7 $.01.“ fix} . . . ,
in ".12 , . ,
5 .

 

." ,

.5 . . .

my? . . . . . v. .. q.
Lfimfi :. , . . . Em. .. _
L

w

 

.. in“. r .. a.
_. ,aggfifi A . mi.
.3. . a fi...w¥¢ . z a

,. fifi..m.$ :7. $5.... yr r».

ya. fififiggr
v.

I

 

ash . , . , 1.“. n .
”my”??? . , 35%....“
.3. .3: 2 . . .
~ 05:2: .\ . .V . . . . .
n..ޤ:.3.. . . . . . ., .
33...... 3.1.3 . , , . .

.. 9953; .. . . w a...
$5.. .3 . . 2.2."...
it. ..

‘3‘}. 2”.

.. . . 3.1..

.41; Ian)... ..
:3... I...

:1 .......h£.. H

 

1.1.5. a
:3... z... . H.
. _ . 7.4....
”1.2.9

. .1 .

. . . 3:
3.3.1.3 .
\ x.

a»...

:3 u
o
c

L. r ..
xvi“...
. . >.

:1...

P}...

{’9

 

 

 

.
.. in. .
.. . .. 3 .
.23... y 2:... .
- : ..3.-.,...-s..-

saai... .. . , 1;.
$.}§¥E§%§k . .. %§%§%§$.§%
I.

tire . . ~

 

f? I

 

 

THEE-it-

200!

This is to certify that the

dissertation entitled

Sustainable Agriculture and Deforestation
in the Peruvian Amazon

presented by

David Russell Yanggen

has been accepted towards fulfillment
of the requirements for

 

 

 

 

 

Ph.D. degree in Agricultural Economics
Major professor
Date Augus c 23 , 2000:"
MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
Michigan State
University

 

 

 

.—__.— --—- ~—.-_‘ ‘— —.

v— F4 -4 # v F i - fir‘. ....

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

 

DATE DUE DATE DUE ' DATE DUE

 

[JUL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

woo enuncroawuepespu

 

SUSTAINABLE AGRICULTURE AND DEF ORESTATION
IN THE PERUVIAN AMAZON

By

David Russell Yanggen

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements I
for the degree of
DOCTOR OF PHILOSOPHY
Department of Agricultural Economics

2000

 

 

5‘» ~
ug...
5

.49’5—
U»\.

I
E”,
M.”

v
'1

.L‘_

“tam ..
5m
H‘I- ..
u‘ld'll '
‘9‘

c0” .3

li‘

 

ABSTRACT

SUSTAINABLE AGRICULTURE AND DEFORESTATION
IN THE PERUVIAN AMAZON

By

David Russell Yanggen

Agriculture accounts for over 80% of deforestation in the Peruvian Amazon.
However, agriculture in the region is not monolithic in nature. Farmers produce different
products and use different production technologies. This thesis analyzes how different
agricultural technologies and outputs affect deforestation.

This thesis uses econometric modeling based on a profit function approach to

fa . ‘ J‘,.‘“.- ““ \
‘\’ I .2 ,

'/ analyze the relationship between agriculture and deforestatiorf There are two levels to
this analysis. The first level analyzes the relationship between agricultural production
determinants (prices and fixed factors) and farmers’ choices concerning production
technologies and agricultural outputs. These econometric models use the methods of
ordinary least squares (OLS) and seemingly unrelated regression equations (SURE). The
second level analyzes how these input and output decisions affect deforestation. These
models are generally recursive in nature but use a two stage least squares approach to
correct for simultaneity when detected by the Hausman specification test.

This thesis found that in an environment where land is abundant relative to labor,
farmer adoption of technologies that enhance land productivity is dependent upon
whether or not these technologies increase returns to labor. The design and promotion of
improved technologies should therefore focus on combining enhanced land productivity

at . . . .
With mcreasmg returns to labor. However, if increased returns to labor are primarily a

 

'3‘"
io-

.n "

-)l.‘

 

4»-

(n
l

u...-

t...‘

result of decreasing per hectare labor requirements, then freed up labor is likely to
increase deforestationfit is therefore necessary to find production practices that “absorb”
labor away from deforestation by increasing labor requirements per hectare while at the
same time increasing returns to labor.

This thesis also found that the use of capital inputs that increase land productivity
diminishes farmer clearing of primary foresti'VIfhe difficulty is that in most cases of
abundant land and scarce capital it is more economical for farmers to use extensive slash-
and-bum practices than to adopt these inputésifJResearch should therefore give priority to
using economic analysis to identify crop and capital input combinations that are both
affordable and profitable for farmers. In addition, capital in the form of credit was found
to have a significant impact (both positive and negative) on forest clearing. Credit should
therefore be tied to production practices that reduce deforestation.

This research also found that agricultural production practices can have different
impacts on different types of deforestation (total, primary and secondary forest clearing).
It is therefore critical for research to take a disaggregated approach to analyzing the
impacts of agricultural on deforestation.

A final finding of this research is a clear evolutionary trend in land use patterns.
Farmers start out clearing primary forest for annual crop production at the forest margins.

In older settlement areas farmers progressively reduce annual crop production and clear

degraded land in secondary forest areas for ranching.

DEDICATION

I dedicate this thesis to my grandmother Helen Yanggen
who passed away at age 100 while I was completing my

thesis research in the Peruvian Amazon.

For her infallible positive attitude, her kindness and
generosity, her desire to see her family happy, her
independent free spirit and her many other qualities I look

to her as one of the great role models in my life.

I know she is now in heaven with a sign on her door that

says “no uninvited visitors” and is reciting by heart verses

of poetry learned in high school as she did at age 99.

iv

.

lav- "ah“

AHA» l-‘VE‘ "
~

r3 [HYQW .

I
u. .1.‘-c5~‘

In
1,1,"...
IUU I llhtooh‘

.“'I ~
LGAI ' eTEII‘D
'

0"“:“-" I" .9.

k’f‘u-‘LLL.

TC mom

i ‘r «‘"v

70 Seen 3

De . L
lumvfrs f

ACKNOWLEDGMENTS

I am grateful to the following institutions for the financial support of this thesis:

The International Center for Research in Agroforestry (ICRAF)
The Inter—American Foundation (IAF)
The United States Agency for International Development (USAID)

The International Food Policy Research Institute (IFPRI)

I am also grateful to the Department of Agricultural Economics at Michigan State
University (MSU) and the USAID Food Security 11 Project for providing me with work

opportunities to support the studies that led to this thesis.

And to the following individuals:

To Thomas Reardon, my advisor, I couldn’t have hoped to have the luck to study under

someone from whom I could learn so much and to whom I owe so much for this

incredible dissertation Opportunity.

To Scott Swinton, Stephen Vosti, Sandra Batie, and Larry Leefers, my committee

members for their guidance and encouragement

 

.‘-0

«hi: :3 A‘-

.‘
-O '3'- -
. A
\- nah—tu-

‘ n '
1.334...- .1.
J§»‘.~ .1"

‘

.,.

In —
“sh-'3 Pa
»

H...
‘ ‘0'...
“use

\

 

To Dale Bandy, John Weber, Wagner Guzman, Carmen Sotelo, Jenny Paz, Leopoldo
Rocca, Julio Alegre, Ricardo Labarta, Anne-Marie Izac, Jackeline Barbaran, and other

colleagues from ICRAF as well as collaborators from CIAT, CIFOR, IN IA, and the

Ucayali Ministry of Agriculture for all their many forms of support

To Valerie Kelly, my research assistantship supervisor, who helped provide me my single
most important learning experience while in residence at MSU working on applied issues

of natural resource management related to agriculture production
To Adolfo Figueroa and Jackeline Velazco of the Universidad Ponticifia Catolica de Peru
for welcoming me at their university and providing me with intellectual support and

feedback on this dissertation.

To Jorge Alva and Angelita Estrella for their superior help in data entry, cleaning, and

analysis.

To Angel Rengifo, Eduardo Lopez, Marcos Tenazoa, Jorge Alva, Renan Shahuano,

Carolina Rengifo, a “dream team” of enumerators

To the Peruvian farmers who participated in the surveys and presurveys for their time,

patience, and insights

vi

Tc m‘. plft‘.’

3m)" Q or
mt‘...s-- ““

. an, "-1
'3‘ MI\ m»...

‘I ' .
- . ~ .- 1
§-‘ ‘k’ 7

. 4 '
L‘...-os ea Usfi‘

T"JJ"V:H~
AL 5¢aab.1
-

Ira" ‘ .; ”l"

. l
~5t.|“~“|\ ‘

To my parents Douglas and Victoria Yanggen for their support during the difficult

moments and my brother Steve and sister Julie who I’m gratefiil to have as siblings

To the many students at the MSU Department of Agricultural Economics who generously

collaborated so that we could succeed together

To Jeffrey Zoeller for going along with my folly

To all the professors, students, friends and colleagues along the way from the University
of Wisconsin-Madison and the Land Tenure Center, L’Institut d’Etudes Politiques-Aix-
en Provence, L’Université de Geneve-L’lnstitut Universitaire d’Etudes du
Developpement, and Peace Corps-Mali who inspired me to follow a career in

international development

And last but not least to farmers such as Zoumana Sanago, Namu Koulibaly, Nvah
Sanago, and Abdoulay Koulibaly who have my complete admiration for their intelligence
and ingenuity. They taught me first hand a great deal about the technical and economic
aspects of agriculture and confirmed my belief in the need to always speak with and
listen to farmers to learn from their wealth of knowledge-a principal that has served me

well

To all of you: Thank you - Gracias — I ni che — Merci

vii

LIST Of I?
UST 0? El

[\TRDDLI

M .;.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................ x
LIST OF FIGURES ........................................................................... xi
INTRODUCTION ............................................................................ 1
Introduction ........................................................................... 1
Overview of Dissertation ............................................................ 1
Thesis Structure ...................................................................... 4
Thesis Data ............................................................................ 5
Summary .............................................................................. 6
ESSAY l
A HOUSEHOLD-LEVEL ANALYSIS OF THE IMPACT OF
AGRICULTURE ON DEFORESTATION IN THE PERUVIAN AMAZON ....... 7
Introduction ........................................................................... 8
Characteristics of the Pucallpa Research Zone ................................... 10
Overview ..................................................................... 10
Evolution of Farm Land Use Patterns .................................... 12
Model .................................................................................. 14
Theoretical Model ........................................................... 14
Econometric Models ........................................................ 16
Results ................................................................................. 20
Output Supply Variables ................................................... 21
Input/Technology Demand Variables .................................... 24
General Conditioning Variables .......................................... 28
Cross-Cutting Results ...................................................... 30
Conclusions and Policy Recommendations ....................................... 32
References ............................................................................. 37
ESSAY 2

KUDZU-IMPROVED FALLOWS IN THE PERUVIAN AMAZON.
A CASE STUDY OF THE IMPACT OF TECHNOLOGICAL CHANGE

ON DEFORESTATION ..................................................................... 42
Introduction ........................................................................... 43
The Pucallpa Research Zone and Kudzu-Improved Fallows ................... 44
Characterization of the Pucallpa Peru Study Zone ..................... 44
Description of Kudzu-Improved Fallows ................................ 46

Comparative Economic Analysis: Kudzu-Improved Fallows vs.
Traditional Slash-and-Burn Agriculture .......................................... 47
Total Land Use ............................................................... 48
Labor Use ..................................................................... 50

viii

 

 

1 fl
5

an
II»

V L'i

Yields ......................................................................... 52

Hypotheses Concerning Kudzu Fallows and Deforestation ..................... 52
Theoretical and Empirical Models ................................................. 53
Theoretical Model ........................................................... 53
Econometric Model ......................................................... 56
Model Results and Interpretation ................................................... 57
Kudzu-Improved Fallow Adoption......................: ................ S7

The Impact of Kudzu-Improved Fallows on Deforestation............59
Conclusions and Implications for Policy and Technology Development. . ...62

References ............................................................................. 67
Appendix 1: Variable Description ................................................. 69
ESSAY 3
THE ADOPTION OF IMPROVED PASTURES AND THEIR IMPACT ON
DEFORESTATION IN THE PERUVIAN AMAZON .................................. 71
Introduction ........................................................................... 72
Environmental and Economic Impacts of Livestock ................... 72
Sustainable Livestock Production in the Amazon? ..................... 73
Paper's Objectives and Organization ..................................... 75
Research Data ............................................................... 76
Characteristics of the Research Zone .............................................. 77
The Evolution of Land Use Patterns in the Pucallpa Zone ..................... 79
Socio-Economic Characteristics of Brachiaria-Improved Pastures ........... 81
The Productivity of Brachiaria-Improved Pastures .................... 82
The Costs of Brachiaria-Improved Pastures ............................ 83
Comparative Labor Use and Financial Returns: Livestock,
Annuals, and Perennials .................................................... 86
Model .................................................................................. 88
Theoretical Model ........................................................... 88
Econometric Models ........................................................ 89
Results ................................................................................. 93
The Impact of Pastures and Livestock on Deforestation ............... 93
Pasture Use and Livestock Production ................................... 94
Correlation Analysis Results ............................................... 97
Conclusions and Policy Recommendations ....................................... 99
References ............................................................................. 104
CONCLUSIONS .............................................................................. 109
APPENDIX A: Main Household Survey ................................................... 121
APPENDIX B: Kudzu-Improved Fallow Survey ......................................... 139
APPENDIX C: Brachiaria-Improved Pasture Survey ................................... 147
APPENDIX D: Map of the Pucallpa Research Zone ..................................... 153

ix

0:33} LE

Eggatrorb "1

APPENDI)

APPENDIX E: Comparison of Input and Output Regression Models Using
Ordinary Least Squares (OLS) and Seemingly Unrelated Regression
Equations (SURE) ............................................................................ 155

APPENDIX F: Results of the Hausman Specification Test and Total
Deforestation Model Using Ordinary Least Squares (OLS) and Two Stage
Least Squares .................................................................................. 158

..i .5
Y IL
.11. AU
.\. ”H“
.\

C l....

Y)
1'

75'

In»;-

a .. «i.
5.5 alw
.n W“
.2. Ti

‘1 ’.‘.— 6 Y
e a» Bk Int.
v I u . . . a.\.
-a fiane Q.
Tl Pr.»

'DZE

‘i

K

ACASE

9‘ Db 05
so .ad .3.
..u .1. .4.
TIA T95 TIA

‘l‘ .‘v
in t
.H .s .h s
A‘- a u
.11.; fill

TL \
v HINIA
.44 re. Fd
.\c r.
\~ ; FL
.3. m D

LIST OF TABLES

ESSAY 1
A HOUSEHOLD-LEVEL ANALYSIS OF THE IMPACT OF
AGRICULTURE ON DEF ORESTATION IN THE PERUVIAN AMAZON

Table 1: Input and Output Regression Specifications, Variable Averages

and Standard Deviations ........................................................... 18
Table 2: Deforestation Regression Model Specifications ................................ 19
Table 3: Previously Undefined Variables ................................................. 19
Table 4: Deforestation Regression Results ................................................ 21
Table 5: Output Regression Results ......................................................... 22
Table 6: Input/'1‘ echnology Regression Results ........................................... 22
ESSAY 2

KUDZU-IMPROVED FALLOWS IN THE PERUVIAN AMAZON.
A CASE STUDY OF THE IMPACT OF TECHNOLOGICAL CHANGE
ON DEF ORESTATION

Table 1: Comparative Analysis of Total Land Use for Differing Fallow Systems. . .50

Table 2: Comparative Use of Labor for Land Clearing .................................. 51
Table 3: Comparative Use of Labor for Weeding ........................................ 51
Table 4: Comparative Yields ................................................................ 52
Table 5: Kudzu Fallow Adoption and Deforestation Regression Results ............. 6O
ESSAY 3

THE ADOPTION OF IMPROVED PASTURES AND THEIR IMPACT ON
DEFORESTATION IN THE PERUVIAN AMAZON

Table 1: Comparative Labor Use and Returns to Different Land Uses ................ 87
Table 2: Regression Variables ............................................................... 90
Table 3: Deforestation Regression Results ................................................. 95
Table 4: Livestock and Pasture Adoption Regression Result ........................... 95
Table 5: Correlation Analysis of the Impact of Livestock and Pastures

on Deforestation ..................................................................... 98

 

 

NRODL'C
FEEL? 1: C3

ESSAY 1

I;
i
L

I V

AHOY: 1
AEMCC I

0733?. l: Tr.

ESSAI'3
THEADOP
DEFOPESI
G:phl:Tt
Gwhlli
02513: La
Gngh4zP;
73;: 5: A\

GE;E6:II

LIST OF FIGURES

INTRODUCTION
Figure 1: Causal relationships leading to deforestation .................................. 3

ESSAY 1
A HOUSEHOLD-LEVEL ANALYSIS OF THE IMPACT OF
AGRICULTURE ON DEF ORESTATION IN THE PERUVIAN AMAZON

Graph 1: The Evolution of Land Use Patterns ............................................ 12
ESSAY 3

THE ADOPTION OF IMPROVED PASTURES AND THEIR IMPACT ON
DEF ORESTATION IN THE PERUVIAN AMAZON

Graph 1: The Evolution of Pasture Use in Pucallpa ..................................... 79
Graph 2: The Evolution of Land Use Patterns on Farmer Land Holdings . . . . 80
Graph 3: Land Use Conversion to Pasture ................................................ 81
Graph 4: Pasture Abandonment and Productivity Declines ............................. 82
Graph 5: Average Installation and Annual Maintenance Costs for Pastures .......... 84
Graph 6: The Costs of Improved and Traditional Pastures .............................. 86

xii

._.,...,,., c-
b»li‘-5|). L
I

”a". ‘1 .‘
.muut Ah“. ‘

' I‘Fma
i-‘unlgne

«‘4; W" '
a'iu BIA-ll!

P»?- 'v- o;
T" 1‘s H

INTRODUCTION

1. Introduction

This dissertation research is part of the Alternatives to Slash and Burn (ASB)
initiative of the Consultative Group of International Agricultural Research (CGIAR)
centers. Components of this initiative have taken place in various countries of Latin
America, Africa, and Asia. The purpose of this research effort is to find economically and
environmentally sustainable alternatives to the ecologically destructive practice of slash-
and-burn agriculture in the world’s humid tropical rainforests. Research results from each
site are pooled together to identify general trends concerning the impact of agriculture on
deforestation. This particular research took place in the lowland Amazon rainforest of
Peru in the zone surrounding the city of Pucallpa. The International Center for Research
in Agroforestry (ICRAF) was the principal sponsor providing financial and logistical
support including office facilities at the Ecoregional Center in Pucallpa.
2. Overview of Dissertation

This thesis provides a case study analysis of the impact of agriculture on
deforestation in the tropical rainforest environment surrounding Pucallpa Peru.
Agriculture accounts for over 80% of deforestation in the Peruvian Amazon. The
predominant agricultural production practice there is slash-and-burn agriculture. This
involves felling and burning an area of rainforest. Ashes provide a brief flush of nutrients
for crops planted by farmers. But infertile tropical soils cause farmers to abandon parcels
after just 1-2 years of cultivation. They then repeat the cycle of slashing and burning a

new area of forest (often fallow areas of secondary forest regrowth) to cultivate a new

‘-

.
"3‘. i '
21k“. I"
A

3:52;: c

l

1 I. .‘
5“; ‘~“
1.“- \bt .

.0
Q

'71?“

I"

b

. -
“0-,... y,
p.531) 4:
.

it: 3-5

--~. ‘

‘

h
.r»

sun

p. 5.1"

‘u

b.
I

I
‘ ’4

parcel. This continual shifting of areas in cultivation into areas of forest is what makes
slash-and-bum agriculture so ecologically destructive.

However, agriculture in the zone of Pucallpa is not monolithic in nature. Farmers
produce different products and use different production technologies. This thesis analyzes
how different agricultural technologies and outputs affect deforestation.

Taking a step further back, it is also necessary to understand the determinants of
farmer decisions about different outputs and inputs. In order to do so, this research uses a
profit function approach. Profits are defined as follows:

(1) it = pq - wx

where p is a vector of prices for outputs, q is a vector of outputs produced by the farmer,
w is a vector of input prices, and x is a vector of inputs. This approach allows the
identification of the optimal levels of input use and output production defined as follows:
(2) q* = ftp,w,2)

(3) X“ = KP,W.Z)

where z is a vector of fixed input factors (e. g., soil type, tenure security, education, etc.).

In sum, there are two levels to this analysis. First there is the causal relationship
between production determinants (prices and fixed factors) and farmers’ choices of
production technologies used and agricultural outputs produced. Next, there is the causal
relationship between these input and output decisions and their impact on deforestation.
These relationships are outlined in the schema represented in figure 1.

As represented in this diagram, agricultural production decisions about input use
and output production are the immediate causes of deforestation. Decisions, for example,

to plant more annual crops or to use fertilizers (which may allow greater reuse of

- ’3
3:373): ares

/\

Figure l:

(“1.
I“

. "V.
Hou...‘
‘ v

I
3
L~».:.

“m-
\..2

I
‘ a

b.) JMVV‘F
“‘5‘...

. 4
AAA.

‘
5‘ ~| "H
C‘“--L..:

‘31:.10 ‘5:
‘4-
‘}n-l '

previously cleared land) are what directly lead a farmer to decide to clear (or not) more

forest area. Production determinants such as a change in prices or variations in soil

 

 

 

 

 

 

 

 

 

Figure l: Causal relationships leading to deforestation.

quality do not directly cause deforestation. Rather, they lead to changes in production
decisions that in turn are the immediate causes of deforestation. These production
determinants are therefore indirect causes of deforestation. This conceptualization lays
the basis for a more coherent and rigorous analysis of the relationship between
agriculture and deforestation.

The principal methodological tool of this dissertation is econometric analysis. The
use of two sets of regression equations reflects the two levels of analysis described above.
In the first set of regressions models, production determinants (independent variables)
lead to farmer’s decisions concerning input use and output production (dependent
variables). The second set of regression models uses farmers’ input and output choices as
independent variables that lead to deforestation, the dependent variable. The structure of
this analysis permits a linkage from production determinants to production decisions

through to the final deforestation outcome.

’ 5..»

I V .
- .‘ '3'
. "Qit‘gltr‘

-l :59 ‘ ”Yi'ur
bioL‘ACETDJHAKIA.

' . 4 1.
:fGI' 8:“; a:
3. Tires
EC 3.
0‘ .I ’
Olifi 12% 01,
. ‘ .
curse. am:
P'“ ‘9? A ' J
4~uby JU.U‘]

93.1mm”, “

:‘hg-a
““fds’t Of T

This dissertation, however, does not rely on regression analysis alone. Extensive
use of descriptive statistics (e. g., the evolution of land use patterns over time, the returns
to land and labor of various production activities, and the costs and benefits of production
technologies) gives further insights into the relationship between agriculture and
deforestation. And, of course, logical argumentation based on a knowledge of economic
theory and agricultural processes is also incorporated.

3. Thesis Structure

The thesis begins with this introduction. This introduction provides a brief
overview of the dissertation’s objectives, methodology, and content. Three essays in a
journal article format then follow. The three essays are entitled as follows: “A
Household-Level Analysis of the Impact of Agriculture on Deforestation in the Peruvian
Amazon”, “Kudzu-Improved Fallows in the Peruvian Amazon: A Case Study of the
Impact of Technological Change on Deforestation”, and “The Adoption of Improved
Pastures and their Impact on Deforestation in the Peruvian Amazon”. A concluding
section synthesizes the key findings and lessons learned from this research. An appendix
section includes material such as the survey instrument, supplementary information
concerning the regression analysis, and a map of the zone.

The first essay, “A Household-Level Analysis of the Impact of Agriculture on
Deforestation in the Peruvian Amazon”, is the overview piece of the dissertation. It
analyzes the determinants of input use and output production and, in turn, the impact of
these production decisions on deforestation. The five input/technologies looked at are
improved fallows, improved pastures, natural pastures, hired labor, and capital inputs.

These are the most common production inputs and technologies used in the zone. Outputs

are grouped in
perennals and

This dé

24.131039. 11". r
p.131? Iorcs
ii mm’ ‘.-n
In: Lc"..r:‘l...:t

o: tact. up:

The se

a... , 4 :
.- a
bv’wku IT}

PM had 1

f0: ' .5 20

I
I?!“ L
..1‘ ”3‘9

\
V
E .-
“35¢ \‘a
P“ MAI-s.

ET. sL

'46

I“: .

4.

are grouped into the three major production categories found in the zone: annuals,
perennials and livestock.

This dissertation also divides deforestation into three categories: primary forest,
secondary forest, and total forest. Primary forests are areas that have never been felled
(although in many cases have been selectively logged), secondary forests are areas of
primary forest that have been cleared and have regenerated into forests, and total forest is
the combination of primary and secondary forest. Deforestation is defined as the hectares
of each type of forest felled by a farm household in 1998.

The second essay, “Kudzu-Improved Fallows in the Peruvian Amazon: A Case
Study of the Impact of Technological Change on Deforestation”, and the third essay,
“The Adoption of Improved Pastures and their Impact on Deforestation in the Peruvian
Amazon”, are both case studies of improved technologies that farmers have widely
adopted in the Pucallpa zone. National and international agricultural research centers
have had limited success in promoting the improved technologies they have developed
for the zone. The focus of these two essays is therefore on understanding i.) the factors
that have led to the successful adoption of these two technologies and ii.) the impact
these two technologies have had on deforestation. A better understanding of these two
questions can provide important insights for the development of sustainable technologies
and the design of policies to promote these technologies.

4. Thesis Data

Virtually all the data used in this thesis is primary data collected by the author in

three household surveys. A general survey of the socio-economic characteristics and

production practices (including deforestation) of 220 households provide the data for the

fiOZCmCmL .1

l',_

n '\F‘
:1?! 0k Ukua...

 

.
a'°;|v-P¢
At¥n AA.

‘ I
affi‘~rl:~‘ 1':
N “Luna ».
U

2. .L 3
ha BL: Pet‘Jxl

econometric analysis. Two mini-surveys concerning the use of kudzu-improved fallows
(n=24) and brachiaria-improved pastures (n=3 8) provide information on the costs and
benefits of these technologies. The general survey took place in the zone surrounding the
city of Pucallpa. This included a stratified sample of farm households based on their
distance to Pucallpa and their type of infrastructure access. Farm households were
sampled along the main paved highway, along the three principal secondary gravel roads,
along numerous dirt feeder roads, and along the three principal rivers in the zone.
5. Summary

This research is part of a global initiative that seeks to identify sustainable

alternatives to the environmentally destructive practice of slash-and-burn
agriculture. This research analyzes the determinants of agricultural production practices
in the Peruvian Amazon and, in turn, the impact of these production practices on
deforestation. In addition, it provides two case study analyses of the adoption of
improved technologies in the zone and their impact on deforestation. It is my hope that
these research results will provide information useful to researchers in their attempts to
design sustainable agricultural technologies as well as to policy makers in their efforts to

promote sustainable agricultural practices.

rm.

A HOUSEHOLD-LEVEL ANALYSIS OF THE

IMPACT OF AGRICULTURE ON DEF ORESTATION

IN THE PERUVIAN AMAZON

ESSAY NUMBER 1

1. lntri

ii'orl
0:: to m 0 1
105166?) 3r

ifTTiiCfi 15 L

-9, «01‘ ..
,rgu I
by. Hussy s

Ffifi‘arfi- -~
“.14". 51 11$ ‘

‘ v .

"'an A

P— sSn ‘3; E.
5.

I .l-
“~.. .. .

(1:1; 9\'2‘!
0».&.~

LA

I I
N .‘ <u.

.‘ “We"
Lick] t ‘.:‘.‘

Isa..-

1 '3‘-
vacuum?“
5".‘369 " "
fsnbg.‘k [1 L
cci‘,H. ‘

L..J‘,35_ I

‘ A
\LU,2:I:L
u

1. Introduction

Worldwide tropical closed forest deforestation is taking place at an alarming rate.
One to two percent of the world’s total rainforest areas, totaling 10-20 million hectares, is
lost every year (Ehui, Hertel and Preckel, 1990). In Peru, 59 percent of the national
territory is covered by Amazon rainforest. 261,000 hectares a year are being deforested.
In 1985, 7.5 percent of the total rainforest area had been deforested. Deforestation is
estimated to reach 12.7 percent by the year 2000 (INRENA, 1995).

Deforestation of the Amazon, the world’s largest rainforest, has led to growing
concerns about possible environmental impacts. The Amazon is believed to mitigate
potential global warming by sequestering carbon from the atmosphere. Furthermore,
deforestation in this region, which has one of the world’s highest concentrations of
biodiversity, threatens the existence of many plant and animal species. Other
consequences of deforestation include increased flooding, erosion, and dam siltation.

Researchers estimate that shifting cultivators are responsible for between 45
percent (UNEP, 1992) and 60 percent (Myers, 1992) of deforestation in developing
countries. In Peru, research puts this figure even higher at 80 percent of total
deforestation (Razzetto, 1997). The primary reason farmers slash and burn forest is not
to clear land but rather to provide nutrients for crops. Most nutrients are contained in the
aboveground biomass and not in the soil (Hecht, 1992). Burning the forest vegetation
provides a brief flush of nutrient to crops. Within a few years, yield declines typically
force farmers to abandon a parcel and clear another area of forest (Sanchez, 1976).

While initial studies of the Amazon concluded that most soils would not permit

continuous cultivation (McNeil, 1967; Goodland and Irwin, 1975), later studies showed

. 4' n9 ' I"
guitar} 3:1 '

Smart 195 I

 

‘ _ i Q *\
CLO“CITI‘A\ \ ”In
H cuss

WI-ACD «A '0-
”his“ ~05.)

.
4' o 4'

. ‘2‘ 1"!" .
ykkdsuklt. \r

, .
“-3“ -
Anne-.EL'II, AK
5

. .
‘3’ 9 -9-9.
.. “ °~ . ~01

-vt.o~t|
v
“$.54 .
I ‘6 9L.)
5 any

{I 14"“ u

H ‘;--qu
{robes Lion i

1“"
U‘
v
ra-
‘~‘\‘."\“ _
v. ‘.:
'3‘

sedentary agriculture was feasible given appropriate soil amendments (Cochrane and
Sanchez, 1981). Nevertheless, the low level of adoption of technological alternatives to
slash-and-bum agriculture indicates this agronomic feasibility has not translated into
economic viability for most farmers.

Households make agricultural production decisions based on calculations of
private costs and benefits (Reardon and Vosti, 1992). And household agricultural
production decisions are the immediate causes of deforestation (Kaimowitz and
Angelson, 1998). However, not all production systems have the same impact on
deforestation. Farmers produce different outputs and use different production practices to
produce them. It is therefore important to understand what determines farmers’
production decisions and how these decisions affect deforestation.

An important body of literature has focused on the national- and regional-level
causes of deforestation (e.g., Barbier and Burgess, 1996; Southgate, Sierra, and Brown,
1991; Deacon, 1995, Rudal, 1989; Southgate, 1994). Others have analyzed the potential
profitability of more intensive and/or sustainable production practices over slash-and-
burn agriculture (e.g., Peters, Bentry, and Mendelsohn, 1989; De Almeida and Uhl, 1995;
Tonolia and Uhl, 1995; Matteos and Uhl, 1994). Still others have focused on general
household determinants of deforestation (Godoy et a1. 1997; Pichon, 1997).

