly the opposite;

total factor productivity increases as farm size increases. This is true both when family











labor is included as a valued input in the total factor productivity measurement (TFP1)

and when it is not (TFP2). These results support the hypothesis that smaller farms

produce more per unit of land because they utilize more inputs per unit of land.



Table 6-2.
Farm Size/Gross Output Value per Hectare/Total Factor Productivity


Average Gross Average Total Average Total
Farm Size Output Value per Factor Factor
Hectare sucress) Productivity 11 Productivity 22

< 2 ha. 5,389,009 1.53 8.57
2 3 ha. 5,314,275 8.21 14.42
3.25 4.75 ha. 4,476,597 16.16 22.36
5 7.5 ha. 4,277,224 30.41 91.56
> 7.5 ha. 3,294,366 104.18 390.19
All Farms 4,716,172 31.44 102.76

'Cost of family labor calculated at average daily rate of pay
2Cost of family labor equals zero

Exchange rate = 7,400 sucres/1 U.S. dollar (February, 1999)



Smaller farms do utilize substantially more inputs per unit of land than larger

farms (Tables 6-3 and 6-4). The average value of non-labor inputs per hectare on the

smallest farms is 2,245,586 sucres, while it is only 584,978 sucres on the largest farms.

In other words, the smallest farms utilize approximately four times the amount of non-

labor inputs per hectare that the largest farms utilize. This difference is even greater

when labor is included among the inputs. The average value of labor and non-labor

inputs per hectare on the smallest farms is 8,448,643 sucres, while it is 1,474,493 sucres

on the largest farms, a difference of approximately six times. Evidently, smaller farms











have a higher gross output value per hectare due to a more intensive use of inputs per

hectare, but this intensive use of inputs leads to a lower total factor productivity.

Average gross crop output value per hectare of cropland also decreases as farm

size increases. Average gross crop output value per hectare of cropland is equal to the

gross output value for crops divided by the number of hectares of land used in crop

production. This measurement differs from gross output value per hectare in that it

excludes dairy revenue from the measure of gross output value and excludes idle land

and pasture land from the land measurement. Idle land and pasture land are still

included in the farm size measurement which determines to which size group a farm

belongs. Average gross output value per hectare of cropland by farm size is given in

Table 6-5.

Table 6-5 also lists average total factor productivity 1 and average total factor

productivity 2 for crop production by farm size. These measurements of total factor

productivity differ from those given in Table 6-2 in that they include only revenue from

the sale of crops and only costs of inputs used in crop production. In contrast to the

positive relationship between farm size and total factor productivity indicated in Table

6-2, no clear relationship exists between farm size and total factor productivity for crop

production.

Smaller farms utilize more inputs for crop production per unit of cropland than

larger farms do, but the inverse relationship between size of farm and crop input use per

hectare of cropland is not as strong as that between size of farm and total input use per

hectare of farmland (compare Tables 6-3 and 6-4 with Tables 6-6 and 6-7). The

average value of non-labor inputs used in crop production per hectare of cropland on the























G 42N t-OV 00

on M -
*T P"< N V O




en0 0
nk




0 eqoO







-8 a 0000
(g OS00 eI 00

m 00oo
14 00 CS W K0


So tn cn t-^ C) W
oo (S ot ch C'4 0


N Io( t 00 4n

t-s o' o C14


>5 o(- o ON 5 CL| 4n (S 4 W) Q *
E



i5 tr; 0 n SO l en, 00,


0 so *r> m as 0 en o
PC a\ p~ 00 t--o
oi- ei~






M -'- 00 tn r- N

~ e %0 enWn n
en0 00 %0 QS(




CB 14s en rtn ^*

NC1
I-n
vi t^ 0 tit~ (





t~- en cw '- '-^ r



U s<

I E.


N a s N
'B S B






mt 0 0 ^t o 0

IWos Cr> s
r 00NC


04
U,
U C
'"U
o -
.0
'U

U


*si
a0 oU 1-- 00





W)C1


en en 00 t 10 en 0-
Z 't en 4 .4


5 o >n ON It m *
:3 0 tn''- 'Rt t- S
- -j o ~ r o 1 -r







0 0
'- jB ^t 0en q S S ^s^ ^' "1'v'1


Ist 0 00 r- s0 t~'
cs4 enC (4 C41 cs











Table 6-5.
Farm Size/Gross Output Value per Hectare/Total Factor Productivity,
Crops Only


Average Gross Crop Average Total Factor Average Total Factor

Farm Size Output Value per Productivity for Crop Productivity for Crop
Hectare of Crops Productioni Production22
sucress)

