• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Table of Contents
 List of Tables
 List of Illustrations
 Foreword
 Acknowledgement
 Summary
 Introduction
 General characteristics of study...
 Time allocation patterns and implications...
 Agricultural production charac...
 Household food consumption and...
 Conclusions and policy implica...
 Bibliography
 Back Cover
 Reprint permission notice














Group Title: Research report - International Food Policy Research Institute ; no. 69
Title: Consequences of deforestation for women's time allocation, agricultural production, and nutrition in hill areas of Nepal
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00085350/00001
 Material Information
Title: Consequences of deforestation for women's time allocation, agricultural production, and nutrition in hill areas of Nepal
Series Title: Research report
Physical Description: 72 p. : ill. : ; 25 cm.
Language: English
Creator: Kumar, Shubh K
Hotchkiss, David
Publisher: International Food Policy Research Institute
Place of Publication: Washington D.C
Publication Date: 1988
 Subjects
Subject: Women agricultural laborers -- Nepal   ( lcsh )
Agricultural productivity -- Nepal   ( lcsh )
Deforestation -- Economic aspects -- Nepal   ( lcsh )
Women agricultural laborers -- Time management -- Nepal   ( lcsh )
Women fuelwood gatherers -- Time management -- Nepal   ( lcsh )
Food supply -- Nepal   ( lcsh )
Travailleuses agricoles -- Népal   ( rvm )
Agriculture -- Productivité -- Népal   ( rvm )
Déboisement -- Aspect économique -- Népal   ( rvm )
Travailleuses agricoles -- Budgets temps -- Népal   ( rvm )
Cueilleuses de bois de chauffage -- Budgets temps -- Népal   ( rvm )
Aliments -- Approvisionnement -- Népal   ( rvm )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Nepal
 Notes
Bibliography: Bibliography: p. 70-72.
Statement of Responsibility: by Shubh K. Kumar, David Hotchkiss.
General Note: "October 1988."
Funding: Research report (International Food Policy Research Institute) ;
 Record Information
Bibliographic ID: UF00085350
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 18628702
lccn - 88028442
isbn - 0896290719

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Title Page
        Page 1
        Page 2
    Table of Contents
        Page 3
    List of Tables
        Page 4
    List of Illustrations
        Page 5
        Page 6
    Foreword
        Page 7
    Acknowledgement
        Page 8
    Summary
        Page 9
        Page 10
    Introduction
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
    General characteristics of study area and households
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
    Time allocation patterns and implications of deforestation
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
    Agricultural production characteristics
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
    Household food consumption and nutrition
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
    Conclusions and policy implications
        Page 66
        Page 67
        Page 68
        Page 69
    Bibliography
        Page 70
        Page 71
        Page 72
    Back Cover
        Page 73
        Page 74
    Reprint permission notice
        Page 75
Full Text
RESEARCH REPORT


s-q, oo0 m Y





CONSEQUENCES OF
DEFORESTATION FOR
WOMEN'S TIME ALLOCATION,
AGRICULTURAL PRODUCTION,
AND NUTRITION IN HILL AREAS
OF NEPAL

Shubh K. Kumar
David Hotchkiss


Octbe 198


























S!- International Food Policy Research
Institute .:- ..:' : '.
ar. i.



: _. r :- .
j,.'-r-' r ':'J F r. 'r *r,-, ,'. 4 ".:



I r '
,7 .r n






J- r .. -


Board cof Trustees


_ I~ II


I I-I -- -















CONSEQUENCES OF
DEFORESTATION FOR
WOMEN'S TIME ALLOCATION,
AGRICULTURAL PRODUCTION,
AND NUTRITION IN HILL AREAS
OF NEPAL

Shubh K. Kumar
David Hotchkiss















Research Report 69
International Food Policy Research Institute
October 1988




















Copyright 1988 International Food Policy
Research Institute.
All rights reserved. Sections of this report may
be reproduced without the express permission
of but with acknowledgment to the International
Food Policy Research Institute.


Library of Congress Cataloging
in Publication Data

Kumar, Shubh K.
Consequences of deforestation for women's
time allocation, agricultural production, and nu-
trition in hill areas of Nepal.

(Research report ; 69)
Bibliography: p. 70
1. Women agricultural laborers-Nepal.
2. Agricultural productivity-Nepal. 3. Defores-
tation-Economic aspects-Nepal. 4. Women
agricultural laborers-Nepal-Time manage-
ment. 5. Women fuelwood gatherers-Nepal-
Time management. 6. Food supply-Nepal.
I. Hotchkiss, David. II. Title. III. Series: Special
report (International Food Policy Research Insti-
tute) ; 69.

HD6247.A4N354 1988 331.4'2572 88-28442
ISBN 0-89629-071-9














CONTENTS



Foreword
1. Summary 9
2. Introduction 11
3. General Characteristics of Study
Area and Households 16
4. Time Allocation Patterns and Im-
plications of Deforestation 24
5. Agricultural Production Charac-
teristics 38
6. Household Food Consumption
and Nutrition 46
7. Conclusions and Policy Implica-
tions 66
Bibliography 70














TABLES


1. Average size of operated land-
holdings of sample farm house-
holds and Nepal, by region,
1982/83
2. Cropping pattern in the study area,
1982/83
3. Yields of principal crops in Nepal,
1975/76-1982/83, and in the
study area, 1982/83
4. Livestock population per house-
hold in study area, 1982/83, and
in selected districts of Nepal, 1977
5. Seasonal pattern of time allocation
6. Comparison of data on patterns of
time allocation from two studies
7. Factors explaining fuel collection
and use, 1982/83
8. Pattern of time allocation and
deforestation, 1982/83
9. Effects of deforestation on input
of household farm labor in hours
per day
10. Effects of deforestation on input
of household farm labor in hours
per person per day
11. Effects of deforestation on input
of household farm labor per day
in hours per hectare
12. Basic crop production data,
1982/83
13. Production functions by crop,
1982/83
14. Marginal products in kilograms
of primary factors for total sam-
ple, 1982/83


15. Factor shares of land, labor, and
livestock in crop production,
total sample, 1982/83
19
16. Characteristics of crop produc-
tion on large and small farms, by
20 cropped area, 1982/83
17. Crop yields by degree of defores-
tation, for lowland and highland
22 sites combined, 1982/83
18. Effects of fuelwood use on cook-
ing time
22
19. Effects of cooking time on cereal
25 composition of the diet
20. Factors influencing dietary calo-
26 ries
21. Anthropometric measurements
32 for preschool children in 1975,
Western Region, and in 1982/
33 83 in study area
22. Anthropometric measurements
for children 6-18 years old,
36 1982/83, study area only
23. Determinants of preschool chil-
dren's nutritional status (probit
37 model)
24. t-values for differences in anthro-
pometric measures for alterna-
37 tive nutritional indexes
25. Types of household constraints
40 possible
26. Agricultural production by type
42 of constraint, major crops, 1982/
83
27. Household characteristics by
43 type of constraint, 1982/83














ILLUSTRATIONS



1. Study area and sites and ecolog-
ical zones of Nepal 17
2. Cropping calendar in the study
area 21
3. Men's time allocation by annual
income per capital 27
4. Women's time allocation by an-
nual income per capital 28
5. Per capital fuelwood consump-
tion, collected by family and non-
family members 30
6. Per capital kerosene consump-
tion 30
7. Fuelwood consumption using
collection time per load as an
indicator of deforestation 31
8. Value of major food items in the
diet by annual income per capital 47
9. Proportion of income spent on
food by annual income per capital 48
10. Nonfood expenditures by annual
income per capital 49
11. Estimated calories available by
annual income per capital 50
12. Sources of income by annual in-
come per capital 51
13. Links between deforestation and
dietary intake 52

















FOREWORD



In the face of growing concerns about the environment, policymakers in developing
countries find themselves increasingly pressured to choose between environmental deterio-
ration in the long run and the growing demands of poor populations in the short run.
Some environmentalists point to new technology-irrigation, fertilizer, and pesticides-
as the basis of ecological decay in rural areas. A number of studies have shown, instead,
that expanding farm yields in less fragile areas through modern technology offers a
viable alternative to stripping the land to expand crop area in marginal soils.
In the hill areas of Nepal, as in many developing countries, women's work is the
key not only to the functioning of the household but also as a necessary supply of field
labor. In Consequences of Deforestation for Women's Time Allocation, Agricultural
Production, and Nutrition in Hill Areas of Nepal, Research Report 69, Shubh K. Kumar
and David Hotchkiss show that the allocation of women's time, as affected by deforesta-
tion, has far-reaching effects on farm output, income, and nutrition. In countries such
as Nepal, where adoption of modern agricultural technology is so low, it seems that
the environment, agriculture, and the quality of life all suffer for this state of affairs.

John W. Mellor

Washington, D.C.
October 1988













ACKNOWLEDGMENTS


The work has benefited enormously from the suggestions and reviews provided by
John Mellor, Dayanatha Jha, Peter Oram, Chandra Ranade, Deepak Bajracharya, Jeffrey
Leonard, Gabriel Campbell, Yair Mundlak, Steve Vosti, and other colleagues at IFPRI.
Ram Yadav, former director of APROSC, provided crucial support for integrating the
nutrition component in the area-level planning exercise that led to the research on
which this study is based. Madhav Gautam, Ramesh Sharma, and other APROSC staff
provided important guidance in carrying out the field work, and FAO staff and consul-
tants, Francoise Petry, Elizabeth Campbell, and Jean-Roger Mercier, provided valuable
input in the design and analysis of data. Most important, however, was the generosity
and trust of the households we visited.

Shubh K. Kumar
David Hotchkiss










1


SUMMARY

The hill areas of Nepal are a prime example of an area in which low-productivity
agriculture is surrounded by rapid environmental degradation as the result of defores-
tation, and the interaction of the two is promoting further deterioration in both. The
following characteristics prevail in the area: low agricultural productivity; high out-
migration; a high reliance on labor input in production, especially given the limited
use of agricultural technology; and pressure to expand cultivated land at the cost of
forest land as population grows. The literature suggests that even in densely populated
hill areas, deforestation does not occur as a result of fuelwood consumption by the
local population. Instead, it is the low agricultural productivity and the inability of the
existing land to sustain the growing population that puts pressure on forestland. Some
estimates suggest that in just over 10 years, from the late 1960s to the early 1980s,
up to half of the forests in some hill regions have been cut down, with the area under
forest reduced from nearly 60 percent to 30 percent of the total area.
In this report, the cost in time spent collecting fuel is used as a measure of the
consequences of deforestation; its effects on time allocation, agricultural output, food
consumption, and nutrition are examined. In particular, the allocation of women's time
is influenced because women are engaged not only in the collection of fuelwood and
other essential forest products affected by deforestation-such as leaf fodder and grass
for livestock feed-but also in agricultural production.
According to the study's hypothesis, deforestation reduces agricultural output from
existing cultivated land by increasing time spent in collecting essential forest products,
which shifts time away from agriculture. As a result, household income from agriculture
is reduced. Unless alternative sources of income increase, food consumption and even-
tually the nutritional status of the population will be adversely affected. Because livestock
production is also an important part of household enterprise in these areas, the destruc-
tion of forests also influences this sector. A reduction in the availability of fodder used
for stall feeding increases the pressure for grazing, which increases soil erosion on
lands that are currently not under cultivation. Also, children who are involved in
collection and livestock grazing activities may experience adverse effects on health and
education, which would ultimately influence the region's prospects for raising the
productivity of labor.
The study is based on results from a year-long survey of 120 households in three hill
districts of the Western Development Region in Nepal, conducted in 1982/83 jointly with
the Agricultural Projects Services Centre of Nepal and the Food and Agriculture Organiza-
tion of the United Nations. Results indicate that when deforestation-represented by
the time required to collect a standard load of firewood-increased by 1.0 percent,
there was a reduction in fuelwood consumption of 0.3 percent and an increase in the
total time required for its collection of 0.6 percent. Assuming a similar response for
other essential forest products, the collection time for fuelwood, leaf fodder, and grass
alone required an additional 1.13 hours per day by women in the high deforestation
sites in the sample. This represents a 45 percent increase, assuming that all workers
engaged in the activity increase their work in proportion to their earlier input. The
effects of this on the amount of labor available for agriculture indicate that women's










farm labor would decrease by 1.4 hours per person per day, or nearly 50 percent. This
decrease is not compensated for by an increase in wage labor or by men's labor, which
may also decrease.
Analysis of the production functions for cropping activities indicates that women
spend the most time on the dry-season crops-wheat, maize, and mustard. But the
time spent for collection of fuelwood is also greater during the dry season because that
is when people collect extra amounts of wood and store it for later use. As may be
expected, the marginal product of upland crops-maize and ragi-is about half that
for the lowland crops-paddy, wheat, and early paddy.
Caloric availability and the ratio of kilocalories from rice compared with other
cereals are positively influenced by the component of household income that comes
from agriculture and time spent on food preparation and cooking. At the same time,
food preparation time is positively associated with the amount of fuelwood used and
negatively with the amount of total time spent in fuel collection. This suggests that in
addition to the effects of deforestation on agricultural production and incomes, secondary
or associated effects could be related to fuel consumption and time spent on food
preparation. The main determinants of preschool child nutrition are degree of defores-
tation, household income, household size, and work loads of women and older children.
In addition, the data indicate that the Tibeto-Burman ethnic groups have better child
nutrition than the Newar-Brahmin groups, when the influences of other household
characteristics are controlled for.
The results suggest that it is not enough to rely on out-migration or reforestation
efforts alone to improve the economy and ecology of the region. (One out of every two
households already has a permanent migrant worker.) Strategies for raising agricultural
productivity need to be considered. In the long run, agricultural products that offer
high value to weight, such as horticultural products, may be a feasible proposition for
the region. This requires investments in research and extension, as well as an efficient
marketing system suited to the primarily smallholder agriculture. At present, production
is largely subsistence-oriented and not very productive; therefore, the degree of rural
market development is limited. In the short run, therefore, it appears necessary to
increase productivity of the traditional crops through the use of improved technologies
that already exist and can be promoted, thus shifting away from subsistence production
and promoting the growth of rural market infrastructure.
This could provide the base for a gradual shift into more specialized horticultural
and livestock products. In order to achieve this, efforts will be needed, first, to promote
improved input use, and, second, to alleviate human labor bottlenecks for both small
and larger farms. Labor-saving technologies for nonfarm activities, such as food process-
ing and water supply, should also be included. Agroforestry programs that reduce
collection time for essential forest products would also complement such an agricultural
intensification effort in the hills of Nepal.










2


INTRODUCTION

Agricultural strategies for areas of low or marginal potential have been receiving
increased attention recently.' This concern has been driven by several factors. Sustain-
ability of agricultural growth requires a constantly expanding frontier beyond the areas
currently experiencing rapid growth. Also, growing regional inequity between rural
areas where rapid agricultural growth is occurring and those left behind is sometimes
pointed to as an undesirable feature of improved agricultural technologies.2 Related to
both is concern for deterioration of the environment, which usually stems from the
effects of excessive growth in input use in intensively cropped areas, but increasingly
also from rampant environmental degradation, including deforestation, in areas where
growth is stagnant.3 All of these concerns point to the need for a closer examination
of production characteristics and constraints to growth in the so-called low potential
areas, so as to identify the best strategy for incorporating them into the overall develop-
ment picture. In Nepal the major focus for agricultural development efforts has been
in the southern plains, the tarai. In the hill areas the focus has been on reforestation
and soil conservation efforts, and agricultural growth incentives have been virtually
nonexistent.
In analyzing the range of situations that arise in the association between labor input
and agricultural output in developing countries, two main groups or types have been
identified. According to John Mellor,4 they can be classified as, first, the labor-surplus
case, in which highly productive nonhuman agricultural resources initially provide food
for more people than are needed to work on the land, eventually leading to a large
pool of underemployed and surplus labor. In such areas, there is a good potential for
widespread adoption of labor-intensive agricultural technologies and for growth in rural
off-farm employment. Second is the "hardworking-peasant" case, in which output is
directly a product of labor input.
The first of these two situations represents most of the high-growth agricultural
areas of Asia and Latin America, while the latter is generally prevalent in African
agriculture. Semi-arid and hill areas in Asia, however, may also be cast in the hardwork-
ing-peasant mold.5 Since the marginal product of labor is higher in the latter situation,
labor constraints are more likely to develop, but it is difficult to support additional
labor, when aggregate productivity is low. The significance of human labor in agricultural

Consultative Group on International Agricultural Research, Technical Advisory Committee, Sustainable
Agricultural Production: Implications for International Agricultural Research (Rome: Food and Agriculture
Organization of the United Nations, TAC Secretariat, 1987).
2 Irma Adelman, "A Poverty-Focused Approach to Development Policy," in Development Strategies Recon-
sidered, ed. John P. Lewis and Valeriana Kallab (Washington, D.C.: Overseas Development Council, 1986),
pp. 49-65.
3 World Commission on Environment and Development, Our Common Future (Oxford: Oxford University
Press, 1987).
4 John W. Mellor, "Determinants of Rural Poverty: The Dynamics of Production, Technology, and Price,"
in Agricultural Change and Rural Poverty: Variations on a Theme by Dharm Narain, ed. John W. Mellor
and Gunvant M. Desai (New Delhi: Oxford University Press, and Baltimore: Johns Hopkins University
Press for the International Food Policy Research Institute, 1986), pp. 21-40.
5 Ibid.









production remains high. Moreover, it is precisely from such areas that there is high
out-migration of labor because of the relatively low productivity of the land. As the
population grows, pressure to expand cultivated area is also high. This promotes defores-
tation and progressively lowers agricultural productivity.
What are the development policy implications for such regions? One approach that
has been favored by many is that of benign neglect or of encouraging migration from
these areas. Another focuses on fostering high-value cash crop production, such as
horticulture and tree crops. This approach can be viable if institutional and infrastruc-
tural developments support the production, and especially the marketing, of these
products. A parallel approach that is receiving increasing attention in this context is
agroforestry development. Forests contribute a wide range of products essential to
household production and survival, but they also contribute to the long-term environ-
mental and agricultural land quality of the region. Growing trees as a crop can, over
the course of years, contribute to household agricultural production and income gener-
ation. In the short run, however, tree crops may compete for land on which food is
grown. It is possible that this competition may be reduced by raising agricultural
productivity, thereby reducing the dual pressure on existing forestland for fuelwood
and for land for cultivation.
Improving agricultural performance requires choosing an appropriate mix of inputs
and technologies. Hence it is expeditious to examine the characteristics of producer
households and to identify their relevant constraints. This provides the basis for appro-
priate policies to supplement market forces in providing the impetus for change.