Existing research, however, does not draw the links between the determinants of
household input and output choices and, in turn, how these household production
decisions effect deforestation. Econometric analysis used in this article first models the
determinants of household-level output supply and input demand and then models the

impact of these production decisions on deforestation. This essay fills an important gap in

the literature by systematically examining the links between household-level production
determinants, production decisions, and deforestation.

The rest of this essay proceeds as follows. Section 2 gives an overview of the key
characteristics of the research zone. Section 2 includes a subsection describing the
evolution of on-farm land use patterns over time. Section 3 presents the modeling
approach. This section includes a theoretical discussion of input technology use and
output production based on a profit function approach. This theoretical discussion is
linked to an empirical econometric model that examines the determinants of production
decisions and how these decisions effect deforestation. Section four analyzes the
econometric model results. The final section presents the conclusions and policy
implications of this paper’s analysis.

2. Characteristics of the Pucallpa Research Zone
2.1 Overview

The research study zone is located in the central lowland rainforest area
surrounding Pucallpa, Peru. Pucallpa’s population of approximately 200,000 makes it the
second largest city in the Peruvian Amazon. Pucallpa’s economy is relatively dynamic. It
is the capital of the region of Ucayali, the largest producer of timber in the country. Oil
and gas exploitation have also expanded greatly in recent years. In addition, it currently
has the best road connection out of the jungle to the coast and capital city of Lima.

Rapid migration has made the Pucallpa zone the fasting growing area in the
jungle. This migration has resulted from the push factors of poverty and land scarcity in
other parts of the country (Riesco, 1993) and the pull factors of cheap land made

accessible by timber and gas feeder roads into the jungle, job opportunities, and until

10

 

1‘33er—

. ABE) A A\"Lfi5¢

1"“ .?.I".'€ 1

3L. .1. .1 t.“
.

K‘-
_ .
t ' w
I
a
u.
:A-
~.
. ‘i

recently coca production (Labarta, 1998). Immigrants to the zone are typically poor in
capital resources for farm-level investment. Rural population density remains low, around
7 persons per kmz.

Agriculture in the Amazon confronts a competitive disadvantage due to jungle
roads that are impassable during much of the rainy season and the need to traverse the
Andes Mountains to commercialize goods outside the region. Import substitution
industrialization policies over the past several decades have also been unfavorable to
agriculture in the Amazon and elsewhere in Peru. Liberalization since the early 1990’s
has partially rectified this urban industrial bias but has greatly reduced credit available to
farmers and has failed to provide the technical, infrastructural, and marketing support
needed to modernize agricultural production (Escobal, 1999; Palomino, 1993). Political
violence in the latter half of the 1980’s and early 1990’s further undermined agricultural
development in the Amazon region.

Low soil fertility combined with high levels of acidity and aluminum content
further limit agricultural development in the Amazon. Farmer use of capital inputs such
as fertilizers, pesticides, and plows is minimal. A median farm is 30 hectares of which
primary forest constitutes 31 percent, secondary forest fallows 30 percent, pasture 25
percent, annual crops 10 percent and perennials 4 percent.

Agricultural production in the Pucallpa zone by and large remains mired in slash-
and-burn agricultural production of semi-subsistence crops, principally maize, rice,
cassava, and plantains. In Spite of low prices, virtually all farmers market a substantial
proportion of these crops to meet cash needs. Cash crop development, in particular

perennials such as rubber, cacao, black pepper, and achiote, has seen a series of booms

ll

. L..: 510W

3
s:
‘5

 

(tr

maemilor.

in sun:

A.
1

airmail} it:
1.1 I) olu
A to:

aren'- .. '
runwdmf. h.

g

p

Vawo‘ u

was it- 2::
~

n ”81
..
E :-

Efllfi “in
\- ~&r‘Aa

\

0.60
«0.50

X

:0.“

a h
3 0.3r
'0

20.2:

-.I
0.1

or:

and busts and remains of limited importance‘. Livestock production is important and
growing slowly but is hindered by a lack of dairy processing facilities and high
transportation costs out of the region.

In sum, migrants coming to Pucallpa are extremely poor. They spontaneously
settle areas of abundant land recently opened up by gas, oil, and timber extraction.
Agriculture remains stagnant due to low prices, high marketing costs, and limited
government support. Capital input use is typically unaffordable for farmers. Farmers are
essentially left with their labor to cultivate abundant land.

2.2 Evolution of Farm Land Use Patterns

A comparison of farms based upon the amount of time they have been in
production helps trace the evolution of land use patterns over time. Graph 1 groups
farms2 together into four categories based on the number of years in production. This

grouping gives insights into farmer production strategies over time.

 

 

 

 

 

 

 

 

0.60 +Primary
Fore t
-0-50 ‘ 8
°\. \ +Secondary
f; 0.40 Forest
in
a 0.30 +Pasture
1: I/ ~
g 0.20 +Perennials
..I
0.10 1 ‘
+Annuals
0.00 r r .

 

 

 

 

 

 

0-5 5-15 15-25 >25

Years in Lot

 

 

 

Graph 1: The Evolution of Land Use Patterns

12

During the first ten years, the principal change in land use involves the conversion
of primary forest into secondary forest fallows. Farmers slash and burn primary forest
areas for annual crops, which after cultivating for 1-2 years, they then leave fallow to
regenerate into secondary forests. After ten years, the principal land use change involves
a continuing strong decline in primary forest cover and a sharp increase in pasture area.
This change, however, does not imply a direct conversion of primary forest into pasture.
The fertility obtained from slashing and burning primary forest gives the biggest
productivity boost to annual crops. Pastures, on the other hand, grow well on land
degraded by repeated cycles of annual cropping and fallowing. What in fact is taking
place is a continued conversion of primary forest into secondary forest in the context of
annual crop production and a conversion of secondary forest areas into pastures.

A second reason for this observed pattern is that migrant farmers initially lack the
needed capital resources to enter into cattle ranching. So, although farmers generally
consider cattle ranching the most profitable activity, they first focus on annual cropping
while acquiring sufficient capital for livestock. Nevertheless, many farmers are not able
to make the transition.

As farmers move progressively into livestock production, annual crop production
falls. By the last time period (>25 years), land dedicated to annual crops had declined
35% relative to its peak at ten years. The progressive degradation of farmland as well as
the increase in livestock production lead farmers to decrease annual cropping.

Perennial crop production, for its part, reaches a peak level of 6-7 percent of land
in the first time period (0-5 years) then stagnates and declines. Because these are long-

terrn investments and often high-value crops, it is reasonable to think they would increase

13

  

on” nine. Ht -

I91

than “9033.7

s11 uni

3. Model

Tilt '51:

\

fi‘fi‘l' "If; H

\1'...»Ioitdi\ 1'!
v

1' . A .. 3
a cup in ;

'fi‘“" ‘1 . "A, q I ,3~
3

~si...Lv x. g...
5

‘ 1
hw‘x H
huh-n.2‘13 c
‘
RA A
- ,v- ”.13...
su..l.,,1U,“n

I

30-5. - p. K
“it. DCs'Und
”1"“ " ~v~

, n
r~~"‘-- !

9‘. no- '
3“." A315 at.

 

over time. However, the history of booms and busts of perennial cr0ps to date has limited
their importance as an alternative to shifting annual crop production.
3. Model
3.1 Theoretical Model

The theoretical model used in this essay describes the relationship between
agriculture and deforestation. The principal cause of deforestation is the growing demand
for crop and pasture land. However, not all agricultural products and production
technologies have the same impact on deforestation. Conceptually, the impact of
agriculture on deforestation can be described on three levels. The first level consists of
conditioning factors. These include natural resources (e.g., farm size, hectares of primary
and secondary forest, soils, etc.), family characteristics (e.g., education, origin, age, etc.)
prices (inputs and outputs) and policies (e.g., credit, extension, land tenure, etc.). These
factors determine the second level, which is farmer production behavior in terms of
output supply and input demand. These production behaviors, in turn, are the immediate
causes of deforestation, the third level of analysis.

More formally, the theoretical framework for understanding farmer behavior uses
a profit function approach. This approach assumes that farmers choose a combination of
variable inputs and outputs in order to maximize profit subject to a technology constraint
i.e. the production function of the farm. Profit maximization should be understood in the
broad sense of farmers using limited resources in as efficient a manner as possible to
meet their livelihood (including social) objectives. Furthermore, strict profit
maximization behavior is not a necessary prerequisite for specifying systems of output

supply and factor demand equations, as long as the behavior of the individual agents is

14

—-~"s-Ean’,}\ $1..

3u.u¥15

(gym-u

aberep IS a x -.

 

 

 

.‘firp qv 3'91!-
-ll‘~‘ob ufih‘bAIKAk‘
A .

5: “p73..-“ a“
s.» F1 b aux UL'I

\-
1‘-

n — .

:11 ‘ 01,2.
.p f
1‘ l

HALTER '5:
Q

0;:qu S:

sufficiently stable over time and can be aggregated over farmers (Sadoulet and DeJanvry,
1995)

The profit function for a farm is specified as follows:
(1) II = pq - wx
where p is a vector of prices for outputs, q is a vector of outputs produced by the farmer,
w is a vector of input prices, and x is a vector of inputs.

F armers' profit maximizing behavior is constrained by the production function
where quantity of outputs q is a function of variable inputs x and fixed inputs 2 used in
the production process:

(2) q =f(x.2)

The farmer chooses the optimal level of inputs x and outputs q in order to
maximize profits. The input demand and output supply functions can be written as
follows:

(3) x = x(p,w,z) and q =q(p,w,z)
These equations indicate that the optimal levels of inputs and outputs are a function of
output price, input price, and fixed factors.

Using the expressions in (3) the profit function can be written as follows:

(4) I! = 10460. w.z) - WXQLWJ)

Using Shepard’s lemma, it is possible to derive the output supply and factor demand
functions by differentiating (4) with respect to output and input prices in the following
manner:

(5) dlo’dp;(p,w,z) = q; dm’dw;(p,w,z) = -x,-

15

- 1
‘1]. - ma ifs,"
“I: \“gL'.b\-

section.

32 [cm

I ‘ J
.""‘.3
EL&.\;\AA\ rub

v
"Tin‘fnh -\‘
5Wb§u~. I...

Pu}- .
5“". f

\(f

"'n
uD'»

ZCCCSS' EV;

~' A ..
”a- “ .3
- a
“5:. .gL.~“

‘l

In sum, the profit-maximizing behavior of farmers determines output choice and
input use. These production decisions are, in turn, the immediate causes of deforestation.
This theoretical framework is used to specify the econometric models in the following
section.

3.2 Econometric Models

A survey conducted in October of 1998 provides the data for the econometric
models used in this research. A team of enumerators led by this author interviewed 220
households concerning their agricultural production practices, the socio-economic
characteristics of their households, and the biophysical characteristics of their farms. The
survey used the distance from the main market town and the type of infrastructural
access3 available to farmers as the criteria for sample stratification.

The econometric models are structured as follows. Two sets of regression
equations have, respectively, input/technology use (x) and output production (q) as
dependent variables. These are modeled as a function of the independent variables of
price (p, w) and fixed factors (2). The regression specifications are outlined in table 1,
which also provides averages and standard deviations of the variables.

Outputs are aggregated into the three principal categories of agricultural
production activities: annuals“, perennials, and livestock. Given the dominance of slash
and burn agriculture, farmers use relatively few inputs or improved technologies. This
research analyzes five of the most common inputs/technologies found in the zone: kudzu-
improved fallows, brachiaria-improved pastures, natural pastures, capital inputs and hired

labor. Because the capital inputs farmers use are varied and overall use is low, I created a

16

(Squall

smile 5301131
in east. hoes.
Pnce

«Ham: 1‘ W
LAILAAIlnuhA - an

is: mazze. 1h
assrxehici
tartar} of:
mkumw

«- ' “
1"“39 Q
1...; ..r 0‘ P

I“.- -

ufis:he()

single capital input variable by summing the total monetary value of capital inputs used
by each household.

Price variation with cross-sectional data as used by this research is typically
minimal. One solution is to calculate farm-gate prices. In the case of outputs, I did this
for maize, the most frequently commercialized product, by subtracting transportation
costs (vehicle plus labor time opportunity cost) from the market price received. The
diversity of inputs and technologies and the variability of their costs of commercialization
made it more difficult to choose a representative product. The distance to the principal
market of Pucallpa therefore serves as a proxy for input costs.

The input use and output choice regression models use the technique of ordinary least
squares (OLS). However, because economic theory indicates that farmer input and output
decisions are simultaneous, a systems approach using seemingly unrelated regression
equations (SURE), was also tested. The SURE modeling approach gave results similar to
OLSS, but sharply reduced the number of available observations". This essay therefore
uses the OLS results for this paper's analysis.

A third regression model examines the impact of agricultural output production '
and input/technology use on deforestation. The relationship of outputs and inputs to
deforestation is hypothesized to be recursive: agricultural production results in
deforestation and not vice versa. Selective extraction of timber leaves the residual forest
largely intact. The clear-cutting type of deforestation examined by this research is an
outcome of agricultural production. I tested the unidirectional causality of agriculture
leading to deforestation using the Hausman specification test to detect simultaneity]. This

test confirmed the recursivity between agricultural production and deforestation with the

17

 

 

......

r
.. ...o .

Lit erroc

 

 

 

 

‘-. .

. lav/«sir

r“i

Table 1: Input and Output Regression Specifications, Variable Averages and Standard Deviations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DEPENDENT VARIABLES'
OUTPUTS (q) INPU TS/T ECHNOLOG Y (x)
Average Std. Dev. Average Std. Dev.
Annuals 3.71 3.97 Improved Fallows 2.07 3.80
Hectares of A nnual Crops Hectares ofjaIlows improved with kudzu
Perennials I 1.47 I 2.09 Improved Pasture [ 7.24 1 15.93
Hectares of Perennial Crops Hectares ofpastures improved with brachiaria
Livestock 1 6.93 1 27.40 Natural Pasture ] .95 1 3.33
Head of Cattle Hectares ofnaturaIJJasture
- - Capital Inputsb ] 216.59 1 734.76
- - Total value of capital inputs used in 1998 (soles)
- - Hired Labor 1 64.05T 142.50
- - Number 9f person days of hired labor used in 1998
- INDEPENDENT VARIABLES'
IOF OUTPUTS (q) I IOF INPUTS/7’ ECHNOLOG Y (x) 9
PRICE (p) PRICE (w)
Average Std. Dev. Average Std. Dev.
Fm-Gate Price Maize .36 .056 Market Distance 60.97 27.76
Maize market price minus transpo costs (soles/kg)25 Distance in km farm to Pucallpa
FIXED FACTORS (z) FIXED FACTORS (1)
Average Std. Dev. Average Std. Dev.
Farm Size 37.40 31.37 Farm Size 37.40 31.37
Hectares of land on farm Hectares of land on farm '
Secondary Forest I 1 1.56 I 16.05 Secondary Forest l 11.56 I 16.05
Hectares of secondary forest on farm Hectares of secondary forest on farm
Primary Forest 1 10.98 I 16.70 Primal)! Forest 1 10.98 I 16.70
Hectares of primary forest onfarm Hectares of primary forest org/arm
Years in Lot 1 17.12 L 11.80 Years in Lot 1 17.12 1 11.80
Years a farm has been in production Years a farm has been in production
Education‘ I 33% I n.a. Education‘ I 33% l n.a.
Household head > primary education (dummy) Household head > primary education (dummy)
Family Labor 1 2.72 1 1.34 Family Labor 1 2.72 1 1.34
Family members over 14 working on the Larm F amw members over 14 working on the farm
Off-farm lncome" 1 1844.6OJ 3074.50 orr-rarm Income” 1 1844.60 1 3074.50
Ojjlfarm income earned by farm household (soles) Ofl-farm income earned by farm household (soles)
Credit‘ 1 33% j n.a. Credit‘ 33% l n.a.
Credit received in the past 5 years (dummy) Credit received in the past years (dummy)
Land Tenure‘ [ 84% 1 n.a. Land Tenure‘ L 84% I n.a.
Farmer perceives tenure as secure (dummy) Farmer perceives tenure as secure
Alluvial Soilsd 1 28% 1 n.a. Alluvial Soils“ ] 28% 1 n.a.
Alluvial soils dominate (dummy) Alluvial soils dominate (dummy)
Extension I 3.28 I 11.74 Sandy Soils‘ I 42% ] n.a
Number of extension visits in the past year Sandy soils dominate (dummy)
Origin I 25% I n.a. Distance Social Servicesl 5.67 I 5.57
Household head not from jugle region (dummy) Dist. in km farm to nearest: (school + health post)
Age 48. l6 1 14.08 - -
Age of household head - -

 

 

 

'All variables are measured on a farm household level for the year 1998

”One nuevo sol = 3.32 in 1998

‘Percentages not averages are presented for dummy variables
‘Percentages are determined by sampling strategy (along rivers) and may not reflect regional average.
‘Percentage of highland (not river) soils. Farmers chose between clayey or sandy soil descriptive.

18

r A T
51.331
55353 H

a
raw qr'f‘.

ml“

ign“. H?

Dinah! 5‘
A $80!

Q. -a
v.1)».

tanuzh
\
Perms;
\
Lhruoc
\
Imnroie
\\
\zi
___E:1L
(lfinzll
Ta?
3
T\u
ft! ‘1' C
%
limlh 1
\
»ihu\111l
'r\
Ldutano
\

”YE/H

"' 'v -—-4
o :17 «a if :1" o
:1 g- o 5 8’
3" ~1— 3 "’ W
.J .a A: ~37 5?
—- '1 .1 __ - ’

at ““01

exception of improved pastures in the total deforestation model. This result may make

sense in the context of pasture installation being an evolutionary process. Land deforested

for annual cr0ps is later transformed into pastures. Deforestation may therefore lead to

land being put into pasture and as well as pasture installation resulting in deforestation.

A two stage least squares approach (ZSLS) was used to correct for simultaneity in this

case.

The deforestation dependent variable is measured as the hectares of forest felled

in 19988 by a farm household. It is divided into three categories: total forest deforestation,

primary forest deforestation, and secondary forest deforestation. Primary forests are

areas that have never been felledg. Primary forests that have been felled and have

Table 2: Deforestation Regession Model Specifications

 

 

 

 

Total Deforestation Primary Forest Deforestation Secondary Forest Deforestation
Annuals Annuals Annuals
Perennials Perennials Perennials
Livestock Livestock Livestock

 

Improved Fallows

Improved Fallows

Improved Fallows

 

Improved Pastures

Improved Pastures

Improved Pastures

 

Natural Pastures

Natural Pastures

Natural Pastures

 

 

 

 

 

 

 

 

 

 

 

 

 

Capital Inputs Capital Inputs Capital Inputs
Hired Labor Hired Labor Hired Labor
Tenure Security Tenure Security Tenure Security
Family Labor Family Labor FamiLy Labor
Alluvial Soils Alluvial Soils Alluvial Soils
Education Education Education
Total Forest - -
- Primary forest Primary forest
- Secondary forest Secondary forest
- Primary Forest Products Primary Forest Products

 

Table 3: Previously Undefined Variables

 

 

 

 

 

 

 

 

 

Average Std. Dev.
Total deforestation Primary plus secondary forest felled (ha '3) 236 244
Primary Forest Deforestation Primary forest felled (ha '3) .66 142
Secondary Forest Deforestation Secondary forest felled (ha '3) 1 .72 2.13
Total Forest Hectares of primary & second forest 22.55 2315
Primary Forest Prod’s Value prim. forest prod ‘s harvested (solesa) 319.27 800.07

 

'One nuevo sol - 3.32 in 1998

19

 

 

ppfi'ad f]
{estuww -
v

 

race A cor:
F
acne acme:
il'
mm: hm.

!' l
y‘q‘A-a I; ‘|
Lusaka-s» If“: .

‘4

A. .H'h‘v’

hLlu‘...‘_§. Si
w

a A 1

V'\_ " v“

“V“‘e‘ie SLA;

regenerated are secondary forests. Total deforestation is the sum of both primary and
secondary forest deforestation.

The distinction between primary and secondary forest deforestation is often not
made. A common perception is that once primary forest has been cleared the forest is
gone forever. However, research by FAO (1996) estimated that in 1990 there existed 165
million hectares of secondary forest in Latin America. In the Pucallpa area, survey results
indicate that secondary forests (30%) virtually equal primary forests (31%) on farm
holdings. Secondary forests on farms are areas of fallow that recuperate the soil fertility
and are subsequently slashed and burned for renewed cropping.

Deforestation is modeled as a function of output choice, input/technology use, and
general conditioning factors. General conditioning factors are those that may directly
affect deforestation i.e. not only through their impact on input and output choices. These
include land tenure, family labor, soils, education, forest area, and forest products.

4. Results

The discussion of regression results proceeds as follows. First I examine the
determinants of output and input/technology variables and the impact these variables
have on deforestation. I then examine the general conditioning factors from the
deforestation models. This section ends with a cross-cutting analysis that summarizes the
key linkages between production determinants, farmer input and output behaviors, and
deforestation.

The following discussion considers variables significant up to the .15 level. The
use of this level reflects, in part, the tendency of cross-sectional data to give lower

significance levels. Furthermore, the choice of a significance level should reflect the

20

  
 

costs of “El:

 

0
133.1 UCC has
kl Out

Anr

costs of making an error (Manderscheid, 1965). For the analysis at hand (identifying
production and deforestation determinants), the consequences of a (type 1) error are not
especially serious"). Exact significance levels are presented for the reader’s consideration.
The predictive power of these models (R2) is generally good for cross-sectional data. The
exceptions are the natural pasture and hired labor models which, nevertheless, provide
useful insights about individual exogenous variables. Table 4 presents the deforestation
model results and tables 5 and 6 respectively provide the results of the output and
input/technology models.
4.1 Output Supply Variables

Annuals-Annual crops (e.g., maize, rice, cassava, etc.) stands out as the only
variable positively correlated with all types of deforestation. They are indeed the only
category of outputs that uses shifting siash-and-burn agriculture. Output regression

Table 4: Deforestation Regression Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total Primary Forest Secondary Forest
Deforestation Deforestation Deforestation
121.54 adj 19.50 also adj 122.23 117.30 adj 122.23
Coef @tat) Signif Coef (t-stat) Signif Coef (t-stat) Signif
Constant . l 8 .74 .74
Annuals .490 (7.59) .00‘" .323 ( 3.69) 00"” .301 (3.42) .00‘“
Perennials -. 102 (4.65) .10“ -.040 (-0.53) .60 -.008 (-0.10) .92
Livestock .039 (0.54) .59 .020 ( 0.21) .83 .197 (2.04) .04
Improved Fallows .049 (0.83) .41 -. 129 (4.80) .07" .137 ( 1.9g .06"
Improved Pastures -.023 (-0.32) .75 -.006 (DIE .95 -.154 (4.53) .13"
Natural Pastures -.094 (4.50) .13" -.061 (0.75) .45 -. 163 (~2.00) .05'"I
Capital Inputs -.153 (-2.18) .03” -.125 (4.41) .16 .181 (2.03) .04”
Hired Labor .337 (5.73) .00‘“ .011 (0.13) .89 .019 (0.23) .82
Family Labor .060 ( 1.01) .31 .021 (0.27) .79 .202 (2.61) .01‘"
Education .129 (2.26) .02“ .003 (0.05) .96 .070 (0.98) .33
Tenure Security .097 ( 1.69) .09‘ .025 (0.32) .75 .030 ( 0.39) .69
Alluvial Soils .028 (0.45) .65 -.193 (-2.61) .01 ”“ .203 (2.75) .01‘”
Total Forest .108 ( 1.72) .09‘ - - - -
Secondary Forest‘ - - -.098 (4.41) .16 .166 (2.38) .02”
Primary Forest - - .254 (3. 30) .00’“ - .093 (4 .20) .23
Prim Forest Prod's .168 (2. 36) .02" - .034 (-0. 48) .63

 

 

 

 

 

 

‘Sigmficant at the .01 level " Sigmficant at the .05 level ""‘ Significant at the 15 level
(a) High secondary forest >5meters

21

scafll

 

 

_..
._.:2.7.

3...? 3 72.. a

3...: r... .39:
lat-0.. 1.—

c:
...223... $3... 3 ..z...

5-... :i 3...:
5...: :5??-

....
192.37.

.{ill 11.

\. I «2:: L 72.9
5...: ,3... 3...:
, I:....:<
n.a.-:6: .2:ro._...=.~: :25...» ..V. 92235 -

938:2». 2.

05 «a assoc—55 e: ._u>o_ no. 05 E “caocEflm : 1_o>o_ S. 2: an E8555.

 

22

Jl

',‘JI ‘1'.’
“1‘15“.5

n15“ [

1.1...55.

V
,«ar’ '
5g.....:

DI .-
013231

1'14 ‘1':
‘19 LIAN

P. a:

.
9.1%. "1r
'5“ LL‘ 1.

5 r0:
\Oia‘h.I

0': E):

122

_-
p.312

indicate that annuals tend to be found where there is more primary forest on relatively
younger farms. They are negatively correlated with the farm-gate price of

maize, the proxy for market access. These results confirm the evolutionary process of
recently arrived migrant farmers first planting annuals on newly opened land on the edge
of the forest frontier.

The negative correlation with off-farm income reflects the competition for labor
and that, due to low returns, off-farm income is not invested in annual crop production.
The positive correlation with family labor indicates both that labor is a constraining
factor of production and that more food is needed to feed larger families in a semi-
subsistence setting. The greater capacity to practice continuous cultivation of annuals on
fertile alluvial soils explains the positive correlation with this variable.

Perennials-Perennial crops such as lemons, mangos, and oil palm require
continuous cultivation of a parcel of land. Their labor-intensive nature leads to a strong
correlation with family labor. Perennials “absorb” scarce labor away from shifting slash-
and-burn agriculture and into sedentary production. This leads to a decline in total
deforestation.

Off-farm income decreases perennial production reflecting once again the
opportunity cost of scarce labor as well as the relatively low returns to most perennials in
the zone. Their long-term nature leads to a positive correlation with household head age.
Also, household heads from outside the jungle have more perennials due to the tradition
of tree plantations in the coast and sierra (where natural forests have largely disappeared)
as opposed to the more extractivist tradition in the jungle. Unlike annuals, most

perennials do not grow well in areas of alluvial soils because of water logging from

23

«62-. 11111.?
1'

   
 

mauve cor

_ .
01.133111: \\ 1'.

process a r:

am» i .
La... it's 3:
’14“ so. _-
p‘kbUL HQ

1:17am“. 2

Y‘P?j.;‘
5‘

qul {6

periodic inundations. And extension’s focus on traditional field crops explains its
negative correlation with perennials.

Livestock-The only statistically significant relationship between livestock and
deforestation is a positive correlation with secondary forest clearing. Livestock are found
on farms with less remaining primary forest. This relationship reflects the evolutionary
process whereby farmers first slash and burn primary forest for annual crop production
and later install pastures in areas of secondary forest. The extensive nature of livestock
leads to a positive correlation with farm size. So, over time, farmers’ desire to raise
livestock may indirectly lead to substantial clearing of primary forest.

The positive correlation with tenure security is due to the need to protect the long-
term investments such as pasture installation, fencing, and wells associated with livestock
production. Family labor’s positive correlation with herd size indicates that labor
remains a limiting factor of production despite ranching’s low labor requirements. The
negative correlation with a household head’s age results from older farmers having less
available family labor and being less physically able themselves to maintain large
herds”. _

4.2 Input/Technology Demand Variables

Brachiaria-Improved Pastures-Farmer planting of grasses from the brachiaria
genus (principally Brachiaria decumbens) increases pasture carrying capacity by over 50
percent relative to natural pastures in the Pucallpa zone. Brachiaria’s vigorous growth,
which reduces weed competition, lowers pasture maintenance costs. Increased
productivity and lower costs explain why brachiaria now dominates, representing 78

percent of the total pasture area in the zone (Yanggen, 2000).

24

   
   
  

As \\ .
seconded
parent of f.
of're's 1.

111311303 1

‘
"\I ‘
um. pnma

.t
‘ 738:5

is, .. .
”$331112

As with livestock, brachiaria pastures’ only significant correlation is with
secondary forest clearing. However, the correlation is negative. In the Pucallpa zone, 54
percent of farms have pastures but no cattle. Surplus pasture capacity dampens the desire
of farmers to fell more secondary forest for pasture. Furthermore, brachiaria pasture is
negatively correlated with the presence of primary and secondary forest. Farmers first cut
down primary forest for annual cropping then progressively install pastures in areas of
secondary forest. But unlike shifting annual cropping, sedentary production on pastures
suppresses the regeneration of secondary forests. In sum, surplus pasture capacity and
pasture’s repression of secondary forest regeneration explain brachiaria pastures’
negative correlation with secondary forest clearing.

Farmers tend to install brachiaria-improved pastures nearer to Pucallpa and other
population centers with social services. This tendency reflects the greater push to
intensify production in areas with better access to markets and where land is scarce due to
population pressures. This intensification with brachiaria nonetheless takes place on
larger farms reflecting the land-using nature and low labor requirements of livestock. The
negative correlation with credit reflects the cost-saving nature of this technology.

Natural Pastures-Natural pastures are similar to brachiaria pastures in that
farmers tend to plant them on larger farms with less secondary forest. However, unlike
brachiaria pastures, they have no statistical association with intensification or credit
constraints. Also they are found on older farms, whereas this correlation is not significant
with brachiaria, which farmers install on relatively newer more dynamic farms.

Kudzu-Improved Fallows-Kudzu (Pueraria phaseloides) is a leguminous vine

that farmers plant or that spontaneously regenerates in areas that have recently come out

25

guano-

a ‘,
01"“

"*"m \’t’

- l
,bnv

I a «95‘
hffliw \

Iv

n.a-you;
3.3“.“

. .
Wit»;

Us... -

“‘90-;

VS...

5%
LL‘
11w

'4'
5».

9h.
\‘1‘

h”.—

9..

1m

‘I‘L
a

[J

of annual crop production. It aggressively spreads throughout areas of secondary forest
fallow vegetation, retarding the growth of trees and shrubs and smothering out
herbaceous weeds. Its fixation of nitrogen and weed suppression increase yields of crops
subsequently planted in the fallow area.

Kudzu-improved fallows are negatively correlated with primary forest
deforestation and positively correlated with secondary forest deforestation. Yanggen and
Reardon (In Press) have shown that when farmers plant kudzu in areas of secondary
forest fallows, this reduces the labor costs of land clearing and weeding and increases
yields”. This reduction in the costs (labor) and increase in the benefits (yields) of using
secondary forest has a ceteris paribus effect of increasing the attractiveness of secondary
forest clearing relative to primary forest clearing.