< 2 ha. 7,544,611 1.91 8.07
2- 3 ha. 8,740,001 7.23 9.12
3.25 4.75 ha. 4,290,641 4.80 6.67
5 7.5 ha. 4,661,204 4.66 7.04
> 7.5 ha. 2,884,840 1.88 2.52
All Farms 5,759,909 4.48 6.77

ICost of family labor calculated at average daily rate of pay
2Cost of family labor equals zero


smallest farms is 2,245,586 sucres, versus 995,364 sucres on the largest farms. The

smallest farms utilize approximately 2.3 times the amount of non-labor inputs per

hectare of cropland that the largest farms utilize. The average value of labor and non-

labor inputs per hectare of cropland on the smallest farms is 8,448,643 sucres, while it is

2,169,340 sucres on the largest farms, a difference of approximately four times. This

more intensive use of inputs in crop production on the smaller farms evidently leads to a

higher gross crop output value per hectare of cropland on those farms. However, the

input use is not so high as to lead to a lower total factor productivity on smaller farms.

The results of estimating the six equations listed at the beginning of this chapter

essentially support the above analysis. All regressions are corrected for

heteroskedasticity using White's method.





















W ) el "t I0*9
Qs el < t,- tn S
V434

-
a Z o 00
0 nM -n 00
HN




ON 3 00ase0n




00 0
00 r- ID 00 pt r-







ao o



ro- C4 o o%
5 000 Ncr Od-
2 N Ot C> ONi 0




.4 n Io.







8 t oo ^ s
oN 'S*T 00 C; o
4W '0 ON







i en O h'4
*a -i r>.


.~.



U G ^N ^ f


v AC
___ ___ rN _


0 ca
si
"a a


C
.0
U

o vi
0


0y 00 tt \C
19t NS NS en


I I


I0NN 0I--


4 I-


e I '.0e e


\.0

ZO0
00'.
* 0'g


0



5.


0

a
.i



0





o

1

a

0
z.


WTn e 0 tn Of
W~ ot f I en. 01r
""

Soo tr. (N ('4 (N
0N ON N
go Bs 00,^ ^ t
O ^ o' rcs cW s s
Ei> Z


-A._________










The estimation of equation one, which analyzes the relationship between farm

size and land productivity, confirms the existence of a significant inverse relationship.

However, when variables on land quality and farmer age and education are added to the

analysis in equation two, the relationship is no longer significant. This lends support to

those who argue that many past studies on the farm size-productivity relationship may be

subject to mis-specification, and should include additional variables (such as land quality

indicators) in the analysis. The results of estimating equations one and two are given in

Tables 6-8 and 6-9.

The estimation of equations one and two with gross output value of crops per

hectare of cropland as the dependent variable gives essentially the same results. A

significant inverse relationship is found between farm size and land productivity when

the partial regression is estimated, but the relationship is no longer significant at the five-

or ten-percent level when variables on land quality and farmer age and education are

added to the analysis. It should be noted however, that the P-value for the farm size

variable is .109 when equation two is estimated in this manner, indicating that an inverse

relationship exists between farm size and land productivity which is nearly significant at

the ten-percent level. The results of estimating equations one and two using gross output

value of crops per hectare of cropland are given in Tables 6-10 and 6-11.

The results of the estimation of equations three and four demonstrate a positive

relationship between farm size and total factor productivity, when family labor is valued

at the wage rate for hired labor. This relationship is significant at the one percent level











Table 6-8 Partial Regression with Gross Output as Dependent Variable

Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error
(Constant) 15.54** 0.32 49.26 [.000]
LOP -0.43** 0.21 -2.05 [.048]

**Signficant at 5% level or better

Adjusted R2 = .083


Table 6-9 Full Regression with Gross Output as Dependent Variable

Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error


(Constant)
LOP
HW
TENPART
TENMIX
ED5
ED7
LESS20
LESS60
PLUS60
ARENOSA
BARROSA
CANGAHUA
LOCI
LOC2
PEND
MIXSLOPE
DAYSOFF


16.91**
-0.37
-0.57**
-0.25
0.21
-0.06
-0.08
-0.58
-0.2
-0.71
0.18
0.2
0.45
-0.25
-0.67
0.02
0.24
-0.03


0.66
0.27
0.17
0.57
0.34
0.29
0.58
0.60
0.45
0.58
0.55
0.48
0.87
0.54
0.60
0.53
0.72
0.11


25.77
-1.37
-3.41
-0.44
0.63
-0.23
-0.14
-0.97
-0.45
-1.24
0.33
0.42
0.52
-0.46
-1.12
0.04
0.34
-0.32