Significance of Human Labor in Traditional Agriculture
In general, it may be useful to consider the issues of poverty, low labor productivity,
and the environmental consequences together. Environmental degradation, whether
it is due to excessive livestock grazing or deforestation, is largely in response to strategies
for providing for human energy needs. However, as a consequence of the destruction
of environmental resources, the human cost of using these resources also increases.
Where productivity is low and the reliance on human labor is high to begin with, this
loss of environmental resources is likely to further reduce agricultural productivity.
Traditional agriculture is characterized by a high proportion of human labor input
and low productivity of land and labor. When the total amount of energy that goes into
farming is calculated, it appears that subsistence farming actually requires more energy
resources per hectare, especially per ton of food output, than technologically advanced
forms of agriculture. However, when the concept of "useful energy" is applied-that
is, energy actually delivered or made available for the production process to occur
(through land preparation, irrigation, fertilizer application, and so forth)-then the
picture is reversed.6 Increases in useful energy input via irrigation, application of
manure and fertilizer, improved management practices, or improvement of plant ef-
ficiencies through breeding provide the main mechanisms for achieving higher yields.
While it may seem that these observations really just state the obvious, the useful
energy concept does serve to highlight the importance of human energy input and
labor constraints in subsistence agriculture, as represented by the hardworking peasant
mentioned earlier. An increase in useful energy input in Third World agriculture has

6 Arjun Makhijani, Energy and Agriculture in the Third World: A Report to the Energy Project of the Ford
Foundation (Cambridge, Mass.: Ballinger, 1975).









also been found to increase the intensity of labor use per hectare, even as the share
of human energy input in production declines.7 These empirical observations are fully
consistent with the theoretical expectations for labor input and productivity in high-
and low-productive agriculture, with the theory that labor demand increases with
technological change, and with the nutrition-wage hypothesis.8
Human labor input is required both for the production of crops and the household's
provision of its nonfarm energy supply that comes primarily from forests, providing
essential fuel and animal feed products. As long as both crop and noncrop energy
supplies for human consumption remain heavily dependent on human labor, labor
productivity remains low, yet the marginal product of labor remains relatively high.9
Land under crop production competes with forestland. This competition is further
accentuated by increasing population pressure. In an eastern hill area of Nepal, Baj-
racharya documented that the forest fuel supply was adequate to meet people's needs,
but arable area could not meet their food needs.'0 He concludes that it is essential to
increase productivity of land if farmland encroachment on forests is to be reduced and
a sustainable agricultural economy attained. If this does not occur, forest resources will
become increasingly scarce, further accentuating the requirement for human labor and
creating a further downward pressure on the productivity of land.
Quality of human labor input may also be adversely affected if low land productivity,
combined with heavy demand for human energy expenditure for both crop production
and provision of noncrop energy, impairs the nutritional status of the workers. Both
chronic and acute food scarcities limit human work capacity, and again it is necessary
to improve agricultural and labor productivity. As noted earlier, these improvements
are usually accompanied with some increase in demand for labor. For smallholders of
the hardworking-peasant type, this involves an explicit cost, and unless credit is avail-
able, it may not be possible to shift labor away from other activities, such as fuel and
fodder procurement and essential household maintenance activities. Further, even if
expected yields are vastly improved, the inherent riskiness of input adoption in the
initial stages could discourage early adoption by smallholders. In order to overcome
this obstacle, the public sector initially may need to play a large role in reducing the
risks involved in changing resource allocation of farm households. Consequently, invest-
ments that reduce a range of household constraints-including labor requirements for
noncrop activities, cash, and animal labor constraints-may need to be incorporated.
Clearly, the availability of improved agricultural technologies would provide the justifi-
cation for making these public investments.
In areas where growth in agricultural productivity by smallholders has occurred, it
has been found that this is followed by expansion of markets and private investments
in both agricultural and supportive nonagricultural sector activities in the area." This


7 Ibid.
8 Uma Lele and John W. Mellor, "Technological Change, Distributive Bias, and Labor Transfer in a Two-Sector
Economy," Oxford Economic Papers 33 (November 1981): 426-441; and John W. Mellor and Robert D.
Stevens, "The Average and Marginal Product of Farm Labor in Underdeveloped Countries," Journal of Farm
Economics 28 (August 1956): 780-91.
9 A comparison of labor production functions for low-productive hilly, dryland areas and high-productive
areas was made by Mellor and Stevens in "Average and Marginal Product of Farm Labor."
10 Deepak Bajracharya, "Fuel, Food, or Forest? Dilemmas in a Nepali Village," World Development 11 (No.
12, 1983): 1057-1074.
" Sudhir Wanmali, Service Provision and Rural Development: A Study of Miryalguda Taluka, Research
Report 37 (Washington, D.C.: International Food Policy Research Institute, February 1983).









not only helps to ease the pressure on land to provide income and employment growth
but also produces a base from which further transformation in the agricultural sector
can be more readily obtained. This could include a transition toward horticultural and
other high-valued but perishable products to which the hill ecology is suited. This
serves to provide an additional justification for judicious public investments to overcome
the initial bottlenecks and thus raise agricultural productivity.


Objectives and Rationale
The objectives of this report are to examine the effects of labor constraints, partic-
ularly of women's labor, on output from small hill-area farms and the direct adverse
consequences of deforestation on such farms. More specifically, the report examines
women's time allocation in a poor agricultural area for its effects on agricultural produc-
tion, food consumption, and the nutritional status of children. It looks at the possible
consequences of deforestation for agricultural output, as a result of competing demands
for women's labor. It uses the labor costs of fuelwood collection as a measure.
It has been argued that deforestation in hill areas can have adverse consequences
for plains agriculture. Flooding and silting, traditionally the source of agricultural poten-
tial for plains agriculture, also cause the rapid blockage of irrigation canals. These
consequences are quite apart from the widespread ecological and environmental con-
sequences of deforestation. These observations have led to concerted efforts to slow
down the rate of forest depletion. Though agroforestry-the incorporation of tree
cultivation in household farming-is often promoted as a means of slowing deforestation,
the explicit linking of this process with existing agricultural activities in crop and
livestock production in these areas is seldom made.
If deforestation leads to reduced agricultural productivity of land, understanding
the mechanisms involved would assist in efforts to improve agriculture as well as in
the forestry efforts. When yields are too low to sustain the growing population, even
with high out-migration, there is inevitably pressure to expand cropland. This is often
at the cost of forests. Thus while forestland may provide a safety valve for the growing
population in the short run, its depletion only promotes the cycle of declining agricultural
productivity.12 In this study it is argued that productivity of agricultural land would
decline, despite new land coming into production, as a result of additional labor demands
created by deforestation. The increasing distance from and hence time required to
collect essential forest products competes with labor input in agriculture. Therefore,
unless measures to improve agricultural productivity are undertaken simultaneously,
the push for clearing new land due to population growth will continue.
The focus in this report is on the issue of labor constraints, particularly on women's
labor, as an intermediary to the adverse consequences of deforestation. Low productivity
agriculture is known to rely heavily on input of human labor.13 In areas where women
provide most of the labor in the collection of forest products, this labor increases with
deforestation. If they also provide labor in agriculture, then the consequences for
production depend on their ability to take on more work and the degree of substitution
between men's and women's work in agricultural production. These are some of the
questions that will be examined.


12 Bajracharya, "Fuel, Food, or Forest?"
13 Mellor, "Determinants of Rural Poverty."









Since women also have the major responsibility for activities directly related to
food consumption, such as food processing and preparation, deforestation and increasing
work loads could also have a direct effect on this activity, with possible adverse conse-
quences. Similarly, preparing food and feeding young children is time-intensive for
women, and direct adverse effects are also possible there. These two effects would be
likely to occur over and above any of the changes in food consumption and nutrition
engendered by the effects on agricultural productivity.
In the analysis of all forest products used by the local population, only changes in
fuelwood consumption and the associated time costs are used to represent the conse-
quences of deforestation. However, other essential forest products, such as leaf fodder
and grass and water, can be expected to be influenced in a similar way. Leaf fodder
and forest grass are an important part of livestock feeding because they reduce pressure
on the limited grazing lands available. With deforestation, there would be either the
additional work of collection or increased grazing, with direct effects on livestock
production and indirect effects on both agricultural production and food consumption.
It is expected that making the explicit connection between patterns of labor allocation,
deforestation, and agricultural productivity will help identify the extent to which labor
constraints are a factor and hence will suggest alternative approaches for increasing
productivity.









3


GENERAL CHARACTERISTICS OF STUDY AREA
AND HOUSEHOLDS

This report presents results obtained from a year-long household survey in a hill
area of the Western Development Region in Nepal. The survey consisted of four quarterly
rounds of data collection from 120 households using a pretested, structured question-
naire. Information was obtained on agricultural production, collection of forest products,
time use patterns for selected activities, expenditures, off-farm employment, migration,
and remittances. Food consumption was estimated using the disappearance method
for crops, livestock, and horticultural products, together with expenditure data. Anthro-
pometric measurements on all household members were taken at the time of the
survey. In order to validate the time allocation measures from the survey, a more
detailed time sampling technique was used on a subsample of 12 households.
The survey was conducted by the Agricultural Projects Services Centre (APROSC)
of Nepal and the Food and Agriculture Organization of the United Nations (FAO) in
1982/83 with assistance from the International Food Policy Research Institute. Its
initial purpose was to help area-level agricultural planners incorporate labor, energy,
and nutrition links into the agricultural production process. That exercise was completed
in December 1985. This report builds substantially on the work done for that phase.14
Nepal is divided into three distinct ecological zones, the tarai, a band of lowland
plains to the south, high mountains to the north, and the hill areas in the middle
(Figure 1). The six wards from which the sample households were chosen were selected
from three panchayats in hill districts of the Western Development Region. The objective
of the sampling was to obtain as wide a representation as possible of the major social,
economic, and ecological characteristics of the area. Thus, panchayats were selected
first and then wards to cover variations in altitude, access to roads and markets, ethnic
groups, and degree of deforestation.
The sites selected were

District Panchayat Wards
Gorkha Chhoprak 1 and 7
Tanahun Manapang 5 and 8
Syangja Bagkhor 2 and 8

The hill areas are, naturally, composed of hills (highlands) and valleys (lowlands).
The altitudes at which farms are located range from a low of about 500 meters in
Manapang to a height of about 1,500 meters in Chhoprak. Road transport and markets
are easily accessible in Bagkhor, but they are a day's hike from Chhoprak. While it
was not possible to obtain every possible type of situation, the distribution was a fair
representation of road and market access at different altitudes. Similarly, the degree
of deforestation is most severe in Chhoprak, and less severe in Manapang. In an effort

14 Shubh K. Kumar and David Hotchkiss, "Energy and Nutrition Links to Agriculture in a Hill Region of
Nepal," International Food Policy Research Institute, Washington, D.C., 1985 (mimeographed).








Figure 1-Study area and sites and ecological zones of Nepal


* Manapang


Tanahun


Study Districts


Nepal
Wve~irn Developmenl
Z- Region


iounwitn
Hills
Tara.


Gorkha


* Study sites









to obtain both lowland and highland agricultural characteristics in each of the selected
panchayats, a somewhat higher inclusion of lowland farmers relative to what generally
prevails in these districts may have resulted. Associated with this was a slightly higher
proportion of Brahmin castes in the study source, as they tend to predominate in
lowland farming. In the hill areas, lowland, called khet, may occur either in valley
bottoms or on mountainsides. It is distinguished from upland, called pakho, by the fact
that it can be irrigated at least during some part of the year. It is expected that this
sampling bias may improve the analysis of production constraints for different crops,
but it will not influence the results of the deforestation links to agriculture and nutrition.
Within these selected wards, a random sample of 20 households (approximately 8
percent of households) was taken from each for a total of 120 households for the
survey. A ward census survey was undertaken to assess the characteristics of the
sample. A comparison of the basic assets of the households sampled with those in the
census survey indicates a close comparability between the two groups.'5


Size and Composition of Households and Landholdings
The average household size was between six and seven members (6.6), with one
permanent migrant worker for every two households.16 The average number of adult
women per household was just slightly higher than the number of men. Children below
six years of age constituted 20 percent of the total population, and those below 15
years of age were 45 percent of the population. This is consistent with the expected
demographic pattern in developing countries.
The average landholding size was 1.45 hectares, of which 0.62 hectare was lowland
or khet, on which water control is possible, and 0.83 hectare was upland or pakho,
on which only rainfed agriculture is possible. Comparison with national surveys suggests
a somewhat higher average farm size in this study sample than might be expected in
the region (Table 1). This is mainly because more lowland was included in the sample,
and lowland holdings are generally larger. This table suggests that the effective popu-
lation density in terms of pressure on arable land may be much greater in the hill
regions, even though the population density is lower than in the tarai in terms of
population to total area.


Cropping Intensity and Patterns
The average cropping intensity of land is about 150 percent, indicating that half
the land is cropped twice. Comparison with estimates made for Nepal as a whole until
the late 1970s shows that cropping intensity in the study area is much higher than
the national average. Moreover, the cropping intensity is nearly as high for the pakho
at 142 percent as for the khet at 163 percent. This indicates an extremely high degree
of cultivation intensity, which is partly a consequence of the effective population
pressure on the arable land. A favorable rainfall distribution in the hills relative to the
tarai is also likely to be a factor. The high cropping intensity for upland plots is
particularly striking and could indicate that the potential for raising agricultural produc-


15 Food and Agriculture Organization of the United Nations, Pilot Study on Energy Use and Nutritional
Status at Farm Level in the Hills of Nepal, Report No. 1 (Rome: FAO, 1984).
16 A permanent migrant worker is one who is away for more than six months at a time.









Table I-Average size of operated landholdings of sample farm households
and Nepal, by region, 1982/83
Operated Holdings
Region Lowland Upland Total
(hectares)
Study area
Total sample mean 0.62 0.83 1.45
Nepal
Hill
Eastern 0.10 0.96 1.06
Central 0.34 0.82 1.16
Western 0.22 0.45 0.67
Midwestern 0.13 0.37 0.50
Far Western 0.22 0.16 0.38
Hill average 0.20 0.55 0.75
Tarai
Eastern 1.28 0.74 2.02
Central 1.18 1.03 2.21
Western 0.36 1.64 2.00
Midwestern 2.42 0.58 3.00
FarWestern 1.84 2.11 3.95
Tarai average 1.41 1.22 2.63

Source: Nepal, Agricultural Projects Services Center, Foodgrain Marketing and Price Policy Study (Kathmandu:
APROSC, 1982), p.81; Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization
of the United Nations; and International Food Policy Research Institute, "Nepal Energy and Nutrition
Survey, 1982/83," Western Region, Nepal.





tivity in hill areas is good. This may be even more so in the eastern half of the country
where rainfall is relatively abundant.
While rice is the major lowland crop, both maize and ragi (finger millet) have a
high share of cropped upland (Table 2). Among the lowland crops in the study area,
early paddy production is much more prevalent than might have been expected, and
in the upland plots, ragi area is higher than expected. It can be seen from the cropping
calendar in Figure 2 that these crops do not displace the other major crops, and they
contribute to the high cropping intensity noted earlier. In addition to these crops, small
amounts of soybeans are planted along paddy ridges. Fruits and vegetables are grown
mainly on homesteads and contribute only about 5 percent of household income. No
farmland was reported to be sown with fodder crops.

Crop Yields
Crop yields in the study area are very similar to those for the country as a whole
(Table 3). Although cropped area has increased for the country as a whole and for hill
areas in particular, yields have declined over time. Between 1975/76 and 1982/83,
for example, area under paddy in the hills increased by 30 percent, wheat by 50
percent, and maize by 10 percent. At the same time, yields for paddy declined from
2.6 to 2.0 tons per hectare and maize from 1.8 to 1.4 tons per hectare. Area increases
and yield reductions are not so marked in the tarai. The only exception to this is wheat.
For this crop, suitable high-yielding varieties (HYVs) have become readily available in
the past 10-15 years, which has led to a sharp increase in cropped area for wheat-71.5









Table 2-Cropping pattern in the study area, 1982/83
Bagkhor Chhoprak Manapang
Crop Ward 2 Ward 8 Ward I Ward 7 Ward 5 Ward8 Total
(hectares)
Lowland crops 0.77 0.35 0.42 1.73 1.10 1.54 0.98
Early paddy 0.04 0.01 0.15 0.41 0.05 0.53 0.20
Paddy 0.33 0.28 0.27 1.11 0.82 0.53 0.55
Maize, irrigated 0.14 0.01 0.00 0.00 0.22 0.04 0.07
Wheat 0.25 0.05 0.00 0.20 0.02 0.45 0.16
Upland crops 0.83 0.83 1.69 1.47 1.36 1.07 1.21
Maize, dry 0.42 0.43 0.73 0.74 0.74 0.34 0.57
Ragi (finger millet) 0.41 0.41 0.70 0.19 0.60 0.16 0.41
Mustard 0.00 0.00 0.13 0.16 0.01 0.04 0.06
Blackgram (legume) 0.00 0.00 0.13 0.38 0.00 0.53 0.18

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.



percent compared with 13.7 percent for all cereals between 1970/71 and 1980/81-as
well as to increases in yields of as much as 44 percent during the same period.17


Livestock Ownership
Livestock is an essential component of agricultural production in Nepal. With in-
creasing intensification of crop production, the absolute requirement for animal labor
increases, and the timely availability of livestock labor for field preparation becomes
more important in maintaining or improving crop yields.
There is a wide variation in the distribution of livestock populations in Nepal.
Though figures for the country as a whole are not available, a comparison of the study
area with selected districts for which figures are available is given in Table 4. In Nepal
as a whole, more livestock is owned in the tarai than in the hills, but livestock ownership
in the study area appears to be relatively high for the hills. In addition to inputs for
agriculture, livestock and livestock products contribute about 15 percent of household
income in the study area.
At the same time that livestock contributes essential labor and products for agricul-
ture, their upkeep entails the use of agricultural by-products such as rice and wheat
straw, forest resources for provision of leaf fodder and grass, other feed inputs, and
human labor and time. Animals are stall-fed throughout the year, supplemented with
grazing in the dry months only. Another important source of animal feed is kundo,
which consists predominantly of oilseed cake and straw cooked for extended periods
of time in water. The making of kundo creates additional demand for firewood.





17 Nepal, Agricultural Projects Services Centre, Nepal: Foodgrain Marketing and Price Policy Study
(Kathmandu: APROSC, 1982).









Figure 2-Cropping calendar in the study area


Wheat

Early paddy

Maize

Mustard

Blackgram

Paddy

Ragi

J F M A M J J A S 0 N D


Forest Resources
Though an explicit measurement of forest resources in the study area was not
undertaken, an estimate for the region indicates that about 22 percent of the area is
under forest cover and 50 percent under agriculture.18 Community forests are largely
used for wood and fodder. At the sites where deforestation is most severe, a higher
reliance on private trees occurs. These trees are generally planted around the homestead
or on the boundary of nearby fields. Although the ownership of private trees for fuel
and fodder is considered a necessity by households in such areas, these trees represent
only a small proportion of total fuelwood used even there. The present situation indicates
that a sharp reduction in forestry resources has occurred since a survey by the Nepal
Forest Resource Survey Office published in the early 1970s. In that survey, total area
under forests in the hill regions was 55-60 percent, while agricultural area was about
14 percent in the far west, about 30 percent in the central and western hills, and
about 36 percent in the eastern hills.
There are several mechanisms by which the gradual reduction in forest cover could
have adverse consequences for agricultural productivity. Deforestation contributes to
soil erosion, and marginal lands are increasingly brought under cultivation as land
becomes deforested. As a result, another set of consequences becomes a factor-the
additional demand on family labor to provide essential fuel, fodder, and water for
humans and livestock. Usually there is an implicit assumption that there is a surplus
of family labor on farms, and the higher labor demands can be easily met. This is the
heart of the issue examined in this study. This constraint cannot be measured only in
fuel energy or food energy: the limits imposed on time and on total human energy for
productive activities as well as for essential consumption activities must also be consid-
ered. All have direct implications for human welfare.


'8 Food and Agriculture Organization of the United Nations, PilotStudyon Energy Use andNutritional Status.