Despite increasing productivity, kudzu-improved fallows do not follow a typical
scenario of intensification associated with land scarcity nearer to population centers
(Boserup, 1965, 1981; Pingali et al. 1987). Instead they are positively correlated to farm
size and farmers tend to adopt them at greater distances from Pucallpa and smaller market
towns where social services are located. They are, however, associated with older farms
and less primary forest, which reflect declines in land quality. Less primary forests
implies a depletion of aboveground soil nutrients, whereas farms in production longer
tend to suffer from below ground soil nutrient depletion. In sum, it is a land quality
constraint and not a land quantity constraint that explains much of the adoption of kudzu-
improved fallows.

A higher opportunity cost to educated farmers’ labor may explain why these

farmers adopt this labor—saving technology. Also the increased management of fallow

26

.~' ,‘
memes.

   
    

trauma: 1

Cap
r)“ "F"???
g g r “an“ ’

, I
107351 21537.?

areas: ‘11:
use eases ;
permit a g
or. prima:

T

Search}

with kudzu may require higher levels of knowledge. The negative correlation with credit
makes sense as this technology requires no capital investment and reduces labor
requirements. And finally, the positive correlation with sandy soils likely relates to the
agronomic conditions in which kudzu flourishes.

Capital Inputs-Capital input use is negatively correlated to total forest clearing
and primary forest clearing (.16 significance level) and positively correlated to secondary
forest clearing. Farmers mainly use land-saving capital inputs such as fertilizer,
improved seed, herbicides, and livestock health products in the zone. These inputs
increase the productivity of land and permit more sedentary production. Sedentary land
use eases pressures on total deforestation. These productivity-enhancing inputs also
permit a greater reuse of degraded secondary forest fallow areas, which reduces pressures
on primary forest clearing.

The positive correlation between capital input use and farm size indicates land
scarcity is not driving intensification with these products. More likely, larger farmers
have greater resources to purchase these inputs. The positive correlation with credit
reflects the importance of the capital constraint to purchased input use.

The positive correlation with family labor reflects the fact that these inputs are of
a land-saving and not a labor-saving nature. Also, farmers use fewer capital inputs in
fertile alluvial areas where there is less need for productivity-enhancing inputs. The
movement away from shifting fallow-based agriculture to more sedentary production
may explain why capital input use is negatively correlated with secondary forest fallows.
And finally, farrners’ use of these inputs on farms with more tenure security likely

reflects increased incentives to maintain land productivity.

27

 

 

  
   

1111013111115

..3~

3121331113
prediction.

is 1:
at credit. 1
prcducuon 2

7303‘s 1L;

ibaiwh U“

E H )3
U\

l A

pm.» by...

5‘ :54 ‘Wltl
-

1‘5“” ,L‘n
d1...

”‘9' V
Ski‘s“:

mitt

Hired Labor-Hired labor is positively correlated to total deforestation. This
demonstrates the key role of labor as a limiting factor of production. A greater
availability of hired labor permits more forest clearing for increased agricultural
production.

As in the case of capital inputs, hired labor is positively correlated to farm size
and credit. Larger farmers may have greater resources to hire labor as well as more
production activities that increase labor needs. The positive correlation with credit
reflects the importance of the capital constraint for purchased inputs.

The positive correlation with family labor may appear paradoxical. However,
larger families with more available labor tend to cultivate more total land. Labor is hired
during the key production bottlenecks periods of land clearing, planting, and harvesting.
Larger families that cultivate more land also need to hire more labor at these times.

4.3 General Conditioning Variables

Apart from input/technology and output variables, a number of exogenous
variables that may directly affect forest clearing also appear in the deforestation models.
These general conditioning factors include: family labor, education, tenure security,
alluvial soils, on-farm forest, and the value of harvested primary forest products.

Family labor is positively correlated to secondary forest clearing. This
relationship further confirms the importance of the labor constraint to production in the
zone. Education’s positive correlation to total forest clearing indicates that it enables
farmers to increase production at the extensive margin rather than to adopt more
sustainable production technologies. Deforestation does not reflect farmer ignorance, but

rather the current lack of profitable alternatives to slash-and-burn agriculture.

28

 

n‘o-aU a: K

«F \vg:
in». Sm

5.3.‘r
bib¥a

Other research, notably in Brazil, has found that insecure tenure often pushes
farmers to clear forest because laws require this as proof of productive land use (Hecht,
1985; Binswanger, 1991). This situation is not the case in the research area where untitled
land is typically surveyed into lots soon afier settlement and occupancy of a farm
generally ensures tenure security. Because land clearing is not a prerequisite for
solidifying tenure security in the case of Pucallpa, tenure insecurity does not lead to
increased deforestation. Rather tenure security encourages greater total deforestation.
Land clearing ofien represents a long-term investment in land and farmers are more likely
clear land where tenure security ensures they receive the benefits of this investment.

Alluvial soils are correlated with increased secondary forest clearing and
decreased primary forest clearing. These soils are more fertile and resilient that highland
soils. Farmers can reuse areas of secondary forest to a greater degree thus reducing
pressures to expand into primary forest areas.

Conservationists often propose the promotion of primary forest products as a
means to induce farmers to conserve primary forests. A study by Pinedo et al. (1992) in
the Peruvian Amazon found that the returns per hectare from primary forest products
were substantially lower than those from agricultural production. A study in the Pucallpa
zone (Smith et. al 1997) found that farmers use and sell more primary forest products in
recently opened areas and progressively less as population and market access increase.
These findings led to the hypothesis that the sale of these products may be used to help
finance the conversion of primary forest to agricultural uses and not increase the

conservation of primary forest during frontier development. The positive correlation

29

 

111-.

A
:1

between the value of primary forest products harvested and primary forest clearing
appears to confirm this hypothesis.

A last set of conditioning variables are the “control variables” of the hectares of
on-farm forested land. Greater availability of forest land should allow farmers to clear
more forest. As expected, each type of forest land is positively correlated with its
respective type of forest clearing.

4.4 Cross-Cutting Results

This subsection synthesizes several broad themes from the regression analysis.
This synthesis more explicitly links together production determinants with deforestation.
The themes focused on are: the evolutionary nature of land use patterns, the role of labor,
the role of capital, secondary vs. primary forest clearing, and patterns of intensification.

The regression results support the analysis of land use evolution in section 2.2.
The results show that annual crop production is most strongly associated with early
frontier development leading to deforestation at the forest margin. Livestock tends to
come later after primary forest has been cleared and leads to clearing of less productive
secondary forest fallow land. A lack of evolutionary indicators for perennials, on the
other hand, reflects their stagnation over time.

The regression results confirm the key role of labor as a constraining production
factor. Farmers with more family labor have greater production for all the principal
output categories: annuals, perennials, and livestock. And hired labor and family labor
respectively lead to greater amounts of total and secondary forest clearing. Furthermore,
off-farm employment’s competition with agriculture for labor decreases annual and

perennial crop production. This competition for labor can reduce pressure to clear forest.

30

4.",

«ha
., ‘Ii
0

5.43

05‘.»
”a

youS

l .
1
1‘.

“(1321.1

These models also capture the importance of the capital constraint as reflected by
the significance of the credit variable in the input/technology regressions. Credit is
positively correlated with those inputs requiring a financial expenditure: capital inputs
and hired labor. And the negative correlation of credit with the technologies that save on
labor and capital, kudzu-improved fallows and brachiaria-improved pastures, may reflect
attempts by farmers to overcome capital constraints. While the impacts of these particular
inputs and technologies on deforestation are not uniform, it is nonetheless clear that the
access or lack of access to credit plays a key role in determining the farmer production
practices that influence deforestation patterns.

Kudzu improved fallows, capital inputs, and alluvial soils increase secondary
forest clearing but decrease primary forest clearing. Increases in land productivity
associated with these variables counteract or mitigate declines in soil fertility that result
from the previous cultivation of a parcel of land. This productivity enhancement enables
farmers to reuse secondary forest fallows, which decreases the need to clear primary
forest.

Intensification does not follow a typical pattern of farmers increasing variable
labor and capital inputs due to land area constraints. Kudzu-improved fallows and
brachiaria-improved pastures increase land productivity but decrease labor needs and
have essentially no capital requirements. Furthermore, these improved fallow and pasture
technologies are associated with larger farm size. By easing these constraining production
factors, farmers appear to be able to increase the land area they put into production
(Yanggen 2000, Yanggen and Reardon, In Press). Intensification of land use does not

result for a scarcity of land nor does it reduce the amount of forest land cleared for

31

agricultural f.

 

effective ca?

   
   
 

require fem.
1112.: p03...
more impo.
52: may 32.
macs... the
6. Co'
Tn
310101 0ft
increases

f0! pow c

0&3? a:
331$ 117.:
131331.163

\C‘m‘l S :1

agricultural production. Declines in land quality in some cases, however, may reduce the
effective carrying capacity of the land and push farmers to adopt these practices.

The case of land productivity enhancing capital inputs is different because they
require farmers to increase capital expenditures and are associated with greater labor use.
Their positive correlation with credit indicates that overcoming capital constraints may be
more important than land area constraints in bringing about intensification. Larger farm
size may also in some cases be associated with wealthier farmers. This result again would
indicate that access to capital and not land area constraints leads to intensification.

6. Conclusions and Policy Recommendations

The production of annual crops using shifting slash-and-burn agriculture is the
motor of deforestation in the Pucallpa research zone. And greater labor availability
increases the extensive production practices that cause deforestation. A unifying theme
for policy is how to reduce the labor availability for extensive annual crop production.

Off-farm income opportunities can siphon labor away from annual cropping and
other agricultural activities that lead to forest clearing. Development of the non-
agricultural sector is therefore key to taking pressures off the natural resource base. This
implies the need for a broad-based development strategy including other sectors such as
tourism, industry, and services.

Research and policy needs to help promote more sedentary annual cr0pping
practices. The use of productivity-enhancing inputs reduces deforestation and encourages
more intensive use of land. However, given low output prices and poor marketing
infrastructure, it is not clear that most uses of these inputs are either affordable or

profitable for farmers. Experiment station research that focuses on increasing yields may

32

    
    

16011103 1
a imarove.

. 4.
15563.11 5..

11711er 0;;

use of cap;

mars. h-

 

mzrlxts is.

for timber.

A 4’
«21113.3

high C01:

be of little relevance to farmers. Research needs to focus on combinations of inputs such
as improved seeds, fertilizer, and pesticides that are profitable for farmers. This type of
research should be conducted on-farrn so that it reflects the conditions under which
farmers operate.

Fertile alluvial soils permit more sedentary production of annual crops without the
use of capital inputs. These soils are negatively correlated with primary forest clearing.
Farmers, however, often prefer to settle along road infrastructure because access to
markets is both quicker and cheaper than with river transportation. Most roads are built
for timber, oil, or gas extraction without regards to the agricultural potential of the land.
Studies should be undertaken to investigate the economic viability of improving road
access to areas of alluvial soils along rivers.

Perennial crop production decreases deforestation by absorbing scarce labor into
sedentary production. Their production has languished due to volatile markets and the
high commercialization costs. The transformation of agricultural raw materials into
products such as oils, preserves, and flour can dramatically lower the transportation costs
relative to output value and refined products tend to suffer less market turbulence than
raw materials. Policies that focus on promoting agricultural industries based in the jungle
can help promote sustainable perennial crop production.

It is important, however, that perennial crop production involve small holder
farmers. Large-scale mechanized plantation agriculture, as~ is typical of perennials in
many areas of the world, is likely to increase deforestation and concentrate wealth in the

hands of a limited number of individuals. Individual farmers, however, typically lack the

33

  

parrot: p1-

isms-stir:

9 1
33w? .0:
1 ¢
(.1
-J, '
711“ «Q..-
QAA.“ 3

' , - . \.
\s‘x‘ 3)“L): ’.

capital and collective organizational capability to identify promising crops and set up
systems for input distribution, output transformation, and commercialization.

Government, private industry, and farmer organizations need to work together to
promote perennial production so that economies of scale necessary for efficient input
distribution and product transformation are achieved while at the same time involving as
many farmers as possible. This approach will allow a greater distribution of wealth and
absorb more farmer labor away from shifting slash-and-burn annual crop production.

Interviews with farmers in the research zone indicate their primary future
production objective is to expand livestock production (Labarta, 1998). So while
livestock may not be directly correlated with primary forest clearing, it does appear to
increase deforestation as part of a long-term strategy of moving from annual cropping to
ranching. Informal observations in the zone indicate that a small number of very large
ranchers do clear extensive tracts of primary forest for livestock production. This clearing
sacrifices large areas of rainforest without reducing the poverty of most farmers. Efforts
by policymakers should therefore focus both on restricting the size of cattle ranches and
as well as limiting production to older settlement areas where degraded land is less
productive for annual crops but suitable for pastures. Given superior market access, these
areas are also the most propitious for more intensive forms of cattle raising using
improved pastures and combined meat and dairy production.

Kudzu-improved fallows and brachiaria-improved pastures both increase land
productivity as well as reduce labor requirements. These technologies bring about an
apparently paradoxical situation of intensification associated with increased land clearing.

However, basic economic theory indicates that farmers maximize the returns to

34

 

comm:
1.1,, '9';
“rank “U

' 1’ . «fir
p“: .4511)

”‘1

ran ‘
hurl.) i
figs-I “‘3-
ii\. 'p' \
I

LEX‘I \l

constraining production factors. These technologies, by improving land productivity
while reducing labor costs, increase the returns to the scarce labor factor. Increased land
productivity is therefore a “byproduct” of farmers’ attempts to increase returns to labor.

The challenge is to promote production that increases returns to labor without
encouraging extensive production. Labor intensive production of high value perennial
crops can do this by absorbing labor into sedentary production while still providing high
returns to labor. Agroforestry techniques that incorporate trees with high value products
into pastures and fallows may therefore increase both land productivity and returns to
labor while absorbing labor away from forest clearing.

This research has demonstrated the key role that credit can play in determining
production practices. In the late eighties, easy access to subsidized credit led to increased
slash-and-burn agricultural production and therefore accelerated deforestation in the
Pucallpa zone (Yanggen, 1999). A focused policy is therefore needed that links credit
with more sedentary agricultural practices such as use of productivity-enhancing inputs
and small holder perennial crop production. A more focused credit policy can both help
reduce deforestation as well as limit the fiscal burden of subsidized credit.

This section has proposed a series of strategies to encourage more intensive
agriculture. However, if these new practices or cr0ps are sufficiently profitable, farmers
may invest in labor-saving capital equipment or simply hire more labor. The cheap
accessibility of land remains a potent incentive to expand land area under production.
Intensive cultivation does not preclude land area expansion. There is a need to

complement the promotion of intensive cropping systems with policies that restrict access

35

 

 

511:. id pt
marries i
stilts impc

P 'tfi"; r‘
1‘uL-5‘I10 ‘

-.a_

 

11:27.1 it :1
1

'l. ' ..-
Likin13e“

1615‘ Wu

tmsu

P

V 1
.. a.
Bul Lfear‘

144‘... v. “
aquth'aa

‘ .
‘3? "H.‘
a -gjkj.l‘~

\

to land. Options such as reductions in new road construction, a progressive tax on farm
size, and percentage limits on on-farm primary forest clearing merit consideration.

This strategy can be undertaken using a carrot and stick approach. Carrot
incentives are offered to induce farmers to adopt more intensive cropping systems and
sticks impeding access to land push farmers to adopt more intensive systems. Land access
restrictions alone are unlikely to be effective if they are not accompanied by viable
intensive production alternatives that enable farmers to make a living using less land.
Likewise, the promotion of intensive cropping systems may not limit expansion into the
forest areas if land continues to be the cheapest production factor.

The focus of this essay has been on reducing deforestation at the household level.
But clearly continued migration into the jungle will accelerate forest clearing. The
adoption of intensive cropping practices can only mitigate increased deforestation
resulting from demographic pressures. The root cause of migration into the jungle is
poverty in other areas of the country. Economic development in poverty-stricken areas of
the country needs to take place in order to reduce migration to the jungle. In this sense, an

important part of the solution to deforestation lies outside the jungle itself.

36

m

REFERENCES

Barbier, EJ. and Burgess, J .C. 1996. Economic Modeling of Deforestation in
Mexico. Environment and Development Economics 1 : 203-239.

Binswanger, Hans. 1991. Brazilian Policies the Encourage Deforestation in the Amazon.
World Development, 19 (7): 821-829.

Binswanger, Hans and John McIntire. 1987. Behavioral and Material Determinants of
Production Relations in Land-abundant Tropical Agriculture. Economic
Development and Cultural Change. Vol 36 No 1. pp 73-99

Boserup, E. 1965. The Conditions for Agricultural Growth: The Economics of
Agrarian Change under Population Pressure. Aldine Publishing Company,
New York.

Boserup, E. 1981. Population and T echological Change: A Study of Long Term
Trends. University of Chicago Press. Chicago.

Cochrane, Thomas and Pedro Sanchez. 1982. Land Resources, Soils and their
Management in the Amazon Region: A state of Knowledge Report, in S.
Hecht (Ed.) Amazonia: Agriculture and Land Use Cali: Centro
Internacional de Agricultura Tropical, 1982. Pp 137-209.

Deacon R.T. 1995. Assessing the Relationship Between Government Policy and
Deforestation. Journal of Environmental Economics and Management. 28:
148.

De Almeida, Oriana-Trinadade, and Christopher Uhl. 1995. Developing a
Quantitative Framework for Sustainable Resource-Use Flaming in the
Brazilian Amazon, World Development, 23 (10): 1745-1764.

Ehui, Simeon, Thomas Hertel and Paul Preckel, 1990. “Soil Resource Depletion, Soil
Dynamics, and Agricultural Productivity in the Tropics,” Journal of
Environmental Economics and Management, 18 (2): 136-154.

Escobal, Javier. 1999. Analisis de 1as Politicas Agricolas Actuales.
GRADE. Lima, Peru. Mimeo.

F A0. 1996. Forest Resource Assessment 1990. FAO Forestry Paper 130. Rome.

Godoy, R. et a1. 1997. Household Determinants of Deforestation by Amerindians in
Honduras. World Development. Vol. 25. No 6 : 977-987.

Goodland, R and H. Irwin, Amazon Jungle: Green Hell to Red Desert?
(Amsterdam: Elsevier, 1975).

37

 

 

1.13.4
dc
11:

use» it
DC
at

 

r .L ..
1.4.3114, i

1111‘“.

P1?»

Hecht, SB. 1985. Environment, Development and Politics: Capital accumulation and the
livestock sector in Eastern Amazonia. World Development 13. 663-684.

Hecht, SB. 1992. Logics of Livestock and Deforestation: the Case of Amazonia. In
Development or Destruction: The Conversion of Tropical Forest to Pasture in
Latin America. Downing T.E., Hecht S.B., Pearson H.A. and Garcia-Downing
(eds). Westview Press. Boulder. pp 405. (95-113).

INRENA (Instituto Nacional de Recursos Naturales) 1995. Monitoreo de los Processos
de Deforestacion en la Amazonia Peruana. Medio Ambiente Rural. August 1995,
Lima Peru.

Kaimowitz, David and Arild Angelsen. 1998. Economic Models of Tropical
Deforestation a Review. Center for International Forestry Research (CIF OR).
Bogor, Indonesia.

Labarta, Ricardo. 1998. Los Productores de la Cuenca Amazonica del Peru y la
Dinamica de Uso de la Tierra: Resultados de la Caracterizacion dc Pucallpa y
Yurimaguas. ICRAF Report. Pucallpa, Peru.

Mattos, Marli Maria and Christopher Uhl. 1994. Economic and Ecological
Perspectives on Ranching in the Eastern Amazon, World Development, 23,
(2): 145-158.

Manderscheid, Lester. 1965. Significance Levels—0.05, 0.01, or?. Journal of
Farm Economics, 47 (5): 1381-1385.

McNeil, M. 1967. Laterite Soils, Scientific American, 211 (5): 96-102.

Myers, Norman. 1992. Tropical Forests; The Policy Challenge. Environmentalist, Vol.
12, No. 1 pp. 15-27.

Palomino, L. 1993. Dos Décadas de Crédito Agrario. Serie Investigaciones No 10.
Universidad de Lima, Facultad de Economia. Lima. 67p.

Peters, Charles, Alwyn Gentry, and Robert Mendelsohn. 1989. Valuation of an
Amazonian Rain Forest. Nature 339 (June): 655-56.

Pinchon, Francisco. 1997. Colonist Land Allocation Decisions, Land Use, and
Deforestation in the Ecuadorian Amazon Frontier. Economic Development and
Cultural Change. Pg. 707-44.

Pinedo-Vasquez, Miguel, Daniel Zarin, and Peter Jipp, 1992. Economic Returns from
Forest Conversion in the Peruvian Amazon. Ecological Economics, 6 (2): 163-
173.

38

Pingali, P., Bigot, Y., and Binswanger, HP. (1987). Agricultural Mechanization
and the Evolution of F arming Systems in Sub-Saharan Africa. John
Hopkins University Press. Baltimore, USA.

Razzetto, Fernando. 1997. Regimen Legal de Aprovechamineto de Recursos F orestales
Para el Peru. Symposium/Workshop Bosques: Bases para una Nueva Politica.
Organized by El Consejo Nacional del Ambiente (CONAM), July 7-8, 1997,
Lima, Peru.

Reardon, T. and S. Vosti. 1992. Issues in the Analysis of the Effects of Policy on
Conservation and Productivity at the Household Level in Developing Countries.
Quarterly Journal of International Agriculture. 31 (4).

Riesco, Alfredo. 1993. Intensification Tecnologica en la Selva Baja: El Caso de Pucallpa.
Chapter in Desarrollo Rural en la Amazonia Peruana. Edited by William Loker
and Steve Vosti. CIAT and IF PR1. Cali, Columbia.

Rudel, T.K. 1989. Population, Development, and Tropical Deforestation: A Cross-
National Study. Rural Sociology 54: 327-38.

Sadoulet, Elisabeth and Alain de Janvry. 1995. Quantitative Development Policy
Analysis. John Hopkins University Press. Baltimore and London.

Sanchez, P. 1976. Properties and Management of Soils in the Tropics. Wiley, New York.

Smith, Joyotee et al. 1997. Bosques secundarios como Recurso para el desarrollo Rural y
la Conservacion Ambiental en los Tropicos de América Latina. Taller
Internacional sobre el estado actual y potencial de manejo y desarrollo del Bosque
secundario Tropical en América Latina. Pucallpa.3 8p.

Southgate D., Sierra R. and Brown, L. 1991. The Causes of Tropical Deforestation in
Ecuador: A Statistical Analysis. World Development. 19: 1145-51.

Southgate D. 1994. Tropical Deforestation and Agricultural Development in Latin
America In: Brown, K. and Pearce D. (eds.) The Causes of Tropical
Deforestation, the Economic and Statistical Analysis of Factors Giving Rise to the
Loss of Tropical Forests, 134-44. University of College London Press, London.

Toniolo, Angelica and Christopher Uhl. 1995. Economic and Ecological
Perspectives on Agriculture in the Eastern Amazon, World Development,
23 (6): 959-973.

UNEP (United Nations Environmental Program) 1992. The World Environment 1972-
1992. Nairobi, Kenya.

39

Yanggen, David. 1999 Agriculture's Impact on Deforestation in the Peruvian Amazon.
An Analysis of Policy Options to Promote Sustainable Alternatives in the Post
Liberalization Era. Conference Paper Presented for La Economia Agraria
Peruana en la Epoca Pos-Liberalizacion: Resultados de Eficencia y Equidad en la
Costa, la Sierra, y la Selva. 10-11 de Septiembre de 1999 Sala de Conferencias de
GRADE. Lima, Peru.

Yanggen, David. 2000. The Adoption of Improved Pastures and Their Impact on
Deforestation in the Peruvian Amazon. Essay Submitted for Doctoral Dissertation
Thesis Requirement. Department of Agricultural Economics, Michigan State
University. East Lansing, Michigan.

Yanggen, David and Thomas Reardon. In Press. Kudzu-Improved Fallows in the
Peruvian Amazon. In Technological Change in Agriculture and Deforestation.
Arild Angelsen and David Kaimowitz (Eds). CAB International Publishers.
London.

40

 

I New perennial cash crops such as oil palm, palm heart, piyuayo, and camu-camu are being promoted and

may be more successful than past experiences.

2 Farms from the research’s general survey of 220 households.

3 Infrastructural access included locations on the paved highway Frederico Basadre, secondary gravel
roads, tertiary dirt roads, and rivers.

‘ Annuals includes “semi-annual” crops notably plantains and cassava which may last more than a year.

5 With the exception of the capital input use model, there were only a total of four sign changes (none of
variables significant at the .15 level or lower). Further (again with the exception of capital inputs), variables
significant at the .05 level or lower in the CL S models remained significant at the .15 level or better in the
SURE systems approach. The standard errors generated by the two techniques were remarkably similar,
although the SURE standard errors were generally slightly higher. In the case of capital inputs a review of a
simple correlation matrix between inputs and outputs indicated that capital inputs have the highest level of
correlation with other inputs and outputs. This may explain the changes shifting from OLS to SURE.

6 With the SURE technique, all missing variables from any equation in the system result in an observation
being deleted. Combining all the missing observations in this way resulted in only 40 of the initial 220
observations being available.

7 I performed the Hausman test in order to check for simultaneity in the deforestation regression models
between the deforestation dependent variables and the input/technology and output independent variables.

In order to do so I first estimated the reduced form equations for all the input/technology and output
variables. I then took the estimated residual from each reduced form equation and plugged it into each of
the deforestation regression models as a right-hand-side variable. I then performed a t-test on the
coefficient of the estimated residual from the reduced form equation. I used a .10 cut-off as a criteria to
judge for significance.

8 Rainfall comes in a bimodal pattern and there are two cropping seasons. Partners fell forest and let the
felled vegetation dry before burning and then planting crops. The first cropping season lasts roughly from
mid February to mid May and is generally preceded by a relatively dry period in January. The second
cropping season lasts roughly from mid September to the end of December and is preceded by a relatively
dry period from June to early September. Deforestation was measured as the sum of the two forest clearing
periods in January and July-August of 1998. This survey was conducted in October of 1998.

A substantial proportion of primary forest on farm land has been selectively logged. Often this was done
by the timber companies that opened the initial feeder roads into the jungle where farmers subsequently
settled or by the farmers themselves. These areas retain most of their vegetative cover and biodiversity and
are typically indistinguishable from undisturbed primary forest to the untrained eye. They are sometimes
referred to as areas of “residual” primary forest.

'° The consequences of making a type 1 error, rejecting the null hypothesis which is in fact true (that a
variable is not significantly different from zero in this case) may be devastating, for example, when
analyzing the impact of a medicine on human health. Understanding the effect a variable has on
deforestation or production decisions 85% or more of the time is reasonable for most policy decisions.

" Farm household are dominated by small nuclear families. Farmers' adult children, if they remain in
agriculture, tend to start their own farms. This is due i.) to the availability of new land and ii.) to the
relativity low production risks of land-abundant humid tropical agriculture that reduce the insurance role
provided by extended family production networks (Binswanger and McIntire, 1987)

2 Kudzu-improved fallows accelerate the process of soil fertility regeneration. This allows a shortening of
the fallow period. Kudzu also acts as an aggressive cover crop climbing over shrubs and trees retarding
their growth. Shorter fallow periods and retarded vegetative growth lessens land clearing costs. Similarly,
kudzu reduces weeding costs by smothering herbaceous weeds. Yields increase due to nitrogen fixation, the
provision of other nutrients, and weed reduction.

41

KUDZU-IMPROVED FALLOWS IN THE

PERUVIAN AMAZON

A CASE STUDY OF THE IMPACT OF

TECHNOLOGICAL CHANGE ON DEFORESTATION

ESSAY NUMBER 2

42

Wit):

1. Introduction

Some 200-300 million people practice fallow-based slash-and-bum agriculture on
roughly 30 percent of the world’s exploitable soils (Crutzen and Andrae, 1990; Sanchez,
1976). The purpose of a fallow is to allow land to recuperate its productive capacity after
cropping. Forest fallow vegetation extracts nutrients from the air and from deep in the
soil beyond the reach of crops. These nutrients are returned to the topsoil through leaf
litter and when farmers burn fallow vegetation (Sanchez, 1976; Nair, 1993). Fallow
vegetation also shades out herbaceous weeds, a major source of productivity decline (de

Rouw, 1995).

In regions where low population densities and abundant land permit adequately
long fallow periods, slash-and-burn agriculture is a sound way to manage land (Nair,
1993; Kleimnan et al. 1995). However, increasing population pressures, shorter fallow
periods, and the removal of primary forest cover tend to cause declines in agricultural
productivity (Nair, 1993; Thiele, 1993). Farmers can slow down or reverse this decline
by seeding or encouraging the natural regeneration of plant species that help land to
recuperate more rapidly such as leguminous nitrogen fixers (Kang and Wilson, 1987).

This is known as improved fallowing.

Most research on alternatives to traditional slash-and-bum fallows has focused on
the agronomics of accelerating soil fertility recuperation during the fallow period.
Empirical research analyzing the economic rationality of farmer management of fallow-
based slash-and-bum agriculture is lacking (Dvorak, 1992). This essay helps fill that gap
by comparing the costs and benefits of traditional and improved fallowing practices and

assessing how improved fallow adoption affects deforestation. Specifically, it analyses

43

the impact of kudzu-improved fallows on deforestation in the lowland tropical rainforest

areas surrounding the city of Pucallpa, Peru.

The data for this research comes from a survey in 1998 of 220 farm households that
focused on input use, output production, and forest clearing and a smaller survey of 24
households that looked at farmer use and management of kudzu-improved fallows,
traditional forest fallows, and primary forest. Section 2 of this chapter describes the
Pucallpa study zone and the kudzu-improved fallows found there. Section 3 uses survey
data to compare kudzu-improved fallows with traditional slash-and-bum practices in
terms of land use, labor use, and yields. This leads to hypotheses about how kudzu-
improved fallows affect deforestation discussed in section 4. Section 5 contains a
theoretical model and three econometric models of how farmer decisions concerning
output production and input/technology use affect deforestation. Section 6 presents the
regression analysis results and interprets them. This essay concludes with a summary of
results and policy recommendations.

2. The Pucallpa Research Zone and Kudzu-Improved Fallows

This research analyzes the impact of kudzu-improved fallows on deforestation in
the lowland tropical rainforest areas surrounding the city of Pucallpa, Peru. This section
presents the key characteristics of the Pucallpa research zone as well as those of kudzu
fallows.
2.1 Characterization of the Pucallpa Peru Study Zone

Pucallpa is located 85 kilometers from the Brazilian border in the middle of the
lowland rainforest (<500 m.a.s.l.). A paved road that crosses over the Andes Mountains

connects it to the capital, Lima. With a population of approximately 200,000, it is the

T—""T .
l.

second largest and the fastest growing city in the Peruvian Amazon. Low per capita
income, land scarcity, and civil unrest in other regions of the country have been the
principal push factors causing rapid migration into the zone. Infrastructural development,
cheap abundant land, job opportunities, and until recently coca production have been the
major pull factors (Labarta, 1998; Riesco, 1993).