[.000]
[.187]
[.003]
[.664]
[.536]
[.824]
[.893]
[.344]
[.657]
[.231]
[.747]
[.678]
[.607]
[.650]
[.275]
[.966]
[.737]
[.751]


**Significant at 5% level or better


Adjusted R2 = .021











Table 6-10 Partial Regression with Gross Output as Dependent Variable
(Crops Only)

Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error


(Constant)
LOP


15.84**
-0.76**


0.31
0.23


51.91
-3.27


[.000]
[.002]


**Significant at 5% level or better
Adjusted R2 = .156


Table 6-11 Full Regression with Gross Output as Dependent Variable
(Crops Only)


Variable

(Constant)
LOP
HW
TENPART
TENMIX
ED5
ED7
LESS20
LESS60
PLUS60
ARENOSA
BARROSA
CANGAHUA
LOCI
LOC2
PEND
MIXSLOPE
DAYSOFF


Estimated
Coefficient
16.69**
-0.54
-0.57**
-0.76
0.92*
-0.43
-1.88**
0.46
-1.17**
-1.26*
1.13
1.75*
0.69
-0.66
-0.84
0.55
1.21
0.03


Standard
Error
0.87
0.32
0.26
0.73
0.51
0.49
0.76
0.94
0.49
0.76
0.97
0.95
1.15
0.62
0.70
0.54
1.08
0.13


T-Statistic P-Value


19.12
-1.68
-2.23
-1.03
1.80
-0.89
-2.42
0.49
-2.40
-1.89
1.17
1.85
0.59
-1.07
-1.19
1.01
1.12
0.25


[.000]
[.109]
[.038]
[.314]
[.088]
[.384]
[.025]
[.631
[.027]
[.074]
[.257]
[.080]
[.559]
[.299]
[.247]
[.325]
[.279]
[.808]


**Significant at 5% level or better
*Significant at 10% level or better
Adjusted R2 =.163










for equation three and at the five percent level for equation four. The variable LESS20,

which represents the farmer operator being less than twenty years old, is also significant

at the one percent level in equation four. The coefficient of this variable is negative,

indicating an inverse relationship between this variable and total factor productivity. The

estimation of equations three and four with total factor productivity of crop production

replacing total factor productivity as the dependent variable yields different results. For

both equations three and four, no significant relationship is revealed between size of farm

and total factor productivity of crop production. The results of estimating equations three

and four are given in Tables 6-12 through 6-15.


Table 6-12 Partial Regression with TFP1 as Dependent Variable

Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error
(Constant) 0.19 0.38 0.51 [.613]
LOP 1.3** 0.34 3.78 [.001]

**Significant at 5 percent level or better
Adjusted R2 = .270

Equations five and six use the measurement of total factor productivity which

values family labor at zero as their independent variable. A positive relationship between

farm size and total factor productivity which is significant at the one percent level exists

for equation five (Table 6-16). For equation six, which includes additional independent

variables, a positive relationship which is significant at the five percent level exists

between farm size and total factor productivity (Table 6-17). In this equation, the ED7

and LESS20 variables are significant at the one percent level. There











Table 6-13 Full Regression with TFP1 as Dependent Variable

Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error


(Constant)
LOP
HW
TENPART
TENMIX
ED5
ED7
LESS20
LESS60
PLUS60
ARENOSA
BARROSA
CANGAHUA
LOCI
LOC2
PEND
MIXSLOPE
DAYSOFF


0.45
1.10**
-0.14
0.65
-0.67
-0.96
1.84*
-3.18**
1.21
0.90
0.52
-0.75
1.88
-1.24*
-0.69
-0.27
0.75
0.01


1.38
0.47
0.24
0.61
0.86
0.74
0.91
0.80
0.81
0.92
1.12
1.02
1.40
0.78
0.79
0.70
1.26
0.14


0.32
2.35
-0.60
1.07
-0.78
-1.30
2.01
-3.96
1.49
0.98
0.47
-0.74
1.34
-1.83
-0.87
-0.38
0.59
0.06


[.750]
[.030]
[.558]
[.298]
[.443]
[.210]
[.059]
[.001]
[.153]
[.338]
[.647]
[.471]
[.195]
[.083]
[.393]
[.705]
[.561]
[.954]


**Significant at 5 percent level or better
*Significant at 10 percent level
Adjusted R2 = .353


Table 6-14 Partial Regression with TFP1
(Crops Only)


as Dependent Variable


Estimated Standard
Variable Estimated Standard T-Statistic P-Value
Coefficient Error


(Constant)
LOP


1.08
-0.38


0.65
0.51


1.66
-0.75


[.106]
[.459]


Adjusted R2 = .001