Table 3-Yields of principal crops in Nepal, 1975/76-1982/83, and in the
study area, 1982/83

Area/Year Paddy Maize Wheat Barley Ragi
(metric tons/hectare)
Nepal
1975/76 2.07 1.65 1.18 0.93 1.14
1976/77 1.89 1.79 1.04 0.83 1.13
1977/78 1.81 1.66 1.12 0.88 1.07
1978/79 1.85 1.64 1.21 0.85 1.08
1979/80 1.64 1.28 1.20 0.90 0.97
1980/81 1.93 1.63 1.22 0.86 1.00
1981/82 1.98 1.58 1.32 0.86 1.00
1982/83 1.45 1.41 1.36 0.87 0.94
Hill area
1982/83 1.97 1.43 1.27 0.85 0.95
Study area
1982/83 1.80 1.20 1.00 ... 0.96

Sources: Nepal, Ministry of Agriculture, Department of Food and Agricultural Marketing Services, Agricultural
Statistics of Nepal 1985 (Kathmandu: MOA, 1985); and Nepal, Agricultural Projects Service Center;
the Food and Agriculture Organization of the United Nations; and International Food Policy Research
Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.



In this study an indirect measure of forest resources was used. Since the main
consideration was the increase in work load associated with reduced accessibility to
forests, it was decided that a measurement of the time spent in collection would be
most appropriate. The time for collecting a standard load of fuelwood (about 20 kilograms)
was recorded for each household during each quarterly survey. The average time taken


Table 4-Livestock population per household in study area, 1982/83, and
in selected districts of Nepal, 1977
Sheep, Live-
Goats, stock
District Bullocks Cows Buffalo and Pigs Unitsa

Study households 1.36 1.75 2.14 3.03 6.93
Selected districts
Hills
Ilam 1.65 1.85 1.74 2.41 6.59
Kavre 0.65 2.36 0.42 2.14 4.07
Dhading 0.99 2.15 2.31 2.69 7.14
Syangja 1.02 1.08 2.69 1.14 6.36
Sallyan 1.95 1.74 1.45 2.06 6.28
Kathmandu 0.23 1.69 0.14 1.08 2.35
Tarai
Chitwan 3.56 3.17 2.67 2.04 11.14
Morang 5.00 2.27 1.00 1.73 9.46
Dhanusha 1.81 0.40 0.74 0.28 3.38
Rupandehi 2.65 0.94 0.99 0.85 5.25
Kailali 4.04 6.02 4.36 4.26 17.45

Sources: For selected districts, Ministry of Agriculture, Department of Food and Agricultural Marketing Services,
Agricultural Statistics of Nepal (Kathmandu: MOA, 1977); and for study households, Nepal, Agricultural
Projects Service Center, the Food and Agriculture Organization of the United Nations and International
Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
a Buffalo are weighted at 1.5 livestock units, and sheep, goat, and pigs at 0.2.









per unit of fuelwood for each site was then computed and used to indicate the degree
of deforestation encountered by the households at that site.
Only fuelwood collection time was used as the indicator for deforestation. Although
leaf fodder, grass, and water were also collected, they were not included in the indicator
because they are likely to be collected from specific locations. Hence their collection
times per unit would be influenced more in a temporal than a cross-sectional way with
deforestation. The time required for the household to collect a unit of fuelwood is used
here to represent the household's effective distance from the forest area where the
bulk of collection occurs. Growing of private trees is too limited to have a significant
effect, although households are likely to shift to private trees as deforestation becomes
more severe. At the household level, use of time per unit of fuelwood collected is likely
to be a better indicator of the opportunity cost (and one that will increase as deforestation
increases, given the stable patterns of habitation) than an area-specific measure of
forested land.19



































~9 These measures are also extremely difficult to collect and interpret without aerial surveys.









4


TIME ALLOCATION PATTERNS AND
IMPLICATIONS OF DEFORESTATION

How much household time is allocated to production and to production support
activities? Production support, or expanded economic activity,20 includes all the activ-
ities for collection of fuel, leaf fodder, grass, and water for family and livestock, as well
as food processing and cooking. These are essential activities, and a reduction in them
without any increase in the productivity of these tasks would entail a real cost to
households. The performance of many of the production support activities is affected
by deforestation. In the following section, the implications of deforestation for the
household are considered from the viewpoint of demand for time to spend on these
tasks, as well as possible decreases in the quantity of forest products used. The specific
case of fuelwood use is taken as an example to illustrate these changes.

Household Time Allocation
The information on time allocation in this study was obtained by setting up a series
of recall instruments geared to the specific activities being measured. Since four quarterly
survey rounds were conducted, it was possible to obtain some ideas about the seasonal
pattern of time allocation. The main drawbacks were the use of the recall method and
the limitation of the recall period to one week for most of the routine collection and
food-processing activities. The data collected were then extrapolated to represent a
much longer period of three months. In the case of agriculture, recall was by crop
following the harvest. Because the recall period here extended over several months,
the recall unit was days of work for various activities by crop. Finally, for both agriculture
and off-farm employment, an assumption of an eight-hour work day, including travel
time, was assumed.
The time allocation data obtained for men, women, and children in Table 5 show
that the most intense period for agricultural work is during the rainy-season quarter
of July to September. During the dry-season quarter, April to June, less time is spent
on agricultural work, but it is the maximum time period for work on production support
activities. Fuelwood and water collection activities, in particular, peak at that time of
the year. Women's total recorded work time is between 150 and 180 percent of men's
recorded work time. Of this total time spent by women, between two-thirds and three-
fourths is spent on production support activities. Within the production support category,
half to two-thirds of their time is spent on collection tasks alone, or up to half of their
total recorded work load. Men spend time equivalent to 50 to 90 percent of their total
recorded field work time in agriculture on support activities such as construction,
collection, and domestic work. Their maximum time in production support activities,
like that of women, is in the dry-season quarter of April to June.

20 The term "production support" is used in the study, The Status of Women in Nepal. A summary report
of that study is found in Meena Acharya and Lynn Bennett, The Rural Women of Nepal: An Aggregate
Analysis and Summary of Eight Village Studies, vol. 2, part 9 (Kathmandu: Centre for Economic Development
and Administration, Tribhuvan University, 1981).







Table 5-Seasonal pattern of time allocation


April-June July-September October-December January-March
Men Women Childrena Men Women Children Men Women Children Men Women Children


Activity


Agricultural work
Field work 2.2 2.1
Employmentc 0.5 0.2
Subtotal 2.7 2.3
Support activities
Fuelwood collection 0.4 2.0
Water collection 0.2 1.6
Grass collection 1.2 0.9
Leaf fodder collection 0.1 0.3
Grazingd
Food processing 0.2 0.7
Cooking 0.4 2.2
Subtotal 2.5 7.7
Total of all re-
corded activities 5.2 10.0


0.1 4.1 3.4
0.6 0.2
0.1 4.7 3.6

0.2 0.0 0.9
0.3 0.0 0.9
0.3 0.1 2.4
0.0 0.0 0.0
2.5 0.0 0.0
0. .2 0.7
0.4 2.4
3.3 0.7 7.3

3.4 5.4 10.9


(hours/person/day)

0.0 3.8 3.4
0.5 0.0
0.0 4.3 3.4

0.2 0.0 0.8
0.4 0.1 0.9
0.7 0.1 0.4
0.0 0.2 0.4
0.0 0.0 0.0
0.2 0.7
0.4 2.1
1.3 1.0 5.3

1.3 5.3 8.7


0.0 2.3 2.1
1.6 0.1
0.0 3.9 2.2

0.1 0.1 0.9
0.1 0.1 1.2
0.1 0.0 0.2
0.1 0.1 0.7
2.1 0.0 0.0
0.2 0.7
0.3 1.7
2.5 0.8 5.4

2.5 4.7 7.6


Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations; and the International
Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
a Children 6-15 years old are included.
b Fieldwork was recorded by crop, which cuts across individual quarters. Therefore, it is aggregated into two semiannual periods: dry-season crops were assigned
the first and fourth quarters, and wet-season crops were assigned the second and third quarters.
c Includes seasonal migration.
d Since grazing of cattle was mostly performed by children and no data on individual grazing time is available, it is all assigned to that category.










Given the drawbacks in methodology for collection of time allocation information
for this study, results were compared with those from the subsample of 12 households,
using a more intensive direct observation and time-sampling method,21 and with results
from the in-depth study on The Status of Women in Nepa.22 The methodology used
in this study was identical to that used for the subsample.
The results obtained with the structured recall method and with direct observation
are remarkably similar. The comparison with The Status of Women in Nepal study also
shows similar results (Table 6). This table indicates the amount of time spent on
activities not enumerated in the present study but picked up by the more comprehensive
methodology. It amounts to an additional 20-25 percent of work time by men and
children and just over 30 percent more for women. This suggests that the percentage
of underreported time in this study due to omitted activities was similar for men,
women, and children.
Given the difference in methodologies used in the two studies, the similarity of
results is somewhat unexpected. This may be because the respondents in The Status
of Women in Nepal study were asked to recall the amount of time required to perform
a task rather than time allocation per se as in the structured questionnaire used in the


Table 6-Comparison of data on patterns of time allocation from two studies
Rural Women of Nepal Study" Nepal Energy and Nutrition Survey
Activity Men Women Children Men Women Children
(hours/person/day)
Agricultural work
Household farm 2.73 2.74 0.99 3.10 2.75 0.05
Employment 1.24 0.46 0.21 0.80 0.13
Subtotal 3.97 3.20 1.20 3.90 2.88 0.05
Support activities
Fuelwood collection 0.24 0.38 0.20 0.13 1.15 0.13
Water collection 0.07 0.67 0.33 0.10 1.15 0.23
Grass collection 0.35 0.98 0.28
Leaf fodder collection 1.43 0.97 1.79 0.10 0.35 0.00
Grazing ... 1.80
(0.90)b (0.90)b
Food processing 0.18 0.97 0.16 0.20 0.70
Cooking 0.27 2.05 0.24 0.38 2.10
Subtotal 2.19 5.04 2.72 1.26 6.43 2.44
(2.16)b (1.54)b
Total 6.16 8.24 3.92 5.16 9.31 2.49
(6.06)b (1.59)b
Others 1.37 2.57 0.69 n.a. n.a. n.a.

Sources: Meena Acharya and Lynn Bennett, The Rural Women of Nepal: An Aggregate Analysis and Summary
of Eight Village Studies, vol. 2, part 9 (Kathmandu: Tribhuvan University, 1981); and Nepal, Agricultural
Projects Service Center; the Food and Agriculture Organization of the United Nations; and International
Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
Notes: The ellipses indicate a nil or negligible amount. n.a. means not available.
a Based on observations from five hill villages and one tarai village.
b The numbers in parentheses are alternate assumptions of cattle grazing time. Under the first assumption all
grazing time is assigned to children.
c Activities recorded by Acharya and Bennett but not included in the present study are manufacturing, hunting
and gathering, house construction, plastering walls, cleaning and laundry, shopping, child care, and other domestic
work.


21 The methodology is derived from Acharya and Bennett, Rural Women of Nepa4 pp. 16-19.
22 Acharya and Bennett, Rural Women of NepaL









present study, which requires a higher degree of knowledge about practices in the area
than is usually available for designing survey instruments.
An examination of the effects of household income on men's and women's work
time shows that agricultural work increases for both sexes, but especially for men, as
household income increases. This may indicate some labor surplus for men at lower
income levels, which may be taken up by an increase in seasonal migration. Women's
labor, especially their total work load, is less variable by income. This suggests that
women are already working long hours at low incomes (and lower food availability).
If there is limited substitutability between men's and women's labor in agriculture,
then this could curtail men's labor use until it is possible to expand hired labor use.
Also, part of the increase in both men's and women's labor with higher income may
be due to improved food availability and intakes. Since household income is mainly
the product of agriculture and farm size, household labor use seems to increase with
farm size. This is confirmed later in the analysis of household labor use.
Time spent on other activities is relatively constant at all income levels for both
men and women (Figures 3 and 4). This indicates that there is probably a high propensity
to spend additional time on agriculture, despite an increase in wage labor use at the
higher income levels. This observation, in conjunction with the high marginal product
of labor in crop production, as seen later, suggests that if mechanisms were available


Figure 3-Men's time allocation by annual income per capital
Hours/Person/Day


Farm labor
Livestock care
Food processing and cooking
Fetching of fuel and water
P ZNX


I I I Ijlloie aia
200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)
Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.










Figure 4-Women's time allocation by annual income per capital
Hours/Person/Day
15 Farm labor
14 Livestock care
13 Food processing and cooking
12 Fetching of fuel
and water
11 -
10 -
9
8
7
6
5
4
3
2


0 -


200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400
Annual Income Per Capita (Rs)


3,800 4,200 4,600 5,000


Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.


to reduce the time spent on production support activities, more time would be spent
on agriculture.



Implications of Deforestation

As deforestation proceeds, access to forest resources becomes more and more
removed for the settled population. In the long run, populations have been observed
to shift increasingly to other sources to substitute for these fast-depleting resources.
Fuelwood is increasingly replaced by straw and dungcakes or other by-products. (At
the time of this study, cow dung was used exclusively for manuring fields and not as
fuel.) Leaf fodder and grass for animals are replaced by increased grazing,23 coarse
grains, and crop by-products. Also in the long run, water from forest streams becomes
more scarce and is replaced by water from more distant or contaminated rivers and


23 This is shown to have occurred in India in Subhachari Dasgupta and Asok Kumar Maiti, The Rural Enery
Crisis, Poverty, and Women's Roles in Five Indian Villages (Geneva: International Labour Organisation,
Rural Employment Policy Research Programme, 1986).









ponds. All of these changes have adverse effects on productivity and health, which can
partly be alleviated by higher incomes and the ability to purchase the alternatives in
the marketplace. However, if aggregate productivity of agriculture is also declining,
then the possibility of higher incomes also declines.
In order to understand the wider implications of household responses, the starting
point for this analysis is the behavioral response of deforestation on fuelwood consump-
tion behavior. Fuelwood demand is expected to be influenced by both income and
price factors, as well as fixed demand factors such as family size. Livestock numbers
also influence fuel consumption in Nepal because livestock are fed a cooked feed
preparation called kundo to increase the nutrient availability of miscellaneous crop
by-products used in livestock feed. The larger the herd the more fuelwood is needed
to cook the kundo. Since fuelwood is seldom purchased, the "price" is based on time
required for collection of a unit of fuelwood (a standard load, a bhari, was measured
to be approximately 20 kilograms of fuelwood). Two alternate price estimates can be
made here-one in terms of time required and the other incorporating the average
wage rate for agricultural labor. The information on wage rates for women in particular
was relatively poor because very few women worked for wages. Consequently, this
derived price is also mainly a function of time required for fetching fuelwood. Time
required to fetch a load of fuelwood is determined by the distance to the source, which
will increase as deforestation progresses.
The relationship of fuelwood use to price (time required to collect a bhari of
fuelwood) and income factors is illustrated in Figures 5-7. With rising household income,
fuelwood consumption generally increases through the first seven deciles, after which
there appears to be a declining trend.24 For kerosene, which is used mainly for lighting,
there is a fairly steady increase in use with income growth. There is a rapid reduction
in fuelwood consumption with increasing time per load. For estimating the magnitude
of these changes and its implication for total time spent on fuelwood collection, two
sets of equations are specified.25 In the first,
FW = f(LogY, FWp, HS, L, MB),26 (1)
where
FW = quantity of fuelwood consumed
(in kilograms per capital per year),
Y = total household income (rupees
per capital per year),
FWp = fuelwood price in time per load
(8-hour days),
HS = household size,
L = livestock units,27 and
MB = maize by-products (in kilograms).

4 The quadratic specification for the income effect on fuelwood use is, however, insignificant. This could
be due to the inexplicably reduced use of fuel in the fifth decile.
25 Although there appears to be some substitution of kerosene for fuelwood at higher income levels,
deforestation is not found to be a factor in the shift.
26 This equation was specified in both semilog and quadratic forms, but the semilog form was a better fit.
The quadratic term was insignificant for both the household and per capital equations.
27 Livestock were weighted as follows: buffalo, 1.5; cows and bullocks, 1.0; and sheep and goats, 0.2.










Figure 5-Per capital fuelwood consumption, collected by family and nonfamily
members
Loads/Year


34 1-


1 2 3 4 5


6 7 8 9 10


Per Capita Income Decile
Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.

Figure 6-Per capital kerosene consumption
Liters/Year
3.8

3.6 -

3.4 -

3.2

3.0

2.8 -

2.6 -

2.4-

2.2

2.0 L -- -- ..
1 2 3 4 5 6 7 8 9 10
Per Capita Income Decile
Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.










Figure 7-Fuelwood consumption using collection time per load as an
indicator of deforestation
Annual Household Fuelwood Use
(in 20-kilogram loads)
280 r-


260 -
250 -
240-
230 -
220
210
200
190
180
170
160
150 -
140 -
130 -
120 -
0.0


0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
Collection Time Per Load
(8-hour days/20-kilogram loads)


Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.
Note: Each point represents one-tenth of the sample.

Second,
TFW = f(LogY, FWp HS, L, MB), (2)
where

TFW = time spent on fuelwood collection
(8-hour days per capital per year).

These equations are analogous to estimating both the consumption and expenditure
consequences of price and income changes. While the second equation is not usually
estimated in standard demand analysis, it is interesting given that both price and
expenditure are specified in terms of time spent. Results are shown in Table 7. Calcu-
lations of elasticities for fuelwood consumption with respect to the increasing price of
wood indicate that a 10 percent rise in price of wood, or the time taken for collection
of one unit, will reduce its consumption by only 2-3 percent but will increase the time
spent for its collection by 4-6 percent. This is consistent with a relatively inelastic
response characteristic of a basic necessity. The income response best captured by a









Table 7-Factors explaining fuel collection and use, 1982/83

Time Spent Collecting
Fuelwood Use Fuelwood
Marginal Marginal
Propensity Propensity
Explanatory Factor Average to Consume Elasticity to Consume Elasticity
(20-kilogram loads/ (8-hour workdays/capita/
capita/year Ix 33.7]) year Ix = 18.3])
Intercept ... 43.76** ... 14.25*
Log of total annual income Rs 1,954.0 2.76t 0.8 0.37
(Rs/capita) (1.72) (0.38)
Average household size 6.6 -3.67** ... -1.56**
(-8.37) (-5.91)
Livestock units 7.0 0.43t ... 0.19
(1.67) (1.19)
Maize by-products produced
per household (kilograms) 820.9 0 ... 0
Time per load to collect 0.61 -14.73** -0.3 16.94** 0.6
(5.44) (10.35)
R2 (adjusted) ... 0.47 ... 0.62

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
Notes: Figures in parentheses are t-values. The size of the sample was 117.
a Time per load is determined using the following equation: FWP, = (Time per load [hours]/8) x (workdays).
*Significant at the 0.05 level.
**Significant at the 0.01 level.
tSignificant at the 0.10 level.


semilog functional form is small but significant. The income elasticity is found to be
between 0.08 and 0.09.
Other results from this estimation indicate that, first, per capital consumption of
fuelwood is lower in larger families, as is the time spent per adult in its collection;
second, livestock has a small effect on raising fuel demand; and third, availability of
maize by-products, often cited as a fuel source in the study area, has no effect on the
total fuelwood demand.
The preceding analysis is restricted to fuelwood use. Other forest products, such
as leaf fodder and grass for livestock and water from streams, would also recede with
deforestation. Because, as mentioned earlier, their location in forests is more localized,
the association of distance and deforestation is difficult to document in this cross-sectional
study. Whereas the demand for water is likely to be as inelastic as that for fuelwood,
the demand for livestock feed is influenced by changes in livestock holdings or feeding
practices. If their consumption response is similar to that of fuelwood, that is, relatively
inelastic, then the overall implications for demand for household and especially women's
time would be correspondingly higher with increasing deforestation.