Pucallpa’s economy revolves primarily around timber, gas and oil extraction,
ranching, and annual crops. The principal semi-subsistence crops in the zone are maize,
rice, cassava, plantains, and beans. Cash crops such as oil palm, palm heart, pineapple,
peach palm, cotton, and camu-camu are being promoted in the zone but remain of limited

importance.

In the early 1990’s, the Peruvian government substantially reduced its support for
the agricultural sector. In the Amazon, this included the virtual elimination of subsidized
farm credit and guaranteed floor prices for major cr0ps such as rice and maize. Budget
cuts and a recent history of political violence have limited extension services in the zone,
but those services are now slowly improving. Apart from areas along the main highway,
poor infrastructure results in high marketing costs, while prices for the principal semi-
subsistence crops are low. To obtain secure land tenure, farmers generally do not need to

clear forest. Occupation of unclaimed land typically suffices.

Low soil fertility and high levels of acidity and aluminum content impede crop
growth in the zone. The above ground biomass contains most nutrients (Hecht, 1992).
Farmers use few capital inputs such as fertilizers, pesticides, and plows. The dominant
production practice is slash-and burn agriculture. Most farmers clear forest areas every

year. They typically cultivate a parcel for six months to two years before leaving it in

45

fallow. Average rainfall of around 2000 mm comes in a bimodal pattern, which permits

two cropping seasons per year.

Median farm size is 30 hectares of which primary forest accounts for 31 percent,
forest fallows 30 percent, pasture 25 percent, annual crops 10 percent and perennials 4
percent. Rural population density is low at 7 persons per kmz. Farmland is abundant and
cheap but of low quality for agricultural production. Farmers’ capital and labor resources

are limiting factors of production relative to plentiful land.

2.2 Description of Kudzu-Improved Fallows

survey results indicate that a majority (52%) of farmers has adopted fallows
improved with kudzu (Pueraria phaseloides) in the study zone. This stands in contrast to
the limited adoption of other improved production practices. I chose kudzu fallows as a

case: study in order to understand why farmers have adopted them on a wide scale while
not adopting other improved practices and the effect this technological change has had on
deforestation.

Kudzu is a leguminous vine that farmers plant or that spontaneously regenerates
in areas coming out of annual crop production. It spreads rapidly throughout the fallow
area covering the ground and climbing up bushes and trees. Eventually, farmers once
again slash and burn fallow areas and plant annual crops there.

Research on kudzu in the Peruvian Amazon indicates that it not only fixes
nitrogen but also increases soil available phosphorus, potassium, magnesium and calcium
(Wade and Sanchez, 1983). Furthermore, kudzu’s aggressive growth impedes the
regeneration of secondary forest and the spread of weeds in fallow areas. This

acceleration of soil fertility recuperation and weed reduction allows a shortening of the

46

fallow period. The reduction in secondary forest regeneration and weed presence also
lowers land clearing and weeding labor costs. And finally, kudzu fallows require no
capital investment.

Farmers first began using kudzu about 50 years ago as a cover crop underneath
rubber plantation trees. When rubber production declined, national and international
research centers actively promoted kudzu as an improved pasture. However, grazing and
trampling by livestock rapidly eliminate kudzu, it dries up during the low rainfall months
between June and September, and Brachiaria decumbens, the dominant improved
pasture, tends to out-compete and displace it.

Farmers themselves have been the principal force in adapting kudzu for use in
improved fallows, its current dominant use. Kudzu’s informal diffusion led some
researchers to question whether farmers were truly “adopting” kudzu-improved fallows
or whether it was simply spreading spontaneously. Survey results, however, indicate that
farmers actively manage kudzu in fallow areas. Further doubts have related to the
negative experience in the southern United States where kudzu has disrupted the
ecological balance. In fact, kudzu does not invade primary forest and only temporarily
slows secondary forest regeneration, although it does occasionally invade annual and
perennial crop parcels. Nevertheless, 94 percent of farmers expressed an overall positive
opinion of kudzu due to its improvement of soils, control of weeds, and its use as fodder.

3. Comparative Economic Analysis: Kudzu-improved Fallows vs. Traditional
Slash-and-Burn Agriculture

This section compares the total land use, labor use, and yields of kudzu-improved
fallows with those of traditional fallows and primary forest used in extensive slash-and-

bum agriculture. Farmers in the zone distinguish between two different types of fallows:

47

low fallows “purma baja” and high fallows “purma alta”. The former has secondary
forest vegetation below 5 meters high, whereas the latter has vegetation above this height.
Farmers also have the option of slashing and burning primary forest areas to grow crops.
A key distinction made by this research concerns the choice of farmers whether to clear
secondary forests (high and low fallowareas) or primary forests.

Farmers classify kudzu-improved fallows as low fallows. Trees and shrubs grow
in these areas but kudzu vines climb over them and slow their regeneration. The
shortened fallow period farmers use when working with kudzu-improved fallows further
reduces secondary forest regrowth. Therefore the tress in these fallows may not
regenerate enough to be considered full-fledged secondary forest by some classifications.
However, they grow in areas that would otherwise become secondary forest and perform
many of the same functions as secondary forest with regards to regenerating fertility and
controlling weeds. Therefore, for the purpose of this research, I consider kudzu-improved
fallow areas to be incipient secondary forest.

3.1 Total Land Use

Since shifiing slash-and-bum agricultural systems require farmers to place a
parcel of land in fallow, they must clear and cultivate other land until the fallow area is
ready to be cultivated once again. I use the term “total land use” to refer to the entire area
cleared and cultivated before the farmer can return to his or her initial parcel and
reinitiate the rotational cycle.

One can calculate the total amount of land needed to maintain a given area of
annual crops in production based on the average fallow period and average time farmers

cultivate a parcel before leaving it fallow. The less time that land must be left in fallow to

48

 

ll

restore its soil fertility, the earlier farmers can return there. Therefore, they need less total
land to produce cr0ps. Similarly, the longer a farmer can cultivate a parcel before
abandoning it, the more time fallow areas have to recuperate. This also reduces total land
needs.

Farmers leave land in high and low forest fallows an average of 6.3 and 2.5 years
respectively. For kudzu-improved fallows, the average is only 1.7 years. The average
time farmers cultivate their fields before leaving them in fallow is 1 year for kudzu
fallows and 1.3 and 0.9 years for high and low forest fallows (see table 1).

Based on these figures, one can calculate the total amount of land a farmer would
need to maintain three hectares1 in production using each type of fallow. The principal
simplifying assumptions are that yields are maintained and that only one type of fallow is
used in each scenario. The following formula calculates total land use:

TL = HP + [(FY/Y C) * HP]
TL =Total Land Use HP = Hectares Planted FY = Fallow Years YC = Years Cultivated

The first HP is the original amount of land in production, in this example, the first
3 hectares. The part in brackets is the average needed fallow years divided by the average
years of cultivation multiplied by the hectares planted. This gives the amount of
additional land needed for cultivation until the initial land has recuperated its fertility and
is once again ready to be cultivated. The sum of these two gives the total amount of land
needed in each type of fallow production system.

The principal result of this calculation is that kudzu fallows substantially reduce

the amount of land needed to support annual crop production compared to high and low

49

 

forest fallows. In this simplified model, kudzu fallows reduce land clearing needs by 1 16
percent over high forest fallows and 40 percent over low forest fallows (table 1).
Reducing the total amount of land use in a shifting agricultural production system
represents an intensification of land use. As long as farmers cultivate the same area of

annuals, this will reduce their forest clearing needs.

Table 1: Comparative Anal sis of Total Land Use for Differing Fallow Systems

 

 

 

 

 

 

 

 

Kudzu-Improved High Secondary Low Secondary
Fallows Forest Fallow Forest Fallow
Average Years
of Fallow 1.7 6.3 2.5
Average Years
Of Cultivation 1.0 1.3 0.9
Total Land Clearing:
3ha Annual Crop System 8.1 17.5 11.3
Comparative Land Use
(Kudzu Fallow - Base 100%) 100% 216% 140%

 

 

An important question is the sustainability of these simplified systems. A study by
CIFOR in the Pucallpa zone (Smith et al., 1998), found that fallow vegetation takes
longer to regenerate afier each cropping cycle. A low forest fallow system is likely to
cause rapid land degradation. A high forest fallow system can be sustainable for a longer
period. The sustainability of kudzu fallows appears relatively promising. Of the 24
farmers interviewed in the improved fallow survey, none reported a decline in
productivity over time. However, the majority had been using kudzu fallows for less than
10 years, while four farmers had been using kudzu for ten years, and one each for 12, 21,
and 30 years. Thus at least the medium-term sustainability of kudzu-improved fallows
appears likely.

3.2 Labor Use

Primary forest and secondary forest fallows require substantial labor inputs for

slash-and-bum land clearing. Since kudzu permits shorter fallow periods and suppresses

50

. alvfim-hs
'.

secondary forest regeneration, it reduces the amount of trees and shrubs in the improved
fallow. This reduces land clearing labor needs (table 2). Further, about half the time
kudzu fallows regenerate naturally and for the other half farmers broadcast the seed, so

seeding requires either no labor or minimal labor.

Table 2: Comparative Use of Labor for Land Clearin v (labor days/ha)

 

 

 

 

 

 

 

 

Average Comparative Use
Use (Kudzu Fallow = base 100%)
Primary Forest 26.5 323%
High Forest Fallow 21.8 266%
Low Forest Fallow 13.0 159%
Kudzu Fallow 3.22 100%

 

 

Kudzu also reduces the amount of labor required for weeding. In general, the
more light that penetrates the forest canopy, the greater the presence of herbaceous
weeds. Since little light penetrates the primary forest canopy, these areas have the fewest
weeds, while those cleared from traditional forest fallows have significantly more.
Kudzu, however, is an aggressive cover crop that smothers herbaceous weeds that invade

fallow areas. Areas cleared from kudzu fallows require less weeding labor than other

 

 

 

 

 

 

forest fallows (table 3).
Table 3: Comparative Use of Labor for Weedingflabor days/hectare)
. Rice (t/ha) Comjarison Maize (t/ha) Comparative Use
Primary Forest 3.6 37% 3.1 24%
Egh Forest Fallows 39.7 409% 20.2 159%
Low Forest Fallows 31.6 326% 23.2 181%
Kudzu Fallows 9.7 100% 12.8 100%

 

 

 

 

 

Labor is a limiting factor of production relative to available land. On average,
farmers cultivate annual crops on only 14 percent of land available on their farms
(cropland, fallows, and primary forest). Forest clearing and weeding are two critical labor

bottlenecks in the agricultural calendar in slash-and-bum fallow systems (Thiele, 1993).

51

 

By reducing labor needs at these key moments, farmers can put more land into
production. This could potentially increase deforestation.
3.3 Yields

In slash-and-bum agricultural systems, yields increase when the amount of
biomass burned is higher and hence the quantity of nutrients made available for crop
uptake increases. Primary forest has the most biomass. But many of the trees are too large
to burn well so their nutrients are not released and tree trunks take up field space.
Therefore, according to farmer interviews, high forest fallows often provide the highest
nutrient flush from burning while low fallows typically provide the lowest. Weeds are the
second principal factor affecting yields. Primary forest has the least weed presence
followed by kudzu then by other forest fallows.

Kudzu fixes nitrogen and provides other nutrients that accelerate the regeneration
of soil fertility. And kudzu’s function as a cover crop is effective at reducing weed

competition. The net result of these two factors is a substantial yield increase over both

traditional forest fallows and primary forest (table 4).

Table 4: Comparative Yields (Kudzu Fallow '3 base 100%)

 

 

 

 

 

 

 

 

 

 

 

 

Rice (t/ha) Comparison Maize (t/ha) Comparison
Primary forest 1.6 76% 1.3 76%
High Forest Fallow 1.9 90% 1.5 88%
Low Forest Fallow 1.0 48% 1.4 82%
Kudzu-Improved Fallow 2.1 100% 1.7 100%
4. Hypotheses Concerning Kudzu Fallows and Deforestation

The hypotheses concerning the impact of kudzu fallows on deforestation are
based upon the analysis of their key characteristics in section 3. To review, these

characteristics are shortened fallows, reduced labor cost, and increased yields. The first

two hypothesized impacts are as follows:

52

W: Thanks to its shorter fallow period, kudzu fallows decrease deforestation
by reducing the total land needed in a shifting slash-and-burn production system.

W: Kudzu-improvedfallows easing of labor constraints increases
deforestation by allowing farmers to put more land into crop production.

These two hypotheses predict that kudzu-improved fallows have two contradictory
effects on deforestation. A priori therefore the net impact on total deforestation is
uncertain.

I have defined fallows in general as secondary forest areas and kudzu fallows as a
type of (emerging) secondary forest. The adoption of kudzu-improved fallows decreases
the costs (land clearing and weeding) and increases the benefits (yields) of clearing
secondary forest relative to clearing primary forest. This leads to hypothesis 3:
W: Changes in the relative costs and benefits of secondary versus primary
forest clearing due to kudzu fallows increase secondary forest clearing and decrease
primary forest clearing.

In sum, the labor cost and fallow length impacts of kudzu fallows operate in
different directions on total forest clearing giving an ambiguous net outcome. However,
kudzu fallows change relative costs and benefits in a way that encourages secondary
forest clearing and discourages primary forest clearing.

5. Theoretical and Empirical Models
5.1 Theoretical model

This theoretical section provides a conceptual framework for understanding the
relationship between agriculture and deforestation. Deforestation is not a primary activity
of farmers. This would be the case if farmer deforested in the process of harvesting

timber or other forest products. Farmers do selectively extract and sell hi gh-value

53

$3.1.

hardwood and other species from primary forest areas of their land holdings. However,
because the timber harvesting is selective, the “residual” primary forest remains intact
with a relatively small proportion of the biodiversity having been removed.

Farmers practice slash-and-bum agriculture in order to convert vegetative biomass
into ashes that provide nutrients accessible for crop uptake. This results in a clear cutting
type of deforestation. Deforestation is therefore an outcome of agricultural activities, a
means to the end of agricultural production. In order to understand the deforestation
outcome, it is therefore necessary to understand how farmers make agricultural
production decisions.

The theoretical framework for understanding farmer agricultural production
decisions uses a profit function approach. This approach assumes that farmers choose a
combination of variable inputs and outputs in order to maximize profit subject to a
technology constraint i.e. the production function of the farm. Farmer decisions about
what to produce (outputs) and how to produce it (inputs/technologies) determine the
impact of agriculture on deforestation. This is the basis for the econometric regression
modeling in the following section.

Profit maximization should be understood in the broad sense of farmers using
limited resources in as efficient a manner as possible to meet their livelihood (including
social) objectives. Furthermore, strict profit maximization behavior is not a necessary
prerequisite for specifying systems of outputs supply and factor demand equations, as
long as the behavior of the individual agents is sufficiently stable over time and can be
aggregated over farmers (Sadoulet and De Janvry, 1995).

The profit function for a farm is specified as follows:

54

(l) 1t=pq - wx
where p is a vector of prices for outputs, q is a vector of outputs produced by the farmer,

w is a vector of input prices, and x is a vector of inputs.

Farmers maximize profits subject to the production function which represents how

farmers use fixed inputs (2) and variable inputs (x) to produce outputs (q):
(2) q = f (m)

Variable inputs are those factors a farmer can change in the short run (e.g., fertilizer,
seed, labor, etc.). Fixed factors are those that can’t be adjusted in the short run. Fixed
factors include private factors (e.g., land, education, family labor, etc.) public factors
(e.g., infrastructure, credit, extension, etc.) and exogenous factors (e.g., soils and market
distance).

The farmer maximizes profits by choosing the optimal level of inputs x and
outputs q. The input demand and output supply functions can be represented as follows:

(3) x = x(p,w,z) and q = q(p,w,z)
This indicates that the optimal levels of inputs and outputs are a function of output price,
input price, and fixed factors. Combining equations (1) and (3), the profit function can be
rewritten in the following manner:
(4) n = pq(p,w,z) - “(p,w,z)
Using Shepard’s Lemma, differentiation of equation (4) with respect to output and

input prices in the following manner derives the output supply and input demand

functions:

(5) dn/dpg (p,w,z) = q, dn/dwj(p,w,z) = -x.

55

The principal cause of deforestation is growing demand for crop and pasture land.
However, not all agricultural products and agricultural production technologies have the
same impact on deforestation. It is therefore necessary to understand the determinants of
agricultural production and how agricultural production affects deforestation. The
econometric modeling that follows analyzes the determinants of kudzu-improved fallow
adoption and its impact on deforestation.

5.2 Econometric model

Based on the theoretical model above, this research uses two sets of econometric
models. The first set analyzes the determinants of output production and input/technology
use. The second set analyzes the impact of output production and input/technology use on
deforestation. Outputs are divided into three principal categories: annual crops, perennial
crops, and livestock. Inputs and production technology include improved fallows,
improved pastures, natural pastures, capital inputs and hired labor. Deforestation is
measured as the hectares of forest felled by a farm household in 1998. Three deforestation
variables are examined: total forest (primary plus secondary forest), primary forest, and

secondary forest deforestation.

The econometric modeling for the input and output regressions uses the ordinary
least square method (OLS). A system approach using SURE gave similar results3 but
sharply reduced the number of available observations‘. The deforestation regressions use
recursive modeling due to the unidirectional causality running from agricultural
production to deforestation hypothesized in the preceding theoretical section. The
Hausman specification test generally verified this hypothesiss. The technique of two-stage

least squares corrected for simultaneity when necessary.

56

The econometric models are specified as follows. Exact definitions of the

variables are found in appendix 1.

OUTPUTij = f (farm sizei, secondary forest,, primary forest;, alluvial soilsi, years in loti,
educationi, agei, family laborg, origini, off-farm incomei, crediti, extensiong, farm-gate price

maizei, land tenure,)

j = annuals, perennials, and livestock
i = 1,..., N observations

INPUTij = f (farm sizei, secondary foresti, primary forest, alluvial soils, sandy soils;, years
1111015, education;, family labori, off-farm incomei, distance social services, credit, market

distancei, land tenure;)

j = kudzu fallows, improved pastures, natural pastures, capital inputs and hired labor
i = 1,..., N observations

DEFORij = f (annuals, perenniali, livestocki, kudzu fallowsi, improved pasturei, natural
pasturei, capital input user, hired labori, family labori, land tenure;, education,, forest,,
alluvial soilsi, forest product income;

j = total, primary, and secondary forest deforestation in 1998.
i = 1,..., N observations

6. Model Results and Interpretation

The results presented here focus on the adoption of kudzu-improved fallows and
the deforestation hypotheses presented in section 4. The four regression model results
presented in table 5 are for kudzu fallow adoption, total deforestation, primary forest
deforestation, and secondary forest deforestation. Due to the cross sectional nature of the
data and the diversity of the agricultural systems, I consider variables as significant up to
the .15 level. I also provide the exact significance levels for the reader’s consideration.
6.1 Kudzu-Improved Fallow Adoption

Kudzu fallow adoption is positively and significantly correlated to farm size,
distance to social infrastructure (schools and health posts located in population centers)

and distance to markets. Adoption therefore does not appear to be due to a closing land

57

frontier. The traditional scenario of intensification due to land scarcity, increasing
population, and proximity to population centers (Boserup, 1965, 1981; Pingali et a1.
1987) does not appear to explain the adoption of this land productivity enhancing
technology.

On the other hand, farms with parcels that have been in production longer and
with less primary forest have a significantly higher probability of adopting kudzu fallows.
Both of these variables are indicative of decreasing land quality. The longer a farm has
been in production, the greater the soil nutrient depletion (fallows take longer to
regenerate). And the less remaining primary forest, the greater the depletion of above
ground nutrient stocks contained in the vegetative biomass.

In the land abundant environment around Pucallpa one can not yet talk of a closing land

frontier in quantitative terms. Farmers typically cultivate only a small pr0portion of their
farms and/or have access to nearby land. However, given that farmers are subjecting the

region’s fragile tropical soils to slash-and-bum agriculture with declining fallow periods,
land quality increasingly constrains production. Much of the adoption of kuzu-improved
fallows can be related to a closing of the land “quality” frontier.

Three other significant variables of note are the positive correlation of kudzu
fallows with education and sandy soils and the negative correlation with credit.
Education’s positive correlation indicates that understanding the process of nitrogen
fixation and the management of kudzu in an improved fallow system requires a relatively
higher level of knowledge. This is particularly true given that kudzu fallows have tended

to spread informally by farmers without much support from extension services.

58

Furthermore, higher educational levels may give a greater opportunity cost to labor. More
educated farmers may adopt labor-saving kudzu fallows in order to free themselves up to
benefit from increased opportunities to work off farm.

The negative correlation with credit makes sense in that kudzu-improved fallows
enhance productivity without using capital and decrease labor requirements. Thus, they
offer an attractive alternative for increasing yields for farmers without access to credit.
The positive correlation with sandy soils probably relates to the agronomic conditions in
which kudzu flourishes.

The overall significance of the model is reasonably good (R2 .39 Adj R2 . 26) for
this type of cross-sectional adoption analysis. This is particularly true given the
multipurpose nature of kudzu in the zone. Farmers use kudzu to improve fallows, as
pasture/fodder and as a perennial cover crop. The research survey may not always have
clearly distinguished between these three uses and that may have reduced the model’s
predictive power.

6.2 The Impact of Kudzu-Improved Fallows on Deforestation

To analyze the impact of kudzu fallow on deforestation rates, I examine the sign
and significance of the kudzu fallow variable in the deforestation regression models. The
deforestation models contain numerous interesting results and significant variables.
However, for the focus of this essay, I limit the analysis to the kudzu fallow variables and
others pertinent to the analysis of the impact of kudzu fallows on deforestation. First, I
briefly review the three hypotheses concerning kudzu fallows.

i.) Decreased fallow periods reduce deforestation.
ii.) Easing labor constraints increases deforestation

59

3.82.. m A. .8.

8 Davaooom am... 3.

.25. 2. 2.. a 685:»...- .

.32 m... 2.. 3 2855.6

**

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

822353 2. 35.0508 .2 a. .22 .o. 2.. a 28.255 t.

m... 3.... N... .2... - - is... .85... a... .
«A. 8...- ..... 42.... - - .85... has...
N... 8.... 2. .3...- - - .22.... 32.8%

- - - - a... 9...... at... .afi a. ......- :33... each
S. 8...... S. 2..-... me. ”8... 23m 32...... m... 2...... 85...... .82....
S. 3....- m... 3.... a... 8...... £53... 22.; ..... 3..-. :85
mm. 2... 3. 8... N... 8...... 558...... 2. ..v... axiom room a...
S. 2...... a... .N... .m. 8.... Se... 3.5... a... 8... 2:8... ESE...
N... a... a... 2... 2.. $2... .33 3.... ..N. a... 3...... N=5...
v... 5.... S. 2..- n... 8...-.. 3...... .535 .... 8...... 5:83...
m... 3..-: m... .2..- m.. 42....-. use... .332 S. .m... .3 5 23>
2. .2.-. m... 2....- m... «.8...- .232... .52: E. 8.... 2.8 Exam
e... - R... a... 3..-. .v. 3.... .32.... .55. a... 8...- 2.8 .22....
v... 3.... m... ..N... ..m. 9...... .338»... a... 2..-. .33... has...
a. .3...- 8. ..2..- .... 8..-. 2322... 3.. 8...- .33... 52.8.5
8. ...n... 2.. 2a.... 2.. 8...... 2.552 8. 2a.... .5 a...
«a. v... S... 282.5 8. 2.3.56

.35 L2... ..Iamw ...5 ..léw ..8... 2% ...So

mu. y. .2. ..m. m 2.... =3 on... on... ...._w 3.... 3.... ...a «at...
5:58.80Q 293... confine-Suva 5:83....qu 833:5 use—Em anvzx 855.3.» «33.5.»
tau—.85 30......” .0255 13.. P

”3:58. :22: oz 5:88.539 2.4... .8: 22 32:... 326— .m 95:.

 

60

iii.) Changing the relative costs and benefits of land clearing in favor of secondary
forest leads to declining primary forest clearing and increasing secondary forest

clearing.

The three deforestation regression models examine total, primary, and secondary
forest deforestation. This permits an examination of the impact of kudzu fallows on each
type of deforestation. While it is not possible to empirically separate out the two opposite
effects posited in the first two hypotheses using regression analysis, the total
deforestation regression model allows an estimation of which effect may be stronger. The
primary and secondary forest deforestation models allow a direct analysis of the third
hypothesis.

The sign of kudzu-improved fallows in the total deforestation model is positive
through not significant. This may indicate that kudzu fallows’ easing of labor production
constraints increases deforestation more than shortened fallow periods decrease
deforestation. The importance of the labor constraint (and hence its casing) is indicated
by the positive and highly significant correlation of hired labor in the total deforestation
model and family labor in the secondary forest deforestation model. These indicate that
increased availability of labor has a strong positive effect on deforestation. Easing the
labor constraint via the adoption of a labor-saving technology should therefore have the
same impact.

When total deforestation is broken down into secondary and primary forest
deforestation, the kudzu fallow variable becomes significant. Kudzu-improved fallows
are negatively correlated with primary forest deforestation and positively correlated with

secondary forest deforestation. This supports hypothesis 3. The changes in the relative

61

costs and benefits of land clearing associated with kudzu fallows lead farmers to reduce
primary forest clearing and increase secondary forest clearing.
7. Conclusions and Implications for Policy and Technology Development

This research has analyzed one particular technology (kudzu-improved fallows) in
one specific setting (the lowland Amazon surrounding Pucallpa, Peru). Perhaps the most
basic conclusion drawn from this analysis is that the impact of technological change on
deforestation depends fundamentally on the profile of the new technology and the
biophysical and socio-economic characteristics of the zone where it is introduced.

It would be wrong to assume that the introduction of any and all improved fallows
in the Pucallpa zone would increase total deforestation. For example, another type of
improved fallow may have greater labor requirements and not free up this limiting
production factor. Or, if kudzu-improved fallows were introduced in a socio-economic
environment where labor was not a constraining production factor, deforestation again
might not increase. There is therefore a need for both site-specific and technology-
specific analysis to understand the impact of a technology on deforestation.

This research has also shown that agricultural technology and land clearing
patterns may interact in a complex fashion. Kudzu fallows simultaneously ease the
limiting labor production factor (increasing deforestation) and shorten needed fallow
times (decreasing deforestation). In addition, they have opposite effects on the clearing of
primary and secondary forests. Researchers need to recognize that new technologies may
simultaneously affect deforestation in several distinct ways and may need to undertake a

more disaggregated analysis of these effects.

62

 

Kudzu improved-fallow adoption has been successful precisely because it fits
with the relative factor scarcity of the zone and provides a superior alternative to slash-
and-burn agriculture (by reducing labor while increasing yields). Many other improved
production technologies focus on reducing deforestation via soil conservation and/or
increasing productivity. While this approach is not wrong per se, it may blind researchers
to the fundamental fact that labor and capital, and not land, are typically the main
production constraints in agricultural frontier contexts. Under such circumstances,
attempts to get farmers to conserve land with technologies that require greater use of
capital and/or labor are likely to fail. The general lesson is that soil conservation is not a
primary objective of farmers in a land abundant environment. Soil fertility enhancing
technologies will only be attractive to farmers when their costs and benefits are superior
to the current practice of extensive slash-and-bum agriculture.

A key factor in this analysis has been labor availability. Labor is a limiting
production factor in the Pucallpa zone. Basic economic theory tell us that farmers attempt
to maximize returns to scarce production factors. Kudzu-improved fallows have been
successful precisely because they increase labor productivity. Herein lies a paradox:
the reduced labor requirements that encourage adoption of this land conserving
production technique also free up labor so that overall deforestation increases. One
possible solution might be to introduce high value crops such as certain perennial and
horticultural crops that demand a lot of labor but still increase the return to scarce labor
resources.

In general, improved fallows appear to be a technology with important potential

for reducing deforestation. In the simple model, kudzu-improved-fallows decreased total

63

land clearing needs by 116 percent and 40 percent compared to high and low forest
fallows respectively. The challenge of future research is to find creative ways to harness
this potential and minimize the negative impacts.

Designing improved fallows that use more labor while increasing returns to labor
might also help resolve the “labor paradox”. Improved fallows that produce a useful
product at the same time they fix nitrogen and provide other nutrients for the soil may
achieve these dual goals. Agroforestry fallows systems merit consideration in this
respect.

For example, farmers can plant fast growing leguminous trees in association with
annual crops. For the first year and a half they weed the trees together with the annual
crops. This reduces the key labor weeding constraint associated with the installation of
tree plantations. By the time the farmer abandons annual cropping, the trees are
developed enough to survive in the fallow without weeding. These trees help recuperate
land productivity while providing useful products that absorb farmer labor. For these
systems to succeed, the initial labor and capital costs should be minimized (e.g., by using
bare root bed tree nurseries, direct seeding, planting in association, etc.) and the
secondary tree products must have a high value for home consumption or commercial
sale. Research should focus on finding creative ways to reduce the costs of improved
fallows and on identifying tree species that combine soil amelioration with the provision
of valuable secondary products.

In the specific case of kudzu-improved-fallows, they increase secondary forest
clearing but reduce primary forest clearing. Although higher secondary forest clearing is

not a desired outcome, the reduction of primary forest clearing is clearly a positive

environmental impact, since these forests typically provide the greatest amount of
environmental services. And even though secondary forest clearing increases, the casing
of the labor constraint allows total production to rise and that helps reduce poverty.

This research, to some extent is a case study of trade—offs often necessary between
different goals. In this case, primary forest deforestation and poverty decreased but
secondary forest increased. An economist cannot scientifically evaluate the worthiness of
these particular trade-offs. This analysis can, however, help policy makers understand the
nature of the trade-offs in order to make better informed decisions. Moreover, this type of
analysis can aid in analyzing strategies for changing trade-offs to achieve outcomes
desired by policy makers.

A final set of conclusions and policy recommendations deals with where kudzu
and other improved fallows are most likely to be successful and how this relates to
deforestation. Kudzu-improved fallows are not associated with the typical intensification
scenario of land scarcity nearer to urban centers. Kudzu-improved fallow adoption
increases on farms that have been in production longer and on those with less primary
forest. It is the land quality constraint and not the land quantity constraint that leads
farmers to adopt kudzu fallows. This knowledge can help extension services save their
limited resources by targeting farms for introduction of kudzu and other improved
falloWs where they are most likely to be successful.

In addition, a CIFOR study (Smith et al., 1998) in the Pucallpa zone and another
by Schelhas (1996) in Costa Rica found that farmers in older more developed frontier
zones with less primary forest put a higher value on preserving it. The products and

services provided by the remaining primary forest acquire a higher scarcity value.

65

Improved fallows are likely to be particularly well received in these zones not only
because of declining soil fertility, but also because of a stronger desire to preserve
remaining primary forest. Thus, farmers in older settlement areas may be particularly

receptive to adopting improved fallows in order to conserve the remaining primary forest.

66

REFERENCES

Boserup, E. (1965). The Conditions for Agricultural Growth: The Economics of
Agrarian Change under Population Pressure. Aldine Publishing Company,

New York.

Boserup, E. (1981). Population and Technological Change: A Case Study of Long Term
Trends. University of Chigago Press. Chicago.