Implications of Deforestation for Time Allocation
In examining the time allocation patterns of the sample households by degree of
deforestation, using simple tabular analysis, time spent on fuel-collection activities is
examined separately for lowland and highland sites (Table 8). The average time per









Table 8-Pattern of time allocation and deforestation, 1982/83

Time Taken to Collect a Load of Fuelwood
Lowlands Highlands
Low High Low High
Time Spent per Deforesta- Deforesta- Percent Deforesta- Deforesta- Percent
Activity ton tion ofChange ton tion ofChange
Time/load of fuelwood
(minutes) 106 193 82 163 270 66
Fuel collection time
(hours/day) 1.5 3.0 100 2.3 2.9* 26
All adults 1.3 2.6* 100 2.1 2.8* 33
Women 1.1 2.5* 127 1.9 2.6* 37
Children 0.2 0.4 100 0.2 0.1 -50
Per capital fuel collection
time (hours/person/
day) 0.3 0.4t 33 0.4 0.5 25
All adults 0.5 0.8* 45 0.7 1.0* 43
Women 1.0 1.6* 60 1.2 1.8* 50
Agricultural labor
(hours/day)
Women's field work
Per capital 2.7 2.7 0 3.3 2.7 -17
Per hectare 2.8 1.5* -46 2.8 1.8* -36
Men's field work
Per capital 1.9 4.1* 116 3.7 2.8 -23
Per hectare 2.2 2.7 23 3.1 1.7* -42
Wage labor 0.7 0.9 29 0.3 0.5* 67
Cropped area
(hectares) 1.4 3.2* 129 2.1 2.5 19

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
*Significant at the 0.05 level.
tSignificant at the 0.10 level.

trip for households in each ward is used as a proxy for the degree of deforestation in
the area. In addition to the pattern of household time allocation for fuel collection,
field work time by men and women is also tabulated by the degree of deforestation
(or time per trip for fuelwood).
Table 8 shows that the time per trip for fuelwood increased 82 percent in the
lowland wards and 66 percent in the highland wards, or an average of about 75 percent
as deforestation increased. According to the earlier regression results, it would be
expected that a 75-percent rise in the time per trip for fuel would increase the time
spent in collection by 45 percent. The difference in increase in time spent for fuel
collection between the low and high deforestation sites, in both the valley and hillside
sites, is in line with the expected values from the earlier regression results. Whereas
the total fuel collection time per household increased 100 percent for the lowland
sites, it rose 26 percent for the highland sites. On a per capital basis, however, the
average increase for lowland and highland sites combined was about 45 percent for
adults and 50-60 percent for women.
Collection of fuel and other forest products, such as leaf fodder and grass, is largely
done by women; women account for 72 percent of the time spent on this work (see
Table 6).28 On average, women spend 2.5 hours per day on such activities. This is


28 This figure includes collection time for fuelwood, leaf fodder, and grass.









only slightly less than the time they spend on farm labor-about 2.8 hours per day
averaged over a year. If the increase in time required for collection is borne entirely
by women, and if they take all of this additional time away from farm labor, it would
lead to a reduction in their farm labor input of 1.4 hours daily or a decrease of 50
percent, leading to a total reduction in farm labor of 24 percent if not substituted for
by other household members or by hired labor.
To see the effect of deforestation on farm labor input, the following analysis is
undertaken. First, simple tabular analysis shows that in the highland sites, where the
cropped area between the two deforestation groups is similar, there is a substantial
reduction in women's field labor on both a per capital and a per hectare basis. The only
exception is their per capital labor input in the lowland sites. The explanation for that
probably lies in the 129 percent higher amount of area cultivated, which would demand
additional labor input. Despite this increase in area, women's labor per capital was no
higher in the high deforestation sites. Surprisingly, men in the highland sites also
showed a reduction in time spent in field labor, which was statistically significant for
their hours spent per hectare. At the lowland sites with the higher amount of deforesta-
tion, there is also a significantly larger amount of area cropped. As seen earlier, household
labor for both women and men increases with income (and area farmed). Consequently,
the results of the tabular analysis reflect the combined effects of both deforestation
and cropped area.
To some extent, there is an increase in wage labor use in the areas with higher
deforestation. Even though the percentage of the increase is quite substantial, in absolute
terms it does not appear to make up for the decrease in household labor input in these
areas. A simple demand function for wage labor, however, does not indicate that the
degree of deforestation was a statistically significant factor in influencing its use, after
controlling for total cropped area and altitude.
In order to control for the main cropping characteristics in the association between
deforestation and household labor input in agriculture, the following household labor
input equation is estimated:
Lh, = f(Lw, OffY, Tcrl, PI, FS, Alt, Df), (3)
where
Lhi = household labor input, females and
males (hours/day);
Lw = wage labor input (hours/day);
OffY = off-farm income, the sum of remittances
and wage incomes (Rs/year);
Tcrl = total cropped area (hectares);
Pll = proportion of lowland in cropped area;
FS = number of resident adults in household;
Alt = altitude (high altitude wards = 1,
others = 0); and
Df = deforestation (wards where time per
trip is high = 1, others = 0).









Total household labor input, as well as that of males and females, is expected to
be a function of, first, household cropping characteristics such as total cropped area
and proportion of lowland, both of which would tend to increase labor input if the
income effect on labor input is greater than the substitution for leisure; second, use
of wage labor and off-farm income, which may be expected to reduce household labor
input; third, number of adults, which would increase total and per hectare labor input;
fourth, deforestation, which would reduce household labor input; and fifth, altitude.
Results of the estimation equations are given in Table 9. They confirm the negative
impact of deforestation on women's farm labor input. Even though men's farm labor
also appears to be slightly reduced, this result is not statistically significant. When other
household variables are controlled for, total household labor is reduced more than four
hours per day, out of which three hours are accounted for by a decline in women's
labor input. Also, both the increase in cropped area and the use of wage labor are
accompanied by an increase in household labor use, indicating a strong income effect
on household labor input. This is not unexpected given the high subsistence orientation
and relatively high incidence of malnutrition among children and adults. The low levels
of landlessness and hired labor use could also contribute to these results. An increase
in the number of adults in the household has the expected positive effect on household
labor input, and at higher altitudes, women's farm labor input is significantly higher.
The effect of deforestation on household labor input when estimated on a per
person basis indicates a reduction in women's farm labor of 1.6 hours per woman and
in men's farm labor of 0.8 hours per man (Table 10). Thus, while the effect for women
is more pronounced, there is also some evidence that men may also cut their farm
labor input. This could partly be the result of their involvement in collection activities
and the higher grazing time required in the high deforestation sites. However, the
limited substitutability between men's and women's activities in crop production at
the household level could also play a role, especially since additional use of wage labor
is limited.
The tabular analysis in Table 8 indicates a somewhat smaller reduction in labor
time per person. There are, however, differences between the low-altitude and high-
altitude sites. These differences can be traced to the larger areas cropped, especially
in the low-altitude areas of the high deforestation sites. This correlation between
deforestation and area cropped in the cross section is likely to be due to a two-way
causality between the two.
Results of the regression analysis show that an increase in cropped land tends to
be associated with higher household labor inputs, which partly offset the effect of
deforestation alone. When Equation 3 is estimated for labor input per hectare, it is
found that deforestation reduces household labor input by about 1.9 hours per hectare
or by about 40 percent (Table 11). This is after controlling for the negative effect of
the increasing cropped area on intensity of labor input.


Net Change in Women's Work Load with Deforestation
Results of the fuelwood equations indicate that a 1 percent increase in total collection
time increases time spent collecting fuelwood by 0.6 percent and decreases the amount
of fuel consumed by 0.3 percent. The time required to collect one load of fuelwood
increased by about 75 percent where deforestation was high. This implies about a 45
percent increase in the time allocated for firewood collection. What is the implication
of this result for the total activities surrounding collection of forest products? These









Table 9-Effects of deforestation on input of household farm labor in hours
per day

Household Labor Input
Variable Mean Total Females Males
(hours/day)
Intercept ... -2.22 -0.91 -1.29
(1.4) (1.2) (1.2)
Wage labor (hours/day) 1.3 1.11* 0.74** 0.38
(2.4) (3.3) (1.2)
Off-farm income (Rs/year) 2,865 -2.88E 05 4.50E 05 -7.01E 05
(0.2) (0.7) (0.8)
Total cropped area (hectares) 2.20 3.49** 1.44** 2.05**
(6.5) (5.6) (5.7)
Proportion of lowland in cropped area 0.41 0.76 0.22 -1.06
(0.28) (0.2) (0.6)
Number of adults per household 3.4 1.36** 0.57** 0.78**
(3.3) (2.8) (2.8)
Altitude (high = 1, low = 0) 0.50 1.47 1.54** -0.07
(1.2) (2.6) (0.1)
Time per trip for fuel (high= 1, low= 0) 0.49 -4.37** -3.21* -1.14
(3.1) (4.8) (1.2)
R2 (adjusted) ... 0.65 0.63 0.55

Notes: Means for the dependent variables are total household labor, 9.7 hours per day; female labor, 4.6 hours
per day; and male labor, 5.1 hours per day. The figures in parentheses are t-ratios.
*Significant at the 0.05 level.
**Significant at the 0.01 level.



products are mainly fodder and grass for animals, but over time the availability of water
would also be influenced by deforestation because the forest streams from which most
household water is obtained would eventually dry up without their forest cover.
Women's time spent collecting fuelwood, fodder, and grass for the sample house-
holds was found to be 2.5 hours per person per day averaged over the year. If the
response function for other forest products is assumed to be similar to that for fuelwood,
one would expect that women's time spent collecting forest products would also increase
by 45 percent (assuming the increased requirement is evenly distributed among all
household members). That would lead to a daily increase in women's collection time
for these products of 1.13 hours per day. It was estimated earlier that women's field
labor decreased by 1.6 hours per day due to the effect of deforestation. However, since
cropped area in the high deforestation sites tends to be higher, and this contributes
to a higher demand for household labor, the actual fieldwork of women decreases by
about an hour. This is reflected in the cross-sectional comparison in Table 8, which
indicates a relatively small reduction in fieldwork per person with deforestation. The
implications of deforestation, therefore, are two-fold: first, the work load of household
members is raised, and second, labor input per hectare is reduced, with possible adverse
yield effects. This reduction occurs even after controlling for cropped area, as shown
in the previous section.










Table 10-Effects of deforestation on input of household farm labor in hours
per person per day

Household Labor Input
Variable Mean Females Males
(hours/person/day)
Intercept ... 2.05** 1.79**
(4.1) (3.8)
Wage labor (hours/day) 1.30 0.10 0.10
(0.7) (0.8)
Off-farm income (Rs/year) 2,918 3.43E -05 1.6 05
(0.8) (0.4)
Total cropped area (hectares) 2.20 0.87** 1.07**
(5.2) (6.9)
Proportion of lowland in cropped area 0.41 1.12 0.89
(1.3) (1.1)
Number of adults per household 3.4 -0.4** -0.47**
(3.0) (3.8)
Altitude (high = 1, low = 0) 0.50 0.49 0.37
(1.3) (1.0)
Time per trip for fuel (high = low= 0) 0.49 -1.57** -0.79*
(3.6) (2.0)
i2 (adjusted) ... 0.31 0.45

Notes: The figures in parentheses are t-ratios. Means for the dependent variables are female labor, 2.8 hours per
person per day and male labor, 2.9 hours per person per day.
*Significant at the 0.05 level.
"Significant at the 0.01 level.

Table I I-Effects of deforestation on input of household farm labor per day
in hours per hectare

Household Labor Input
Variable Mean Coefficient
(hours/hectare/day)
Intercept ... 3.27**
(5.26)
Wage labor (hours/hectare/day) 0.57 1.08**
(3.0)
Off-farm income (Rs/year) 2,865 6.19E 06
(0.1)
Total cropped area (hectares) 2.20 -0.18
(1.2)
Proportion of lowland in cropped area 0.41 0.63
(0.63)
Number of adults per household 3.4 0.41**
(2.7)
Altitude (high = I, low 0) 0.50 0.57
(1.2)
Time per trip for fuel (high = 1, low = 0) 0.49 -1.86**
(3.6)
R2 (adjusted) ... 0.18

Notes: The mean for the dependent variable is 4.5 hours per hectare per day. The numbers in parentheses are
t-ratios.
**Significant at the 0.01 level.









5


AGRICULTURAL PRODUCTION
CHARACTERISTICS

Implications of Deforestation for Production
What are the consequences for production of increasing the demand on women's
time of collection activities? Not only does their total work load increase, but time
spent on directly productive activities is reduced. This could have adverse consequences
on income, primarily from agricultural production.
The previous section confirmed that household labor input, and especially the
women's labor component of time spent in farming on a household and per-person
basis, declines with a higher degree of deforestation, as reflected in the time per trip
for fuel collection. At the same time, men's labor input also decreases. This could be
due at least in part to lack of perfect substitutability between women's and men's labor.
Detailed studies of time allocation from Nepal suggest that there are some activities
for which sexual division of labor is more rigid than for others. Agricultural activities
described as being almost exclusively carried out by women are seed selection, weeding,
and application of organic manures.29 If the extent to which men's labor can be
substituted for women's is limited, then reduction in women's labor could also reduce
men's labor if men must wait for women to complete their tasks. Other contributing
factors may be more time spent by men on livestock grazing and increased seasonal
migration by men.
The extent of this decline in agricultural labor is greater when controlling for
cropped area in the regression analysis than in the cross-sectional comparisons. This
is partly explained by the larger amount of area cropped at higher deforestation sites,
and it is especially pronounced for the lowland sites. These larger land areas are
consistent with the central thesis that excessive land clearing contributes to deforesta-
tion. An increase in cropped land is found to have a strong income effect on labor
input. An increase of 1 hectare in cropped area increased household labor by 3.5 hours
per day, women's labor by 1.4 hours per day, and men's labor by 2.1 hours per day
(Table 9).
In sum, high deforestation sites have greater demand for labor both for collection
activities and for agricultural labor. This is reflected in a significantly larger household
size and a lower number of permanent and seasonal migrants from these sites. The
possible long-term demographic implication of this kind of response to deforestation,
while not possible to explore in this report, should be examined further. Labor input
per hectare is significantly lower in the high deforestation sites, and both female and
male labor are reduced when cropped area and proportion of lowland cropped are
controlled for.
What are the implications for agricultural production? The production response to
these reductions in labor input is indicated by the factor shares and the marginal

29 Bina Pradhan, The Newar Women ofBulu, vol. 2, part 6 of The Status of Women in Nepal (Kathmandu:
Centre for Economic Development and Administration, Tribhuvan University, 1981); and Sidney Schuler,
The Women of Baragaon, vol. 2, part 5 of The Status of Women in Nepal (Kathmandu: Centre for Economic
Development and Administration, Tribhuvan University, 1981).









product of labor. Women's labor is primarily influenced by deforestation, but the degree
of substitutability between men's and women's labor influences the reduction in men's
labor. These issues are examined in the next section.


Significance of Labor to Agricultural Production
The main crops grown in the study area are wheat, early paddy, and paddy for
lowland plots, and maize and ragi (finger millet) for upland plots. A small proportion
of maize is planted on lowland plots. Mustard and blackgram are also grown, and both
provide essential dietary ingredients. Production data recorded for each crop were area;
labor input in days, by household and by wage labor of men, women, and children;
days of animal labor used in crop production; purchased fertilizer use; and output.
Aggregate observations for each crop are given in Table 12. Levels of labor input and
yields for cereals are within the range expected for hill regions.30 The yields for both
mustard and blackgram are on average substantially lower than expected, and the
variation in their yields is also very high. This would indicate either extreme riskiness
in production of these crops or some underlying problem in reporting. The ratio of
output to total labor days, indicating the average product of labor, is highest for early
paddy at 7.7 kilograms and lowest for blackgram at 1.2 kilograms.
In order to estimate the marginal product of land, human labor, and animal labor
in the production of each crop, translog or specific production functions are used. In
the first equation, an interaction term for aggregate labor input and area is included
to test whether the effects of each are influenced by the level of the other. This is used
later to determine the factor shares and marginal products for small and large farms.
The equation is
LogP = f(LogA, LogLt, LogLv, LogALogLt), (4)
where
P = crop output (in kilograms),
A = crop area,
Lt = total human labor input (in days
of family and hired labor), and
Lv = animal days used.
Next, the marginal product of men's and women's labor input in crop production
is tested. This is considered important for two reasons. First, there is reason to expect,
given the nature of sexual division of labor in Nepal, that there may not be perfect
substitutability between men's and women's labor. Second, it has been documented
that women carry nearly the entire work load for fuel, fodder, grass, and water collection
as well as food processing and cooking. So if "women's tasks" in crop production are
not easily assumable by men, then it would indicate that, in order to increase agricultural
productivity, an improvement in productivity of women's tasks in nonagricultural activ-
ities may also be essential to increase agricultural production.
To test the homogeneity between men's labor and women's labor, the production
function is tested to see if it is separable into these subaggregates. Separability implies

30 Nepal, Agricultural Projects Services Centre, Nepal: Foodgrain Marketing and Price Policy Study.








Table 12-Basic crop production data, 1982/83

LaborUse
Numberof Household Wage Average
Cultivating Labor Labor Total Fertilizer Production/
Crop Households Area Yields Men Women Men Women Labor Ox-days Used Labor Day

(N = 118) (hectares) (kilogram/ (days/hectare) (days/ (kilogram/ (kilogram/
hectare) hectare) hectare) labor day)
Wheat 46 0.40 1,008 78 82 12 5 177 9 112.4 5.7
Maize 113 0.66 1,183 77 82 17 10 186 28 ... 6.4
Earlypaddy 54 0.40 1,751 108 103 8 9 228 23 .. 7.7
Paddy 97 0.67 1,863 165 105 37 11 318 67 ... 5.9
Ragi 92 0.50 955 111 162 7 8 288 27 ... 3.3
Mustard 38 0.18 263 46 59 15 10 130 74 ... 2.0
Blackgram 53 0.40 210 54 71 23 26 174 53 ... 1.2

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations; and International Food Policy Research Institute,
"Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
Note: The ellipses indicate a nil or negligible amount.








that marginal rates of substitution between pairs of factors in the separated group are
independent of the levels of factors outside that group.31 In this case, the subaggregates
are men's (Lm) and women's (LI) labor, and area is used as the factor outside the group.
If the true production function is weakly separable into Lm and L, and the separable
form of a translog function is interpreted as an exact production function used as a
Cobb-Douglas function of translog subaggregates, then
LnP = f[LnLt(LnIL, LnL,), LnA...], (5)
where L, is the aggregate labor input. For f(LnLt, LnA) to be linear in LnLt and LnA
only, then it has to be shown that LnLt is quadratic in LnLm and LnL,. For the argument
of separability, that is, the homogeneity of men's and women's labor, to hold, the
constraints to apply are
LnLm LnA = LnL- LnA = 0. (6)
It is assumed that production is a linear function of both land and labor, with a declining
marginal product for both. The coefficients for the squared terms are therefore set
equal to 0. The equation estimated is
LnP = f(LnA, LnLm, LnL, LnLm* LnA, LnL,- LnA, LnL,), (7)
where Lm is men's labor input for the crop (in days of family and hired labor), and L,
is women's labor input for the crop.
Results from these three equations are used to calculate the marginal products and
factor shares for each input at its mean value in use. These are shown in Tables 13,
14, and 15. The main observations from these tables are as follows:
1. The production response of labor is consistently significant, and its factor share is
higher than that for land for all crops except early paddy. Also, the marginal product
value for most crops is equal to or greater than the prevailing wage rates, indicating
that despite the small size of average holdings, hill agriculture does not represent a
labor surplus situation, and labor is a significant constraint in the production of most
crops grown.
2. Wheat and paddy have the highest marginal products with increasing labor input.
These are predominantly lowland crops. The upland crops, maize and ragi, have a
lower marginal productivity of labor. A comparison with local wage rates-Rs 8.00
per day for men and Rs 5.00 for women-suggests that wages are comparable to
the marginal product of labor for maize and ragi, but are lower than those for wheat
and paddy if rice is priced at Rs 5.00 per kilogram and wheat, ragi, and maize at
Rs 3.00 per kilogram.
3. The marginal product of men's and women's labor is found to be independently
significant for all crops except early paddy, often with different marginal products.
The separability test indicates that except for the dry-season crops, wheat and mus-
tard, men's and women's labor are weakly separable, that is, they have different
marginal rates of substitution with land. This suggests that, at least for some crops,
there is little justification for aggregating men's and women's labor.