Crutzen, PJ. and Andreae, MO. (1990). Biomass Burning in the Tropics: Impact on
Atmospheric Chemistry and Biogeochemical Cycles. Science. 250, 1669-1678.

De Rouw, Anneke (1995). The fallow Period as a Weed-Break in Sifting Cultivation
(Tropical Wet Forest). Agriculture, Ecosystems and Environment 54, 31-43.

Dvorak, K.A. (1992). Resource Management by West African Farmers and the
Economics of shifting agriculture. American Journal of Agricultural Economics

74, 809-814.

Hecht, SB. 1992. Logics of Livestock and Deforestation: the Case of Amazonia. In
Development or Destruction: The Conversion of Tropical Forest to Pasture in
Latin America. Downing T.E., Hecht S.B., Pearson H.A. and Garcia-Downing

(eds). Westiver Press. Boulder. Pp 405. (95-113).

Kang, RT. and Wilson, GP. (1987). The Development of Alley Cropping as a Promising
Agroforestry Technology. In: Steppler, H.A. and Nair, P.K.R. (eds.),
Agroforestry: A Decade of Development. ICRAF, Nairobi, Kenya, pp. 227-243.

Kleinman, P.J.A., Pimentel, D., and Bryant, RB. (1995). The Ecological Sustainability of
Slash-and-Burn Agriculture. Agriculture, Ecosystems and Environment 52, 235-

249.

Labarta, Ricardo (1998). Los-Productores de la Cuenca Amazonica del Perri y la
Dinamica de Uso de la Tierra: Resultados de la Caracterizacion de Pucallpa y

Yurimaguas. ICRAF Report. Pucallpa, Peru.

Nair, P.K. Ramachandran (1993). An Introduction to Agroforestry. Kluwer
Academic Publishers in Cooperation with the International Center for Research in

Agroforestry. London.

Pingali, P., Bigot, Y., and Binswanger, HP. (1987). Agricultural Mechanization and the
Evolution of Farming Systems in Sub-Saharan Africa. John Hopkins University

Press. Baltimore, USA.

67

Riesco, Alfredo. (1993). Instensificacion Tecnologica en la Selva Baja: El caso de
Pucallpa. In: Loker W. and Vosti, S. (eds.) Desarrollo Rural en la Amazonia
Peruana. CIAT and IFPRI. Cali, Colombia.

Sanchez, RA. (1976). Properties and Management of Soils in the Tropics. John Wiley,
New York, USA.

Sadoulet, Elizabeth and Alain De Janvry. (1995). Quantitative Development Policy
Analysis. John Hopkins University Press. Baltimore.

Schelhas, J. (1996). Land Use Choice and Change: Intensification and Diversification in
the Lowland tropics of Costa Rica. Human Organization 55, 298-306.

Smith, Joyotee. Petra Van der Koop, Kenneth Reategui, Ihnacio Lombardi, Cesar
Sabogal, Armando Diaz. (1998). Dynamics of secondary forest in slash-and-bum
farming: Interactions among land use types in the Peruvian Amazon. Proyecto
Manejo de Bosques Secundarios en America Tr0pical. Convenio
CIFOR/CATIE/BID. Pucallpa, Peru.

Thiele, Graham (1993). The Dynamics of Farm Development in the Amazon: The ‘
Barbecho Crisis Model. Agricultural Systems 42, 179-197.

Wade, MK. and Sanchez, RA. (1983). Mulching and Green Manure Applications for
Continuous CrOp Production in the Amazon Basin. Agronomy Journal 75, 39-45.

68

Years in Lot

Sandy Soils
Alluvial Soils
Farm size
Secondary Forest
Primary Forest
Forest

Family Labor
Education

Forest Prod. Income

Off-farm Income

Age

Annual
Perennial
Livestock
Kudzu Fallows

Nature Pasture
Improved Pasture

Capital Input Use
Hired Labor

Land Tenure
Market Distance

Dist Social Serv’s
Credit
Extension

F m-Gate Pr Maize

APPENDIX 1: VARIABLE DESCRIPTION

Number of years a farm has been in production (past & current
owners)

Dummy variable, sandy soils dominant on farm

Dummy variable, farm located in alluvial soil zone

Total hectares of a household’s land holdings

Hectares of secondary forest fallow on a household’s land holdings
Hectares of primary forest on a household’s land holdings
Hectares of total forest on a household’s land holdings

Number of family members 14> years old working on the farm
Dummy variable, household head has secondary education or
higher

Value (in soles) of products harvested from the primary forest in
the previous year

Value (in soles) of off-farm income earned by a household in the
past year.

The age of a household head

Hectares of annual crops on a household’s land holdings

Hectares of perennial crops on a household’s land holdings

Head of cattle owned by a household

Hectares of kudzu-improved fallows on a household’s

land holdings .

Hectares of natural pasture on a household’s land holdings
Hectares of pasture improved with Brachiaria on a household’s
land holdings

Value (in soles) of capital inputs used in the past year by a
household

Days of paid labor used on farm (includes labor exchange “minga”)
Dummy variable, household head perceives tenure as insecure
Household’s distance in kilometers to the principal market
Pucallpa

Household’s distance in kilometers to nearest (school + health post)
Dummy variable, household received credit in the last five years
The number of extension visits a household received in the past
year

The price for maize received by a household minus transportation
and labor marketing cost

69

 

' This is the average amount of annual crops cultivated by a farm household in the study area

2 This includes labor for collecting kudzu seeds and seeding a kudzu fallow.

3 With the exception of the capital input use model, there were only a total of four sign changes (none of
variables significant at the .15 level or lower). Further (again with the exception of capital inputs), variables
significant at the .05 level or lower in the CL S models remained significant at the .15 level or better in the
SURE systems approach. The standard errors generated by the two techniques were remarkably similar,
although the SURE standard errors were generally slightly higher. In the case of capital inputs a review of a
simple correlation matrix between inputs and outputs indicated that capital inputs have the highest level of
correlation with other inputs and outputs. This may explain the changes shifting from OLS to SURE.

‘ With the SURE technique, all missing variables from any equation in the system result in an observation
being deleted. Combining all the missing observations in this way resulted in only 40 of the initial 220
observations being available.

5 I performed the Hausman test in order to check for simultaneity in the deforestation regression models
between the deforestation dependent variables and the input/technology and output independent variables.
In order to do so I first estimated the reduced form equations for all the input/technology and output
variables. I then took the estimated residual from each reduced form equation and plugged it into each of
the deforestation regression models as a right-hand-side variable. I then performed a t-test on the
coefficient of the estimated residual from the reduced form equation. I used a .10 cut-off as a criteria to

judge for significance.

7O

THE ADOPTION OF IMPROVED PASTURES

AND THEIR IMPACT ON DEF ORESTATION

IN THE PERUVIAN AMAZON

ESSAY NUMBER 3

71

1. Introduction
1.1 Environmental and Economic Impacts of Livestock

Most rainforest cleared in the tropics of Latin America eventually ends up as
pasture (FAO, 1993; Houghton et al., 1991). The environmental impact of this
transformation is stark. The conversion of one hectare from rainforest to pasture results in
the reduction of hundreds of tons of biomass containing thousands of species over 40
meters of vertical space into a field containing a few native and exotic species compacted
one to two meters high. Some 170 tons of carbon per hectare are released into the
atmosphere potentially aggravating global warming (Mattos and Uhl, 1994).

The social and economic benefits derived from tropical forest clearing for
ranching appear limited. According to Ledec (1992), per land area deforested, cattle
raising produces much smaller benefits than other production systems-whether these
benefits are measured as foreign exchange, financial revenues, employment, calories, or
animal protein. His data indicate that a dollar's worth of beef requires the deforestation of
between 37 and 119 times more land than a dollar's worth of coffee, 15-24 times more
land than sugar, 34—43 times more land than cotton, and about 145 times more land than
bananas.

Nevertheless, cattle raising in tropical Latin America offers advantages that make
it attractive to small and large farmers alike. Cattle is a relatively low risk economic
activity compared to other land uses both in terms of production and market price
(Feamside, 1990; Seré and Jarvis, 1992, Schelhas, 1996). In addition, cattle have more
flexibility in the timing of their sale than most agricultural crops and farmers can walk

the product to market in areas with inadequate infrastructure (Hecht, 1993; Seré and

72

 

 

 

 

that

not

 

 

Jarvis, 1992). Furthermore, cattle production can take place on degraded land that is
unproductive for other agricultural activities (Hecht, 1993; Sere’ and Jarvis 1992; Thrupp,
1980).

At the heart of the mixed picture of cattle's environmental and economic impacts
in the tropics lies cattle's extensive use of land and low labor requirements. In Brazil for
example, a typical large cattle ranch uses only one cowboy for every 1500 hectares of
cleared land (Hecht, 1993). So while returns to land may be low, returns to labor tend to
be quite high. This use of resources reflects the fact that land is relatively abundant and
cheap relative to available labor. Maximizing returns to the limiting factor of production
is a rational response that maximizes profitability. This private rationality nevertheless
entails a heavy social cost particularly in environmental terms.

In the Peruvian Amazon, 46 percent of farmers own livestock and 95 percent of
these have less than a hundred head of cattle (IN El, 1986). These farmers tend to be very
poor and depend upon livestock to meet basic material subsistence needs. Thus short-
terrn survival needs take precedence over longer-terrn environmental costs. The reality is
that, for better or for worse, ranching has an established presence in the Amazon that will
not soon go away (Mattos and Uhl, 1994; Serrao and Toledo, 1990; Larbarta, 1998).

1.2 Sustainable Livestock Production in the Amazon?

Given the likelihood that ranching will remain in the region, is there a hope for

improving the sustainability of livestock production? This section introduces the key

issues related to sustainable livestock production in the Amazon.

73

1.2.1 Intensification in the Land-Abundant Tropics

Research has shown that intensification of agriculture and ranching is
agronomically feasible on lowland tropical soils (Rao et al., 1993; Serrao and Toledo,
1992; Smith, 1992). Seré and Jarvis (1992) and Ledec (1992), however, point out that the
track record of livestock intensification efforts in tropical Latin America is poor and that
extensive land-using production practices remain the norm. Numerous studies have
shown that, given the abundance of cheap land, extensive production practices are more
profitable than intensification. (Hecht et. al 1988; Serrao and Toledo, 1990; Griffith and
Zepeda, 1994; Munoz, M., P. Oderrnatt, and J. Reyes, 1994)

Boserup (1965 and 1981) maintains that intensification of agricultural production
results from increasing population pressures on a limited land base. As a result, land
becomes more valuable relative to other inputs, and increases in land productivity
become desirable. The increase in the value of land relative to labor stimulates an
increase in input use per hectare of land (Smith et a1. 1994).

In the Peruvian Amazon, however, land remains abundant and cheap relative to
labor and capital. Induced technological innovation theory (Hayami and Ruttan, 1985)
states that the adoption of agricultural technologies will tend to reflect the factor scarcity
of the zone. Farmers tend not to adopt technologies that conserve an abundant resource
(land) by using relatively scarce resources (capital and/or labor). Indeed Boserup (1965)
argues that agricultural intensification is unlikely as long as a frontier of unexploitated
agricultural land persists. Nevertheless, widespread adoption of pastures improved with
grasses from the brachiaria genus (principally Brachiaria decumbens) has substantially

increased average pasture can'ying capacity in the zone of Pucallpa, Peru.

74

“$1.14 nit-

1.2.2 The Impact of Improved Technologies on Deforestation

Pasture degradation has been the rule throughout the Amazon region (Serrao and
Toledo, 1990). This degradation has led to the frequent abandoning of pastures within ten
years of their installation (Mattos and Uhl, 1994; Hecht, 1982, 1985). This abandonment
often forces ranchers to move further out into the agricultural frontier and clear more
forest for pastures. Buschbacher (1986) has referred to this dynamic of land use as
"shifiing ranching".

Numerous authors have stated that intensification of livestock production has the
potential to reduce deforestation (Serrao and Toledo, 1992, 1993; NRC, 1993; Steinfeld
et a1. 1998; Seré and Jarvis, 1992). They argue that improved pastures would obviate the
need to abandon degraded pasture and clear new forest. In addition, adapted forage
gerrnplasm that permits farmers to raise more cattle using less land would further
reducing forest clearing (Serrao and Toledo, 1990; Seré and Jarvis, 1992).

Other authors, however, point out the potential of improved pastures (Kaimowitz,
1996) and productivity-enhancing technologies in general (Tomich et al., 1999) to
increase deforestation. If these technologies increase the profitability of agriculture, they
also increase the opportunity cost of conserving natural forest. Increased returns may
encourage both current farmers and new migrants to expand production into the forest
margins. In fact, very little empirical work documents the impact of technological change
in agriculture on deforestation (Kaimowitz and Angelsen, 1998).

1.3 Paper's Objectives and Organization
Given that agricultural intensification typically does not take place in land

abundant areas, the first objective of this paper is to understand the widespread adoption

75

of improved pastures by farmers in the Pucallpa zone. Secondly, given the lack of
empirical evidence concerning the impact of technological change in agriculture on
deforestation, what effect has the adoption of this land-intensifying technology had on
forest clearing?

This paper is organized in the following manner. Section 2 describes the key
characteristics of the Pucallpa, Peru research zone. An analysis of the evolution of land
use patterns on farm land holdings in section 3 gives insights into the process of
deforestation over time. Section 4 compares the socio-economic characteristics of
brachiaria-improved pastures to natural pastures and livestock production with
brachiaria pastures to annual and perennial crop production. Section 5 presents a
theoretical model that uses a profit function approach to outline the relationship between
the determinants of farmer adoption of livestock and pasture and how these production
decisions in turn effect deforestation. This theoretical discussion provides the basis for
the specification of three econometric models that examine the determinants of livestock
and pasture adoption and the effect these have on deforestation. The results of the
econometric models are discussed in section 6. Because the econometric models use
cross-sectional data, they capture the current dynamic of agricultural production and
deforestation. Section 6 therefore complements these models with an examination of the
historical impact of livestock and pastures on deforestation using correlation analysis.
Section 7 presents conclusions and policy recommendations.

1.4 Research Data
The data used in this essay comes from two surveys led by this author in 1998.

The main survey interviewed 220 households concerning their agricultural production

76

practices, the socio-economic characteristics of their household, and the biophysical
characteristics of their farm. This survey used the distance from the main market town
and the type of infrastructural accessl available to farmers as the criteria for sample
stratification.

This research principally uses the main survey data for the econometric modeling.
A second follow-up survey of 38 households with pasture collected more detailed data
concerning farmers’ experiences with natural pastures, brachiaria-improved pastures,
and livestock. This second survey provides information for this essay’s section on the
socio-economic characterization of natural and brachiaria-improved pastures.

2. Characteristics of the Research Zone

This essay’s research zone is in the lowland tropical rainforest surrounding the
city of Pucallpa, Peru. This relatively new frontier area was first connected to the capital,
Lima, approximately 60 years ago when Pucallpa was little more than a village of 2000
inhabitants. Pucallpa's population has since grown to approximately 200,000 making it
the second largest urban center of the Peruvian Amazon.

Nevertheless, the Pucallpa zone remains a land abundant environment. Population
density is low at 7 individuals per kmz. Data from this research's main survey indicates
that median farm size is 30 hectares. Only 39 percent of farmland is in some form of
agricultural use (pastures 25%, annuals 10%, perennials 4%) with the remaining 61
percent in secondary forest fallows (30%) and primary forest (31%). In addition, new
lands areas are continuously made accessible by feeder roads opened up for timber, gas,

and oil extraction.

77

Virtually all farmers continue to practice shifting slash and burn agriculture for
annual crop production. The predominant crops are maize, rice, cassava, and plantains.
23 percent of farmers have livestock holdings. Of these 96 percent are small holders
owning 50 head of cattle or less and 92 percent of cattle owners have mixed farming
systems including the production of annual crops.

Although only 23 percent of all farmers own cattle, 49 percent have installed
pasture on their farmland. A majority of farmers with pasture (52%) therefore have no
cattle. Virtually all these farmers stated their intention was to acquire cattle. In general,
farmers state that their primary future production objective is to increase cattle production
(Labarta, 1998). There is little evidence of farmers planting pasture for land speculation,
as was frequently the case in Brazil (Hecht, 1985). Numerous well located ranches on
the principal highway leaving Pucallpa had for sale posted signs during 1998-99 without
evidence of many sales taking place. And, apart from a small program of rotating in-kind
livestock creditz, the Peruvian government does not actively promote cattle production in
the zone.

Pucallpa does not have a milk processing plant. Most milk sold in the zone is
brought in from other parts of the country in evaporated form. Efforts by both national
and international agricultural research centers to promote the production of milk have met
with limited success. Most cattle are raised for beef production and milk is generally for
home consumption. The evolution of pasture use in the zone has been markedly in the
direction of a reduction in natural pastures and an increase in pastures improved with
brachiaria. Brachiaria-improved pastures have also come to dominate other types of

improved pastures (graph 1).

78

In sum, the zone surrounding Pucallpa can be characterized as a relatively early
frontier area. Population density remains low, land is cheap and abundant, and market
development is limited. Furthermore, virtually all livestock owners are small holders with
mixed production systems. Yet despite this relatively early stage of frontier development

and a large surplus capacity of pastureland, farmers have intensified the productive

 

o 9 Evolution of Pasture Types capacity of their pastures

 

a Brachiaria Pasture

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.8 -- . .
8 o 7 - I Natural Pasture -:-.=;._ by adopting the improved
a o 6 g C] Other Improved .. . .
g 0.5 . pasture grass brachiaria.
‘8 0.4 - .. . . . .
t o 3 , 23; Given that rntensrficatron
o . --.
=2 0.2 « :31 43’? ' ' 11 ' t d 'th

0.1 . m 1??? rs typrca y assocra e wr

0 #33. "2.3m-
1982 1989 population increases, land

 

 

 

Graph 1: The Evolution of Pasture Use in Pucallpa (Riesco 1993 and author survey 1998)

scarcity, and market integration, the rest of this essay will attempt to explain the
widespread adoption of this improved production technology in the early stages of
frontier development and the impact it has had on deforestation.
3. The Evolution of Land Use Patterns in the Pucallpa Zone

An examination of the evolution of land use pattems gives insights into how cattle
production fits into farmers’ production strategies over time and its relation to
deforestation. Graph 2 shows that during the first ten years the principal change in land
use patterns involves the conversion of primary forest into secondary forest fallows. This
conversion is the result of annual crop production using shifting slash-and-bum
agriculture. Primary forest areas are cut down for annual crops, cultivated for 1-2 years
and then left fallow to regenerate into secondary forests. The amount of land put into

pasture remains low and increases only slightly.

79

 

 

 

0.60 + Primary
Forest
0.50

—l— Secondary
0.40 Forest
0.30

+ Pasture
0.20

 

 

 

 

+ Perennials

 

K
\
0'10 m —)l(— Annuals

0.00 f I I
0-5 5-1 5 15-25 >25

Years in Lot

 

 

 

 

 

 

 

 

 

Graph 2: The Evolution of Land Use Patterns on Farmer Land Holdings (Author Survey 1998).

After ten years, the major change involves a continuing strong decline in primary forest
cover and a sharp increase in pasture area. However, primary forest areas are rarely
converted directly into pastures. Graph 3 shows that pastures are rarely planted in areas
cleared from primary forest, rather they are almost always installed in areas coming out
of annual crops or in secondary forest.

Several reasons explain this transformation in land use. First, most migrants come
to the Pucallpa zone due to poverty in their region of origin. They initially lack the
needed financial resources to enter into ranching (cattle purchase, fencing, etc.). Farmers
begin practicing slash and burn annual crop production because it has low capital start-up
costs. Farmers need time accumulate capital to purchase cattle and ranching infrastructure
such as fencing and wells. Secondly, farmers prefer to install pastures on degraded areas
with falling annual crop yields. Pastures have lower soil fertility requirements than

most crops and therefore extend the effective utilization of degraded areas (Seré and

80

Jarvis, 1992; Hecht 1992). Finally, farmers install pastures in cropland because this
avoids the double labor of clearing land for both pasture and annual crops.

The importance of the capital constraint is demonstrated by the fact that 85
percent of farmers who were able enter into livestock production used off-farm sources of

income. Almost no

 

 

Areas Where Farmers "Always" or "Frequently"
Install Pastures famers reported

 

90

so , ,

m financrng livestock
to D Brachiaria Pastures
; 5° lAll Pastures . _ _
E so acqursrtron solely from
12 40
V n
33 3° farm sales. Thrs further

20

1" l: explains why more than

0 . ' a
Cropland Secondary PrlmaryForoat Other Pasture
Fm“, half of

 

 

Graph 3: Land Use Conversion to Pasture (Author Survey 1998).

farmers have pasture but have not yet been able to accumulate enough capital to purchase
cattle.

In sum, the sharp increase in pasture land use begins after approximately 10
years when farmers have accumulated enough capital to purchase cattle and when
declining soil fertility decreases crop yields. A similar evolution of land use patterns has
been observed in other tropical regions of Latin America (Thiele, 1993; Ledec, 1992;
Fujisaki, 1996).
4. Socio-Economic Characteristics of Brachiaria-Improved Pastures

An analysis of the key socio-economic characteristics of brachiaria-improved
pastures provides insights concerning their widespread adoption and impact on

deforestation. This sections examines the productivity and costs of brachiaria improved

81

pastures relative to natural pastures and compares livestock production (with brachiaria),
annuals, and perennials in terms of their use of labor and returns to labor and land.
4.1 The Productivity of Brachiaria-Improved Pastures

Grasses of the Brachiaria genus have low soil fertility requirements and good
nutritional quality (Lapointe and Miles, 1992). A study by Abanto and Vela (1997) of
pastures dominated by brachiaria in the Pucallpa zone found that they substantially

increase fodder production

 

 
    

 

  

 

 

 

Pasture Abandonment and
moo - Productlvlty Declines relative to na 1 pastures
osoo - mm
0.000 ..r .
o m (known as "tourouco" m the
5 0.000 4 _ _ . “m.
-. om. ctr-chur- Mm zone). Natural pastures
“ I htunl Pasture
‘5 om .- _ .
* 0.300 -. produced 807 kg/ha during the
one i . __
0.100 . '_ rainy season compared to
om ._-_,_ ~ 2 ..
Abandonment NoAbendoun at ”0:1an P 06 W ' '
' mm Jufim 1337 kg/ha for brachiaria.

 

 

Graph 4: Pasture Abandonment and Productivity Declines (Author Survey 1998).
This increase in fodder production raises the carrying capacity of pastures in the zone by
53 percent (from 1.01 to 1.55 tropical animal units).

Abanto and Vela also identify other productivity benefits associated with
brachiaria-improved pastures relative to natural pastures. Calf production, for example,
increased by 26 percent whereas mortality rates declined by 70 percent. In addition, total
annual meat production increased by 31 percent from 68.6 kg/ha to 89.6 kg/ha.

Another important indicator of productivity is the sustainability of production

over time. In the Amazon, degraded pastures have frequently been abandoned after only

82

10 years. Pastures of the brachiaria genus, however, persist for greater periods of time in
the phases of high and intermediate productivity as long as insect pests such as
spittlebug do not damage them (Serrao and Toledo, 1990).

In the case of Pucallpa, there is very little evidence of pastures being abandoned
due to productivity declines. Less than 10 percent of farmers reported ever having
abandoned a brachiaria-improved pasture due to productivity declines (graph 4).
However, approximately a third of farmers had noted some decline in productivity in
these pastures. Interestingly, however, these results are virtually the same for tourouco
natural pastures. In fact, a study by Toledo and Serrao (1987) found that natural pastures
in the Peruvian Amazon generally maintain a stable carrying capacity.

4.2 The Costs of Brachiaria-Improved Pastures

As mentioned in section 1.2.1, it is typically less profitable to augment production
with intensive technologies than with an increase in the amount of land under production
using extensive systems. Mattos and Uhl (1994), on the other hand, calculate that some
intensive and specialized systems may be more profitable than extensive systems but
their high costs combined with the capital constraints faced by farmers make them
unaffordable.

Graph 5 compares the per hectare costs of natural pasture with those of
brachiaria-improved pastures. There are two key differences. First, the installation costs
(land clearing, seeding, etc.) associated with brachiaria pastures are substantially higher
than those associated with natural pastures. Tourouco pasture is “natural” in the sense
that it spontaneously appears when an area of land is cleared. Installation costs are

therefore negligible. Brachiaria, on the other hand, needs to be planted, requires more

83

careful land clearing, and in some cases needs an initial weeding to facilitate its
installation.
Nevertheless, the installation costs of brachiaria are still quite low. Capital

requirements are typically nil (1 percent of total installation costs) and labor requirements

 

 
 
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Average Installation and Annual Maintenance Costs
160 -
140 ‘ ' DBrachIarIa Pasture
120 -- - - , [Natural Pasture
g 100
o 30 ..
m
60 +-
40
20 .-
0 .
Land Clearing Seedlng Other Installation Total Installation Annual
Malntenanee

 

 

 

Graph 5: Average Installation and Annual Maintenance Costs for Pastures (Author Survey 1998).
(One nuevo sol =$.32 in 1998) .

are 18.7 person days per hectare. Use of areas coming out of annual crop production
minimizes land clearing costs and seasonal labor demands are typically low after harvest
so labor is generally available. Brachiaria, which is typically transplanted by hand in
small clumps spaced at regular intervals in a field, rapidly spreads and takes over a field
and only occasionally requires weeding during installation. One of the most commonly
mentioned advantages of brachiaria by farmers in the survey was the ease of its
propagation.

This manner of installing brachiaria-improved pastures makes it a completely
divisible technology. There are no economies of scale due, for example, to tractor use or
bulk fertilizer purchases. It is proportionally as cheap to plant 1 hectare as it is to plant

10. Farmers are thus able to take a progressive approach to pasture installation. F arrners

84

typically only install from 1-3 hectares per year. This progressive approach spreads out
labor requirements so they do not impede adoption.

The second difference between brachiaria and traditional pastures is that the
recurrent annual costs of pasture maintenance (weeding, cutting, and burning) are nearly
twice as high for natural pasture as those of brachiaria-improved pastures. Brachiaria is
a much more aggressive pasture than tourouco and therefore minimizes weed regrowth.
Farmers mentioned pasture weed reduction a key advantage of brachiaria. A study by
Riesco (1992) also identifies weed reduction as the principal advantage of brachiaria.

Graph 6 shows that, when one sums the initial installation and annual
maintenance costs, in the Span of a few years the cumulative total costs of brachiaria are
less than those of natural pasture. The line marked natural pasture 1 shows that in
approximately 4 '/2 years the cumulative costs of one hectare of natural pasture surpass
those of one hectare of brachiaria pasture. By the end of ten years, the cumulative costs
of natural pasture are 43 percent greater than those of brachiaria pasture.

Furthermore, this one-to-one comparison is biased in favor of natural pasture. The
carrying capacity of brachiaria pastures is slightly over 50 percent higher than that of
natural pastures. Farmers need about 1.5 hectares of natural pasture to feed the same
quantity of livestock as one hectare of brachiaria pasture. Accounting for this difference,
the line labeled natural pasture 2 shows that the cumulative costs of natural pasture
surpass those of brachiaria pasture within about 2 ‘/2 years and are 118 percent higher by

the end of ten years.

85

 

Cumulative Costs of Irmroved and Traditional Pastures

 

 

 

 

 

 

 

1400 1
1200 -
+Brachiarla Pasture
1000 l + Natural Pasture 1
g 800 4 +Natural Pasture 2
E
8 6004
400 « I
zoo ‘ /
0 4 v - , . fl 7 1 v r
1 2 3 4 5 6 7 8 9 10
Year

 

 

Graph 6: The Costs of Improved and Traditional Pastures (author survey 1998)

Discounting these costs over the ten year time span shows that brachiaria -
improved pastures remains less expensive up to a discount rate just over 25 percent.
Researchers generally believe that small holder farmers in developing countries have
high rates of discount. However, given the low post-harvest opportunity cost of labor
when pastures are installed and maintained, farmers are likely to have a low discount rate
for these expenditures. The cost advantage of brachiaria pastures thus appears to be
relatively robust.

In sum, although the initial installation costs of brachiaria pastures are higher
than those of natural pasture, the annual maintenance costs make it a cheaper option
within a brief period of time, particularly when one controls for pasture carrying capacity.
Also, installation costs do not represent a constraint to adoption because brachiaria
pastures are a completely divisible technology.

4.3 Comparative Labor Use and Financial Returns: Livestock, Annuals, and
Perennials

This section compares the average labor use and the financial returns to land and

labor for the principal output categories-cattle, annuals, and perennials-in the Pucallpa

86

area (table 1). More specifically, these categories consist of cattle ranching using

brachiaria pastures; a proportional mix of the dominant annual crops3 in the zone (maize,

rice, plantains, and cassava); and oil palm, the current dominant perennial cash crop. This

comparison gives insights into farmer logic concerning resource use and land clearing

patterns. The main focus of this study, however, is on comparing brachiaria-improved

pastures with natural pastures.

Annuals and perennial oil palm production require respectively 148 percent and

218 percent more labor per hectare than cattle with brachiaria pastures. Although reputed

to give low returns to land, cattle’s returns are 49 percent higher than annual crop

production.

Table 1: Comparative Labor Use and Returns to Different Land Uses (author survey 1998).

 

 

 

 

 

 

 

 

Cattle (Brachiaria) Annualsa Perennials (Oil Palm)
Labor days/ha/yr 12.3 30.5 39.1
Solesb/ha/yr 497.4 334.9 1 125
Soles/labor day 40.4 1 1.1 28.8

 

‘Shifting fallow system requires an average of 4.8 ha’s to maintain I ha of annual crops in production
”One Sol = $.28 US in 1998

Perennial crops, however, provide returns to land 126 percent higher than those of cattle.

The returns to labor for cattle are greater than both annuals and perennials by 236 percent

and 40 percent respectively.

The low labor requirements of cattle allow an expansion of pasture into forested

areas in a zone where land is abundant and labor is a constraining factor of production.

Cattle’s returns to land indicate that it provides a more productive use of land than the

dominant practice of slash-and-bum annual crop production. And finally, given that

farmers maximize returns to scarce factors of production, cattle’s high returns to labor

make it an attractive option relative to the other alternatives.

87

 

5. Model

This research uses a theoretical model based on a profit function approach to
describe farmer production decisions. These production decisions are the direct causes of
deforestation. The research then uses empirical regression analysis based on this profit
function approach to model the determinants of production decisions and, in turn, how
these production decisions affect deforestation.
5.1 Theoretical Model

A profit function approach provides the theoretical foundation for explaining

farmer production behaviors. A farm’s profit function can be represented as follows:

(1)7r=pq-wx

where p is a vector of prices for outputs, q is a vector of outputs produced by the farmer,
w is a vector of input prices, and x is a vector of inputs.

F anners' profit maximization is constrained by the production function. This
function represents how farmers use variable inputs (x) and fixed inputs (2) to produce a

level of output (q):

(2) q =f(x.2)

Variable inputs are factors the farmer can change in the short term (e.g., fertilizer,
seed, labor, etc...) Fixed factors are those that can not be adjusted in the short term. These
factors include private factors (e. g., land, education, family labor, etc.), public factors
(e.g., infrastructure, credit, extension, etc.) and exogenous factors (e. g., soils and market

distance).