31 Michael Denny and Melvyn Fuss, "The Use of Approximation Analysis to Test for Separability and the
Existence of Consistent Aggregates," American Economic Review 67 (une 1977): 404-418.







Table 13-Production functions by crop, 1982/83

f LLabor LArea x LLabor LAnimal R'
Crop Intercept LArea Total Men's Women's Total Men's Women's Labor N (Adjusted)


Wheat
Equation (4) 3.834 0.525 0.521* ... ... -0.081 ... ... -0.033
(1.44) (2.82) (-0.86) (0.62)
Equation (7) 4.632 0.78* ... 0.37* -0.014 ... 0.116* -0.326* -0.002
(2.99) (2.13) (-0.14) (2.03) (-3.72) (-0.04)
Early paddy
Equation (4) 5.284 0.800 0.515 ... ... -0.018 ... ... 0.028
(1.49) (1.60) (-0.13) (0.30)
Equation (7) 5.369 0.708 ... 0.454 -0.053 ... 0.020 -0.027 0.026
(1.42) (1.43) (-0.35) (0.11) (-0.18) (0.26)
Maize
Equation (4) 3.699 -0.026 0.393* ... ... 0.047 ... ... 0.326*
(-0.08) (2.53) (0.63) (2.54)
Equation (7) 4.453 0.153 ... 0.031 0.193* ... 0.049 -0.024 0.421*
(0.73) (0.20) (2.25) (0.51) (-0.25) (3.33)
Ragi
Equation(4) 2.802 0.796* 0.617* ... ... -0.171* ... ... 0.001t
(2.71) (5.62) (-2.71) (1.97)
Equation (7) 3.536 0.723* ... 0.409* 0.151 ... 0.000 -0.167 0.057t
(2.57) (2.00) (0.77) (0.0) (-1.10) (1.79)
Paddy
Equation (4) 3.640 0.340 0.638* ... ... 0.1010 ... ... 0.095
(1.40) (4.63) (0.19) (0.47)
Equation (7) 4.039 0.353t ... 0.455* 0.175 ... -0.119 0.147 0.055
(1.68) (3.15) (1.09) (-1.01) (1.11) (0.57)
Blackgram
Equation (4) -1.733 -0.49 1.02* ... ... -0.01 ... ... 0.28
(-1.27) (2.15) (-0.15) (0.70)
Equation (7) -0.855 -0.46 ... 0.10 0.83 ... -0.25 0.23 0.33
(-1.28) (0.10) (0.93) (-0.49) (0.46) (0.77)
Mustard
Equation (4) 1.364 0.23 0.43 ... ... -0.002 ... ... 0.51*
(0.73) (1.21) (-0.02) (2.04)
Equation (7) 2.210 0.32 ... 1.35* -1.14* ... 0.58* -0.57* 0.53*
(1.49) (3.62) (-2.55) (3.37) (-3.41) (2.27)


Notes: Equation (4) is LogP = f(LogA, LogL,, LogLv, LogALogL,), where P is crop output, A is crop area, L Is total human labor input, and Lv is animal days
used. Equation (7) is LnP = f(LnA, LnL., LnL,, LnL.- LnA, LnLr LnA, LnL}), where Lm is men's labor input for the crop and L is women's labor input
for the crop.


46 0.59

46 0.64

54 0.56

54 0.56

112 0.62

112 0.65

92 0.67

92 0.66


97 0.79

97 0.79

52 0.47

52 0.46

37 0.62

36 0.74


*Significant at the 0.05 level.


t~igniflcant at the 0. 10 level.









Table 14-Marginal products in kilograms of primary factors for total sample,
1982/83
Total Days of Labor Days of
Crop Area Men's Women's Both Animal Labor
(hectares)
Wheat
Equation (4) 242.2' ... ... 4.5 -2.6a
Equation (7) 242.2 2.5 7.4 ... 0.0a
Early paddy
Equation (4) 1,259.1' 3.7a 2.4a
Equation (7) 1,311.6' 7.3a 1.7 .. 2.4a
Paddy
Equation (4) 719.5a .. ... 3.8 1.5'
Equation (7) 738.0 20.6 0.9a ... 1.8a
Blackgram
Equation (4) -634.9 .... 0.9 1.0
Equation (7) -57.3 0.9 0.8 ... 1.2
Maize
Equation (4) 220.7' ... ... 2.1 12.4
Equation (7) 302.0a -0.7' 2.4 ... 15.7
Ragi
Equation (4) 52.8 .. ... 2.5 2.0
Equation (7) 79.2 3.0 1.9a ... 2.0
Mustard
Equation (4) 30.8 ...... 0.8 2.1
Equation (7) 44.2 0.2 0.5 ... 2.1

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
Notes: Equation (4) is LogP = f(LogA, LogL,, LogLv, LogALogL,), where P is crop output, A is crop area, L, is
total human labor input, and Lv is animal days used. Equation (7) is LnP = f(LnA, LnLm, LnL,, LnLm- LnA,
LnLf LnA, LnL ), where L is men's labor input for the crop and L, is women's labor input for the crop.
The ellipses indicate a nil or negligible amount.
a These figures are marginal; they are not based on any significant coefficients.


4. The marginal product of women's labor is found to be substantially higher than that
of men's labor for wheat, maize, and mustard. It should be noted that these are all
dry-season crops-a time when there could also be increased competition from fuel
and other collection activities, all of which are performed mainly by women.

5. For the paddy crop, men's labor input is the main constraint. For ragi, the other
main wet-season crop, men's and women's labor is equally important. For wet-season
crops, it is possible that women's labor may be more readily available because the
burden of collection activities is reduced.

6. Among the cereal crops, animal labor is a significant constraint only in maize produc-
tion. Incremental use of animal labor is also indicated as beneficial for production
of both mustard and blackgram. This finding, however, cannot be seen as conclusive
and needs to be investigated further.
High factor shares for labor and the significantly positive marginal product for labor
found in the above analysis point to labor input as a constraint in agricultural production.
This is despite the low levels of improved technological inputs used and the widespread
perception of hill agriculture in Nepal as being labor surplus. Factor shares and marginal










Table 15-Factor shares of land, labor, and livestock in crop
total sample, 1982/83


production,


Total Days of Labor Days of
Crop Area Men's Women's Both Animal Labor
(hectares)
Wheat
Equation (4) 0.24a ... ... 0.63 -0.03a
Equation (7) 0.24 0.20 0.45 ... 0.00'
Early paddy
Equation (4) 0.721 ... ... 0.41' 0.03'
Equation (7) 0.75 0.42a -0.09 ... 0.03'
Paddy
Equation (4) 0.39a ... ... 0.63 0.05'
Equation (7) 0.40 0.55 0.05' ... 0.06'
Blackgram
Equation (4) -0.52 .. ... 1.04 0.28
Equation (7) -0.50 0.45 0.48 ... 0.33
Maize
Equation (4) 0.19' .. ... 0.36 0.33
Equation (7) 0.26a -0.06' 0.20 ... 0.42
Ragi
Equation (4) 0.06 ... ... 0.80 0.06
Equation (7) 0.09 0.41 0.34' ... 0.06
Mustard
Equation (4) 0.23 ... ... 0.43 0.51
Equation(7) 0.33 0.05 0.14 ... 0.53

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
Notes: Equation (4) is LogP = f(LogA, LogL,, LogLv, LogALogLt), where P is crop output, A is crop area, L, is
total human labor input, and Lv is animal days used. Equation (7) is LnP = f(LnA, LnL., LnL, LnL LnA,
LnLf LnA, LnL ), where L is men's labor input for the crop and L is women's labor input for the crop.
a These factor shares are not base on any significant coefficients.



products were also calculated for small and large farms and were found to be similar in
the present specification (Table 16). Sample sizes were too small for separate estimations.
The results of this analysis in conjunction with the earlier results on the reallocation
of household labor with deforestation provide evidence of a labor constraint and of a
declining aggregate productivity in agriculture resulting from reductions in labor input
This is further confirmed in this study by the significantly lower yield for the dry-season
crops in the high deforestation sites (Table 17). The results also suggest that, first, adverse
effects of deforestation on agricultural productivity will be relatively more pronounced on
the smaller farms; second, introduction of improved agricultural technologies that require
higher labor input may be problematic on smaller farms unless improvements in labor
productivity are accessible for other essential tasks being performed by household members;
and third, if improvements in agricultural labor productivity are possible, there is less
likely to be a push for expanding area under cultivation than there would be in a labor-surplus
situation.








Table 16-Characteristics of crop production on large and small farms, by cropped area, 1982/83

LaborUse Marginal Marginal
Numberof LaborUse ma
Households Household Labor Wage Labor Total Animal Factor Share Product
Crop Cultivating Area Yield Men Women Men Women Labor Labor Fertilizer Land Labor Land Labor
(hectares) (kilograms/ (days/hectare) (days/ (kilograms/ (kilograms/ (kilograms/ (Kg/ (Kg/
hectare) hectare) hectare) hectare) day) hectare) day)
Smaller farms
Wheat 25 0.18 1,209 94 110 11 5 226 4 152 0.26 0.69 244 3.26
Maize 56 0.39 1,169 72 91 21 9 193 25 ... 0.16 0.34 138 1.86
Earlypaddy 16 0.24 1,679 86 101 6 6 199 20 ... 0.74 0.54 997 4.19
Ragi 52 0.28 856 115 169 9 11 304 19 ... 0.12 0.87 74 2.28
Paddy 41 0.28 2,024 170 109 38 12 328 62 ... 0.38 0.62 637 3.48
Larger farms
Wheat 28 0.52 831 65 67 18 6 156 12 82 0.22 0.61 145 3.14
Maize 57 0.92 1,198 92 83 13 13 201 32 ... 0.20 0.38 174 2.02
Earlypaddy 38 0.46 1,782 117 104 9 10 239 24 ... 0.73 0.54 857 3.55
Ragi 52 0.66 1,071 124 165 5 4 299 35 ... 0.01 0.77 67 2.30
Paddy 56 0.96 1,745 161 103 37 11 311 70 ... 0.39 0.64 602 3.66

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations; and International Food Policy Research Institute,
"Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
Notes: Total cropped area for smaller farms is less than 1.10 hectares. The ellipses indicate a nil or negligible amount.

Table 17-Crop yields by degree of deforestation, for lowland and highland
sites combined, 1982/83

Low High
Crop Deforestation Deforestation
(kilograms/hectare)
Paddy 1,802 1,909
Ragi 1,009 907
Wheat 1,255 754*
Maize 1,365 998t

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
*Significant at the 0.05 level.
tSignificant at the 0.10 level.










6


HOUSEHOLD FOOD CONSUMPTION
AND NUTRITION

Relationship Between Income and Food Consumption
In addition to a general description of patterns of consumption and associated
characteristics of income and expenditure in the study area, two aspects of food con-
sumption are examined. First, levels of calorie intake32 are analyzed as a function of
income, agricultural production, time allocation patterns for women, and quantity of
fuelwood used. Then cereal composition of the diet and the possible implications of
deforestation on it are analyzed, using a two-step approach. First, the effect of quantity
of fuelwood used on cooking time is examined. (There was no variation in the type of
stove used in the sample, and cooking time includes all time spent in food preparation.)
Second, cooking time is incorporated into an equation to explain the ratio of rice to
total calories in the diet.
Rice and maize are the chief dietary staples, with rice consumption expenditure
rising rapidly with income (Figure 8).33 Ragi has a smaller though relatively constant
place in the diet. Wheat is consumed in negligible amounts and is largely produced
for sale or for wage payments. Expenditures on other foods also increase with income,
but not as rapidly as those for rice.
Home-produced items contribute greatly to the household's food basket. Surpris-
ingly, the reported value of purchased foods does not appear to increase much as per
capital income rises, but home-produced consumption does. Milk, milk products, and
eggs constitute the major part of noncereal food expenditures in home-produced consump-
tion. Total cereal consumption and home-produced cereal consumption are virtually
identical.
Allocation of total income to food shows the expected pattern of gradually declining
proportions, commonly referred to as Engel's Law (Figure 9). An inflection point occurs
at the second income decile, which is a phenomenon that has been observed in other
low-income countries.34 This is shown by a rising proportion of income spent on food
at very low income levels, before it starts to decline as expected. This occurs only in
a seriously deprived population, in which practically all income increments are spent
on food. It is hypothesized that there are essential nonfood fixed costs that must be
incurred even at the lowest income levels, and these prevent a higher proportion of
income being spent on food. The proportion of income spent on food rises from about
72 percent in the first income decile to more than 90 percent in the second income
decile, before declining to about 40 percent for the highest income decile.
Cash expenditures for food are a much larger share of income at lower income
levels, and the share is reduced as incomes rise. However, nonmarket value of consump-


32 All calories referred to in this report are kilocalories.
33 Rice consumption expenditure refers to the total value of food in the diet, purchased and home-produced.
34 Neville Edirisinghe and R. M. K. Ratnayake, "A Preliminary Analysis of the Determinants of Nutritional
Welfare in Sri Lanka," International Food Policy Research Institute, Washington, D.C., January 1984
(mimeographed).










Figure 8-Value of major food items in the diet by annual income per capital
Annual Value of Food Items
(Rs/capita)
2,250 r-


2,000

1,750

1,500

1,250 -

1,000 -

750 -

500-


250



200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)

Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.

tion in relation to total income roughly follows the same pattern as did total food
expenditures.
Nonfood expenditures show an interesting pattern in which wage payments clearly
become the major increasing expenditure item in this category (Figure 10). This reflects
the relatively small amount of other nonfood expenditures incurred, given the limited
use of wage labor on aggregate. This observation again emphasizes the labor constraint
faced by these households, where an increase in household labor in production is
paralleled by an increase in hired labor use.
Consumption reported for individual food items is converted to dietary calories.35
The annual average daily per capital consumption is estimated at 2,137 calories. The
distribution by income of caloric consumption is shown in Figure 11.
Finally, the contribution of the main income sources, farm production and remit-
tances, is examined (Figure 12). Though the absolute values of both remittances and
farm production rise with income, the ratio of farm production to total income is



35 The estimate of food consumption is based on amounts reportedly kept for home consumption for all
field crops, livestock, horticultural produce, food purchases, and foods received as wage payments or gifts.
A major disadvantage of this method is that seasonal fluctuations in food consumption cannot be identified.










Figure 9-Proportion of income spent on food by annual income per capital
Ratio of Annual
Expenditure to Income

0.9 -
Numbers indicate income deciles.
0.8
4 Total value of food
3 Value of purchased food
0.7 16

0.6 5

0.5 -
9
0.4 -
10

0.3 -

0.2 -

0.1 .
0.0 I I I I I I I I I I I
0.0 1
200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)
Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.

calculated to be about 60 percent for different income levels, and the proportion of
remittances varies from about 20 percent of income at the lower income levels to
about 10 percent for the middle income group and 16 percent for the highest income
group, with no perceptible trend.


Links Between Deforestation, Diet Composition,
and Intake

The effects of deforestation on collection time for fuelwood and, by inference, other
essential forest products, and the possible effects on agricultural production via a
substitution of production time for collection time have already been examined. Here,
these effects of deforestation are extended to see if they also affect household food
consumption and its composition. Figure 13 shows the pathways by which such effects
might occur.
There are several simultaneously occurring mechanisms by which deforestation
can be expected to have a direct effect on dietary intake and composition (the indicator
of dietary composition used here is the ratio of rice calories to total cereal calories in
the diet).










Figure 10-Nonfood expenditures by annual income per capital
Annual Expenditure
PerCapita (Rs)
900

800 -
Nonfood consumption items
700 -
Livestock
600 Purchase of by products
Fuel purchase
500 Wage payments

400 -

300 -

200 -

100 -



200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)

Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.


As shown earlier, households use less fuelwood as their time costs rise. The higher
collection time may reduce the time available to spend on food preparation and cooking,
which in turn may influence both the amount of food consumed and diet composition.36
While higher fuelwood consumption is likely to be associated with higher cooking
time, higher collection times may reduce cooking time. The net effect of deforestation
would then be to reduce cooking time, because deforestation both reduces fuelwood
use and increases time spent in collection activities. To the extent that diets are
influenced by the time spent in the kitchen, both qualitative and quantitative aspects
of the diet would be relevant in the analysis. If fuelwood shortage affects the composition
of the diet, it would be mediated by differences in cooking time for different products
or commodities, given no change in the cooking technologies available. Also, if additional
time for collection activities with deforestation reduces time available for cooking, the
amount of food being consumed may also be reduced.


36 This has been shown for Mexico in Margaret Evans, "Change in Domestic Fuel Consumption in Central
Mexico and Its Relation to Employment and Nutrition," working paper prepared for World Employment
Research Programme, International Labour Organisation, Geneva, 1986; and Elizabeth Cecelski, "The Rural
Energy Crisis, Women's Work, and Family Welfare: Perspectives and Approaches to Action," working paper
prepared for World Employment Research Programme, International Labour Organisation, Geneva, 1984.










Figure 1 1-Estimated calories available by annual income per capital
Calories/Capita/Day
9,000 r-


8,000

7,000 -

6,000 -

5,000 -

4,000 -

3,000 -

2,000 -

1,000 -


- Calories based on reported consumption
Calories based on availability from retentions


woo Oso


0 1 1 1 1 1 1 1 1 1 1 I 1 I
200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)

Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.



Also, with deforestation, the increased time spent in collecting fuelwood and other
forest products, such as fodder and grass, may reduce the time spent on agricultural
production, thereby reducing income from agriculture. This would have a direct adverse
effect on food consumption.


Cereal Composition in the Diet
Although maize and rice are the two main sources of calories in this region, the
households in the study have an overwhelming preference for rice. More than any
other food, rice consumption value increases rapidly with income. Rice preparation,
however, requires longer cooking time than other cereals. If advancing deforestation
means that more time must be allocated to collecting fuelwood, it is hypothesized that
cooking time will decrease and, as a result, the amount of rice consumed will be
reduced relative to other cereals.
The multivariate analysis consists of two models: a cooking time function that
includes independent variables relating to deforestation-amount of fuelwood used
and total time spent in its collection, plus household characteristics-and a diet composi-
tion function that includes the independent variable, total cooking time. The functional
form of the first model is










Figure 12-Sources of income by annual income per capital
Annual Income Per Capita
from Component (Rs)
3,300

3,000 Crop value
--- Remittances
2,700 -- ---- Cropby-products

2,400

2,100

1,800

1,500

1,200

900 .

600 0 00

300 -


200 600 1,000 1,400 1,800 2,200 2,600 3,000 3,400 3,800 4,200 4,600 5,000
Annual Income Per Capita (Rs)
Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.


C = f(L, T, H, A, AR, DP, WR), (8)
where
C = total cooking and food preparation time per day,
L = total 20-kilogram loads of fuelwood consumed
peryear,
T = total number of eight-hour days spent on fuel-
wood collection per capital per year,
H = household size,
A = livestockunits,
AR = farm size dummy (farm in larger half of farms = 1),
DP = ratio of children to adults, and
WR = ratio of women's farm labor time to total farm
labor time.