88

A farmer maximizes profits by choosing the optimal level of inputs x and outputs
q. The optimal levels of inputs and outputs are a function of output price, input price, and

fixed factors.

(3) x = X(P-W.Z) and 6] =CI(P.W-Z)

Combining equations (1) and (3) the profit function can be rewritten as follows:

(4) 7! = PIMP-Ml) - W109. W-Z)

Shepard’s lemma allows the derivation of the output supply and input demand functions

by differentiating (4) with respect to output and input prices in the following manner:

(5) Wrap-(p,w,z) = q, (170/de (p,w,z) = -x,-

Equation (5) provides the basic structure for the econometric modeling of the
determinants of output supply and input demand in the econometric models.
5.2 Econometric Models

The econometric regressions model two levels of relationships. The first level is
the impact of production determinants on input use and output supply. The second level
is the impact of inputs use on output supply on deforestation. Table 2 provides a
description of the variables used in the regression analysis.

The first two sets of econometric models have, respectively, input/technology use
(x) and output production (q) as dependent variables. As described in the theoretical
framework, these input and output regressions are specified as a function of the
independent variables of price (p, w) and fixed factors (2). Table 2 provides the averages

and standard deviations of the variables.

89

Table 2: Regession Variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OUTPUTS FIXED FACTORS cont.
Average Std. Dev. Average Std. Dev.

Annuals 3.71 3.97 Education‘ 33% n.a.
Hectares of Annual Crops Household head > primary education (dummy)
Perennials I 1.47 I 2.09 Family Labor I 2.72 I 1.34
Hectares of Perennial Crops Family members over 14 working on the farm
Livestock 1 6.93 I 27.40 Off-farm Income} I 1844.60 I 3074.50
Head of Cattle Off-farm income earned by farm household (soles)

INPUTS/TECHNOLOGY Credit‘ 1 33% l n.a.
Improved Fallows I I Credit received in the past 5 years (dummy)
Hectares of follows improved with kudzu Land Tenure“ I 847W n.a.
Improved Pasture I I F armer perceives tenure as secure (dummy)
Hectares of pastures improved with brachiaria Alluvial Soils‘I I 28% I n.a.
Natural Pasture I I Alluvial soils dominate (dummy)
Hectares of naturaLpasture Extension I 3.28 I 11.74
Cgpital Inputsb I I Number of extension visits in the past year
Total value of capital inputs used in l 998 (sales) PRICES
Hired Labor I I F m-Gate Price Maize I .36 I .056
Number of person days of hired labor used in 1998 Maize market price minus transpo costs (soles/kg)b

FIXED FACTORS Market Distance 1 60.97 I 27.76

Farm Size I 37.40 I 31.37 Distance in km farm to Pucallpa
Hectares of land on farm DEFOREST A T ION
Secondary Forest I 11.56 I 16.05 Prim. Forest Clearing I .66 I 1.42
Hectares of secondary forest on farm Primary forest felled (ha 's)
Primary Forest J 10.98 I 16.70 2nd. Forest Clearifl I 1.72 I 2.13
Hectares of primary forest opfarm Secondary forest felled (ha ’3)
Years in Lot I 17.12 I 11.80 Total Forest Clearing I 2.36 I 2.44
Years a farm has been in production Primary plus secondary forest felled (ha 's)
Sandy Soils' I 42% I n.a OTHER CONDITIONING FACTORS
Sandy soils dominate (dummy) Total Forest I 22.55 I 23.15
Distance Social Services I 5.67 I 5.57 Hectares of primary & second forest
Dist. in km farm to nearest: (school + healtLposU Primary Forest Prod’s i 319.27 I 800.07

 

 

 

 

 

 

Origin I 25% I n.a. Value prim. forest prod 's harvested (soles‘)

Household head not from jungle region (dummy) - - -
Age I 48.1 6 I 14.08 - - -

Age of household head - - -

 

 

 

'All variables are measured on a farm household level for the year 1998

I’One nuevo sol = 5.32 in 1998

cPercentages not averages are presented for dummy variables
“Percentages are determined by sampling strategy (along rivers) and may not reflect regional average.
°Percentage of highland (not river) soils. Farmers chose between clayey or sandy soil descriptive.

Outputs are aggregated into the three principal categories of agricultural activities:

annuals, perennials, and livestock. This research analyzes five of the most common

inputs/technologies used by farmers in the Pucallpa zone: improved fallows, improved

pastures, natural pastures, capital inputs and hired labor. The focus of the essay is on

comparing the results of brachiaria and natural pastures.

90

 

The cross-sectional data used by this research has minimal price variation. The
effect of prices on production is therefore difficult to estimate. To deal with this situation,
I chose to use farm-gate prices in order to capture greater variation in these variables. For
outputs, I did this with maize, the most frequently commercialized product, by
subtracting transportation costs (vehicle cost plus labor time opportunity cost) from the
market price received. In the case of inputs and technologies, it was not possible to
choose a reasonably representative product due to the variability of their costs of
commercialization. The distance to the principal market of Pucallpa therefore serves as a
proxy for the variation in input prices.

The input use and output choice regression models used both ordinary least
squares (OLS) and a systems approach using seemingly unrelated regression equations
(SURE). The SURE modeling approach gave results similar to OLS“, but sharply reduced
the number of available observationss. This paper therefore uses the results from the OLS
analysis.

The third set of regression models examines the impact of agricultural output
production and input/technology use on deforestation. In the Pucallpa zone, forests are
clear cut and then burned for agricultural production. Logging companies, on the other
hand, are not allowed to clear cut forests but rather practice selective extraction that
leaves the forest largely intact. Timber extraction therefore does not clear land that
farmers then choose to cultivate. For these reasons, the relationship of outputs and inputs
to deforestation is hypothesized to be recursive: agricultural production results in

deforestation and not vice versa.

91

 

lu
barren.
pastures
inputs 3]
evohnu
and 12.11
exntir
can re:
shnuh
()ITI

educe
maiz-

S€C(
Prir

del}

fOFe.

I used the Hausman specification test6 to examine the hypothesis of recursivity
between agricultural production and deforestation. With the exception of improved
pastures in the total deforestation model, this test confirmed the recursive nature all
inputs and outputs with deforestation. This exception appears logical given the
evolutionary process of pasture installation. Farmers first clear forest for annual crops
and later plant pastures in areas of previously cultivated land. Simultaneity may therefore
exist in so much as deforestation can lead to pasture installation and pasture installation
can result in deforestation. A two-stage least squares approach (ZSLS) corrects for
simultaneity in this case. The input and output regression models are specified as follows:
OUTPUTu= f(farm sizei, secondary forest, primary forest, alluvial soilsi, years in loti,

educationi, agei, family labor,, origini, off-farm income,, crediti, extension, real price
maizei, land tenure,)

annuals, perennials, and livestock

j =
i = 1,...,Nobservations

IN PUT” = f(farm sizei, secondary forest,, primary forest,, alluvial soilsi, sandy soils-,years

in loti, education,, family labori, off-farm incomei, distance social services, crediti, market
distance,, land tenure,)

j = improved fallows, improved pastures, natural pastures, capital inputs and hired labor
i = 1,..., N observations

The deforestation regression models measure the dependent variable as the
hectares of forest cut in 19987 by each farm household. This research distinguishes
between three types of deforestation: total deforestation, primary forest deforestation, and
secondary forest deforestation. Primary forests are areas that have never been felleds.
Primary forests that have been felled and have regenerated are secondary forests. Total
deforestation is the sum of both primary and secondary forest deforestation.

The literature on deforestation rarely distinguishes between different types of
forest clearing. People frequently equate deforestation with primary forest clearing and

assume that once a primary forest has been cleared the forest is gone forever. Research by

92

FAO (1996), however, estimated that in 1990 there existed 165 million hectares of
secondary forest in Latin America. In the Pucallpa area, survey results indicate that
secondary forests virtually equal primaryforests on farm holdings and that 72 percent of
forest clearing is secondary forest and only 28 percent is primary forest.

Deforestation is modeled as a function of output choice, input/technology use, and
general conditioning factors. General conditioning factors are those that may directly
affect deforestation i.e. not only through their impact on input and output choices. These
include land tenure, family labor, soils, education, forest area, and forest products. The

deforestation model is specified as follows:

DEFOR, = f(annual,, perenniali, livestocki, kudzu fallowsi, improved pasture,, natural
pasturei, capital input use,, hired labori, family labori, land tenure,, educationi, ;, foresti,
alluvial soils,, forest product income,)

j = total, primary, and secondary forest clearing in 1998.
i = 1,..., N observations

6. Results

The resultsipresented here focus on the determinants of the adoption of
brachiaria-improved pastures, natural pastures, and livestock and the effect these have
had on forest clearing. Because cross-sectional data tends to give less significant results
and due to the diversity of production systems, variables are considered significant up to
the .15 level. Furthermore, for the purposes of policy analysis in this context, this level is
reasonable in most instances. However, the exact degree of significance of each variable
is reported for the reader's consideration.
6.1 The Impact of Pastures and Livestock on Deforestation

Brachiaria-improved pastures, natural pastures and livestock are all significantly

correlated with secondary forest clearing and not statistically correlated with primary

93

forest clearing (table 3). This result reinforces the temporal analysis of land use changes
in section 3 which found that primary forest is initially cleared by new migrant farmers
for annual cropping. Farmers then progressively install pastures on degraded areas that
have already been cropped and have regenerated into secondary forest. Increased
livestock is therefore positively related to secondary forest clearing.

Brachiaria and nature pasture, on the other hand, are negatively correlated to
secondary forest clearing. The surplus capacity of pastures in the zone dampens the need
for those farmers who already have pastures to clear more forest area. Furthermore,
unlike shifting fallow-based annual cropping, sedentary production on pastures
suppresses the regeneration of secondary forests. In sum, pasture’s surplus capacity and
repression of secondary forest regeneration explain the negative correlation of both types
of pasture with secondary forest clearing.

A comparison of the regression results for natural pastures relative to brachiaria
pastures shows that the magnitude and significance of the decrease in deforestation is
consistently greater for natural pastures. The greater dynamism of brachiaria pastures
due to their lower cost and higher productivity appears to mitigate the decrease in
deforestation.

6.2 Pasture Use and Livestock Production

Results from the brachiaria pasture regression (table 4) show that its adoption is
significantly correlated to larger farms with less remaining primary and secondary forest.
They are most often found nearer to the main market town of Pucallpa (market distance)
and to smaller population centers with schools and health posts (dist. soc'l services). The

greater presence of brachiaria nearer to population centers indicates farmers have more

94

Table 3: Deforestation Regression Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total Primary Forest Secondary Forest
Deforestation Deforestation Deforestation
111.54 adj 18.50 111.30 adj 111.23 18.30 adj 121.23
Coef (t-stat) Signif Coef (gust) Signif Coef (t-stat) Signif
Constant . 18 .74 .74
Annuals .490 (7.59) 00“" .323 (3.69) 00*" .301 (3.42) 00*"
Perennials -.102 (-1.65) .10“ -.040 (-0.53) .60 -.008 (0.10) .92
Livestock .039 ( 0.54) .59 .020 (0.21) .83 .197 ( 2.04) .04"
Improved Pastures -.023 (-0.32) .75 -.006 (0.06) .95 -.154 (1.53) .13“
Natural Pastures -.094 (1.50) .13‘ -.061 (0.75) .45 -.163 (-2.00) .05“
Improved Fallows .049 (0.83L .41 -.129 (-l.80) .07“ .137 (1.91) .06*
Capital Inputs -.153 (-2.18) .03“ -.125 (1.41) .16 .181 (2.03) .04"
Hired Labor .337 (5.73) 00"" .011 (0.13) .89 .0194 0.23) .82
Family Labor .060 ( 1.01) .31 .0214 0.27) .79 .202 (2.61) 01*”
Education .129 (2.26) .02“ .003 (0.05) .96 .070 (0.98) .33
Tenure Security .097 ( 1.69) .09“ .025 (0.32) .75 .030 (0.39) .69
Alluvial Soils .028 (0.45) .65 -.193 (2.61) 01"" .203 (2.75) 01""
Total Forest .108 ( 1.72) .09“ - - - -
Secondary Forest‘ - - -.098 (1.41) .16 .166 (2.38) .02"
Primary Forest - - .254 i 3.30) 00*“ -.093 (-1.20) .23
Prim Forest Prod's - - .1684236) .02" -.034 @148) .63

 

 

 

 

 

‘High secondary forest >5meters

Table 4: Livestock and Pasture Adoption Regression Result

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OUTPUT Livestock INPUT Improved Pastures Natural Pasture
REGRESSION R258 g: 18.40 REGRESSION R280 adj R2.78 111.19 adj 12.2.12
Coef t-stat Signif Coef Signif Coef Signif
Constant .21 Constant .08 .01
Farm Size .912 (3.97) "‘.00 Farm Size l.362(21.48) "£00 .448 (3.46) °”.00
Secondary Forest - .052 (0.36) .72 Secondary Forest - .575 (-12.43) "'00 - .186 (2.03) ".05
Primary Forest -.588 (-2.93) "'01 Primary Forest -.907(14.40) "£00 -.164 (-1.29) .20
Years in Lot .139 (0.99) .33 Years in Lot .053 (-1.I4) .26 .198 (2.18) ".03
Education -.058(0.45) .65 Education -.053(-1.28) .20 .098 (1.19) .24
Family Labor .267 ( 1.81) ‘08 Family Labor - .047 ( -1.I3) .26 .028 (0.35) .73
Off-Farm Income .028 (0.19) .84 Off-Farm Income - .005 (-0.11) .91 .077 (0.92) .36
Credit - .102 (-0.80) .43 Credit - .087(-2.06) ".04 -.021 (025) .81
Tenure Security .313 (2.23) ”.03 Tenure Security .061( 1.42) .16 .033 (0.38) .70
Real Price Maize 07240.55) .59 Market Distance - .084 (-1.97) ".05 .029 (0.35) .73
Origin not jungle .123 (0.92) .36 Dist Soc'l Services -.110(-2.40) ".02 -.089 (-0.98) .33
Extension ‘ - .070 (-0.46) .64
Age -.204 (-1.50) '.14
Alluvial Soils .125 (0.90) .37

 

 

’Significant at the .01 level *“ Significant at the .05 level “* Significant at the .15 level

incentive to intensify production due to better access to markets and other elements of

social infrastructure.

This pattern of land use coincides closely with the barbecho9 crisis model (Thiele,

1993). This model predicts that near population centers small holder slash-and-bum

95

 

annual cr0p production uses up primary forests and subsequent production cycles using
secondary forest fallows leads to progressively declining yields. This loss in yields forces
many farmers to move further out to the forest margins in search of more productive land.
Farmers remaining in older frontier areas tend to specialize in cattle production on larger
consolidated farmland holdings. This pattern of land use is driven by declining soil
fertility, the ability of livestock to occupy degraded land, and the land-using nature of
livestock production.

Natural pastures and brachiaria-improved pastures are both positively correlated
with farm size and negatively correlated with the presence of secondary forests.
However, there are several important differences between them. Natural pastures tend to
be found on older lots whereas this correlation is not significant for brachiaria pastures.
This is because brachiaria pastures are progressively replacing natural pastures and
farmers are installing few new areas of natural pasture. In addition, unlike brachiaria
pastures, natural pastures are not found nearer to markets where farmers have greater
incentives to intensify production. The cheaper costs of brachiaria-improved pastures
are reflected by their negative relationship with credit which may indicate that farmers
adopt them when faced with a capital constraint. This relationship is not found with
natural pastures.

LivestoCk and brachiaria pastures are both positively correlated to farm size and
negatively correlated with on-farm primary forest. This correlation reflects both the
evolution of land use and the extensive nature of livestock production. Livestock is also
positively correlated with tenure security reflecting the greater need to protect land

investments associated with ranching such as pasture installation, fencing, and wells. Its

96

negative correlation with household head age is due the fact that older farmers are likely
to find themselves with less available family labor and are less physically able
themselves to maintain large herds. Household structure is dominated by small nuclear
families. Farmers' adult children, if they remain in agriculture, tend to start their own
farms. This is due to the availability of new land and the relativity low production risks
associated with land-abundant humid tropical agriculturelo (Binswanger and McIntire,
1987)

An important relationship is the positive correlation between livestock and family
labor. This correlation may appear paradoxical given the low labor requirements of
livestock. However, compared to annuals and perennials, this relation is weaker both in
terms of it magnitude and significance (Yanggen, 2000). This relative weakness
demonstrates that despite livestock’s low labor requirements, labor remains a key limiting
factor of production.

I have demonstrated (Yanggen 2000) the importance on the labor constraint in the
Pucallpa zone. As just mentioned, family labor is positively correlated with the three
principal agricultural activities. But when farmers have access to off-farm employment,
this competition for scarce labor leads to a reduction in annual and perennial cropping.
Also, hired labor and family labor are positively correlated with total and secondary
forest clearing respectively. Put simply, more labor increases agricultural production and
more agricultural production increases deforestation.

6.3 Correlation Analysis Results
The regression analysis uses cross-sectional data that identifies the determinants

of deforestation at a given moment in time. The correlation analysis done here introduces

97

three new deforestation variables that complement that analysis by giving a historical
picture of livestock and pasture’s impact on deforestation.

The first variable is the annual rate of primary forest clearing over time. This rate
is calculated as: (total farm size minus farm area in primary forest)/(years farm in
production). The second variable is aggregate deforestation, which measures the total
hectares of primary forest cleared on a farm. This is the total farm size minus the hectares
of remaining primary forest. The last variable is the percentage of the farm in secondary
forest. This indicates to what extent cattle production tends to eliminate secondary forest
regeneration.

In general, both head of cattle and pasture are associated with higher average
annual rates of primary forest clearing over time, larger aggregate area deforested, and a

diminished presence of secondary forest. The one exception is natural pastures with the

Table 5: Correlation Analysis of the Impact of Livestock and Pastures on Deforestation

 

 

 

 

 

 

 

 

 

Head of Cattle Brachiaria Pastures Natural Pastures

Coef. Signif. Coef. Signif. Coef. Signif.
Annual Deforestation Rate .58 .00 .40 .00 .02 .85
Aggregate Deforestation .43 .00 .60 .00 .18 .01
% Secondary Forest -.19 .00 -.36 .00 -.16 .02

 

 

 

 

annual rate of deforestation. This result can be explained by the fact that natural pastures

tend to persist on older and less dynamic farms. The greater age and slower current

growth tend to reduce the average annual deforestation rate.

In general, the brachiaria pasture variable has a higher degree of significance and

a coefficient of greater magnitude than the natural pasture variable. This indicates that the

introduction of brachiaria-improved pastures tends to increase deforestation relative to

natural pastures. This coincides with the evidence of the regression analysis.

98

 

 

 

Thea

    
  

counterimurt
forest clear-r
capacity and

deforestatitr:

adopted an
cheap and
technolog}
evidence c
recomme-

A

relative tr
Saving na
benefits 1
171165, am

A
filter Of
maxlmjzt

These results further clarify the impact of pastures on deforestation. The rather
counterintuitive result in the regression analysis was that pastures decrease secondary
forest clearing whereas livestock has the opposite effect. The current excess pasture
capacity and pasture’s suppression of secondary forests lead to a reduction in this type of
deforestation. The correlation results show that, given a longer time framework, pastures
are indeed associated with higher rates of secondary forest clearing as they should be if
they are a complement to cattle production.

7. Conclusions and Policy Recommendations

This essay has examined two principal questions. First, why have farmers
adopted an intensive technology that increases pasture carrying capacity in a zone with
cheap and abundant land? Second, what impact has the adoption of this improved
technology had on deforestation? This concluding section examines the weight of the
evidence concerning the adoption and deforestation questions and proposes a series of
recommendations to promote more sustainable production practices.

A key finding of this research is that brachiaria-pastures reduce costs over time
relative to natural pastures by over 50 percent. This outcome results from the labor-
saving nature of brachiaria pastures. Brachiaria pastures also provide productivity
benefits including increased pasture carrying capacity, low mortality and higher calf
rates, and increased meat production.

A technology that increase land productivity while reducing labor costs has the
effect of increasing returns to labor. Basic economic theory indicates that farmers
maximize returns to scarce production factors. The adoption of brachiaria-improved

pastures can be explained by farmer’s attempts to increase returns to scarce labor.

99

Furthermore, the intensification of land, which is not a limiting factor of production, is
not the principal motivation of farmers but rather a secondary outcome of maximizing
returns to labor.

A slight nuance, however, can be added to this conclusion. Farmers tend to install
pastures in degraded areas where land is becoming scarce in terms not of its quantity but
rather it terms of its quality. Livestock production with improved pastures allows
productive use of degraded land and raises returns to land relative to shifting annual crop
production. This is one manner of overcoming qualitative land scarcity.

This "qualitative” land scarcity may be compounded by localized quantitative
land scarcity near to the principal market Pucallpa and smaller population centers where
brachiaria pastures tend to be installed. Land with superior infrastructure and access to
markets is relatively scarce. Most farms in the Pucallpa area have serious problems of
accessibility to markets during much of the 8 month rainy season due to impassable
roads. Farmers therefore have greater incentives to increase the productivity of these well
located farmland areas.

There are nevertheless many examples of farms that have installed and continue to
install brachiaria-improved pastures without having any cattle. Land scarcity thus
appears to be a secondary explanation of adoption. The driving force behind the adoption
of brachiaria-improved pastures is that they help maximize returns to the scarce labor
factor of production.

The weight of the evidence indicates that brachiaria-improved pastures do not
reduce deforestation. The key limiting factor of production in the zone is labor.

Brachiaria- improved pastures ease this constraint and therefore allow a greater

100

expansion of land under agricultural production. Over time, brachiaria and natural
pastures are both positively correlated to deforestation, but the magnitude and statistical
significance of the impact on deforestation is greater for brachiaria pastures.

A key argument supporting improved pasture germplasm’s ability to reduce
deforestation is that it avoids pasture abandonment due to degradation, which forces
farmers to clear more land at the forest margins. However, the abandonment of natural
pastures has been negligible and no greater than that of brachiaria-improved pastures.
The introduction of improved pastures therefore does not have the potential to reduce this
source of deforestation.

In sum, the introduction of brachiaria-improved pastures has tended to accelerate
deforestation over time. Brachiaria eases the key labor constraint to increased pasture
expansion and increases the returns to forest conversion for livestock production.
Furthermore, brachiaria pastures do not reverse a situation where farmers abandon
degraded pastures and clear new ones at the forest margins.

The paradox of the success of brachiaria-improved pastures is that a principal
reason they have been adopted—their reduction of labor costs—has also led to an
aggravation of livestock’s impact on deforestation. Reduced labor inputs and greater
pasture productivity increase the returns to labor. The challenge is therefore to find ways
to increase the returns to labor without freeing labor up for increased forest clearing. An
example of how this can be done is with silvopastoral systems.

A small number of farmers in the zone have begun integrating livestock with fruit
tree and oil palm production. The principal cost of maintaining most tree plantations is

weeding labor. If cattle are able to graze among the trees, this can greatly reduce weeding

lOl

labor for the perennials while providing income from the livestock. Livestock can benefit
from the shading'of the intense tropical sun and nutritionally from fruit that falls to the
ground.

Overall labor requirements per hectare of land are increased due to the
management of both the livestock and the trees. This “absorbs” labor away from
extensive agricultural production into the forest margins. Yet silvopastoral systems with
these types of complementarities can increase the retums to labor by providing high value
output of both livestock and perennials. In order for these systems to work, more research
needs to be done in order to find systems where the tree and livestock components are
truly complementary and avoid problems such as soil compaction by livestock and
pasture fires damaging trees.

Nevertheless, as long as abundant low cost land remains readily available, farmers
will have a potent incentive to expand production into the forest margins. Restrictions on
land access therefore merit consideration as a complementary policy to improved
technologies. Indeed, limiting access to land gives farmers a greater incentive to adopt
intensive production practices such as silvopastoral systems, improved pastures, legume
associations, and pasture rotation because they make land artificially scarce and therefore
more expensive. Greater pasture scarcity and improved production practices would also
push farmers to increase the current low stocking rates to higher levels. This would limit
deforestation as well promote a more productive use of land cleared for pasture.

One policy option would be to limit livestock to areas where it is most
appropriate: older frontier areas no longer productive for annual crop production. Another

potential restriction would be to limit the size of ranches. Larger ranches have a high cost

102

in terms of deforestation, but the benefits are concentrated in the hands of a few
individuals. Another option is to focus on improving existing infrastructure and limiting
the construction of new roads. This would increase the value of existing land and limit
access to new land. Issues of the institutional and political feasibility of these policies
require careful consideration of the most appropriate option.

While livestock in general and the use of brachiaria-improved pastures in
particular accelerate deforestation in the zone, they also have a series of advantages that
make them attractive to farmers. The reality is that livestock has established a long-term
presence in the zone and by all indications is likely to expand. The challenge is to find the
best ways to manage the development of the livestock sector to minimize its negative

impact on deforestation and to maximize the social returns that it provides.

103

REFERENCES

Abanto, CG. and J. Vela. 1997. Evaluacion Te'cnica y Economia de Sistemas de
Produccion Ganadera en Fundos de la Region Ucayali. Instituto Nacional de
Investigacion en Agricultura Pucallpa, Perri. 44p.

Binswanger, Hans and John McIntire. 1987. Behavioral and Material Determinants of
Production Relations in Land-abundant Tropical Agriculture. Economic
Development and Cultural Change. Vol 36 No 1. Chicago, 73-99.

Boserup, E. (1965). The Conditions for Agricultural Growth: The Economics of
Agrarian Change under Population Pressure. Aldine Publishing Company,
New York.

Boserup, E. (1981). Population and T echological Change: A Study of Long Term
Trends. University of Chicago Press. Chicago.

Buschbacher, R]. (1986). Tropical Deforestation and Pasture Development. Bioscience
36, 22-28.

F AO (Food and Agriculture Organization of the United Nations) (1993). Forest
Resources Assessment 1990, Tropical Countries. F AO Forestry Paper 112. FAQ.
Rome.

FAO. 1996. Forest Resource Assessment 1990. FAO Forestry Paper 130. Rome.

Feamside, P. (1990). Environmental destruction of the Amazon. In The Future of
Amazonia. Goodman D. and Hall A (eds). St. Martins. New York. pp 179-226.

F ujisaka, S., W. Bell, N. Thomas, L. Hurtado, and E. Crawford. (1996). Slash-and-Bum
Agriculture, Conversion to Pasture, and Deforestation in two Brazilian Amazon
colonies. Agriculture Ecosystems & Environment 59 (1996) 115-130.

Griffith, K., and L. Zepeda. (1994). F arm Level Trade-Offs of Intensifying Tropical Milk
Production. Ecological Economics. 9. 121.

Hayami Y. and V. Ruttan (1985). Agricultural Development: An International
Perspective. 2nd ed. Johns Hopkins Press, Baltimore MD.

Hecht, SB. (1982). Cattle Ranching Development in the Eastern Amazon. Doctoral
Dissertation, University of Califomia, Berkelely.

Hecht, SB. (1985). Environment, Development and Politics: Capital accumulation and
the livestock sector in Eastern Amazonia. World Development 13. 663-684.

104

Hecht, S.B., R. Norgaard, and G. Possio. (1988). The Economics of Cattle Ranching in
the Eastern Amazon. Interciencia 13, pp 233-240.

Hecht, SB. (1992). Logics of Livestock and Deforestation: the Case of Amazonia. In
Development or Destruction: The Conversion of Tropical Forest to Pasture in
Latin America. Downing T.E., Hecht S.B., Pearson H.A. and Garcia-Downing
(eds). Westview Press. Boulder. pp 405. (95-113).

Hecht, S. (1993). The logic of livestock and deforestation in Amazonia. Bioscience v 43,
no. 10 . pp 687-694.

Houghton R.A., D.S. Lefltowitz, and D.L. Skole. (1991). Changes in the Landscape of
Latin America between 1850 and 1985. Progressive Loss of Forests. Forest
Ecology Management, 38. Pp 143-172.

INEI (Instituto Nacional de Estadistica) (1986). Encuesta Nacional de Hogares Rurales
(ENAHR): Resultados Definitivos. Lima, Perr'r.

Kaimowitz, D. (1996). Livestock and Deforestation Central America in the 1980s and
1990s: A Policy Perspective. International Food Policy Research Institute.
CIF OR, Indonesia. 88p.

Kaimowitz, D. and A. Angelsen. (1998). Economic Models of Deforestation. Center for
International Forestry Research. Bogor, Indonesia. p139.

Labarta, R. (1998). Los Productores de la cuenca Amazonica del Perri y la dinamica de
uso de la tierra: Resultados de la caracterizacion de Pucallpa y Yurimaguas.
International Center for Research in Agroforestry. Pucallpa, Peru.

Lapointe, 8., Miles, I. (1992). Gerrnplasm case study : Brachiaria species. In Pastures
for the Tropical Lowlands CIA T ’s Contribution. Cali, Colombia. 43-55.

Ledec, George. (1992) New directions for livestock policy: An environmental
perspective. In Development or Destruction: The Conversion of Tropical Forest
to Pasture in Latin America. Downing T.E., Hecht S.B., Pearson H.A. and
Garcia-Downing (eds). Westview Press. Boulder. pp 405. (27-65).

Mattos, Marli Maria and Christopher Uh1.(l994). Economic and Ecological Perspectives
on Ranching in the Eastern Amazon . World Development, 22, (2): 145-158.

Munoz, M., P. Odennatt, and J. Reyes. (1994). Retos y Oportunidades del sistema leche
de Mexico ante e1 TLC (Tratado Trilateral de Libre Comercio). Universidad
Autonoma de Chapingo, Chapingo, México.

NRC (National Research Council). (1993). Sustainable Agriculture and the Environment
in the Humid Tropics. National Academy Press. Wahington DC.

105

Rao, I.M., R.S. Ziegler, R. Vera, and S. Sarkarung. (1993). Selection and Breeding for
Acid-Soil Tolerance in Crops: Upland Rice and Tropical Forages as Case Studies.
BioScience, 43 (7). pp 454-465.

Riesco, Alfredo. (1993). Intensification Tecnologica en la Selva Baja: El Caso de
Pucallpa. In Desarrollo Rural en la Amazonia Peruana. Edited by William Loker
and Steve Vosti. CIAT and IF PR1. Cali, Columbia.

Schelhas, John. (1996). Land Use Choice and change: Intensification and Diversification
in the Lowland Tr0pics of Costa Rica. Human Organization. Vol 55 #3. 298-306.

Seré C. and Jarvis LS. (1992) Livestock Economy and Forest Destruction. In
Development or Destruction: The Conversion of Tropical Forest to Pasture in
Latin America. Downing T.E., Hecht S.B., Pearson H.A. and Garcia-Downing
(eds). Westview Press. Boulder. pp 405. (95-113).

Serrao, EA, and Toledo, J .M. (1990) The Search for Sustainability in Amazonian
Pastures. In Alternatives to Deforestation. Anderson A. (ed) Columbia University
Press. pp 196-211.

Serrao, EA, and Toledo, J .M. (1992) Sustaining Pasture-Based Production Systems for
the Humid Tropics. In Development or Destruction: The Conversion of Tropical
Forest to Pasture in Latin America. Downing T.E., Hecht S.B., Pearson H.A. and
Garcia-Downing (eds). Westview Press. Boulder. pp 405. (257-280).

Serrao, EA, and Toledo, J.M. (1993). The Search for Sustainability in Amazonian
Pastures. In Alternatives to Deforestation: Steps Toward Sustainable Use of the
Amazon Rain Forest. A.B. Anderson (ed.) Columbia University Press. New York.
195-214.

Smith, Joyotee, A.D. Barau, A. Goldman, and J.H. Mareck. (1994). The Role of
Technology in Agricultural Intensification: The Evolution of Maize Production in
the Northern Guinea Savanna of Nigeria. Economic Development and Cultural

Change. Pg 538-54.

Smith, Nigel J .H. (1992). Despite Grave Losses of Forest, Some Good Things are
Happening in Amazonia. Environmental Conservation l9 (4) pp 295-295.

Steinfeld, H., C. de Haan, and H. Blackburn.(1998). Livestock and the Environment:
Issues and Options. In Agriculture and the environment: Perspectives on
sustainable Rural Development. Edited by Ernst Lutz. A World Banck
Symposium 1998. The World Bank. Washington DC. 283-299.

Thiele, Graham. (1993). The Dynamics of Farm Development in the Amazon the
Barbecho Crisis Model. Agricultural Systems 42 : 179-197.

106

Thrupp, LA. (1980). Deforestation, Agricultural Development, and Cattle Expansion in
Costa Rica, An Integrated Approach to Problems of Land Use Transformation.
B.A. Thesis, Latin American Studies, Stanford University.

Toledo J .M. and E.A.S Serrao. (1987). Pasture after Humid Tropical Forest Clearing in
the Amazon. Paper Presentedat the International Symposium on Grassland in
Forest Area. Harbin, R.P. China. Dates Unspecified.

Tomich, T., et a1. (1999). Agricultural Intensification, deforestation, and the
Environment: Assessing Tradeoffs in Sumatra, Indonesia. Forthcoming Paper in
Proceedings of AAEA Conference on Agricultural Intensification, Economic
Development, and the Environment. David Lee and Chris Barret (eds.). Revised
May 20‘h 1999.

Yanggen, David and Thomas Reardon. (In Press). Kudzu improved fallows in the
Peruvian Amazon: A case study of the impact of technological change on
deforestation. In Technological Change in Agriculture and Deforestation.
Kaimowitz D. and Angelsen A. (eds.)

Yanggen, David. (2000). A Household-Level Analysis of the Impact of Agriculture on
Deforestation in the Peruvian Amazon. Unpublished Essay Submitted for
Doctoral Thesis Requirement in the Department of Agricultural Economics,
Michigan State University. East Lansing.

107

 

' Infrastructural access included location on the paved highway Frederico Basadre, secondary gravel roads,
tertiary dirt roads, and rivers.

2 Farmers are given a small number of “seed” cattle to begin a herd. They return these initial head of cattle
after approximately 5 years but retain the offspring from these cattle.

3 Plantains and cassava can last for more than one year and are sometimes referred to as “transitory” crops.
’ With the exception of the capital input use model, there were only a total of four sign changes (none of
variables significant at the .15 level or lower). Further (again with the exception of capital inputs), variables
significant at the .05 level or lower in the OL S models remained significant at the .15 level or better in the
SURE systems approach. The standard errors generated by the two techniques were remarkably similar,
although the SURE standard errors were generally slightly higher. In the case of capital inputs a review of a
simple correlation matrix between inputs and outputs indicated that capital inputs have the highest level of
correlation with other inputs and outputs. This may explain the changes shifting from OLS to SURE.

5 With the SURE technique, all missing variables from any equation in the system result in an observation
being deleted. Combining all the missing observations in this way resulted in only 40 of the initial 220
observations being available.

6 I performed the Hausman test in order to check for simultaneity in the deforestation regression models
between the deforestation dependent variables and the input/technology and output independent variables.
In order to do so I first estimated the reduced form equations for all the input/technology and output
variables. I then took the estimated residual from each reduced form equation and plugged it into each of
the deforestation regression models as a right-hand-side variable. I then performed a t-test on the
coefficient of the estimated residual from the reduced form equation. 1 used a .10 cut-off as a criteria to
'udge for significance.

Rainfall comes in a bimodal pattern and there are two cr0pping seasons. Farmers fell forest and let the
felled vegetation dry before buming and then planting crops. The first cropping season lasts roughly from
mid February to mid May and is generally preceded by a relatively dry period in January. The second
cropping season lasts roughly from mid September to the end of December and is preceded by a relatively
dry period from June to early September. Deforestation was measured as the sum of the two forest clearing
periods in January and July-August of 1998. This survey was conducted in October of 1998. s

A substantial proportion of primary forest on farm land has been selectively logged. Often this was done
by the timber companies the opened the initial feeder road into the jungle where farmers subsequently
settled or by the farmers themselves. These areas retain most of their vegetative cover and biodiversity and
are typically indistinguishable from undisturbed primary forest to the untrained eye. They are sometimes
referred to as areas of “residual” primary forest.

9 Barbecho is Spanish for fallow.

10 The diversity of crops and low probability of drought reduce the need for families to work together in an
extended family unit as a risk sharing mechanism.

108

CONCLUSIONS

This dissertation consists of three essays examining agriculture’s impact on
deforestation in the lowland rainforest zone surrounding Pucallpa, Peru. The first essay is
an overview piece examining household-level determinants of agricultural production
practices and how these practices in turn effect deforestation. The following two essays
are case studies of agricultural technologies that farmers have widely adopted in this
same zone: kudzu-improved fallows and brachiaria-improved pastures.

These two technology adoption case studies are of interest due to the limited
success national and international agricultural research institutions have had in promoting
the adoption of technologies they have developed for the zone. This research therefore
examines two key questions. First, why have farmers adopted these two technologies and
not others? Second, what impact have these two technologies had on deforestation?

In general, most sustainable technology development efforts have focused on
reducing deforestation by improving land productivity. The rationale of this approach is
that increased land productivity allows more production with less land and therefore
farmers clear less forest. Or by simply maintaining land productivity, farmers are not
forced to practice shifting slash-and-bum agriculture in forested areas.

The problem with this approach is that increases in land productivity typically
require greater per land unit use of capital and/or labor inputs. In the Pucallpa area, land
is abundant and cheap relative to capital and labor. It is generally not economical for
farmers to use relatively expensive labor and capital inputs in order to increase the

productivity of the relatively cheap land input.

109

Historically, intensification of land use has taken place in land scarce areas.
Pucallpa is a land abundant area. What then explains the adoption of these land
productivity enhancing fallow and pasture technologies? The answer lies in their use of
labor.

This dissertation has shown that both kudzu-improved fallows and brachiaria-
improved pastures not only increase land productivity but also reduce labor requirements
relative to their traditional counterparts without increasing capital requirements. Labor is
scarce relative to land. Basic economic theory indicates that farmers maximize the retums
to scarce production factors. By definition, a technology that decreases labor
requirements and increases land productivity (crop yields or pasture carrying capacity)
increases returns to labor. Farmers’ primary objective in adopting these technologies is
not to increase land productivity but rather to increase labor productivity. Increased land
productivity in these cases is a secondary result of fanners’ primary focus on increasing
labor productivity.

This finding has important policy implications for the development and promotion
of improved technologies. Traditionally, agronomic research efforts have focused their
energies on increasing yields (i.e. land productivity) with greater input use. Where land is
abundant relative to labor and capital, these efforts have typically failed and will continue
to fail. If farmers are to adopt “improved” technologies in these conditions, the primary
focus of research must be on increasing returns to labor.

Unfortunately, the story is not as “simple” as that. This dissertation has shown
that labor availability in the Pucallpa area is positively correlated to increased

deforestation. The reduction in labor requirements that leads to the successful adoption of

110

these technologies also frees up labor for increased deforestation. Essay 2 on kudzu-
improved fallows referred to this situation as the “labor paradox”. Brachiaria-improved
pastures led to a similar phenomena. The challenge is to find way of increasing returns to
labor without freeing up labor for increased deforestation.

Essay 3 on brachiaria-improved pastures compared the average per hectare labor
requirements, returns to land, and returns to labor of cattle, annuals, and perennials. This
comparison found that while the shift from annuals to perennials (oil palm) increased per
hectare labor requirements by 28 percent, returns to labor nevertheless also increased by
159 percent. Increasing per hectare labor requirements while increasing financial returns
to labor are not mutually exclusive objectives. Furthermore, regression results indicate
that perennial crop production does indeed decrease deforestation.

The key to success in this case is that perennial crops are relatively demanding in
labor and, by their nature, “sedentarize” farmers on a given parcel of land. Shifting
annual crop production, on the other hand, requires on average a total of 4.8 hectares to
maintain one hectare in production over time. Perennial crop production therefore
“absorbs” scarce labor away from clearing new forest areas.

The key lesson here is not simply about the potential of perennials to decrease
deforestation under certain circumstances. Other agricultural production may also
increase both per hectare labor requirements and returns to labor. Horticultural
production typically is intensive in its labor use per hectare of land but, because they are
higher value crops, may also increase returns to labor. Dairy production may follow a

similar pattern.

111

These findings also have interesting implications for agroforestry and
agrosilvopastoral systems. Adding a tree component to an annual crop or livestock
system may have the same effect of increasing both per hectare labor requirements and
increasing returns to labor. Because these systems are more complex, they typically have
greater per hectare (labor) management requirements as well as increasing the
productivity of land (i.e. total output per unit of land).

There are two keys to the success of these systems. First, there must exist some
sort of synergy between the trees and the other component of the system (otherwise they
could be produced On separate land units). For example, trees shade livestock from the
harsh tropical sun and fruit that falls to the ground provides a nutrition supplement.
Grazing by livestock controls the competition of other plants with the trees and reduces
weeding needs. Second, the trees must provide some product of value so that increased
per hectare labor requirements are accompanied by increasing returns to labor.

In sum, these two technology adoption case studies provide useful insights for
technology development and policy design. In an environment where land is abundant
relative to labor, farmer adoption of technologies that enhance land productivity is
dependent upon whether or not these technologies increase returns to labor. The design
and promotion of improved technologies should therefore focus on combining enhanced
land productivity with increasing returns to labor. However, if increased returns to labor
are primarily a result of decreasing per hectare labor requirements, then freed up labor is
likely to increase deforestation. It is therefore necessary to find production practices that
“absorb” labor away from deforestation by increasing labor requirements per hectare

while at the same time increasing returns to labor.

112

This thesis has found that labor availability plays a key role in determining land
use patterns. The case study essays on kudzu-improved fallows and brachiaria-improved
pastures just described have already provided numerous insights into labor’s key role.
The following paragraphs briefly expand upon these'findings.

Greater availability of hired and family labor are positively correlated to total
forest clearing and secondary forest clearing respectively. This is logical given that
regression analysis also shows that households with more family labor have more land in
annual and perennial crops and have larger cattle herds. In short, more labor leads to
more agricultural production which in tum leads to greater deforestation.

It is therefore important to orient labor away from extensive slash-and-bum
agriculture and towards more sedentary forms of agriculture. Two examples of more
sedentary agriculture mentioned previously include perennial crops and agroforestry
systems. Another potential solution involves greater exploitation of areas of alluvial soils
along rivers. These areas are much more fertile than “highland” soil areas because
periodic inundations deposit nutrient rich sediments. This greater fertility allows farmers
to cultivate fields for greater periods of time and thereby decreases pressures to clear new
areas of primary forest.

Farm households typically prefer to settle along areas of road infrastructure
because this allows quicker and cheaper access to markets. New roads into the jungle,
however, are most often constructed by timber, gas, and oil companies without regards to
the productive potential of the agricultural land they are opening up. Research should be
undertaken to examine the costs and benefits of road infrastructural pOIicies that provide

better access to areas of alluvial soils.

113

Another key labor-related finding of this research is that off-farm income
opportunities decrease the amount of land put into annual and perennial crop production.
Off-farm employment therefore has the potential to take pressure off the natural resource
base and decrease deforestation. So while slash-and-bum agriculture is the main cause of
deforestation in the Peruvian Amazon, improving the sustainability of agriculture is not
the only means of addressing this problem. Economic development policies that provide
off-farm income opportunities can also help alleviate deforestation.

Although this concluding section noted that both labor and capital are limiting
factors of production relative to abundant land, the focus until now has been on labor.
This focus has been due to the fact that, in spite of labor being relatively scarce, all
households have some access to labor whereas purchased input use is quite limited.
Nevertheless, this research found that when farmers do have access to capital this plays
an important role in determining land use patterns.

Use of capital inputs diminishes farmer clearing of primary forest. Most of the
capital inputs used in the zone (e. g., fertilizers, herbicides, improved seed, and
phytosanitary livestock products) enhance land productivity. Increased land productivity
permits a greater reuse of previously cleared land which, in turn, decreases the need for
farmers to clear new areas of primary forest.

This impact of capital inputs on deforestation bears out the rationale of the
traditional focus of improved technology development and promotion as previously
described in this concluding section. The problem with this approach is that in most cases
of abundant land and scarce capital it is more economical for farmers to use extensive

slash-and-bum practices than to adopt these technologies. Research should therefore give

114

priority to using economic analysis to identify crop and capital input combinations that
are both affordable and profitable for farmers.

This research also found a significant correlation between farmer access to capital
in the form of credit and farmer use of production technologies. Credit is positively
correlated with hired labor and capital inputs. This is logical given that these inputs
require a financial expenditure. In addition, credit is negatively correlated to farmer
adoption of kudzu-improved fallows and brachiaria-improved pastures. This again is
logical because these technologies conserve on scarce labor and capital resources. A lack
of access to credit may therefore push farmers to adopt these technologies.

In sum, access (or lack of access) to capital plays an important role in the decision
of farm household to use hired labor, capital inputs, brachiaria-improved pastrrres, and
kudzu-improved fallows. The use of these inputs and technologies has an important
impact on deforestation outcomes as previously described in this concluding section.
However, the impact on deforestation of these technologies is not uniform. There is
therefore a potential for using linked credit to promote agricultural production that
reduces deforestation such as the use of land productivity enhancing capital inputs or
perennial crop production.

Another key finding of this research concerns the relationship between
agricultural production practices and different types of deforestation. The literature on
deforestation often does not clearly define what is meant by deforestation. Often the
perception is that once primary forest is cleared the forest is gone forever. In recent years,

however, there has been a tremendous amount of secondary forest regeneration in Latin

115

America. In the case of the zone surrounding Pucallpa, secondary forest area is
approximately equal to primary forest area on farm household land holdings.

This thesis defined three different types of deforestation: primary forest clearing,
secondary forest clearing, and total forest clearing. Each of these was measured as the
hectares of forest cleared by a farm household on their land holdings during 1998.
Primary forests are areas that have never been felled in a clear cutting manner (though
often selective extraction has taken place), secondary forests are areas that have been
clear cut and have regenerated back into forest, and total forest is the sum of primary and
secondary forest.

This research found that kudzu-improved fallows, capital input use, and
agriculture in alluvial areas decreased primary forest clearing and increased secondary
forest clearing. The common thread in each of these cases is enhanced soil fertility.
Agricultural cultivation rapidly degrades most soils in the Peruvian Amazon. Secondary
forests are fallow areas that are recuperating soil fertility of degraded land in a system of
shifting agriculture. Greater soil fertility allows farmers to increase the reuse of these
degraded secondary forest fallow areas. Greater reuse of secondary forest fallows in turn
reduces the need for farmers to clear new areas of primary forest.

Increased use of secondary forest and decreased use of primary forest constitute
an intensification of land use. By reusing previously cultivated land (secondary forest)
there is less need to clear new areas of previously uncultivated land (primary forest). This
allows greater agricultural production using less total land over time. In terms of
deforestation, the ideal situation would be if farmers were able to continuously cultivate a

given parcel of land. Nevertheless, the reuse of secondary forest areas and decrease in

116

primary forest clearing is a step in the right direction of decreasing the negative
environmental impact of agriculture.

This finding demonstrates the need for research to distinguish between the
different types of deforestation. In the case of this research, several agricultural variables
had opposite effects on primary and secondary forest clearing. A failure to make this
distinction would have missed a critical insight into the dynamics of how agricultural
production affects forest clearing. This finding also suggests that, where economically
feasible, technology development should focus on land productivity increasing practices
that allow greater reuse of secondary forests in order to reduce primary forest clearing.

A final finding of this research involves the identification of an evolutionary trend
in land use patterns from annual crops to livestock. New migrant farmers tend to settle on
the edges of the agricultural frontier and slash and burn primary forest in order to
cultivate annual cr0ps. During the first ten years of settlement, this leads to a decrease in
primary forest area and an increase in secondary forest area as trees regenerate in the
fallow areas of the shifting agricultural system.

In the time periods 10-20 years and greater than 20 years after farmers settle a
plot of land, there is a continued decline in primary forest and an increase in pasture area.
However, this trend does not reflect a direct conversion of primary forest into pasture.
Rather, farmers continue to slash-and-bum primary forest for annual crop production and
install pastures in degraded areas of secondary forest.

This pattern of land use is due to two factors. First, migrant farmers rarely have
enough capital to immediately invest in cattle ranching. They therefore initially cultivate

annual crops, which require much lower start-up capital outlays. Cattle is considered a

117

more profitable and less risky activity so many farmers progressively accumulate capital
in order to purchase livestock. The time needed to accumulate capital explains why
pasture areas do not significantly increase until this second time period.

A second explanation is that pasture grows well on degraded land whereas annual
crops need relatively more fertile land. Farmers therefore first cultivate more fertile land
with annual crops and then progressively install pastures as land begins to degrade. Herd
size’s lack of correlation with primary forest clearing and positive correlation with
secondary forest clearing further reflects the evolutionary nature of agricultural
production from annuals to livestock (and the importance of distinguishing between types
of deforestation).

The fact that pastures tend to be installed on previously cleared land and allow a
productive use of degraded areas is a positive result. Furthermore, pastures in the
Pucallpa area maintain their productivity and do not require shifting slash-and-bum
practices. For a given year, cattle is not correlated with primary forest deforestation.
Correlation analysis of cattle and deforestation over time, however, indicates that
ranching both increases primary forest clearing and decreases the presence of secondary
forest.

This outcome is due to the fact that livestock is less intensive in its use of labor
and more extensive in its use of land. The labor requirements for a hectare of land in
cattle ranching for a given year of production are only about 8.4 percent those of annual
crops. This allows an expansion of production onto more land.

In sum, the effect of cattle ranching on deforestation is mixed. Over time it does

increase deforestation. However, it does permit a productive use of already cleared land

118

that is less suitable for annual cr0pping. If farmers were not able to move into cattle
ranching they would probably eventually be forced to abandon degraded land and move
out to the agricultural frontier and clear more primary forest for annual cropping.
Furthermore, essay 3 found that ranching increases financial returns to both labor and
land relative to annual cr0pping. This indicates that livestock is a relatively productive
use of land and can have a positive impact on reducing poverty.

This dissertation has focused on analyzing ways of promoting more sustainable
production practices that reduce deforestation. However, if these improved production
practices are sufficiently profitable, then farmers may hire more labor or invest in labor-
saving inputs such as chain saws and tractors that allow more total land to be brought into
production. As long as land is cheap and abundant, farmers have a strong incentive to
increase the amount of land in production even when using more sustainable production
practices. This can have the perverse effect of accelerating deforestation. Restrictions on
access to land may therefore by a necessary complement to any attempts to promote
sustainable agriculture in the Peruvian Amazon. Policy research needs to investigate what
types of land use controls are feasible and effective.

This dissertation thesis has analyzed deforestation at the household level. This is a
relevant level of analysis because household agricultural production decisions are the
direct causes of deforestation. However, if immigrants continue arriving in the jungle to
farm, it is clear that deforestation on the landscape level will accelerate. The principal
cause of migration into the jungle is land scarcity and poverty in regions outside the
jungle. Programs of economic development are therefore necessary in these regions to

provide economic opportunities to individuals that would otherwise migrate to the jungle.

119

In this sense, an important part of the solution to deforestation of the rainforest lies

outside of the Amazon region.

120

APPENDIX A

Main Household Survey

121

 

 

AGRICULTURA SOSTENIBLE EN LA AMAZONIA PERUANA
INVESTIGACION DE ASB-ICRAF— 10-98.

 

Encuestador .......................... Fecha ..................... Encuesta deigo ..........
Coordenada X ............ Coordenada Y .................
1 - r = 2 2 n = 22

W
I. Nombre completo del infonnante: .............................................................. (NOMINF)

2. Nombre completo de la persona que toma decisiones en la parcela.
......................................................................................... (N OMJEF)

(SEXJEF) Hombre( 1 ) Mujer( 2)
3. Relacion del infonnante con la persona que toma decisiones en la parcela:
(RELINF)

4. Sitio donde se encuentra e1 predio: km. Fed. Basad ................. (KMFB)
km. 1 entrada .................... (KMENTI)
km. 2 entrada .................... (KMENTZ)
km. 3 entrada .................... (KMENT3)
Izquierda( 1) Derecha( 2) (LADO)

5. Nombre del caserio: ..................................................................................... (CASERI)
Colono ( l ) Nativo ( 2) (ETHNIA)

W (USA UN CROQUIS)

6. LCuantas hectareas totales tiene su fundo incluyendo todas las parcelas (que estén en
produccién o no, incluso: purmas, cultivos anuales, monte alto, cultivos perennes,

pastas, etc.)

Parcela l ............ Has.
Parcela 2 ............ Has.
Parcela 3 ............ Has.
Parcela 4 ............ Has.
Iota} .................. (HTOTAL)

122

 

7. (Cuantas hectareas estén en purrna? (HPURMA): . .Has.
0 LCuantas hectareas estén en punna baja de menos de 5 metros incluso kudzales? (haz
un punto de referencia de 5 metros tal como la altura de la casa 0 un tirbol)

(HPURBA)

...Has.
0 gCuantas hectareas estén en purrna alta de mas de 5 metros?(HPURAL) . ..Has.
- LHay problemas de incendios incontrolados en sus purmas Si ( 1 ) No ( 2)
0 Observaciones ......................................................................................
8. (,Cuantas hectareas estan en monte alto? (HMONTE) .Has.
9. LCuantas hectareas estén en pastos? (HPASTO) Has
10. LCuantas hectareas estén en cultivos anuales? (HAN UAL) .............. Has.
0 gCuantas hectareas son de cada cultivo anual y asociaciones.
CULTIVO AN UAL Has.
Platano (HPLAT)
Yuca (HYUCA)
Maiz (HMAIZ)
Frijol (HFRIJ 0)
Arroz (HARROZ)
Asociacionl Anual-Anual:
Asociaci6n2 Anual-Anual
Otro:
Otro:
ll. gCuantas hectareas (o arboles) en total estén en cultivos permanentes?
(HPERM) .. ............ Has.

0 LCuantas hectareas (o arboles) hay de cada cultivo permanente que Ud. produce.
CULTIVO PERMANENTES Has. Arboles Total

123

Limon

Naranja

Papaya

Palrna Aceitera

Palmito

Asociaci6n1 Perm-Annual
Asociaci6n2 Penn-Perm
Otro:

Otro:

Otro:

Total:

12. (,Cuantas hectareas de reforestacion tiene Ud.? ............... Has. (HREF OR).
0 Cuales especias tiene Ud. ...............................................................

o Observaciones .............................................................................

13. (,Ud. dijo que tiene hectareas en punna no? LUsa Ud. algo para mejorar la

recuperacion de los suelos en punna por ejemplo kudzu u otra cosa?
Si( 1 ) No ( 2) (RECUP)
211mm _ Has.
Kudzu (HKUDZU) ‘
Otro:

Otro:

o (Ud. siembra kudzu o crece solo? (SEMKUD) Siembra( l ) Crece s010( 2)
o LPiensa Ud. que el kudzu es sobre todo una planta positiva con beneficios o sobretodo

negativo con inconvenientes (KUDZOK): Positivo( 1 ) Negativo( 2)

124

o Cuales son los aspectos positivos imporlames de kudzu para su produccion
(KUDPOS)

! | R 'I' l K 1 1/
Abona la tierra( l)

Disminuye 1as malezas(2)

Forraje(3)

Otro:

Otro:

Otro:

0 Cuales son los aspectos negativos 11112011311128 para Ud. en su produccion
(KUDNEG)

WW! 1/
Invade areas de otros cultivos(1)
Otro:
Otro:
Otto:
14. LUd. dijo que tiene _ hectareas de pastos, no?

0 (Que tipos de pastos hay? LCuantas hectareas hay de cada uno?

Tipo de Pastos Has.
Pasto Natural (tourouco) (PASTN A)
Braquiaria (PASTBR)
Brachiaria-Kudzu (PASTBK)
Otro:

15. LCuantas cabezas de ganado tiene usted (CABGAN) ............... cabezas

125

0k ' 0, 0 9.. 0'1. -‘._|-_ -0. a- Li;

(Explica al agricultor que queremos saber co'mo se ha cambiado el uso de sus tierras al
través del tiempo)

 

l6. (,Alguien cultivo éste lote antes que usted? Si No
a LPor cuanto tiempo? (LOTANT) ....................... Afios
l7. gCuanto tiempo ha estado su familia cultivando esta tierra sin contar afios de
ausencia? Parcela 1 (T LOTPl) ................. Afros
Parcela 2 (TLOTP2) ................. Afios
Parcela 3 (TLOTP3) ................. Afios
Parcela 4 (TLOTP4) ................. Afios
0 Cuando llegaron cuantas hectareas de habia:
Uso de Tierra PARC] PARC2 PARC3 PARC4
Purma (alta y baja) (LPURMA)
Pastos (LPASTO)
Cultivos Pennanentes (LPERM)
Cultivos Anuales (LANUAL)
Monte Alto (LMONTE)

o Hace cinco afios en 1993 (menciona un acontecimiento) cuéntas hectareas de

 

. habia: (si el agricultor no estaba, pasa a la siguiente pregunta).
Uso de Tierra . PARC] PARC2 PARC3 PARC4
Purma (alta y baja) (SPURMA)
Pastos (SPASTO)

Cultivos Pennanentes (SPERM)
Cultivos Anuales (SANUAL)
Monte Alto (SMONT E)

18. LCuales son los cambios mayores que han ocurrido en la zona desde su llegada y

que han tenido un impacto sobre la produccion en su finca?

CAMBIQS Emu)“- neg, igual)
PRODUCCION INSUMOS MONTE ALTO
(+ - =) (+ - =) (+ - =)' TUMBADO

126

19. LQué precios recibio Ud. para los productos agricolas siguientes:

PRODUCTO LUGAR DE UNIDAD DE PRECIO: SI. TRANSPO:
AGRICOLA VENT A MEDIDA ; POR UNID. » S/. POR ‘
MEDIDA UNIDtMED.

YUCA

MAIZ

ARROZ

PLATANO

127

F RIJOL

OTRO
rMpnn'rAN'rrr

CULTIVO
PFDM A N1

CULTIVO
mm M A N?

CULTIVO
PFDM A we

GANADO
'VAPIINO

20. Si Ud. solo va con al mercado, LQué cantidad llevarias en promedio?

PRODUCTO CANTIDAD UNIDAD DE TIEMPO AL
‘ MERCADO
MEDIDA (Ida y Vuelta)

 

Yuca
Platano

Maiz

128

Arroz

Cultivo Pennl:
Cultivo Perm2:
Cultivo Perm2:
Ganado

Otro:

Otro:

SECCION 5: CARACTERISTICAS DE LA FAMILIA

21.

23.

24.

25.

26.

27.

28.

29.

LCuanto tiempo hace que vive usted en la zona de Pucallpa?
(ANOZON) ....................... Afros
gReciben Uds. mas que 10% de sus ingresos de personas que no viven en el predio?
(REMI'I) Si ( 1) No (2)
LMandan Uds. mas que 10% de sus ingresos a personas que no viven en el predio?
(MAND) Si (1)No (2)
LBS la persona que toma 1as decisiones en la parcela originalmente de la costa, sierra
o selva? (REGNOR) Costa ( 1) Sierra (2) Selva (3)
(,La persona que toma 1as decisiones en esta parcela antes de venir acé, ha vivido
alguna vez en una chacra o fundo? (AGEXP) Si ( 1) No (2)
gCuantos meses (semanas, dias) durante el afio, no hay mucho trabajo que
hacer en su proprio fundo: Meses.... . . . . . . . .. Semanas ............. Dias ................
(ESTBAJ) Total ............... dias

gHa habido momentos en los r'rltimos 5 afios cuando la produccion en su chacra no
bastaba para 1as necesidades de alirnento de la familia? gCuantas veces?
(SEGALI) ............ veces

gCree Ud. que su familia va a estar cultivando este mismo frrndo en 5 ar‘ios?
Si ( 1) No ( 2)

Porqué

 

129

SECCION 6: CARACTERISTICAS DEL PREDIO/SITIO
31. (,Qué tipo de suelo predomina en su predio.(SUELOS)

Wine- Parcl Parc2 Parc3 Parc4
CARREIERA:
Arcilloso:
(Gredoso; tierra roja,blanca, amarilla,gris,
negra)
Arenoso:

Limoso
(bajiales, aguajales, suelos inundables)
RIOS;
Franco arcilloso-arenoso:
(Restinga alta)
Arenoso:
(Restinga baja)
Arenoso limoso:

(Barrizal, suelos aluviales)
32 (Hay mas tierras fuera de su predio disponibles que Ud. podria adquirir para cultivar?

(TERNVA) Si (1)No (2)

33.1,Cuantos kilometros hay entre su predio y Pucallpa, usando la ruta que tomas para ir a

Pucallpa? (DISMKT) .................. km.
34. LCuantos kildmetros hay entre su predio y: la escuela mas cerca? ............. km.
e1 puesto de salud mas cerca? ............. 1cm.
(DISIN F) total .............. km.

35. OJOII Encuestador llena lo siguiente (no se lo pregunta al agricultor! !)

Tipo de Infraestructura donde se 1/
Encuentra e1 Predio (INFRA)

Federico Basadre (1)
Carretera Secundaria (2)
Carretera Terciaria (3)

Rio (4)

 

_ SECCION 7: APOYO INSTITUCIONAL

130

36. LCuantas visitas ha tenido usted de cualquier tipo de agente de extension agropecuaria
durante e1 ar’io pasado? (EXTEN S) ....................... Visitas
37. (Ha recibido crédito en los riltimos cinco afios incluso crédito en especie?
(CREDIT) Si( 1 ) No( 2)

TIPO DE CANTIDAD INSTITUCION O REQUISITOS
CREDITO PERSON A QUE
OTORGA

SECCION 8: SEGURIDAD DE TENEN CIA
38. gTiene Ud. algr'rn documento de propiedad de su chacra-para cada uno de sus parcelas

si tiene mas que una?