The results in Table 18 indicate that cooking time is positively associated with
fuelwood consumption and negatively associated with fuelwood collection time per









Figure 13-Links between deforestation and dietary intake


I erorestaton I


Positive effect Fuel Amount
-- Negative effect


Other Collection Fuel Collection Ratio of Upland
Times Time to Lowland,
I Land Quality
II 1I

Cooking Time



Labor in Dietary Riceto
Agriculture Calories Cereal Calories



] Income ,


Source: Based on data collected by Nepal, Agricultural Projects Service Center; the Food and Agriculture Organiza-
tion of the United Nations; and the International Food Policy Research Institute, "Nepal Energy and
Nutrition Survey, 1982/83," Western Region, Nepal.



capital. The coefficients of both these independent variables are significant. The results
indicate that the additional time necessary for the collection of fuelwood has a negative
effect on time for cooking, after controlling for the other variables in the equation. It
is also likely that cooking time may be negatively affected by time spent on other
activities, such as fuel, fodder, and water collection, that increase with deforestation
and place additional burdens on women's time.
Assuming that increasing levels of deforestation decrease time allocated to cooking,
a diet composition function is estimated to determine the effect of cooking time on
composition of cereal consumption. The functional form of this model is


RRICE = f(C, DP, Y, RAR, WR, H), (9)
where
RRICE = ratio of calories from rice to calor-
ies from other cereals,
Y = total income per capital, and
RAR = ratio of lowland area to total area.










Table 18-Effects of fuelwood use on cooking time
Cooking Time Cooking Time
Variable (Model 1) (Model 2)
(minutes/day)
Intercept 211.11 222.85
Total loads of fuelwood peryear 0.13* 0.13*
(20-kilogram loads) (2.10) (2.12)
Time spent on fuelwood collection -0.09
(-1.18)
Time spent per capital on fuelwood ... -0.69t
collection (-1.71)
Household size 3.68* 2.35
(2.15) (1.24)
Livestock units 1.82t 1.72t
(1.73) (1.68)
Dummy for farm size -4.53 -5.04
(-0.50) (-0.56)
Age dependency (children/adults) -4.52 -5.04
(-0.43) (-0.49)
Ratio of women's farm labor to -27.81 -25.04
total farm labor (-0.84) (-0.76)
A2 (adjusted) 0.27 0.28

Notes: Cooking time includes all time spent in food preparation. Figures in parentheses are t-values. Ellipses
indicate variables that are not included in the equation.
*Significant at the 0.05 level.
tSignifcant at the 0.10 level.

The results of the regression procedure show that additional amounts of cooking
time across income groups significantly increase the ratio of rice calories to other
calories (Table 19). In addition, the variables of child dependency, income per capital,
and the ratio of lowland area to total area have significantly positive associations with
the dependent variable, while the ratio of women's farm time to total farm time has
a significantly negative relationship.
Taken together, the results of the two models above indicate that there is an indirect
relationship between deforestation and diet composition.

Dietary Calories
Food consumption behavior in terms of total intake of calories is analyzed in this
study to see if it is affected by changes in household income, its composition, adult
literacy, and cooking time. Even though caloric intake is made up of a variety of foods,
the demand for which can be altered by a change in income, the net effect is reflected
in caloric or protein levels. Energy is a basic dietary requirement, a deficiency of which
is felt as hunger. Therefore, to some extent, it may be said that people do purchase
food for its calories, more perhaps than for any food component other than "taste,"
which is largely undefinable.37

37 Howarth Bouis, "An Agricultural Sector Model for the Philippines Identifying Localized Shortfalls in
Household Food Security and Evaluating Policy Options Using Data from Ongoing Quarterly Surveys," a
paper presented at a workshop of the International Food Policy Research Institute-Food Studies Group,
Oxford, England, July 7-9, 1987.









Table 19-Effects of cooking time on cereal composition of the diet
Ratio of Rice Calories
Variable to Cereal Calories

Intercept 0.034
Cookingtime 0.0003t
(1.80)
Age dependency 0.09t
(1.70)
Income per capital 0.00005*
(3.37)
Ratio of lowland area to 0.55*
total area (5.58)
Ratio of women's farm labor -0.54*
to total farm labor (-3.18)
Household size 0.01
(1.34)
i2 (adjusted) 0.48

Notes: Cooking time includes all time spent in food preparation. Figures in parentheses are t-values.
*Significant at the 0.05 level.
tSignificant at the 0.10 level.

There has been much debate in the literature on the size of the caloric response
to income.38 It has been observed that lower-income households and those with gen-
erally calorie-deficient diets have a higher income elasticity for caloric intake than the
rest of the population. Nevertheless, income elasticities for more expensive foods are
higher than those for calories, even for these households. In other words, people are
likely to spend incremental income on more expensive food than on larger quantities
of high-calorie but less appealing foods. Rural households, which generally expend
more energy and therefore have higher requirements than urban households do, also
have a higher level of caloric intake and a higher income elasticity for calories at all
levels of intake and incomes than their urban counterparts. These findings suggest
that, while households with inadequate consumption have a high propensity to consume
additional calories, they are also buying taste. There have been some attempts to
quantify the extent to which households buy taste rather than calories at different
income levels, and the implications of this for diet quality, but there is still little known
about this component of food demand.
Caloric consumption functions are estimated as follows:

LCAL = f(LY, FY/Y, ADLIT, CPC), (10)
where
LCAL = log of calories per capital per day,
LY = log of total annual per capital house-
hold income,


38 For a review of the debate on methods and results obtained for individual foods, overall food expenditures,
and energy intakes, see Harold Alderman, The Effect of Food Price and Income Changes on the Acquisition
of Food by Low-Income Households (Washington, D.C.: International Food Policy Research Institute, May
1986).









FY/Y = ratio of farm income to total income,
ADLIT = percent of literate adults in the house-
holds, and
CPC = women's cooking time divided by the
number of household members.


The equation is estimated, first, for the total sample; second, for households for which
preschool nutritional status is low for the year as a whole (average weight-for-age for
four rounds of quarterly measurements is less than 80 percent of the standard); and
third, for households for which preschool nutritional status is in the normal range, that
is, more than 80 percent of the standard weight-for-age. This indicator combines both
short- and long-term effects of malnutrition.
The results are shown in Table 20. The income elasticity for calories is 0.51 for
the total sample, 0.57 for households with malnourished children, and 0.44 for the
others. More than half the households had undernourished children, using the weight-
for-age indicator. These households had a 20 percent lower per capital caloric availability
than the rest of the households. The difference in terms of calories per adult equivalent
was 12 percent. The relatively small difference in dietary calories between the two
groups is consistent with the generally poor association between household food avail-
ability and child nutrition found in other studies.39 The coefficient or farm income ratio
is significant and similar for the three equations, as is the mean value for the ratio
itself. Cooking time per capital is significant for the sample as a whole and in the
subsample without malnourished children. Adult literacy in the present specification
is not significant, but it has a positive sign in the total sample and in the sample with
malnourished children.
These findings on the size of the income elasticity of dietary calories are consistent
with those from other rural areas.40 The differences in dietary calorie elasticities be-
tween those households with malnourished children and those without are noteworthy.
As mentioned earlier, similar differences are found when households are grouped by
low and high levels of caloric intake. They indicate a significantly higher propensity
to increase dietary calories among households where less food is available. Finally, on
the income effect, the farm income component appears to be responsible for most of
the income effect on consumption. This is a reflection of the strong orientation to
subsistence agriculture of households in which own-production contributes the major
part of the diet. The relationship between household dietary intake and child nutrition
will be analyzed in the next section.
Time spent on cooking by women, weighted by the number of household members,
is a significant factor in the level of dietary intake only for those households with higher
income, higher calorie intake, and higher nutritional levels, suggesting that at lower


39 See, for example, Eileen T. Kennedy and Bruce Cogill, Income and Nutritional Effects of the Commer-
cialization of Agriculture in Southwestern Kenya, Research Report 63 (Washington, D.C.: International
Food Policy Research Institute, 1987); Howarth Bouis and Lawrence Haddad, "A Case Study of the Com-
mercialization of Agriculture in the Southern Philippines: The Income, Consumption, and Nutritional Effects
of a Switch from Corn to Sugar Production," International Food Policy Research Institute, Washington,
D.C., 1988 (mimeographed); and Joachim von Braun, David Hotchkiss, and Maarten Immink, "Nontradi-
tional Export Crops in Traditional Smallholder Agriculture: Effects on Production, Consumption, and Nutri-
tion in Guatemala," International Food Policy Research Institute, Washington, D.C., 1988 (mimeographed).
40 Ibid.









Table 20-Factors influencing dietary calories
Households Households
with Without
Sample Total Sample Malnourished Sample Malnourished
Independent Variable Mean Sample Mean Children Mean Children

Intercept ... 2.77 ... 2.36 ... 3.29
Logoftotal household income 1,970 0.51** 1,684 0.57** 2,287 0.44**
(annual Rs/capita) (10.29) (7.12) (6.82)
Farm income ratio 0.75 1.02** 0.74 0.99** 0.76 1.03**
(7.83) (5.20) (5.75)
Cookingtime 0.62 0.33** 0.59 0.13 0.64 0.41**
(2.88) (0.65) (2.97)
Adult literacy (percent of 42.4 1.25-03 42.2 1.91-03 42.6 -5.06
literate adults) (1.07) (1.18) (0.3)
R2 (adjusted) ... 0.66 ... 0.67 ... 0.63

Notes: Households with malnourished children are those with average weight-for-age of preschool children less
than 80 percent of the standard over four quarterly measurements. The anthropometric standards used
in this study are those recommended in World Health Organization, Measurements of Nutritional Impact
(Geneva: WHO, 1979). Figures in parentheses are t-values. Ellipses indicate a nil or negligible amount
**Significant at the 0.01 level.

income levels, where the bulk of the nutrition problems are likely to be, it is the time
spent on production activities that appears to be the primary limiting factor. It was
shown earlier that even though the marginal product of labor for the small farm
households is as high as that for larger farms, their time spent on production is lower
than that on larger size farms. Could it be that the poorer level of nutrition of the
working population is a contributing factor in the time allocation pattern of the smaller
farm households? This question is outside the scope of this report. The implication
here is that for all households, food consumption is adversely affected by deforestation
because additional collection time detracts from food production activities. This is
especially pronounced for the low-income households with smaller farms. For the rest
of the households, production time is a less significant factor, but time spent in other
essential consumption-related activities, such as cooking, becomes increasingly important


Effects on Nutritional Status
Nutritional status of all household members is assessed using heights and weights
measured quarterly over the survey year. For children up to 18 years, the World Health
Organization (WHO) growth standards for international use are used to determine
those who are below the expected growth curve distributions.41 This section examines
the nutrition situation of children in the study area and some of the main determinants
that are of relevance in the present study.
Anthropometric measurements for two age groups, children less than 6 years,
designated as preschoolers, and those 6-18 years old, are shown in Tables 21 and 22.
For children less than 6 years old, results are compared with those of the 1975 Nepal
Nutrition Survey. Both low weight- and height-for-age are more prevalent in the present



41 World Health Organization, Measurements of Nutritional Impact (Geneva: WHO, 1979).










Table 21-Anthropometric measurements for preschool children in 1975, Western Region, and in 1982/83
in study area

Total of
Sample Syangja District, Tanahun District, Gorkha District,
1975 Obser- Bagkhor Panchayat ManapangPanchayat Chhoprak Panchayat
Measurement Survey nations Ward 2 Ward 8 Ward 5 Ward 8 Ward 1 Ward 7
(percent)
Weight-for-age
More than 75 percent of standard 51.4 48.9 33.0 50.8 64.8 30.4 72.4 41.2
60-75percent 44.8 39.8 51.0 40.0 25.0 53.9 25.0 44.1
Less than 60 percent 3.8 11.3 16.0 9.2 10.2 15.7 2.6 14.7
Height-for-age
More than 90 percent of standard 44.9 33.6 28.0 40.0 56.7 19.1 31.9 30.7
85-90 percent 36.8 35.1 36.0 23.1 22.8 41.6 44.8 35.7
Less than 85 percent 18.4 31.3 36.0 36.9 20.5 39.3 23.3 33.6
Weight-for-height
More than 90 percent of standard 60.5 69.1 61.0 67.7 75.0 61.8 88.0 60.8
80-90 percent 34.1 22.6 26.0 24.6 15.9 28.1 10.3 30.1
Less than 80 percent 5.4 8.3 13.0 7.7 9.1 10.1 1.7 9.1
Less than 85 percent of height-for-
age with less than 80 percent of
weight-for-height 3.7 6.2 9.0 6.2 5.7 6.7 1.7 7.7

Sources: 1975 results for the Western Region are taken from U.S. Agency for International Development, Office of Nutrition Development Support Bureau, Nepal
Nutrition Status Survey, January-May 1975(Washington, D.C.: USAID, 1975); those for the sample observations for this study are from Nepal, Agricultural
Projects Service Center; the Food and Agriculture Organization of the United Nations; and International Food Policy Research Institute, "Nepal Energy
and Nutrition Survey, 1982/83," Western Region, Nepal.
Notes: Preschool children includes children from 6 months to 6 years of age. The anthropometric standards used in this study are those recommended in
World Health Organization, Measurements of Nutrition Impact (Geneva: WHO, 1979).









Table 22-Anthropometric measurements for children 6-18 years old, 1982/83, study area only

Total of
Sample Syangja District, Tanahun District, Gorkha District,
Obser- Bagkhor Panchayat Manapang Panchayat Chhoprak Panchayat
Measurement nations Ward 2 Ward 8 Ward 5 Ward 8 Ward 1 Ward 7

(percent)
Weight-for-age
More than 75 percent of standard 17.4 3.3 26.2 25.8 9.1 22.7 11.4
60-75 percent 46.3 37.4 47.5 46.8 53.0 53.9 37.7
Less than 60 percent 36.3 59.3 26.3 27.4 37.9 23.4 50.9
Height-for-age
More than 90 percent of standard 22.3 6.6 28.7 66.1 21.2 26.9 14.0
85-90percent 36.0 41.8 31.3 29.0 31.8 43.3 40.4
Less than 85 percent 41.7 51.6 40.0 4.9 47.0 29.8 45.6
Weight-for-height
More than 90 percent of standard 75.6 48.3 86.2 85.5 81.8 83.7 60.6
80-90percent 21.2 45.1 10.0 13.7 15.9 14.9 33.3
Less than 80 percent 3.2 6.6 3.8 0.8 2.3 1.4 6.1
Less than 85 percent of height-for-
age with less than 80 percent of
weight-for-height 2.0 5.5 2.5 0.0 2.3 0.0 3.5

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations; and International Food Policy Research Institute,
"Nepal Energy and Nutrition Survey, 1982/83," Western Region, Nepal.
Note: The anthropometric standards used in this study are those recommended in World Health Organization, Measurements of Nutrition Impact (Geneva:
WHO, 1979).









study than in the earlier one. About 11 percent are severely underweight for age and
more than 8 percent are severely underweight for height. This compares with about
4 and 5 percent respectively for the 1975 survey. Low height-for-age is found in 31
percent of the children observed in this study compared with 18 percent in the earlier
one. These figures suggest a somewhat higher level of malnutrition than in the earlier
period. Though similar standards were used in both surveys, it is not clear whether
the differences are due to sampling differences, or if there has been a decline in
nutritional status since the time of the earlier survey. Assuming that no major measure-
ment problems were involved in either of the surveys and that the sample from the
national survey was representative of the area, the evidence could lend support to the
secular decline hypothesis. This conclusion is also supported by a somewhat larger
average landholding size reported for this sample than the average figures for the area.
No comparable figures from the earlier period for the older children are available.
Although wasting (low weight-for-height) is less prevalent in the older group, weight-for-
age and height-for-age indicators suggest that there is continued slow growth and no
evidence of catch-up growth.
In order to examine the factors associated with child malnutrition, two probit
models are estimated for the age group below six years.42 The two dependent variables
are low height-for-age-an indicator of longer-term malnutrition-and low weight-for-
height-an indicator of short-run malnutrition. The dependent variable is equal to 1
if the child's nutrition is in the normal range and 0 if malnourished. The cutoff points
used are 90 percent of the WHO reference height-for-age and 95 percent of the reference
weight-for-height.
Explanatory variables for child nutrition are based on household income, agricultural
characteristics, household size, ethnic characteristics, relative work load of women in
agriculture, the amount that older children work in collection and agricultural activities,
and deforestation. These characteristics are expected to influence the direct determi-
nants of nutritional status such as the household's and children's diets and time available
for child care. The ethnic variable is introduced to capture differences, if any, between
the Tibeto-Burman and the Newar-Brahmin groups. Two of the six sites were exclusively
of Tibeto-Burman extraction, while the rest were from the Newar-Brahmin groups.
The deforestation variable is expected to capture the combined effects via the time
allocation and agricultural productivity effects that have been found to be associated
with it. The variable on children's work load is included because if older children are
extensively engaged in work, less child care may be available for the younger children.
A similar interpretation may be made for the share of women's work load in agriculture.
Household income and landholding are expected to influence nutritional status via
their impact on diets, quality of housing, health, water supply, and sanitation. These
indicators may also be expected to be associated with the level of education of household
members.
The equation estimated is

NS = F(Df, HS, Y, Tcrl, ETHNIC, WR, ChL), (11)
where
NS = height-for-age more than 90 percent of the reference and
weight-for-height more than 95 percent of the reference
are 1; others are 0;

42 In this model the dependent variable represents the probability of overcoming low nutritional status.









Df = deforestation (wards where time per trip is high are 2,
others are 1);
HS = household size;
Y = household income (in annual rupees per capital ;
Tcrl = total cropped area (in hectares);
ETHNIC = ethnic group (Tibeto-Burman groups are 1, others are 0);
WR = women's share of household labor in agriculture; and
ChL = time spent by older children in collection activities,
grazing of cattle, and agricultural production (in hours
per day).
The results indicate that for preschool children, the longer-term nutrition indicator,
height-for-age, is positively influenced by increasing per capital income in the household,
even though the size of the coefficient is small and is negatively influenced by increasing
household size (Table 23). Deforestation has a significantly negative influence on this
nutritional status indicator. The more time spent by children working in collection
activities, livestock grazing, and field work, the lower the level of nutrition of their
preschool-age siblings.
Another indicator of nutritional status uncovered in the study is the ethnic back-
ground of the children. The ethnic groups classified as Tibeto-Burman in the sample
appear to be taller than would be expected for people of their group using the height-for-
age indicator after controlling for household income and other characteristics.43 The
others, which are largely from the Newar and Brahmin groups, would normally be
taller, but they clearly rated below the Tibeto-Burman groups on this indicator. The
reasons for the anomaly are unclear, but a future study might examine the effects of
cultural differences in the area on nutritional status.
Poor nutritional status of preschoolers in the short term, as reflected in their
weight-for-height, is also found to be significantly associated with deforestation, with
high deforestation areas showing reduced weight-for-height. The ethnic group variable
showed a similar result as the height-for-age equation, with the Tibeto-Burman groups
having better child nutrition. Overall, the degree of predictability of weight-for-height
is relatively lower than that of height-for-age. The time allocation variables-relative
women's work load and degree of work involvement by older children-are found to
be negatively associated with preschoolers' weight-for-height.
The results point to five main observations: first, high deforestation areas have
poorer child nutrition when controlling for other household characteristics; second,
household income is positively associated with good long-term nutrition indicators for
preschool children; third, household size has a significantly negative effect on the
longer-term nutrition indicator, height-for-age, but is insignificant for short-term nutri-
tional status; fourth, the ethnic groups classified as Tibeto-Burman have better child
nutrition than other groups; and fifth, the degree of involvement of older children in
collection, grazing, and agricultural activities is negatively associated with child nutri-
tion, and so, less significantly, is the relative workload of women in agricultural production.


43 The Tibeto-Burman groups in the sample were Gurungs and Magars. These ethnic groups usually reside
at higher altitudes in the hill and mountain regions of Nepal and are usually shorter than the Newar and
Brahmin groups.









Table 23-Determinants of preschool children's nutritional status (probit
model)
Height- Weight-
Variable for-Age for-Height

Intercept 1.92 2.14*
(1.53) (2.01)
Household size -0.25**
(2.77)
Household income (Rs/person/year) 0.0005**
(2.83)
Total cropped area (hectares) 0.27 0.15
(1.63) (1.25)
Ethnic group (Tibeto-Burman = 1, 1.34** 0.76*
other = 0) (3.34) (2.33)
Deforestation (high = 2, low = 1) -0.92* -0.68*
(2.43) (2.09)
Ratio of women's agricultural work ... -2.22
(1.59)
Children's work (hours/day) -0.29* -0.17
(2.08) (1.46)
Significance of chi2 0.89E- 6 0.39 01

Notes: Figures in parentheses are t-ratios. The ellipses indicate t-ratios less than 1. Preschool children include
those from 6 months to 6 years old. Cutoff points for probit analysis for preschool children are 90 percent
of the standard height-for-age and 95 percent of the standard weight-for-height. The anthropometric
standards used in this study are those recommended in World Health Organization, Measurements of
Nutrition Impact (Geneva: WHO, 1979).
*Significant at the 0.05 level.
**Significant at the 0.01 level.


Implications for Area-Level Programs
The results of this analysis indicate that overall there are strong possibilities that
a labor constraint, especially of women's labor, is limiting the production and consump-
tion potential in the area and that deforestation further aggravates this problem. In
terms of policies and programs to alleviate the problems, this suggests that the payoff
from labor-saving technologies for women, in conjunction with production technologies
that raise labor productivity, could be high. Labor-saving technologies would be espe-
cially relevant if desired agricultural improvements require an expansion of labor input
in production and postharvest processing. Labor-saving technologies for women would
also benefit the nutrition of preschool children.
However, within an area, there is the potential for several alternative types of
constraints to be present for different households. Where there is a significant labor
constraint, there are also likely to be a number of farmers who face a land constraint
and who would benefit from an increase in off-farm activities or from the opportunity
to increase productivity from their land. Therefore, characteristics in the study area
were examined to identify areas and households with different types of constraints to
assist area planners in deriving an appropriate development package for different
localities. Whereas for most areas inputs for increasing crop yields are desirable, it is
also necessary to consider labor-saving technologies for labor bottlenecks, including
those created by essential activities such as collection of fuel, fodder, or water, or food
processing.









Another aspect of area-level programming is the issue of targeting publicly subsidized
inputs and services. It has been widely acknowledged that, unless agricultural technology
and input delivery institutions explicitly consider the needs of small farmers, benefits
are likely to reach only a few well-off farmers in the initial stages of adoption of new
technology or when there is a scarcity of these inputs and services."
In keeping with the previous arguments, it is considered desirable to examine the
characteristics of nutritionally at-risk households to facilitate the targeting of alternative
program components to them. In order to identify these households, several alternative
criteria could be used. It is generally accepted that preschool-aged children form the
most vulnerable segment of the population. However, older children represent a con-
siderable opportunity for catch-up growth, and nutrition of the working-age population
has direct productivity implications with indirect consequences for other household
members. Thus, nutrition of all household members is relevant in its own way. In
order to test which single indicator is positively correlated with various nutritional
status indicators for all other groups, a simple test is conducted using all three (height-for-
age, weight-for-age, and weight-for-height) for children below 6 years old and children
6-18 years old and the weight-for-height measure for adult men.45 The results, shown
in Table 24, indicate that adult men's weight-for-height is positively correlated consis-
tently with all indicators of the younger age groups. Whereas preschool children's
nutritional status is positively correlated with men's weight-for-height, it is not corre-
lated with the nutritional status of older children. Based on these results, and also
because about 30 percent of households did not have any preschool-aged children,
men's weight-for-height is used as the criterion for selecting at-risk households.

Table 24-t-values for differences in anthropometric measures for alternative
nutritional indexes
Nutritional Index
Anthropometric Measure NSTAT I NSTAT2 NSTAT3

Children less than 6 years old
Height-for-age 3.8 -0.8 1.6
Weight-for-age 8.0 0.6 1.7
Weight-for-height 5.8 2.4 1.7
Children 6-18 years old
Height-for-age -0.4 -0.6 1.4
Weight-for-age 1.5 4.3 3.1
Weight-for-height 0.1 3.8 1.4
Adult
Weight-for-height (males) 1.8 0.5 12.3

Notes: Anthropometric measures are given as a percentage of the standards recommended in World Health
Organization, Measurements of Nutrition Impact (Geneva: WHO, 1979). NSTAT 1 is households with
children 6 months to 6 years old who are less than 80 percent of weight-for-height and 85 percent of
height-for-age. NSTAT 2 is households with children 6-18 years old who are less than 80 percent of
weight-for-height and 85 percent of height-for-age. NSTAT 3 is households with adult males who are less
than 80 percent of weight-for-height.


44See G. Goodell, "Bugs, Bunds, Banks, and Bottlenecks: Organizational Contradictions in the New Rice
Technology," Economic Development and Cultural Change 33 (October 1984); and Peter B. R. Hazell and
C. Ramasamy, Green Revolution Reconsidered: The Impact of High-Yielding Rice Varieties in South India
(Baltimore: Johns Hopkins University Press for the International Food Policy Research Institute, forthcoming).
45 The measure for adult women could not be used because records of their physiological status were
inadequate.








Within the at-risk group thus selected, three main subgroups are identified using
principal components analysis.46 These are land deficit, labor deficit, and excess live-
stock. Based on these earlier results, cut off points for these indicators are defined as
less than 0.2 hectare (4 ropani) of land per capital, household size-to-worker ratio
greater than 1.3, and less than 0.25 hectare per livestock unit kept by a household.47
Generalizing from these results to the entire population, it is possible to identify a total
of eight groups with various combinations of constraints. These are indicated in Table
25. From the prevalence of different constraints, the three district samples in the study,
as represented by the panchayatsof Bagkhor, Manapang, and Chhoprak, show a differing
pattern of constraints. In Bagkhor, there is predominantly a land constraint, with smaller
degrees of excess livestock and labor constraints. In Chhoprak, there is mainly a labor
constraint, again with smaller degrees of excess livestock and land constraints. In
Manapang, there is predominantly an excess of livestock, with no land constraint and
some labor constraint. In this case, a secondary labor constraint may arise from the
need for additional labor for grazing of livestock or for fodder and grass collection.
In predominantly land-constrained areas, deforestation is likely to be more severe
in that area-expansion possibilities are probably minimal. Whereas intensification of
input use to raise agricultural productivity in such areas is clearly indicated, indirect
labor bottlenecks may also exist because the labor demand for collection activities is
high. In the Bagkhor area, however, collection time was relatively low because the
planting of private trees for fuel and fodder was expanded. Agroforestry efforts are
most likely to succeed in such areas. Other areas, where landholdings are more adequate
but labor is constrained, are likely to face ongoing deforestation. In these cases, input
use should be intensified to stem further area expansion, especially on steeper slopes.
Here tackling labor bottlenecks and intensifying input use in agriculture should both
be integral parts of a program to improve production and consumption potentials of
households.
Finally, some of the production characteristics of different household types and
locations are examined in order to determine where improvements may best be made
and the kinds of programs that may be adopted. In order to simplify the tabulation of
all observations, the eight original household types are regrouped into four. In this
process, livestock numbers in excess of cultivated land capacity are assumed to add to
the labor requirement. If there is no existing labor constraint, then this type of constraint
is not dropped from consideration. The new types are N1, which has both land and
labor constraints (former types 1 and 3); N2, which has a land constraint only (former
types 2 and 4); N3, which has a labor constraint only (former types 5 and 7); and N4,
which has neither land nor labor constraints (former types 6 and 8).
The crop production characteristics of each of these four groups are shown in Tables
26 and 27. Overall, N2 households-those with only a land constraint-achieve the
best production picture. They have the lowest amount of land under each crop but the
highest yields and the largest amounts of household and wage labor inputs, fertilizer,
and animal labor. Nevertheless, their annual per capital incomes are low. It is N3
households-those with a labor constraint-that have the lowest average production
performance in terms of yields, labor, and fertilizer inputs. But because they have the
largest amount of land under cultivation, they achieve the highest level of farm produc-

46 Food and Agriculture Organization of the United Nations/Nepal, Agricultural Projects Services Center,
District Planning Manual for Nepal (Kathmandu: APROSC, 1984).
47 Food and Agriculture Organization of the United Nations/Nepal, Agricultural Projects Services Center,
District Planning Manual for Nepal.










Table 25-Types of household constraints possible

Landholding Landholding
Less Than Household Size- Less Than 0.25
0.2 Hectares to-Worker Ratio Hectares Per
Type of Constraint PerCapita More Than 1.3 LivestockUnit

1 X X X
2 X ... X
3 X X ...
4 X
5 ... X X
6 ... ... X
7 ... X ...
8 ... ... ...
8




tion and household per capital income. The N1 households-those with both land and
labor constraints-have an average-to-low production picture and are not able to achieve
the productivity levels of N2 households, which also have a land constraint but no
labor constraint. Both types of land-constrained households have more migrant workers
and income from remittances.


Table 26-Agricultural production by type of constraint, by major crops,
1982/83

Total Total
Constraint Number of Family Wage Total
Type/Crop Households Area Yield Labor Labor Labor
(hectares) (kilograms/ (days/hectare)
hectare)
Type Na
Wheat 14 0.37 987 155 18 173
Maize 32 0.49 1,078 146 15 161
Ragi 31 0.33 781 259 15 274
Paddy 26 0.35 1,857 266 49 315
Type N2b
Wheat 13 0.13 1,322 234 16 250
Maize 24 0.42 1,169 196 35 231
Ragi 23 0.30 987 353 13 366
Paddy 19 0.32 2,058 285 49 334
Type N3c
Wheat 16 0.47 825 137 31 168
Maize 31 0.74 1,082 147 33 180
Ragi 25 0.54 956 234 17 251
Paddy 28 1.05 1,951 254 54 308
Type N4d
Wheat 10 0.47 929 141 12 153
Maize 26 0.99 1,447 200 32 232
Ragi 25 0.72 1,176 314 14 328
Paddy 24 0.86 1,612 282 40 322

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
a Type NI has both land and labor constraints (N = 33).
b Type N2 has a land constraint only (N = 26).
c Type N3 has a labor constraint only (N = 31).
d Type N4 has no land or labor constraint (N = 28).










Table 27-Household characteristics by type of constraint, 1982/83
Type of Constraint
Land No Land
and Land Labor or Labor
Household Characteristic/ Labor Only Only Constraint
UnitofMeasure (N1) (N2) (N3) (N4)

Household size (average number) 6.9 5.7 6.6 7.1
Adults (working age) 2.8 3.4 2.8 4.7
Children (up to 15 years) 3.8 2.1 3.5 2.2
Migrant workers (number) 0.6 0.6 0.4 0.4
Land cultivated (hectares)
Upland area 0.5 0.4 1.1 1.3
Lowland area 0.3 0.3 1.0 0.9
Cropping intensity (percent)
Upland intensity 159 169 141 143
Lowland intensity 170 156 167 157
Livestock units 4.9 5.7 8.1 9.3
Annual total income (Rs/capita) 1,421 1,427 2,573 2,394
Farm production 715 654 1,813 1,765
Employment 171 133 85 72
Remittances 382 442 138 323
Farm by-products 166 215 403 257
Food consumption
Calories (per capita/day) 1,431 1,758 2,926 2,449
Protein (grams/capita/day) 41.8 50.4 78.7 68.5
Anthropometry (percent of children)
Children less than 6 years
Less than 85 percent height-
for-age and less than 80
percent weight-for-height' 8.6 5.8 4.6 4.2
Children 6-18 years
Less than 85 percent height-
for-age and less than 80
percent weight-for-height 4.8 1.0 0.0 1.0

Source: Nepal, Agricultural Projects Service Center; the Food and Agriculture Organization of the United Nations;
and International Food Policy Research Institute, "Nepal Energy and Nutrition Survey, 1982/83," Western
Region, Nepal.
Note: The anthropometric standards used in this study are those recommended in World Health Organization,
Measurements of Nutritional Impact (Geneva: WHO, 1979).
a This measurement combines stunting and wasting to indicate both long- and short-term malnutrition.



Though their overall income levels are comparable, both food consumption and
child nutrition levels are significantly lower for N1 households, which may be attributed
to the higher work load of adults. In addition, the nutrition of the 6-18 year olds is
the worst in this group, perhaps because their work load may also be high. This age
group does best in N2 households, which may also be linked to their relative work
loads. Also, preschool nutritional status for N2 households is closer to that of the
higher-income N3 and N4 households.
Thus, it is N1 and N3 households that need improvement, both in production and
nutrition. These are essentially labor-constrained households, either with or without
a land constraint. Deforestation would further exacerbate the situation for the groups
that are worst off.









7


CONCLUSIONS AND POLICY IMPLICATIONS

The area described in this report is typical of the hill areas of Nepal, which are
characterized by low levels of agricultural productivity, limited adoption of new technol-
ogy, and a high rate of natural resource degradation, particularly deforestation. The
study area is typical of many rural areas that have been bypassed by improvements in
agricultural technologies and related benefits during the past two decades. However,
in one respect it may be quite different. The pace of deforestation here has direct
repercussions on the vast foodgrain belt of the plains. The evolution of a sustainable
and stable agriculture in the hill areas may be, therefore, necessary for the welfare of
more than these areas alone.
A policy prescription for such areas is simply to leave things alone, using them as
a reservoir for out-migrating labor and letting remittances into the area provide the
means to an effective demand for the inflow of food and other products needed to
maintain their standard of living. Despite high out-migration-an average of one migrant
worker for every two households-remittances have so far been small and have not
contributed to raising food consumption of the households. This may change when the
workers return, either with accumulated savings or retirement pensions. Alternatively,
efforts in recent years have focused primarily on reforestation and other measures
designed to preserve the natural resource base.
These approaches alone, however, are inadequate. Bajracharya has argued persua-
sively that it is initially the low productivity of agriculture in these areas that promotes
deforestation.48 According to his analysis, the inability to provide for food needs from
the returns to agriculture provides an incentive for increasing land under cultivation.
The results confirm that agriculture in the area is highly dependent on labor, and there
is increasing competition between productive and other uses for labor, possibly con-
tributing to inefficient methods of collection of forest products, primarily fuelwood.
(For example, concentrating collection efforts on the most accessible forest, rather than
spreading out the collection efforts, can promote rapid deforestation, despite the overall
adequacy of forest resources in the region as a whole.)
In other words, the low agricultural productivity derived from a high dependence
on human labor contributes to the relatively higher population pressure on arable land
and to the erosion of the natural resources that are essential for maintaining the
production base.49 Clearly, a vital question in this context is whether the introduction
of improved agricultural technologies in this area will reduce or increase the pressure
on land and the encroachment on forest lands. Most likely the answers will differ
depending on the crops and type of technologies promoted, the institutional setup used
to make them available to smallholders, the pace of infrastructure development in the
region that can help to channel resources to the off-farm sector, and the presence of
forestry and related efforts.


48 Bajracharya, "Fuel, Food, or Forest?"
49 Consultative Group on International Agricultural Research, Technical Advisory Committee, Sustainable
Agricultural Production.








Results of this study suggest that deforestation has adverse effects on agricultural
production, food consumption, and nutrition that occur as a result of the additional
work loads entailed in the collection of essential forest products on which the households
depend, primarily fuelwood, fodder, and grass. In the longer run, water availability for
both household and irrigation purposes could also be affected. The adverse effects on
irrigation potential will of course be far-reaching geographically.
Women, in particular, have a high and increasing work load as deforestation expands,
and this burden reduces labor in agriculture. Limited availability of labor and low
substitutability between women's and men's labor can lead to a much larger overall
decline in labor input in agriculture with deforestation. This is a seldom acknowledged
factor in the decreasing agricultural productivity of hill agriculture, where the high rate
of out-migration has usually been interpreted as a sign of an underutilized work force,
when instead it is the low returns to labor in agriculture that encourages out-migration.
The evidence indicates that deforestation is directly and indirectly linked with levels
of food consumption and nutrition for the households.
The following observations have emerged from this study:
Deforestation, which represents a 1 percent increase in time spent for collection
of a unit of fuelwood, leads to a reduction in fuel consumption of 0.3 percent
but an increase in the total time required for collection of 0.6 percent.5
The sites in the study sample where deforestation is extensive require 75 percent
more time spent for collection per load of fuelwood, which suggests a 45 percent
increase in total collection time. With a similar response for other forest products
that are collected on a regular basis, such as fodder and grass for animals, and
assuming that men, women, and children increase their time in proportion to
their existing time allocation, women would need an additional 1.13 hours per
day to collect these products. This estimate is consistent with the increases in
actual time spent by women.
Estimation of the change in household labor supply to agriculture, given the above
magnitude of deforestation, shows a reduction in women's field labor of about
1.5 hours per person per day. Further, this reduction is accompanied by a reduction
in men's agricultural labor of about 0.8 hours per person per day. As a result,
household labor per hectare decreases by about 40 percent in high deforestation
areas. The small increase in wage labor that occurs does not compensate for the
reduction in household labor input.
Estimates of physical production functions show the significance of labor input
in crop output. Women's labor has the highest marginal product for dry-season
crops. On a seasonally adjusted basis, the dry season is also when more time has
to be spent on collection of forest products. This suggests that there are losses
in agricultural productivity and real income with time allocation patterns that
accompany deforestation. This is likely to be only partly offset by higher seasonal
out-migration and income from such employment.
Food consumption, in terms of caloric consumption and the ratio of rice calories
to cereal calories, is a significantly positive function of household income (the


50 All figures are based on quarterly surveys, which have been aggregated to represent time allocation on
an annual basis.









farm income component is the only statistically significant component) and food
preparation time. Food preparation time, however, is a positive function of fuel-
wood used and a negative function of the total collection time, suggesting these
factors may also influence food consumption via less time spent on cooking and
food preparation. This is in addition to the effect of deforestation on agricultural
production and hence on household farm income.
Preschool child nutrition is improved by raising household income, and it is lower
in areas with higher levels of deforestation. An increase in household size and
in children's work is also associated with inadequate nutrition of preschool chil-
dren. Controlling for other household characteristics, the Tibeto-Burman ethnic
groups have better child nutrition than other groups.
An examination of the characteristics of the nutritionally at-risk households con-
firms that these are more likely to be labor constrained than land constrained,
and they have lower levels of agricultural productivity.
The results clearly show that under current productivity, households continually
need to expand cultivated area. To the extent that this contributes to deforestation,
this expansion reduces household income, food consumption, and nutritional status.
Under these circumstances, relying only on out-migration in these areas is unlikely to
improve living conditions or stem the environmental degradation through deforestation.
It is also questionable whether reforestation efforts by themselves will suffice, though
this would be a useful question to examine empirically. In the past, settlement schemes
in the tarai have drawn a fair share of settlers from the hill areas. But if deforestation
continues at the present scale, en masse movement of population from the hills may
occur at some future point in time. It seems evident that there is a need to increase
agricultural productivity, preferably on soils that are better endowed. This would con-
tribute to improvements in household welfare and reduce deforestation.
Another policy option that has frequently been proposed for the development of
these hill areas is the growth of horticulture production.51 The rationale stems from
both the agroclimatic characteristics of the hill regions of Nepal and the poor road
transport facilities that prevail. Products with high value to weight are consequently
appealing. An IFAD study recommends the following steps be taken: a comprehensive
ecological classification of areas for production of specific horticultural products, a
survey of national and international markets, a research program focused on selected
species, human and physical capital development to support the extension effort, re-
search on desirable processing technologies, and support of commercial enterprises for
setting these up.52
For smallholders' horticultural production to be successful would require not just
an outlet where the products would be competitive with existing suppliers, but also
an extremely efficient marketing system in the hills that would be readily accessible
to the scattered smallholders and would avoid the high perishability inherent in mar-
keting of horticultural products. Given the largely subsistence orientation of agricultural
production at present in these areas, the degree of rural market development is too
severely limited to be able to handle a large outflow of horticultural produce. Public
marketing is probably not a good idea, as the degree of investment required and the

51 International Fund for Agricultural Development, Report of the Special Programming Mission to Nepal
(Rome: IFAD, 1979), p.38.
52 Ibid.









relative inefficiency of this sector will likely lead to monopolistic tendencies that will
be unproductive in the long run.
The issue of the need for efficient rural markets to handle the horticultural products
is also related to the need to be competitive with the existing suppliers to the major
metropolitan centers of the subcontinent, such as Kathmandu, New Delhi, Calcutta,
Islamabad, and others. These centers already have a well-established network of
suppliers, so that products from the Nepal hills would have to be produced very cheaply,
or they would have to aim for the high price end of the horticultural products market.
The latter option is probably preferable but will require many years of research on
commodity and market development. Such an effort clearly needs to begin.
In the meantime, efforts for increasing productivity of traditional crops in hill regions
need to continue. This is important not only for maintaining consumption without
substantial area expansion, but also for promoting an expansion away from largely
subsistence production. Production for the market should promote the growth of rural
markets in the hill areas, providing the basis for the growth of a privately financed
rural services infrastructure. The growth of such a base will facilitate the gradual shift
into more specialized production of horticultural and livestock products.
The results of the study suggest that in order to increase agricultural productivity,
a combination of efforts for increasing input use and reducing labor bottlenecks is
necessary. These findings are consistent with the IFAD recommendation promoting
the use of selected labor-saving technologies such as power tillers as an alternative to
oxen for easing labor bottlenecks and for spreading employment on small farms more
evenly.53 These tillers are relatively inexpensive and simple to operate and maintain.
They could also reduce demand for work oxen, thus reducing human labor needed for
feed collection and grazing, and alleviating pressure on the environment. Other labor-
saving technologies for improvement in operations such as food processing or water
supply should also be explored, as well as agroforestry efforts that reduce collection
time for forest products such as fuel and fodder.




