Nro. DE TIENE TIPO DE FORMA DE mugggggfim
PARCELA DOCUME DOCUMENTO (PROPIEDAD . .CULTIVANDO eon
N?" - - p...- =1 -‘ rte-9.3339“
$1 = 1, . “(111118sz . Si=1 N0=2
N0 = 2 Prestado=3
Posseslouario Precarla-nt
Parcela 1
Parcela 2
Parcela 3

Parcela 4

131

. ..A "u
.‘l

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

SECCION 9: INGRESOS SUPLEMENTARIOS
40. {,Qué productos ha cosechado su familia en el r'rltimo afio del W que se

encuentra en su predio? (dile toda la listall)

PRODUCTO UNIDAD CANTIDAD PRECIO/ POR VALOR
DE MEDIDA UNID. DE TOTAL
MEDIDA SOLES

F ruta l :
Fruta2:
F ruta3:
Maderal :
Madera2:
Madera3:
Madera4:

Lefra

Carbon

Miel de Abej a
Hojas

F ibra
Serrrillas
Otro:

Otro:

Otro:

Otro:

Otro:

132

 

41. (,Qué OTROS PRODUCTOS PRODUCE EN LA CHACRA su familia que no son

los cultivos anuales, cultivos perrn's, o ganado vacuno? Por ejemplo: (dile la lista.’.’)

ACTIVIDAD r/
Gallinas
Chancho

Leche
Piscigranja
Otro:
Otro:
Otro:
Otro:
Otro:
Otro:
Otro:
Otro:

42. LCuantos dias (o semanas o meses) en el afio cada individuo del hogar TRABAJA
FUERA DE LA CHACRA y qué actividades hacen por ejemplo: trabajo jomal
agricola, trabajo jomal agricola, extraccion madera fuera de su chacra caza firera su
chacra, fi'uta frrera su chacra? (obligatorio decirle la listall)

PERSONA ACTIVIDAD porejernplo: trabajojomal TIEMPO SOLES/ TOTAL

agricola, peace, extraccion madera fuera do an chacra TIEMPO
cazafuerasuchacra, fruta fuerasuchacra?

133

SECCION 10: USO DE INSUMOS
43. LDurante e1 riltimo ar’io han venido personas que no son de su familia a trabajar en su
fundo? (que sean jomales contratados, minga, u otra). Si No

Actividades N 0 Personas Cantidad de dias Pago por dia
(jomal, minga, etc.) por persona

Tumba y
Quema

Siembra
Deshierbo
Cosecha
Otra:

44. We, ha Ud. usado los siguientes insumos:

TIPO DE INSUMO UNIDAD DE CANTIDAD COSTO ' .
MEDIDA TOTAL SOLES (S/.) ESPECIES CULTIVO

F ert Quirnl:
Fert Quim2:
Fert Quim3:
F ert Quim4:
Fert. Orgl:
F ert. Org2:
Fert. Org3:
Herbicl:

134

Herbic2:
Semilla Cert:

Insecticides
Fungacidas
Otro:
Otro:

SECCION 11: PRACTICAS DE TUMBA Y QUEMA
45. (,Cuantas hectareas en total tumbaste y quemaste en el r'rltimo afio?
(HT&Q) Has.

46. (De 1as hectareas en total que tumbaste y quemaste e1 r'rltimo afio, cuantas
hectareas tumbaste y quemaste de

HECTAREAS
TUMBADA Y
QUEMADA ULTIMO
AN 0

Monte Alto (MTT&Q)

Purma Alta (PAT&Q)

Punna Baja (PBT&Q)

Pasto (PST&Q)

TOTAL (TLT&Q)
47. (,En promedio, durante los r’rltimos tres afros, cuantas hectareas en total tumbaste y

quemaste cada afio? (3HT&Q) ................... Has.
0 (De 1as hectéreas en promedio que tumbaste y quemaste cada afio durante

los riltimos tres afios, cuantas hectareas en promedia tumbaste y quemaste de:

PRpMEDro DE HECTAREAS POR
ANO TUMBADA Y QUEMADADE
L083 ULTIMOS ANOS

Monte Alto (3MTT&Q)
Purrna Alta (3PAT&Q)
Purma Baja (3PBT&Q)
Pasto (3PST&Q)
TOTAL (3TT&Q)

135

48. (De la cantidad total de tierras trabajado en la campafia pasada cuanta de ellas
va Ud. a dejar enpurrnar?
SUPERFICIE PARA EN PURMAR
1':er ANO arAs.)
Cultivos anuales
(ANPRM)
Cultivos permanentes
(PEPURM)
Pastos (PAPURM)
Otro:
Otro:

49. LCuales son los rendimientos promedios por hectérea de sus cultivos principales?
CULTIVO AN UAL Rendimientos lHas.

Platano (RPLAT)
Yuca (RYUCA)
Maiz (RMAIZ)
Frijol (RFRIJ 0)
Arroz (RARROZ)
Cultivo Perml:
Cultivo Perm2:

Cultivo Perrn3:
1 36

Otro:
Otro:

Otro:

137

N "oz

a "a
~33...
.2. 95...?
«Hannah.

"Samoa... 4.. z... 872 Symon: gong

N 0:7.

w "mm
25.0.:—
_o.. 955..
«mane-.8

.. . "45.0... 87.... Evan...

9.0.2.5
«.0 3.53.
cacao nu Egon .
a. 8 a...» 2.5.8 .05.... _ _ .
2:. 35. on 2:. afloaoaum ween..— H 8333...... seem-3n a. 2. 9:.an

Cum—HM.— Am Zm— HE HBO m<ZOmflm mean HG m<U~HmEHHU<M<U .cm

138

APPENDIX B

Kudzu-Improved Fallow Survey

139

 

 

ENCUESTA SOCIO—ECONOMICO SOBRE BARBECHOS MEJORADOS
ICRAF-ASB PUCALLPA PERU ENERO 1999

 

Encuestador .......................... Fecha ..................... Encuesta Codigo ..........

Coordenada X ............ Coordenada Y .................

W

3. Nombre completo del infonnante: ..............................................................
(NOMIN F)

4. Nombre completo de la persona que toma decisiones en la parcela.

(N OMJEF)
3. Relacion del infonnante con la persona que toma decisiones en la parcela:
(RELINF)

15. Sitio donde se encuentra el predio: km. Fed. Basad ................. (KMFB)
km. 1 entrada ....................
(KMENTI)
km. 2 entrada ....................
(KMENTZ)
km. 3 entrada ....................
(KMENT3)
Izquierda( 1 ) Derecha( 2) (LADO)
l6. Nombre del caserio: .....................................................................................
(CASERI)

140

 

.22.... 3.. 325...... a. An. 8838 up 82.88.. a. N. «82... a «no.

.m m .m m m .m
.N N N .N .N N
A . _ ._ A _
.m m .m .m .m .m
.N N N .N .N N
._ _ _ ._ ._ _
.m m m .m .... a
.N N .N N .N N
._ _ A _ .— —
.m m m .m .m m
.N N N .N .N N

as 8883
8.5.52 2....

3.5304 22

88881." ages-nu
Segue-n 2....

3>=.=U-N 2:,

33an

.388 0.. 0.88... _o 2. “mo 0.... o. :o .385. 8.3.. «8...... .8... «88.. 68081.3. £083 8: 888: 880 «6.5.5.2.. a. mo 3:0... A... .N

"85‘
33—

83.4
8.851. a... 2.5...

.n

.85..
«magi.- 82 «8.5..

82:

83.504 82 8.82

. 88.5.. 3:.
288.98.. 83:3 x 8.28 8208.

 

0.. 8.... m8. .388 0.. $88.. one .88 82.3.38. 0.. 5.80 ..o .33 5.58% 8.8.68.8 83:8 mo. :8 3.80... 3.

2888 28.3.. 0.. 8.8.

 

up «2% 8. 0.. 2.8: 2... 28.8.. «8. .5 «>23 2380.... 8. menu 88.8.. 088.5... A... .—

I41

..«N...... 0 .«..«.. «8...... .0..«
«8...... .0..« 0808 0.. «0.... 8. .«...8.. 0.. «65.0.0... 0. «.00..« 0.... ..«.......0.. .. .m«~0.«8 .«00. 0.. 0.8%.. 0.... 9.8 .0.0«.. 8.0 ..«m... A...

 

N . , . .55.
. m N «..«m «8......
N N m «.3. «8......
m v v 0.3. 02.0.).
88000 m .003.
.0>«.0 .80... 0.... ..«.......0... 280.3). 0.. «.0080... 808080.. 0.2.0....
32...... 0......20.
60.0.50.— 60.02.00

6.32.. v m N . 6.22...
"9.2 "008.2

 

 

 

 

 

 

 

 

142

.303.

.30...

«..< «8...... «..«m. «8......
a... 3:080.—
.«587503 «8000

0.. «0.800.. «8.

.N
..
0..< 0803.
g 05 «2.0
€805 0.08 0.080.... .0... «7.0.0.. «8. 3.08.. 0...... «.80 «..0 «.«0 .03 «w«..... A... m
«.5805 .305.
A...
«8.....—
5?.
«8......
82.
02.0.2
3... «2.3.0.. 2...— 2.03.0.—
.«euoifioc «00m Aug—«50c «08:...

«.«.. .03 ..«ms ««.0 8.8.30 08800 30080.. «88.00... A... .m

143

ma:—

0000

000.9

.320

"m 00¢.

"N 9¢< Aug—«9.093
.0005—
: 0¢< 0000.309

Aug—00003
.. .5 0.5
a 22 «E009
"— 00< 00005009
"m 0.2
a 22 €68..an

52 00000.9

3 00¢. 0000.309
"n 0.2

"N o¢< Adamo-0E9:
82 00002
"9 00¢ 0000.309

0000> 00.9 000.— 0000> 00.9 0.00.— 0000> 209 0009 m000> 00.9 0.009 0000>

25:0 30> 0..: "20¢

_0 g :0 :0 00000800 00 .0: 000 gum 00 0003000 00: .03 0000300 m000> 080000

.0

144

145

"m 05.. ”m 00¢. “n 05.. ”n 05.. ”n 05..
"a 00< “N 00< ”N 05. ”N 05. "N 05.. Ana—000.00.
.0005.
0. 00¢. u. 00< u. 0.00. n. 00< n. 05. 0.000.009 800.00.003—
”m 05. “m 00< “m 00< ”m 00¢. ”m 00¢.
"N 05. ”N oc< ”N 05.. ”N 05.. ”N 00< 92000.00.
0.09 00.009
H. 05‘ u. 05. n. 05. n. 00< n. 00< 0.000.009 0.00.00.00.09
"m 05. ”m 0.04. "m 00< ”m 0..... ”m 00<
0 2.2 0 22 ”N o.< 0 22 "a 2.2 0.53250
52 2.0....
u. 00< u. 05. 0. 00¢. n. 05. u. 0..< 0.000.009 0.00.0..0009
”m 0...< ”m 00< "m 05.. ”n 05‘ "n 0..<
”N 05. “a 00... "N 0.2 ”N 004. ”N 00< ?.<000.00.
82 3.32
n. 05.. n. 05. n. 0..... H. 0..< u. 00< 0.000.009 0.00.00.00.09
.03.... 0050.9 «00> 0.02 000.3
000.00...
0. 00 0000000 000 0000 000 00 0000 00 00.00... 00 0000000 0030.00 00.00.0m.m mo. 00 0.000.000 0.00.8000. 0.0. m0 .000 .m
.m
.N
..
.0003. 0..< 00.009 00.9 00000 0..< 0.002
30.903 0000

20.20.0000 0.0.00.3 .00. 0.00000 000 00000000 0000 0000 0.00 0000 .03 0w0n...v ..0 v

 

0000
.302

0000 00.0000 00 0000000
sz 00000 00 0000000 0.00.0 0.00000 .0008. 00 0000 00 0000000 00.00.00.008 00 00.000000 00 .03 000.00 0....0 ..0

 

 

 

0000 0000.. 00 00 00.00000. .03 000.. 0000 00.0000 00009..0 ..0
0000 00.00.0030 0 00>.0> 00 00.00 .0008. 00 00.00 00.. >00 0000 0000000... A...
0000 0000.0 0.0000

 

0000 000.000 0.0000 ”000 0.0000000 00 0000.. 0000 0000000..0 ..0 .0

146

APPENDIX C

Brachiaria-Improved Pasture Survey

147

 

ENCUESTA SOCIO-ECONOMICO SOBRE PASTOS MEJORADOS
ICRAF-ASB PUCALLPA PERU ENERO 1999

 

 

 

Encuestador .......................... Fecha ..................... Encuesta Cédigo ..........
Coordenada X ............ Coordenada Y .................

5. Nombre completo del- infonnante: ..............................................................
(N OMINF)

6. Nombre completo de la persona que toma decisiones en la parcela.

 

(NOMJEF)
3. Relacién del infonnante con la persona que toma decisiones en la parcela:
(RELINF)
17. Sitio donde se encuentra e1 predio: km. Fed. Basad ................. (KMFB)
km. 1 entrada ....................
(KMENTI)
km. 2 entrada ....................
(KMENTZ)
km. 3 entrada ....................
(KMENT3)
Izquierda( 1 ) Derecha( 2) (LADO)
l8. Nombre del caserio: .....................................................................................
(CASERI)
l9. éCuantas hectareas de pastos tiene Ud.? has.
0 (Que tipos de pastos hay? LCuantas hectareas hay de cada uno?
Tipo de Pastos Has.
Pasto Natural (torouco) (PASTNA)
Braquiaria (PASTBR)
Brachiaria-Kudzu (PASTBK)
Otro:
7. LCuantas cabezas de ganado tiene usted? (CABGAN)
............... cabezas

8. LDénde instala Ud. un nuevo pasto en su fundo? (donde habia:)
Cultivos Purma Baja Puma Alta Monte Otro:

148

Brachiaria
Torouco

Otro:

Siempre (1) Frecuente (2) A Veces (3) Muy Poco (4)

(,Cuantas personas durante cuéntos dias usan Uds. para IN STALAR una nueva

hectarea de pasto de

 

MAN 0 DE OBRA DE LIMPIEZA:

:BBACHIAEIA
Limpieza de

 

Personas:
Dias:

Total:

IQRQLLCQ
Limpieza de

 

Personas:
Dias:
Total:

QIRQ;

Limpieza de

 

Personas:
Dias:

Total:

Limpieza de
Personas:
Dias:

Total:

Limpieza de
Personas:
Dias:

Total:

Limpieza de
Personas:
Dias:

Total:

Mano de obra de siembra:

SIEMBRA
Brachiaria
Personas:
Dias:
Total:

Mano de obra de otra actividad de instalacibn:
WW

Torouco

Brachiaria
Personas:
Dias:

Total:

Torouco

Personas:
Dias:

Total:

Personas:
Dias:

Total:

149

 

 

 

Limpieza dc

 

Personas
Dias

Total

Limpieza de

 

Personas
Dias

Total

Limpieza de

 

Personas
Dias

Total

Otro:
Personas:
Dias:

Total:

Otro
Personas:
Dias:

Total:

10. a. (Hace Ud. MANTENIMIENTO de sus pasto
(por ejemplo desyerbo, quema, etc...)

No(2)

b. LQue’ tipo de MANTENIMIENTO hace Ud. en su sus pastos?

Desyerbo s i

Quema s i
(,Cada cuanto tiempo? LCada cuanto tiempo?

Otro:

(,Cada cuanto tiempo?

Otro: s i
(,Cada cuanto tiempo?

 

 

 

 

c. LCuantas personas durante cuantos dias usan Uds. para MANTENER una
hectarea de pasto de

Deshierbo
Personas
Dias

Total

Deshierbo
Personas
Dias

Total

Deshierbo
Personas
Dias

Total

Quema
Personas
Dias
Total

Quema
Personas
Dias
Total

Quema
Personas
Dias
Total

en un afio?

BRACHIARIA

Otro:
Personas
Dias

Total

TOROUCO

Otro:
Personas
Dias

Total

OTRO:
Otro:

Personas
Dias

Total

Otro:
Personas
Dias

Total

Otro:
Personas
Dias

Total

Otro:
Personas
Dias

Total

11. LUsa Ud. algunos insumos o méquina para el pasto como fertilizantes, tractor

(para el pasta no para el ganado)
Brachiaria (Soles/halaflo) Torouco (Soles/ha/aflo) Otro(Soles/halafio)

Insumol :
InsumoZ:
Insumo3:

12. LUn hectarea de

 

150

puede alimentar cuantas cabezas de ganado

13.

14.

15.

Brachiaria Tourouco Otro:

 

 

Cabezas: Cabezas: Cabezas:
a. LCuantos afios tienen sus mas antiguos pastos de ?
Brachiaria Tourouco Otro:
afios afios afios
b. (A su conocimiento, hay que abandonar un pasto de
despue’s de cierto numero de afios porque ya no produce

suficiente forraje?
Brachiaria T ourouco Otro:
Si( 1 ) afios Si( 1 ) afios No( 2) Si( 1 ) afios No( 2)
No(2)

c. LSe baja la produccion de forraje despue’s de cietta cantidad de afios en un

pasto de:
Brachiaria Tourouco Otro:
No(2) No(2) No(2)
Si: un poco mucho Si: un poco mucho Si: un poco mucho
Después de afios Después de afios Después de aflos
0 Si si, Lpor que'?

 

 

Si Ud. tiene pasto sin ganado, aporque' ha sembrado pasto?

o Rango de importancia: mas importante (1) ........... (5) menos importante
Compraré ganado Ayuda a obtener Anmenta valor de la lnvadio cl por Otro:
con ahorro personal crédito para ganado {undo para venderlo Por motivo de _

RANGO: RANGO: RANGO: RANGO: RANGO:
Observaciones: Observaciones: Observaciones: Observaciones: 0bserv.:

Si Ud. tiene ganado, Lcomo lo obtuvo?

o Rango de importancia: mas importante (l) ........... (5) menos importante
Ahorro personal de la venta de productos del fundo Crédito del gobierno
Ahorro personal de trabajo fuera de la flnca Otro:
Ayuda de pariente/amlgo ' Otro:

151

Observaciones

 

 

 

16. LCuales son las ventajas y desventajas de:
BRACHIARIA TOROUCO OTRO:

152

APPENDIX D

Map of the Pucallpa Research Zone

153

Aguatia River
Nueva Requcn
Curimana Ucayali River

Ncshuy (60km) CamnV ’0' . .. Q Pucallpa

(0km) _______f ,

O
Federico Basadre Highway

W8

Tournavista

 

Pachcitea River

154

APPENDIX E

Comparison of Input and Output Regression Models
Using Ordinary Least Squares (OLS) and
Seemingly Unrelated Regression Equations (SURE)

155

OUTPUT REGRESSION RESULTS USING OLS AND SURE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ANNUALS (OLS) ANNUALS (SURE)
R2 .54 Adj R2 .35
Coefficient Stand Err Signif Coefficient Stand Err Signif
CONSTANT 2.3107 2.7267 0.403 1.9484 2.4799 0.433
FARM SIZE 0.0193 0.0286 0.504 0.0306 0.0273 0.263
SECONDARY FOREST 0.0472 0.0506 0.357 0.0295 0.0523 0.573
PRIMARY FOREST 0.0521 0.0339 '0.134 0.0353 0.0321 0.272
ALLUVIAL SOILS 1.5672 0.7492 “0.044 1.5082 0.8202 '0067
YEARS IN LOT 00654 0.0442 '0.148 00596 0.0476 0.211
EDUCATION 0.3694 0.7260 0.614 0.0376 0.7480 0.960
AGE -0.0181 0.0284 0.528 0.0052 0.0269 0.846
FAMILY LABOR 1.2812 0.3673 "'0001 1.3439 0.3619 ""0000
ORIGEN (not jungle) -0.5672 0.7947 0.480 0.4650 0.8164 0.570
CREDIT 0.7699 0.7173 0.291 0.1904 0.7754 0.806
OFF-FARM INCOME 00003 0.0001 “0.040 00003 0.0001 “'0.006
EXTENSION -0.0652 0.0588 0.275 -0.0651 0.0563 0.249
REAL PRICE MAIZE -9.0094 5.6918 '0.123 -6.2960 5.1639 0.224
TENURE SECURITY 1.8436 1.8945 0.337 -2.6879 1.5196 '0.078
PERENNIALS (OLS) PERENNIALS (SURE)
R2 .50 Adj R2 .30
Coefficient Stand Err Signif Coefficient Stand Err Signif
CONSTANT -3.6355 1.8869 ‘0.062 -2.3615 1.8417 0.201
FARM SIZE -0.0140 0.0198 0.484 00253 0.0204 0.217
SECONDARY FOREST 0.0206 0.0350 0.559 0.0371 0.0392 0.345
PRIMARY FOREST 0.0322 0.0235 0.179 0.0432 0.0240 '0.073
ALLUVIAL SOILS -1.1651 0.5185 “0.031 -1.1109 0.6149 ‘0.072
YEARS IN LOT -0.0215 0.0306 0.488 -0.0286 0.0357 0.423
EDUCATION 0.1351 0.5024 0.790 0.0348 0.5613 0.951
AGE 0.0691 0.0197 “'0001 0.0700 0.0200 ““0001
FAMILY LABOR 0.6514 0.2542 “0.015 0.4071 0.2706 '0.134
ORIGEN (not jungle) 1.1024 0.5499 “0.053 1.0765 0.6053 '0.077
CREDIT 0.6360 0.4964 0.209 0.5633 0.5832 0.335
OFF-FARM INCOME 00001 0.0001 '0.132 00001 0.0001 0.333
EXTENSION 00770 0.0407 '0.067 00396 0.0417 0.344
FM-GATE PR MAIZE —0.1838 3.9387 0.963 -0.1540 3.8249 0.968
TENURE SECURITY 0.6538 1.3110 0.621 -0.6622 1.0687 0.536
LIVESTOCK (OLS) LIVESTOCK (SURE)
R1758 Adj R5540
Coefficient Stand Err Signif Coefficient Stand Err Signif
CONSTANT 40.0943 7.8384 0.207 4.973432 8.253966 0.547
FARM SIZE 0.3262 0.0820 “0.000 0.3199041 0.0932539 “’0.001
SECONDARY FOREST 00556 0.1541 0.721 0.1021822 0.1849813 0.581
PRIMARY FOREST 02996 0.1022 "‘0006 -0.2700278 0.1127295 “0.01 7
ALLUVIAL SOILS 2.3259 2.5750 0.373 2.617264 2.930292 0.373
YEARS IN LOT 0.1375 0.1386 0.328 0.1180356 0.1702934 0.489
EDUCATION -1.0624 2.3444 0.653 0.6935119 2.668329 0.795
AGE 01511 0.1003 '0.141 0.2461703 0.0906417 "'0007
FAMILY LABOR 1.9733 1.0900 ‘0.079 1.805333 1.283831 0.161
ORIGIN (not jungle) 2.3586 2.5465 0.361 0.412634 2.621776 0.875
CREDIT -1.8472 2.2939 0.426 -2.615343 2.836407 0.358
OFF-FARM INCOME 0.0001 0.0004 0.844 0.0001541 0.0004553 0.735
EXTENSION 00925 0.1980 0.643 0.1803244 0.1776334 0.311
FM-GATE PR MAIZE 9.8958 17.9265 0.585 0.6506359 16.12668 0.968
TENURE SECURITY 5.5646 2.4881 “0.032 -6.615253 2.514173 “0.009

 

 

 

 

 

“‘Significant at the .01 level

”Significant at the .05 level
Coefficients are unstandardized

‘Significant at the .15 level

156

 

INPUT/TECHNOLOGY REGRESSION RESULTS USING OLS AND SURE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IMPROVED FALLOWS (OLS) IMPROVED FALLOWS (SURE)
R2 .33 Adj. R7r .25
Coefficient Stand Err Signif Coefficient Stand Err SIgnIf
CONSTANT -3.7446 1.4882 0.013 4.794328 2.775335 0.086
FARM SIZE 0.0646 0.0162 "'0.000 0.060682 0.0414451 0.145
SECONDARY FOREST 0.0245 0.0304 0.421 01981938 0.108374 '0.069
PRIMARY FOREST 00332 0.0258 0.201 0.0216657 0.0496884 0.663
ALLUVIAL SOILS 04832 0.7259 0.507 -1.971985 1.371748 '0.152
SANDY SOILS 1.1615 0.5916 “0.052 3.379892 1.378916 “0.015
YEARS IN LOT 0.0421 0.0288 '0.147 0.2015417 0.0747249 “'0008
EDUCATION 1.4630 0.5749 “'001 2 2.365908 1.125242 “0.037
FAMILY LABOR 0.2880 0.2416 0.236 0.313289 0.495412 0.528
OFF-FARM INCOME 0.0000 0.0001 0.952 0.0000646 0.0001925 0.738
DIST SOC'L SERVICES 0.0811 0.0522 0123 0.0927358 0.0805555 0.251
CREDIT -1.0076 0.5739 ‘0.082 0.8415701 1.28104 0.512
MARKET DISTANCE 0.0216 0.0112 '0.057 0.003346 0.0214731 0.876
TENURE SECURITY 00323 0.9557 0.973 0.4651375 2.871959 0.871
IMPROVED PASTURES (OLS) IMPROVED PASTURES (SURE)
R2 .80 Adj. R2 .78
Coefficient Stand Err Signif Coefficient Stand Err Signif
CONSTANT 1.8618 3.1693 0.558 4.1291 5.3539 0.441
FARM SIZE 0.7880 0.03607 “'0000 0.5658 0.0955 “'0.000
SECONDARY FOREST 08445 0.0679 “0.000 05155 0.1777 “'0.004
PRIMARY FOREST 08022 0.0557 "'0000 06292 0.1143 "'0.000
ALLUVIAL SOILS - - - — - -
SANDY SOILS - - - - - -
YEARS IN LOT 00674 0.0591 0.256 00295 0.1437 0.837
EDUCATION -1.6209 1.2645 0.202 -1.1193 2.3544 0.635
FAMILY LABOR 06221 0.5470 0.258 -1.0336 1.0247 0.314
OFF-FARM iNCOME 0.0000 0.0002 0.910 0.0013 0.0004 “'0.002
DIST SOC'L SERVICES 02780 0.1155 “0.018 03796 0.1739 "0.030
CREDIT -26357 1.2741 “0.041 -4.0337 2.6629 '0.131
MARKET DISTANCE 00441 0.0223 ”0.050 00044 0.0439 0.921
TENURE SECURITY 3.1313 2.2075 0.159 03548 5.9530 '0.118
NATURAL PASTURES (OLS) NATURAL PASTURES (SURE)
R2 .19 Adj. R2 .12
Coefficlent Stand Err Signif Coefficient Stand Err Signif
CONSTANT -1.2139 1.0211 0.237 05747 0.4257 0.179
FARM SIZE 0.0494 0.0143 “'0.001 0.0219 0.0070 “‘0.002
SECONDARY FOREST 00518 0.0255 “0.045 00352 0.0141 “"0013
PRIMARY FOREST 00277 0.0214 0.198 00216 0.0087 “"0013
ALLUVIAL SOILS - - - - - -
SANDY SOILS - - - - - -
YEARS IN LOT 0.0483 0.0221 “0.031 0.0148 0.0116 0.202
EDUCATION 0.5695 0.4780 0.236 0.1872 0.1901 0.326
FAMILY LABOR 0.0639 0.1831 0.728 0.0598 0.0822 0.468
OFF-FARM INCOME 0.0001 0.0001 0.362 0.0000 0.0000 0.317
DIST SOC'L SERVICES 00429 0.0436 0.326 00015 0.0142 0.919
CREDIT 01190 0.4811 0.805 00514 0.2189 0.814
MARKET DISTANCE 0.0030 0.0086 0.729 0.0055 0.0035 '0.119
TENURE SECURITY 01812 0.4750 0.703 01025 05150 0.607

 

 

 

 

 

 

 

”‘Slgnificant at the .01 level

"Significant at the .05 level
Coefficients are unstandardized

’Significant at the .15 level

157

 

 

APPENDIX F

Results of the Hausman Specification Test and Total Deforestation Model Using
Ordinary Least Squares (OLS) and Two-Stage Least Squares (ZSLS)

158

 

Hausman Specification Test Results

 

HAUSMAN SPECIFICATION TEST RESULTS

 

(Significance of Reduced Form Error Term in Initial Regression)

 

 

 

 

 

 

 

 

 

Total Defor Primary Defor Second Defor
ANNUAL 0.237 0.346 0.112
PERENNIALS 0.166 0.305 0.209
LIVESTOCK 0.225 0.558 0.326
IMPROVED FALLOWS 0.631 0.994 0.238
IMPROVED PASTURES 0.078 0.756 0.646
NATURAL PASTURE 0.806 0.167 0.692
CAPITAL INPUTS 0.562 0.945 0.370
HIRED LABOR 0.987 0.576 0.503

 

 

 

 

 

159

 

3?... 2. ”a a .fiofiém-

.32 8. as .5 28.-«Ea:

woufiuavafimg Ea 3:22.080
:32 8. us a ascfiat.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

000.0. 0500.0 0N _.0.0 #000. 0000.0 0? ..0.0 h0mm0n. -_<._.O.P_
«00.0 000N.0 000—20 N050 0NON.0 N000.0 0..—.00 -_<_>D._-_<
000.0. 0900.0 3 0.0 09.0. 0000.0 N000 E13000 mmazmb
0N0.0: Nmmwd w0u0.0 h 5.0: v0v~.0 F ..00.0 zO_._.<ODDm
0 «0.0 000 _. .0 N00 P .0 000.0. 5000.0 500 w .0 1005 >.=E<n_
000.02.. 0 50.0 v0 V0.0 000.0:- 0 .000 N0 F00 moms ONE:
0000:. «000.0 v000.0- N000: N000.0 3000.0- 0.252. -_<.:m<o
v0 P .0. 0000.0 V0000- N00.0. #0000 {-00.0- m¢3h0<m 4§Dh<z
00¢.0 N000.0 00N0.0 00N.0 00 _.0.0 0? 5.0- 0mmak0<a om>0¢l§
V050 0? 5.0 0000.0- 00v.0 0000.0 vomod 0304-2“. om>Omn=§
000.0 0000.0 N000.0 00N.0 0000.0 0000.0 x00h0w>3
V0 _. .0. #0000 000 ...0- 000.0. 0 ..00.0 000 _. .0- 04<_zzm¢ma
000.0... 053.0 500.0 000.0:- 0030 n ..000 4<Dzz<
N0 _. .0 000v.0 00N0.0- 00 _. .0 0V0¢0 V0000- hz<h0200
3:90 tw 0:80 030 2:20 :0 0:80 0600 m..m<.m<>
0v. um 460 00. um 00. N”. ...60 #0. N1
0403 .0400
zO_._.<._.0mmOu_wo -_<._.O._. zo_._.<.r0me-._mn -_<._.O._.

 

 

 

 

 

8.5.3.— 3335— ..8 3:5 cos—582.2.— 53 A0405 «9:300 3.3-— «0&0 95,—.
E... 649 8.5.5 «So-— bus—:0 9:5 .252 3:809: 5:88.89: .58.

160

I'IIII’IIIIEEIIIZII