53 Ibid.









BIBLIOGRAPHY

Acharya, Meena, and Bennett, Lynn. The Rural Women of Nepal: An Aggregate Analysis
and Summary of Eight Village Studies, vol. 2, part 9 of The Status of Women
in Nepal Kathmandu: Centre for Economic Development and Administration,
Tribhuvan University, 1981.
Adelman, Irma. "A Poverty-Focused Approach to Development Policy." In Development
Strategies Reconsidered 49-65. Edited by John P. Lewis and Valeriana Kallab.
Washington, D.C.: Overseas Development Council, 1986.
Alderman, Harold. The Effect of Food Price and Income Changes on the Acquisition
ofFood byLow-Income Households. Washington, D.C.: International Food Policy
Research Institute, May 1986.
Bajracharya, Deepak. "Fuel, Food, or Forest? Dilemmas in a Nepali Village." World
Development 11 (No. 12, 1983): 1057-1074.
Bouis, Howarth. "An Agricultural Sector Model for the Philippines: Identifying Localized
Shortfalls in Household Food Security and Evaluating Policy Options Using Data
from Ongoing Quarterly Surveys." Paper presented at IFPRI-Food Studies Group
workshop, Oxford, England, July 7-9, 1987.
Bouis, Howarth, and Haddad, Lawrence. "A Case Study of the Commercialization of
Agriculture in the Southern Philippines: The Income, Consumption, and Nutri-
tional Effects of a Switch from Corn to Sugar Production." International Food
Policy Research Institute, Washington, D.C., 1988 (mimeographed).
Braun, Joachim von; Hotchkiss, David; and Immink, Maarten. "Nontraditional Export
Crops in Traditional Smallholder Agriculture: Effects on Production, Consump-
tion, and Nutrition in Guatemala." International Food Policy Research Institute,
Washington, D.C., 1988 (mimeographed).
Cecelski, Elizabeth. "The Rural Energy Crisis, Women's Work, and Family Welfare:
Perspectives and Approaches to Action." Working paper prepared for World
Employment Research Programme, International Labour Organisation, Geneva,
1984.
Consultative Group on International Agricultural Research/Technical Advisory Commit-
tee. Sustainable Agricultural Production: Implications for International Agricul-
tural Research. Rome: Food and Agriculture Organization of the United Nations,
TAC Secretariat, October 1987.
Dasgupta, Subhachari, and Maiti, Asok Kumar. The Rural Energy Cris, Poverty, and
Women's Roles in Five Indian Villages. Geneva: International Labour Organisa-
tion, Rural Employment Policy Research Programme, 1986.
Denny, Michael, and Fuss, Melvyn. "The Use of Approximation Analysis to Test for
Separability and the Existence of Consistent Aggregates." American Economic
Review 67 (June 1977): 404-418.
Edirisinghe, Neville, and Ratnayake, R. M. K. "A Preliminary Analysis of the Determi-
nants of Nutritional Welfare in Sri Lanka," International Food Policy Research
Institute, Washington, D.C., January 1984 (mimeographed).








Evans, Margaret. "Change in Domestic Fuel Consumption in Central Mexico and Its
Relation to Employment and Nutrition." Working paper prepared for World
Employment Research Programme, International Labour Organisation, Geneva,
1986.
Food and Agriculture Organization of the United Nations. Pilot Study on Energy Use
and Nutritional Status at Farm Level in the Hills of Nepal. Report No. 1. Rome:
FAO, January 1984.
Food and Agriculture Organization of the United Nations/Nepal, Agricultural Projects
Services Center. District Planning Manual for Nepal Kathmandu: APROSC,
1984.
Gautam, Madhav. "Nutrition and Environment in a Nepalese Village." Paper presented
at the Fourth Asian Congress of Nutrition, Bangkok, November 1983.
Goodell, G. "Bugs, Bunds, Banks, and Bottlenecks: Organizational Contradictions in
the New Rice Technology." Economic Development and Cultural Change 33
(October 1984).
Hazell, Peter B. R., and Ramasamy, C. Green Revolution Reconsidered: The Impact of
High-Yielding Rice Varieties in South India. Baltimore: Johns Hopkins University
Press for the International Food Policy Research Institute, forthcoming.
Kennedy, Eileen T., and Cogill, Bruce. Income and Nutritional Effects of the Commer-
cialization of Agriculture in Southwestern Kenya. Research Report 63.
Washington, D.C.: International Food Policy Research Institute, 1987.
Kumar, Shubh K., and Hotchkiss, David. "Energy and Nutrition Links to Agriculture
in a Hill Region of Nepal." International Food Policy Research Institute,
Washington, D.C., 1985 (mimeographed).
Laufer, Leslie A. "The Substitution Between Male and Female Labor in Rural Indian
Agricultural Production." Center Discussion Paper No. 472. Economic Growth
Center, Yale University, New Haven, Conn., April 1985.
Lele, Uma, and Mellor, John W. "Technological Change, Distributive Bias, and Labor
Transfer in a Two-Sector Economy." Oxford Economic Papers 33 (November
1981): 426-441.
Makhijani, Arjun. Energy and Agriculture in the Third World: A Report to the Energy
Project of the Ford Foundation. Cambridge, Mass.: Ballinger, 1975.
Mellor, John W. "Determinants of Rural Poverty: The Dynamics of Production, Technol-
ogy, and Price." In Agricultural Change and Rural Poverty: Variations on a
Theme by Dharm Narain. Edited by John W. Mellor and Gunvant M. Desai.
New Delhi: Oxford University Press, and Baltimore: Johns Hopkins University
Press for the International Food Policy Research Institute, 1986, 21-40.
Mellor, John W., and Stevens, Robert D. "The Average and Marginal Product of Farm
Labor in Underdeveloped Countries." Journal of Farm Economics 28 (August
1956): 780-791.
Nepal, Agricultural Projects Services Center. Nepal: Foodgrain Marketing and Price
Policy Study. Kathmandu: APROSC, July 1982.









Nepal, Agricultural Projects Services Center, the Food and Agriculture Organization of
the United Nations, and International Food Policy Research Institute. "Nepal
Energy and Nutrition Survey, 1982/83." Western Region, Nepal.
Nepal, Department of Forestry/U.S. Agency for International Development, Forest
Resource Survey Office. "Forest Statistics of the Hill Region." Kathmandu, 1973.
Nepal, Ministry of Agriculture, Department of Agricultural Marketing Services. Agricul-
tural Statistics of Nepal. Kathmandu: MOA, 1977, 1983, and 1985.
Nepal, Ministry of Water Resources, Water and Energy Commission. "The Forests of
Nepal: A Study of Historical Trends and Projections to 2000." Kathmandu, 1983.
Nield, R. S. "Fuelwood and Fodder-Problems and Policy." Working paper for the
Water and Energy Commission Secretariat, Kathmandu, November 1985.
Oram, Peter. "Africa: An International Food Research Strategy." International Food
Policy Research Institute, Washington, D.C., October 1987.
Pradhan, Bina. The Newar Women ofBulu, vol. 2, part 6. The Status of Women in
Nepal. Kathmandu: Centre for Economic Development and Administration,
Tribhuvan University, 1981.
Schuler, Sidney. The Women of Baragaon, vol. 2, part 5. The Status of Women in
Nepal. Kathmandu: Centre for Economic Development and Administration,
Tribhuvan University, 1981.
U.S. Agency for International Development, Office of Nutrition Development Support
Bureau. Nepal Nutrition Status Survey, January-May 1975. Washington, D.C.:
USAID, 1975.
Wanmali, Sudhir. Service Provision and Rural Development: A Study of Miryalguda
Taluka. Research Report 37. Washington, D.C.: International Food Policy Re-
search Institute, 1983.
World Commission on Environment and Development. Our Common Future. Oxford:
Oxford University Press, 1987.
World Health Organization. Measurements ofNutritionalImpact Geneva: WHO, 1979.










RECENT IFPRI RESEARCH REPORTS (continued)


51s: GCOV ERNM ENI EXPEND ITULRES ONAGRRICUL TURE .4NDA.4GRICL'l TURAL GROWTH IN L4TIN
.-LAMERICA-, i.Llobr 1 85. by Victor .1 Ellas
4'9 LI'ESTO-A PRODUCTS IN THE IHIRLD I ORLD. PAST TRENDS AND PROJECTIONS TO /IW0
AND 2000. April I -''., by J. 5 Sa'ma and Patrick Ye-ung
48 RURAL HOLUEHOLD USE OF SERL ICES. 4 STUDY OF LI/R VAL GULLA 141 UA4. INDIA, March
1085. jy Sudhir Wannali
4-7 EL 'OL LVNG FOOD G.4FS IN THE .A/DI I EA ST. NORTH -ITRIC4. PROSPECTS AND POLICY
I/1 PL1_AT4IONS. Deci mber 1j84, hy Nat-il Khaldi
4., THE EFFECTS ON INCOMEf DISTRIBURITNAND NUTRITION OF AL TERNA1 'E RICE PRICE
PO LIC/ES IvN FH414. ND, Novembc-r 1',I4, ty Prasarn Tralratvorakul
4S THE EFFECTS OF THE EGYPTIA.N FOODT RATION .AN)n sU S/I SYS-IF4A ON INCOME DISTRI/
BU.TION ANO CONSUtfIPTION. Jul, I, '4, b, Harl.-1 Aldernman and Jracnim vn-.n Braun
44 CONStTR -INTS ONA hTN' t 41 SFOU)I' 4NI AhFL 'FRA- _EXPt. XpRI 1, April ii o1, byv ichaeil 'chlui r
43? CLOSIN G THF CFRJLA/ G4P KITH FRI A/ .4-AD FOOCDAij lanuarn 19'84. by Bar bara Huddleston
42 THE EFFECTS OFr f.-COD PRICl 4AN!' 'It IS/,) PuOlCiF.5 ON EGYPTIAN .GRICUL TURE.
Iri'.loeniber 1) 8'3. \.\ I..,a himh vlni Braun aid Harrwil de Hari-l
41 RURAI GROliTH17 LINKA.4G;ES HO')tFHOL/'L, E.P F'.DITHRE PATTERNS IN MIAL YYS/A AND
NIGER/I S"p., mb_. r ij83 b, F'iP r B. R Ha:P!l and Aild,, P.-.Ill
40 FOOD (SUBI'DIES IN/ EG) P 7: THEI/ IrA 1'P4CT ON CORI IGN; t1 CH.AIGE AND TRADE. Au ust
1'83. t.v i-.ranr %. S inLj
:'. THF Ii(RLD RiCE MA.4RKFT: STRU-CTULRE. CONDI'CT. AND PERFORMANCE. June 1'83, by
Arrmar ,lainv.alla and St.rpnen IHa~, is
3P POL1/C .ViOLELIN.G I'F 4 DU4L GRAIN .'tRK- I/Hi .lASL OF ItHF.l41 1 IN /ND/A, May
l'-'J 3. L., R.ij Kri.hria irld Alay Chlhbber
3" SERVICE 'RO il7Oi, 4ANP[ RLR1. OIf L I OPMAENT It/ INDII4 .4 i fID)Y' O' .11tR} ALC/UL
141C'L A. February, I '-:3, ty ,u.dhir Wanmal
3 .4 GRICUL' TURE .4AND ECONO.I/C GROi'7TH IN .AN' OPEN CONO41Y: THI CASE OF ARGFN
TIN4, [.ec, nimb,-r IO82, t,' Dornmin., Civall.v and 'iair MNin.Jlal
35 POLK. OPI !O/PIN' /I / /Hf- GRAIN ECONOfl } Oi.F THF EII OPEN COM.lIMUllN/Ti': LI PLICA
T/ON'S FOR DEELFOP/ING COUNTRIES, rln.enmber 1''8. by 1' lrch KoI-sitr
34 EGYPT'S FOOD 5(BSWD)Y At/[ R1A.4TIONIN.; SYSITE 1 4 I itRIPTION, i.cirLber 1 )82, by
Harold Ald.jrman. Joiilahiirrn .rn Braun, and r &hmn: d Sakr
33 AGRIC(UL IUR4L GROlt TH AND- I/NUSIRHIA41 PERFORMIANLZ IN I/N/A, IUcinotr 1'82, by
IC Hangaralar!
32 F/ -fl/ CONISU.PIIN P4RA.4 It'R1 f-OR BRAZIl IN/I' IHlR APPLICATI/ TO FOOD
POL/ICY. eptpemLernr IoA'. by lheryli Wiiliam-on GIr
31 SUSTAINING RAkP/F GRil'7ITH I IN IND4I FERTILIZER CONSIUAPTION: 4 PERSPECTII'E
E4Sf-l (O / CO.MtP(.II/OHfN OF UI', ,Augjt 10,82, i,uruvantr M Desai



'hubh K. Kumar has been a research fellow at iFPRI since 1078.
David Hotchkiss, formeriv a research assistant at IFPRI, is a research
associate at Gallaudet Research Institute






RECENT IFPRI RESEARCH REPORTS

68 COFFEE BOOM, GOVERNMENT EPENT XPENDITURE, AND AGRICULTURAL PRICES: THE COLOAM.
BIAN EXPERIENCE, August 1088, by Jorge Garcia Garcia and Gabriel Niontes Llamas
67 NATURE AND IMPACT OF THE GREEN REVOLUTION IN BANGLADESH, .uly 1988, by Mahabub
Hossain
00 THE BRAZIlA4N I'HEAT POLICE) ITS COSTS, BENEFITS, AND EFFECTS ON FOOD CONSUMP
lION, May 1088, by Geraldo M Calegar and G. Edward Schuh
65 CREDIT FOR ALLE NATION OF RURAL POVERTY THE GR.AIFEEN BANK IN BANGLADESH,
February 1088, by Mahabut Hossain
04 COOPERATIVE DAIRY DE ELOPMENT IN K4Rv.4TAA,~-, INrDIA- N 455ESSME.ENT, December
1987. bv Harold Alderman
63 INCOME 4ND NUTRITIONAL EFFECTS OF THE CO, .MERCIALIZA4TION OF AGRICUL TURE
IN SOUTHWESTERN KENYA November 1087, b Eillcen T. K rinn-ay anJ Bruce Cogill
02 AGRICULTURAL RESEARCH IN NEPAL RESOURCE 4LL OCA TIOF.. STRUCiTURE, ND INCEN
TIVES, September 1987, by Ram P 'tadav
61 THE PILOT FOOD PRICE SUBSIDY SCHEME IN THE PHILIPPF'!ES: T5 I IMPACT 0,N INCOME,
FOOD CONSUM'PTION, AND N,'UTRITIO.'..4 STA.'S, 4 .s; i'-E8' by Maritc. l _.aria and Per
Pinstrup-4ndersen
cO POPULATION POLICY AN.D INDI IDA4 L CHOICE .4 THEOfET7C.,41A I.'ESTICA.4OT/. iu.un
'187, byv iarc Nerio\%. A:,af dazin. and Efa'rn Saika
59 PRODUCTION, !NCEN'TIL ES IN PHILIPPtIE' .A-1R/CI,'Tl.-L.E- EFFECTS OF TRADE AvD EX-
CHANGE RATE POLICIES. Mav lo-87. cv Prome.-. Baundit.
58 THE FOOD STAMP SCHEME IN SRI LA4_4A: COSTS. BENEFITS, 4V' OPTIONS FOR .ODIFI
CATION, March lo'". by eville Ediriiingh-
5S" CEREAL FEEL USE IN THE THIRD I ORLLD. PAST TRE.'.ES 4.' PRO/ECTIOnVS TO 2000,
December 1o.8,, b J S S rmr
So THE EFFECTS OF TRADE 4,ND EXCHA..'GE RATE rf, L,',_'IS 7-. .GRICUL Ti'RE IN Z4JRE,
November 1'Q$80, by Tshikala B Thicbaka
55 THE EFFECTS OF TRADE .4.D E .CH.AVGE RA TE POLICEiES ;. 4CRICUL TLRE IN NIGERIA,
OcLiober O80,t. by T Ader-ola I)ytlde
54 .EATHER AND CR4/N YIELDS 1:'N THE 50t /ET UC.!OP. Str, W .t-,- 1*8,o. Paarma DesaL
53 REGIONAL COOPER.4 TIOA TO IMPROVE FOOD 'i.- /'R.!T !:' jL'THE1RN 4A'D EASTERN
AFRICA4N C'OUNTRIES. lulv i'1-)8 .t i llr;nch koesre-
52 FOOD IN THE TH!RI' LI ORL'. PAST TRE1:' 4,:r' .'ClECTIO.VS TO 2000. lune io80. bv
Leonardo A Pai'n',
51 DETERM.INA.NTS OF 4 GRIC,- iTRAL POl/C/EfU'. LRHE L',i' TEC TA TE5.4D THE EUROPEAN
CO,tLUNMT), in.ember !085. :', Micr.e! Fi- ,
..-*-* *.- ?-a ..* r


Wahntn 6 .C 203SS









Reprinted
with permission of the
International Food Policy
Research Institute




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs