University of Southern California Dissertations and Theses

Community forest management in Nepal: saving the forest at the expense of the poorest

(USC Thesis Other)

Community forest management in Nepal: saving the forest at the expense of the poorest

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

COMMUNITY FOREST MANAGEMENT IN NEPAL:

SAVING THE FOREST AT THE EXPENSE OF THE POOREST

by

Christine Cooper

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ECONOMICS)

August 2008

Copyright 2008 Christine Cooper
ii

Acknowledgements

I would like to express grateful appreciation to my advisor Jeffrey Nugent, without whose
indefatigable patience and incomparable grasp of detail and nuance I would ever have
completed this work.

I also thank Jennifer Wolch for her indomitable energy and encouragement, and John
Strauss for his invaluable input.

I acknowledge with great appreciation Samar Datta, whose insights and experience allowed
a more richly defined project and whose good humor enlivened all our meetings.

I am indebted to colleagues who gave generous input on my work in progress, especially the
dedicated participants of our Thursday morning workshop: Beth Armey, Dollie Davis, Matt
Shapiro and Medhi Majbouri.

My deepest and warmest gratitude is reserved for Constantine Glezakos, whose fine
instruction, early advice and generous direction sent me on this enjoyable journey. Thank
you, my friend.

iii
Table of Contents

Acknowledgements ii

List of Tables v

List of Figures vii

Abstract viii

Chapter 1 Introduction 1

Chapter 2 Literature Review 6
2.1 Enabling Conditions 8
2.2 Group Dynamics 9
2.3 Institutional Details 10
2.4 Limiting Factors 12
2.5 Successes 14
2.6 The Nepalese Experience 17
2.7 Policy Interventions 30
2.8 Summary 31

Chapter 3 Forestry Co-Management Model and Simulation 32
3.1 General Specification 33
3.2 Functional Forms and Calibration 41
3.3 Baseline Model 42
3.4 The Introduction of Co-management 45
3.5 Policy Experiments and Comparative Statics 48
3.6 Conclusions 76

Chapter 4 Behavioral Responses to Forestry Management Regimes 80
4.1 Data 82
4.2 Estimation 99
4.3 Conclusion 111

Chapter 5 Welfare Effects of Community Forestry 112
5.1 Data 113
5.2 Estimation 128
5.3 Conclusions 157

iv

Chapter 6 Conclusions 159

Bibliography 167

Appendix Model Specification 175
A.1 Functional Forms 175
A.2 Variables 179
A.3 Parameter Values 180

v
List of Tables

Table 3.1 Variable values in baseline model with no co-management 44

Table 3.2 Effect of introducing co-management, with proportional sharing of benefits 47

Table 3.3 Effect of co-management, experiments I 51

Table 3.4 Effect of co-management, experiments II 66

Table 4.1 Sampled Forest Dataset Descriptive Statistics – Household Characteristics 86

Table 4.2 Sampled Forest Dataset Household Characteristics (Pooled) 88

Table 4.3 Descriptive Statistics – Forest Benefits (Pooled) 89

Table 4.4 Sampled Forest Dataset – Household / Forest Benefits 90

Table 4.5 Calculated average benefits from forests by land-owning group 96

Table 4.6 Calculated average benefits from forests by land-owning group,
by management regime 98

Table 4.7 Estimation results for household labor allocation function for firewood
from community forest 102

Table 4.8 First stage probit results for FUG formation 103

Table 4.9 Estimation results for labor allocation function for fodder from
community forest 106

Table 4.10 Estimation results for labor allocation function for all resources from
community forest 108

Table 4.11 Gini Indices 110

Table 5.1 Dependence on Forests 115

Table 5.2 Household Income and Asset Holdings 116

Table 5.3 Housing Characteristics by Management Regime 118

Table 5.4 Household Assets by Management Regime 120

vi

Table 5.5 Household Assets by Management Regime - Panel 121

Table 5.6 Household Asset Growth by Management Regime - Panel 122

Table 5.7 Inequality (Ginis) in Rural Hill Communities 125

Table 5.8 Inequality (Ginis) in Rural Hill Communities 126

Table 5.9 Inequality (Ginis) in Panel Rural Hill Communities 127

Table 5.10 Estimation Results (OLS) for Per Capita Real Expenditure 132

Table 5.11 Probit Results for FUG formation 136

Table 5.12 Estimation Results (2SLS) for Per Capita Real Expenditure 137

Table 5.13 Estimation Results for Household Housing Values 142

Table 5.14 Estimation Results (OLS) for Household Total Asset Holdings 144

Table 5.15 Estimation Results for ∆ ln Per Capita Real Expenditure (1995-2003) 147

Table 5.16 Estimation Results for ∆ Net Earnings (1995-2003) 149

Table 5.17 Estimation Results for ∆ ln Total Asset Holdings (1995-2003) 150

Table 5.18 Estimation Results for Inequality (1995) 153

Table 5.19 Estimation Results for Inequality (2003) 154

Table 5.20 Estimation Results for Change in Inequality (1995-2003) 156

Table A.2 Variables 179

Table A.3 Parameter Values 180

vii
List of Figures

Figure 3.1 Forest-Based Economy Model ............................................................................ 34

Figure 4.1 Symmetry Plots .................................................................................................... 91

Figure 4.2 Deviation from Normal Distribution.................................................................... 93

viii
Abstract

Globalization, population pressures and consumptive choices coincident with higher
incomes have induced an urgent search for optimal management techniques as indigenous
resources have become increasingly accessible and hence more vulnerable to over-
exploitation. Devastating losses of forest cover and persistently high rates of deforestation in
south Asia have led governments to devolve authority over forest management to the
community level.

The case study literature suggests that community forestry has managed to halt degradation
but that the expected benefits from community level resource management and exploitation
have been unevenly shared within the communities, with the poorest most negatively-
impacted, as unexpected consequences of behavioral responses to the institutional shift
percolate throughout the economy.

This study presents a model of a forest-based community that is heterogeneous along several
axes and undergoes an institutional change in its resource management regime. Calibrated to
yield results emerging from south Asian experience, it allows us to experiment with
parameter changes to simulate policy options with the aim of understanding how the
implementation of community forestry affects household behavior, household welfare and
community distributional equity.

The inferences are tested on empirical data from Nepal, which implemented a widespread
institutional change in the early 1990s and which thus becomes an ideal laboratory for

ix
research on outcomes of that change. First, an analysis of household behavioral responses to
varying levels of devolution is conducted, followed by an examination of the effect of this
institutional shock on household welfare and distributional equity.

Our findings indicate that for a transitional period there is a substantial loss in terms of
income growth but that this loss is gradually eliminated. Further we find some evidence to
support the hypothesis that inequality in several income and asset categories increases in the
short term.

Given that the authority for forest management is increasingly being devolved to the local
level, these results suggest that welfare and equity considerations should take a more
prominent role, particularly in developing countries where those populations are dependent
on natural resources for their livelihoods and thus have little welfare to cede, even in the
short term.

1
Chapter 1
Introduction

With just under one third of the earth’s land mass covered with forests, how this valuable
resource is conserved and protected is a concern for researchers from many disciplines, not
only for the purposed preservation but also for the livelihood and environmental services that
forests provide. From sustainable exploitation of forest products for commercial purposes, to
the provision of non-timber products for consumptive and energy uses, to prevention of
erosion and the provision of biodiversity shelter, wildlife habitat, water purification and
carbon storage, forests provide a wide range of commodities and services. Most importantly,
an estimated 1.6 billion people rely on forests for at least a part of their livelihoods, with as
much as half of these living in or entirely dependent on them (PROFOR, 2007; FAO, 2003;
WRI, 1995).

The loss of forestland, therefore, and the degradation of the remaining forest resources, is a
growing cause for concern. It is estimated that during the period 2000 to 2005, net global
deforestation occurred at an annual rate of 0.18 percent, a loss of more than seven million
hectares per year (FAO, 2007). More than forty percent of this loss of forested land occurred
in Brazil alone, which lost more than three million hectares per year. Another fifty-five

2
percent occurred in Africa where together Sudan and Zambia accounted for a loss of over
one million hectares per year. In Asia, Indonesia has increased its deforestation rates since
the 1990s, losing over one million eight hundred thousand hectares of forest area annually, a
rate of 2 percent of its forested land per year; this rate of deforestation was matched by the
Philippines, Cambodia and Pakistan. During this same period, China increased its forested
land by four million hectares annually through the expansion of plantation lands.

The FAO points to population pressures and growing per capita consumption as the
proximate cause of most deforestation, particularly in developing countries, as agricultural
production expands onto newly cleared lands (FAO, 2003). However, there is widespread
recognition that there are a number of causes of deforestation in addition to population
pressures, including increased affluence, agricultural technology, national debt, commercial
logging, national government policies, forest accessibility and political instability (FAO,
2007; Gibson, McKean and Ostrom, 2000; Pearce and Brown, 1994). In spite of identifying
these causes, however, there is disagreement about the processes under which these factors
contribute to loss of forested land, and about the complex interactions between them
(Gibson, McKean and Ostrom, 2000).

Moreover, many of these factors occur at the macro level and the associated research fails to
adequately account for local level influences and responses, as if the communities whose
livelihoods depend on forest products and processes are somehow silent and play no
behavioral role in deforestation. Yet the FAO estimates that, by 2050, forty percent of all
forests will be managed or owned by individuals and communities (FAO, 2003). As a result
of this growing importance of community involvement, research has more recently been

3
aimed at uncovering the behaviors and incentives affecting local level communities in their
interactions with the natural resources which they own or manage.

Accordingly, there have been a number of anthropologic, sociologic and economic case
studies covering many communities and a variety of natural resources. While these studies
have been interesting and revealing in their diversity, comprehensive understanding of
causal factors of deforestation is still elusive. What has emerged from the empirical evidence
is that, contrary to prior assumptions that centralized coordination would be necessary for
the preservation of large scale resources, in many cases local communities have been
successful in the management of forests (among other resources), despite the lack of formal
property rights. Indigenous knowledge, social norms and cultural traditions have all played
significant roles in determining the effectiveness of local institutional development and
implementation, and thus in the success of local level involvement in the management of
natural resources.

Yet to achieve the declines in resource extraction, use and access restrictions have been
imposed which may negatively impact those households who are the most dependent on this
common pool resource for risk diversification and poverty alleviation. On the other hand, the
institutional design may unintentionally favor households with access to substitute resources
and exit opportunities, exacerbating distributional inequality within the community, a result
which may in turn have implications for community cohesiveness and cooperation which is
needed to produce the collective action necessary to support the institutional framework.

4
With these considerations in mind, this thesis begins with the development of a general
equilibrium model to portray a forest community characterized by heterogeneity across
several axes. Beginning under conditions of centralized management, a co-management
regime is introduced and alternative policy options simulated. The general equilibrium
effects are traced through the model to explain the observed (simulated) outcomes.
Inferences are then made on how policy interventions can improve outcomes under the three
criteria of resource protection, welfare improvement and distributional equity.

The model reveals how households respond to the given institutional framework by altering
their allocation of labor to management or protection activities (which improve the resource
and hence future household outcomes) and to resource extraction for household consumption
purposes or income-generation (which improve current household outcomes at the expense
of current and future resource health). Clearly, details of the institutional framework such as
the effectiveness of enforcement, the relative magnitude of penalties, the distribution of
benefits, and so on, have profound effects on the household’s behavior. To test labor
allocation responses to differential institutional structures, an examination is made of a
unique data set collected from a nationwide sample of forest communities in the rural hills
region of Nepal. Results reveal that competitive rather than cooperative behavior occurs
under a regime with less than complete property rights, exacerbating community inequality.

While resource protection was the primary goal of the policy of devolution, household level
welfare effects of the consequent restriction from resource use, at least in the early resource
regeneration period, have only begun to be investigated. Localized empirical inquiry into the
effectiveness of communities in protecting their resources revealed somewhat less than

5
desirable welfare effects, in particular on the relatively more poor within the community as
their access to livelihood support from the resource was curtailed without the adequate
provision of substitutable opportunities. Concern is now centered on distributional equity
and fairness in addition to livelihood support for the poor.
While case studies reveal incipient welfare and equity considerations, these remain in large
part anecdotal and may fail to record positive effects evident in a larger sample. The final
section in this thesis fills this evidentiary gap using two waves of a nationwide household
survey administered in Nepal in 1995 and in 2003. These two waves essentially book-end
the large-scale institutional reform implemented in 1995 in the devolution of forest
management to the community level.

The paper is therefore organized as follows. Chapter 2 reviews the literature pertaining to
common-pool resource management and discusses the experience of Nepal in its
implementation of community forestry. The general equilibrium model is described in
Chapter 3, and several policy interventions simulated. An analysis of household behavioral
responses is conducted in Chapter 4 using data collected from a sample of forest
communities in Nepal. Chapter 5 discusses the empirical approach to uncovering welfare
and equity effects of community forestry and the econometric results. The final chapter
draws conclusions from the analyses.

6
Chapter 2
Literature Review

Early work of resource economists such as Gordon (1954) and Schaefer (1957) used
comparative public goods analysis to demonstrate that individual users of a common-pool
resource would extract for their own use more than the sustainable yield would allow
because they would fail to take into account the costs that their extraction imposed upon
other users.

This led to the thinking that individualizing the ownership of the resource would internalize
such externalities and hence lead to more efficient use (Crutchfield, 1964; Demsetz, 1967).
Given that forests are physically large and house multiple ecosystems, making them difficult
candidates for subdivision into smaller parcels, it seemed that the central government would
be the most appropriate owner and consequently the policy of the 1960s and 1970s was to
nationalize forestry assets.

As it turned out, however, centralized management was not always successful. The size of
the resource made it impractical to monitor, particularly in developing countries with

7
budgetary challenges, and restrictions on usage were not enforced leading to de facto open
access regimes. Worse yet, the imposition of centralized control often trampled on
traditional systems of monitoring and enforcement that evidence from sociological and
anthropological case studies revealed may have been quite effective at preventing resource
degradation.

That local communities may overcome resource overexploitation is consistent with game
theory literature suggesting that the open access “tragedy of the commons” is in effect a
prisoner’s dilemma, which can be avoided if communication is permitted among the
participants, or if the game is repeated (Axelrod, 1984). The communities featured in much
of the case study literature were rural communities with long histories of repeated
interaction, established communication and norms of reciprocity (Runge, 1981, 1984).
Experimental work on group cooperation has shown that discussion among participants
yields positive cooperative effects (Ostrom et al., 1992; Ostrom, 1999; Cardenas, 2003). An
explanation of this success is the influence of communication in enhancing group identity (or
solidarity) or by the content of discussion encouraging commitment to cooperation
(Kopelman et al., 2002).

Thus this more recent acknowledgment of the role that local communities can play has led to
the implementation of a variety of management and co-management regimes where the legal
property rights remain with the national government but control and management is turned
over to local communities (in return for some compensation) such that by the end of the
twentieth century over fifty countries were involved in the devolution of at least some
management of their national forests to local community control (FAO, 1999).

8
Yet empirical evidence on the success of these types of sharing regimes has been mixed; this
has motivated the development of theory about the necessary conditions under which co-
management might work, much of it focused along the two dimensions which influence the
perceived costs and benefits to induce more cooperative behavior: the nature of group
dynamics; and the structure and details of the incentive scheme involved.

2.1 Enabling Conditions

After a careful examination of data from community-managed irrigation systems in thirty-
one villages in South India, Wade (1988) developed a set of enabling conditions for
successful commons management regimes, including: small group size (both in number and
in geographic dispersion), interdependence among group members, clear resource
boundaries, high level of resource dependence, simple detection of rules violations, and ease
of enforcement.

Similarly, Ostrom (1990) reviews fourteen case studies to devise a set of design principles,
“essential element[s] or condition[s] that helps to account for the success of these institutions
in sustaining the [common-pool resources] and gaining the compliance of generation after
generation of appropriators to the rules in use.” These principles include well-defined
resource boundaries, well-defined group boundaries; ease of rules enforcement, graduated
sanctions, and local authority free from state interference.

In a wide-ranging synthesis of numerous empirical studies, Baland and Platteau (1996)
conclude with much agreement with Wade and Ostrom on crucial conditions: small group

9
size, shared norms and interdependence, non-complex rules, ease of enforcement; and so
forth. Baland and Platteau also suggest that heterogeneity in endowments is conducive to
successful governance while heterogeneity in identities (or interests) is a hindrance.

2.2 Group Dynamics

Many of the group characteristics determined to be conducive to successful commons
management are simply those that are more likely to enable collective action, such as smaller
group size and group homogeneity (Ostrom, 1962; Olson, 1965; Sandler, 1992) since these
characteristics would also be more likely to permit a group to recognize incipient resource
problems, to agree on the proximate causes and effects of individual actions and to decide
what action to take to ameliorate whatever condition needs attention.

Yet empirical evidence suggests that not all characteristics identified for cooperative group
behavior need be present for sustainable common-pool resource management, nor are the
effects of the interactions among these characteristics always clear. For example, while small
group size may be more conducive to successful collective action, Agrawal and Goyal
(2001) on evidence from the Kumaon Himalaya find that the relationship is not linear, and
Ostrom (1999) suggests that other factors mediate the disadvantages of large group size,
such as the state of production technology, the endowment heterogeneity of the group, and
the ease of excludability.

Similarly, empirical evidence of the effect of heterogeneity on collective action is not
conclusive: cultural or ethnic homogeneity may facilitate cooperation through shared norms

10
(Baland and Platteau, 1996); and economic heterogeneity may lead to one group
internalizing a large enough share of the cost of the resource extracted to lead to a
sustainable yield (or, alternatively, efficient provision of maintenance).

Yet Varughese and Ostrom (2001) find in a study of Nepali villages that despite
heterogeneity in a community being a potential hurdle to collective activity, groups can
develop institutional innovations to overcome these hurdles provided that the benefits from
coping with heterogeneities outweigh the costs. Others note that community elites may be
able to capture a larger share of resource benefits (Agrawal, 2001; Platteau, 2003). In a
theoretical exercise, Dayton-Johnson and Bardhan (2003) show that the relationship between
wealth inequality and resource conservation is non-linear such that increased inequality
worsens conservation but once a critical imbalance is attained, further inequality will
actually improve conservation.

Baland et al. (2002) investigates the role of inequality in Nepal’s deforestation by asking
whether inequality impedes the collective action necessary to limit resource extraction as
measured by household labor allocated to firewood collection. They find evidence that
collective action reduces firewood collection and that inequality leads to lower collection but
fail to find support for the hypothesis that inequality itself impedes collective action.

2.3 Institutional Details

In addition to group dynamics, the structure of the incentive scheme has also been a subject
of research. Under a co-management regime, the legal property rights to the natural resource

11
remain with the central government while control and management is delegated to the
communities or users groups. In exchange for management services, the community receives
some type of compensatory exchange; for example, direct wage payment for rules
enforcement activity, a share of the profits from resource extraction, or a degree of de jure
property rights.

Many instances of co-management have been implemented worldwide in reference to
various resources: forests in India (Richards, 2000; Kumar, 2002), Nepal (Agrawal and
Ostrom, 2001; Edmonds, 2002), Bolivia (Becker and Leon, 2000) and Mexico (Klooster,
2000); fisheries in Japan (Baland and Platteau, 1996); woodlots in Ethiopia (Gebremedhin et
al., 2000); and land and a variety of wildlife and natural resource endowments in Canada
(Rusnack, 1997). Incentives under these regimes have varied. In India, forest communities
may receive wage payments for protective services, or they may receive a share of the
government revenue from sales of harvested timber, or they may receive a combination of
both. Forest user groups in Nepal receive reimbursement for enforcement costs and direct
ownership to village land. In Mexico, forest ejidos are entitled to manage their own
resources under license from the government.

Success in these cases has also varied. Reasons advanced for this echo the factors given
above for success in collective management regimes such as group homogeneity (Klooster,
2000; Kumar, 2002; Kant, 2000), group heterogeneity (Varughese and Ostrom, 2001), and
dependence on the resource (Kant and Berry, 2001; Kant, 2000; Cardenas, 2003).

12
2.4 Limiting Factors

But the strength of the incentives remains important and if insufficient this can become the
limiting factor. Prior to the implementation of JFM in Tanzania in 2002, for example,
harvesting was unregulated and an important source of income for villagers. Under
management agreements, this has been regulated and this lost income has not been replaced
with compensatory levels of benefits through the sharing scheme, undermining the villagers’
commitment to co-management (Akida and Blomley, 2007).

In other cases it is the degree of resource degradation that robs the incentivizing power of
co-management. When highly degraded resources turned over, the economic potential from
exploitation is limited and regeneration requires high level of inputs which must be funded
externally (FAO, 2006).

The degree of state control retained over the resource or over the communities’ abilities to
manage their own resource can also affect local level incentives to participate in
management regimes. In Pakistan, for example, in spite of private ownership of forests, by
individuals or communally, the Forest Department retains significant control in that owners
need approval prior to harvesting, marketing and daily usage of timber and firewood, leaving
them with little incentive to manage their forests (Nasir, 2006). Joint forest management in
India is proving successful in terms of forest management yet there exists substantial Forest
Department control over activities and benefit-sharing and a reluctance to cede control to
communities, threatening the sustainability of the program (Singh, Singh and Sinha, 2006).

13
Ambiguity in the benefit-sharing arrangement with the state can also be an impediment to
successful management. Such ambiguity may be the result of inadequate delineation and/or
protection of property rights. China’s forestry reforms in the early 1980s defined forest
ownership rights but failed to adequately define forested land and forests, thus ownership
remains ambiguous and increased deforestation and illegal cutting continue (Zheng, 2006).
Simorangkir and Sardjono (2006) demonstrate how lack of clarity over ownership and rights
over land and natural resources in Indonesia has caused an escalation of conflicts between
villages since decentralization.

A second dimension of ambiguity in benefit-sharing is elite capture, where, for example, in
Pakistan the rights of local communities to receive shares of proceeds of timber sales are
diverted by powerful timber traders (Nasir, 2006) who purchase the rights of poor
communities many years before they prepare their working plans. In Nepal, elite groups
holding key posts in forest user group executive committees get most of the benefits and
opportunities (Singh and Chapagain, 2006). The active participation of users, especially the
poor disadvantaged groups and women is difficult to achieve, particularly in decision-
making processes and benefit-sharing.

In a study of eight villages in the eastern Indian state of Jharkhand, five of which are
managing their forests under a joint forest management regime and three not, Kumar (2002)
uses a cost-benefit analysis to measure the impact of joint forest management on households
separated into land-wealth categories. He finds that the distribution of benefits is so uneven
that the improvement in forest cover comes at the expense of the poor, leading to the

14
likelihood that the poor will abandon participation in these schemes, further distancing
themselves from potential future benefits because of the unequal burden of current costs.

Finally, limited capacity can hinder all stages of implementation. The drawing of
management plans requires technical assistance and forest knowledge that may be absent at
either the state or community level (or both), resulting in delay, poor planning or poor
execution. Local communities may not be aware of their responsibilities and rights under
new systems of land or forest tenure. Application procedures for establishing community
groups to which management and/or ownership is to be devolved can be complex, unclear
and costly to execute and time-consuming for applicants and for state officials.

Thus in practice, failures in community forestry can be segregated by source: those
attributed to external causes, such as the degree of state interference, security of property
rights, resource conditions, capacity at governance levels; and those resulting from internal
characteristics, such as the identification of community membership and participation, elite
capture, benefit-sharing arrangements, and capacity limits at the local level.

2.5 Successes

Nevertheless, it is clear that various combinations of factors can lead to incentive responses
that yield successful outcomes.

For community forestry and joint forest management, success is most probable when rights
are granted on a long-term basis and are clearly defined; titling reforms can lead to secure

15
tenure and improve incentives to maintain responsibility over resources, especially when
done as in Tanzania and the Gambia, using a phased approach to allow for local capacities to
build, and with clear mechanisms for benefit-sharing (FAO, 2006). In Mozambique,
communities enter into long-term concessions with timber companies, receiving twenty
percent of all revenues collected on wildlife and natural forest exploitation (Sitoe and
Tchaúque, 2007). The transfer of state-owned plantations to the private sector in South
Africa in 2001 through lease agreements increased investment in the sector, improved the
condition of the forests and provided employment opportunities (Clarke, 2007).

Similarly, private sector investment in Uganda after the land reform that provided secure
tenure yielded employment opportunities and opened land for crop cultivation (Kigenyi,
2007). In Vietnam, forests are managed by individual households and joint venture
enterprises by allocation for long-term renewable leases of fifty years. The legal land use
certificate confers the right to exchange, transfer, lease, inherit and mortgage their forest
land and to capture commercialization gains, while granting the responsibility to protect their
forests against unauthorized use and to plant trees where needed (Nguyen, 2006). Mexico’s
experience with community forest enterprises under ejido management are yielding positive
economic benefits amid forest regeneration (Bray et al., 2005).

These experiences lead us to conclude that there exists a complex interplay among factors
such that outcomes may hinge on the effect of one or more factors in the presence of a
threshold state of other causal factors. Moreover, positive outcomes can be measured along
several dimensions, including not only resource conservation but also institutional

16
sustainability, livelihood protection (or improvement) and distributional equity; not all
outcomes are equally desirable or achievable.

Thus policy innovations may influence the state of underlying factors such that the
community can proceed toward more sustainable outcomes. For example, if a negative
relationship is found to hold between wealth inequality and successful resource management
at some critical level of income, then a policy of redistribution can improve outcomes.
Similarly, a policy aimed at strengthening community norms or promoting cooperation in the
enforcement of rules violations should also increase the probability of successful resource
management.

Yet the literature on these multi-variable relationships is unclear. Charkravarti et al. (2005)
present a model that encompasses many of the features found in the empirical literature in
order to predict individual effects. Their model allows for heterogeneity in the forest
community, both in outside opportunities and in the share of costs and benefits from the
resource, with an incentive scheme that distributes a share of the revenues from timber sales
to the forest community. The predicted results vary depending on production technologies
and preferences but point to the possibility that the sharing of harvest revenues with forest
communities will motivate increased maintenance effort and therefore improve forest
conservation.

One advantage of employing such a model is that it permits an examination of general
equilibrium effects of policy changes that may not be immediately obvious, and to trace the
source of observed outcomes to specific channels through which the policy operates.

17
Although the initial predictions were promising, a more thorough presentation as presented
below allows for more clearly-defined outcomes and policy interventions. Before turning to
this analysis, however, we turn to a discussion of the experience of forest management in
Nepal.

2.6 The Nepalese Experience

Nepal is one of the poorest nations in the world, with a per capita gross national income of
$290 (or PPP $1630) in 2006 (World Bank, 2008). From the latest data available, in 2003-4,
twenty-four percent of the population of 28 million lived on less than $1 a day, sixty-eight
percent on less than $2 per day; forty-five percent of children under the age of five were
malnourished. Almost thirty-five percent of the rural population was living under the poverty
line. This in itself was an improvement over the 1995-6 survey period in which over forty-
three percent of the rural population lived under the poverty line and the prevalence of child
malnutrition under the age of five was forty-nine percent. Agriculture is the main economic
activity, contributing almost forty percent of GDP and employing seventy-six percent of the
population.

Geographically, Nepal lies on the transition zone between the Gangetic Plain and the high
Himalaya mountains and is therefore characterized by a high degree of geographic and
climatic variation. Its southern zone, the Terai, is a lowland tropical and subtropical belt of
flat, alluvial land, lying from about 300 meters above sea level along the border with India
and rising to about 1,000 meters above sea level at the foot of the Hills region. The Terai
contains Nepal’s richest economic region having availability of agricultural land. Regional

18
demand for its commercially exploitable forests and for agricultural land left the Terai
deforested by the early 1990s.

Along Nepal’s northern border lies the mountain region situated at 4,000 meters or more
above sea level. This range includes Mount Everest and seven others of the world’s ten
highest peaks. The region is characterized by inclement weather and rugged topographic
conditions rendering human habitation and economic activities limited and arduous.

It is the hill region, lying between the Terai and the mountain region that is the subject of
this study. This area lies between one and four thousand meters above sea level and has
traditionally been the most heavily populated, home to over forty percent of the nation’s
population. The hills are heavily sculpted into terraces to maximize cultivable land.
Agriculture is the predominant activity, supplemented by livestock raising, foraging and
seasonal migration of laborers. The vast majority of households are land-hungry. The
growing season is short thus limiting the ability to grow multiple crops. As altitude
increases, crop productivity decreases and the importance of livestock as a livelihood pursuit
increases.

Due to the topographic nature of the region and the lack of infrastructure and roads, the
forests have not been commercially viable and therefore not exploited as in the Terai.
Nevertheless, population increases have led to increasing dependence for subsistence on
forest products and a consequent denudation of existing forests.

19
Since 1978, forest cover in Nepal has declined at an annual rate of 1.7%. During the period
2000 to 2005, the annual rate was 1.4%, amounting to a loss of 53,000 hectares per year
(FAO, 2007). Although serious, this was an improvement over the decade of the 1990s
during which the annual rate of deforestation was 1.8%, or a loss of 78,000 hectares per
year. This improvement is indicative of the more recent success in Nepal as its forest policy
has broadly shifted from what turned out to be failed centralized control since the 1950s
towards local-level management in the last decade of the twentieth century.

2.6.1 Historical Background

The experience of community forestry in Nepal is summarized as follows. Early
nationalization of forests in 1957 in the post-feudal period was meant to prevent destruction
of forests, which had until then suffered from a policy of agricultural expansion as a means
to increase food production and to generate state revenue through land taxes (Gautam et al.
2004). Prior to 1957, forests had been under private management by elite groups, including
members of the royal families and their relatives, who exploited forests as a source of timber
for export to India. Nationalization was meant to prevent further destruction by encouraging
investment in and sustainable use of privately-owned forests, but it failed to recognize
traditional use rights of local communities, and as a result forest destruction continued apace
as conversion to agricultural land ensured land claims (Acharya, 2002).

Under the Forest Policy of 1961, and Forest Protection Act of 1967, the Department of
Forests was mandated to protect forests; offenses were listed and penalties prescribed, yet
enforcement was weak and forest area continued to decline. The 1976 National Forestry Plan

20
recognized the role of local communities and emphasized their participation in forest
management; by 1978 village panchayats were permitted to manage barren or degraded
lands for forest production. The initial policy emphasis of this community forestry was on
resource creation through afforestation and reforestation such as large-scale plantations.

But it was the Master Plan for the Forestry Sector in 1988 that marked a turning point in the
history of forest policy in Nepal, which after twenty-five years of flirting with local resource
management through half-measure policy initiatives mandated the transfer of forests to local
communities and shifted the goal of the Department of Forests from forest protection to
forest transfer.

The transfer of forests which had begun in 1976 with the turnover of forests to village
panchayats accelerated after rules were written and codified in 1995. In an admirable display
of administrative efficiency, by the end of 2006 over 14,300 forest user groups (FUGs) had
been formed and been given operational control over 1.2 million hectares, or almost 25% of
the total forest land of Nepal.

The middle hills of Nepal are especially suited to the introduction of community forestry.
Forests in this region are not of commercial value due to their degree of degradation and
their lack of easy accessibility, thus the government of Nepal does not view its transfer of
authority as a loss of potential revenue and is willing to enlist the sizeable rural labor pool
for protection, monitoring and enforcement of use restrictions (Kumar, 2002). At the same
time, the significant dependence of local communities on forest resources for subsistence

21
agricultural inputs and firewood (among other uses) provides substantial incentive for
households and communities to take appropriate collective action.

The formation of FUGs is initiated by local forestry staff of the Department of Forests
(DoF), which first identifies a forest for transfer and enumerates all users using a procedure
similar to Participatory Rural Appraisal (PRA). After certification by the DoF, an
operational plan is formulated consistent with an outline provided by the DoF which defines
the area to be managed and describes the practices for management and harvesting,
including protection methods, penalties, etc. Decisions regarding these matters are ostensibly
made by consensus.

Early examination suggests that ecological sustainability is occurring (Gautam et al., 2004;
Yadav et al., 2003)), but often at the expense of the poorest members of the communities
whose livelihoods are most affected by resource availability (Malla et al., 2003; Neupane,
2003; Acharya, 2002; Kumar, 2002; Bhattarai and Ojha, 2000). Use restrictions and
mandatory participatory labor requirements would seem to disproportionately disfavor the
landless and those with limited opportunity for livelihood diversification or alternatives.

Additionally, it is argued that the distribution of benefits yielded from community forests
tend nevertheless to be weighted more towards those least likely to need them. Thus these
researchers have noted that welfare considerations were being ignored and that equity
concerns were becoming critical as the poor were becoming even more marginalized having
been left out of community forestry operation and benefits.

22
2.6.2 Resource Dependence

Although households in the middle hills in Nepal are among the poorest, with over 40% of
the population living below the poverty line (HMGN, 2005), Springate-Baginski et al.
(2003) describe the differential dependencies on community forests of community segments
as follows: All households are highly dependent on forest resources for cooking fuel.
Agriculture is the primary livelihood activity, which is carried out on small terraces of
irrigated or non-irrigated land. The richest in the community have extensive landholdings
with their own tree resources, so that their dependence on community or government forest
resources may be relatively low. They may own grazing land and private forests and own a
substantial number of livestock. They often have land outside the village and will usually
have one or more members who earn an income from service employment. Middle-class
households are often the most dependent on community forests because they have
landholdings and cattle but few private tree resources and less grazing land to supply leaf
fodder, grass and leaf litter for compost-making. The poorest households commonly have
little or no land and depend on tenuous livelihood strategies such as providing agricultural
labor, portering, collection and sale of firewood and other non-timber forest products
(NTFP), blacksmithing and artisanal production.

Community forest management begins with the preparation of an operation plan, which
specifies the objectives of forest management, which products are to be produced and which
may be harvested or collected. Although the outline of the plan is provided by DoF forestry
staff, the community itself determines the specifics, with collective decision-making
negotiated among local parties.

23
2.6.3 Implementation Results

Case study literature has much to say about the implementation results. Maskey,
Gebremedhin and Dalton (2003), using survey data from 443 households in a community
forest in the middle hills, find that wealthier households are more likely to participate in the
decision-making and in the management activities themselves, given the higher relative
opportunity costs of labor of poorer households. They also find that the distribution of
benefits from forests is determined by the level of participation in management; therefore not
only are the imperatives of the poorest ignored because their voices are not heard during the
decision-making and negotiation, but also distribution is not made in accordance with need.

Bhattarai and Ojha (2000) estimate the costs and benefits received from the forests by the
households of two FUGs in the Koshi district: one a small user group of 44 households, the
second a larger group of 180 households. Enumerating operation and transactions costs per
household and quantifying the benefits received in forest products, with wealth classes
determined through participatory assessment, they find that net benefits accruing to the poor
are actually negative for these same reasons, a finding concurred by Neupane (2003) in his
study of 128 households in four FUGs in the districts of Parbat and Myagdi in the Western
Development Region. Factors identified as limiting the access of the poor to community
forestry benefits include: insufficient support from the District Forest Office; inadequate
transmission of new information and knowledge (including forest management techniques);
exclusion of the poor from decision-making processes; and inappropriate arrangements for
the distribution of forest products.

24
A comprehensive research project investigating the impact of forest user groups on forest
resources and livelihoods was conducted in 1997-2000 by the University of Leeds in the
United Kingdom, the Nepal U.K. Community Forestry Project (NUKCFP), Natural
Resources International and the Department for International Development (DFID). To
evaluate results, the project used a participatory action research methodology in fourteen
communities, eleven of which were governed by forest user groups, in the Koshi Hills area
in the Eastern development region. The average number of households in these user groups
was one hundred and thirty-five, with an average duration of four and a half years at the time
of the initial survey in 1998.

Through careful measurement of randomly-selected plots within community forests,
including categorization of trees according to genus and size, preliminary examinations of
the forest resource base conducted by Yadav et al. (2003) show that regeneration occurs
unambiguously under forest user group management, specifically through resource
protection and replanting. At the same time, however, the methods used to protect the
resource are presenting challenges for community participants, including the unequal
distribution of benefits due to varying wealth-based needs, uneven household satisfaction
with the implementation of community forestry.

In the same research study, Dev et al. (2003) investigates the impact and distributional
implications of community forestry on a full range of community features, including changes
in product flows, social capital, infrastructure, credit opportunities, human capital and
livelihood opportunities, in the eleven communities under forest user group management.
Using a livelihood assessment approach, they find that although there is a variety of

25
experience, many changes due to the institutional change seem to have benefited the
relatively wealthy more than the poor and landless. For example, timber distribution from
the community forest, in spite of increased extraction controls, have tended to be skewed
towards the elite, an outcome of implementing a flat-rate fee for product extraction wherein
the fee for timber is relatively higher given its scarcity value which discriminates against
poorer households. Similarly, the benefits of improved community infrastructure benefit the
relatively more wealthy as, for example, school investments favor those able to send their
children to school, which exclude the poorer households.

Credit opportunities are also cited as a potentially significant outcome of community forestry
if the generation of financial capital from the institutional implementation is put to good use
in micro credit schemes. Although few in number, those communities with established credit
opportunities are administered by the wealthier people who are hesitant to lend to the poor
for fear of default. These findings are corroborated by Malla (2000) and Malla et al. (2003)
(see below), but Dev et al. (2003) also point out that potential exists for improvement in
these sectors with the introduction of further institutional changes, such as, for example,
improved forest product processing and marketing, developing livelihood skills and the
promotion of marginal groups such as the very poor and women.

That this extensive research study found uneven results in the distribution of the benefits of
user group management, even given resource improvement, is not cause for optimism. The
Koshi Hills area has been the recipient of significant intervention in the form of aid and
educational projects in particular by the DFID since the early 1970s, including agricultural
extension, an integrated development program, and programs specifically involved with

26
livestock management and community forestry. Given the level of external support and
technical assistance rendered through these facilities, one might have expected that
livelihood improvements and intra-community distributional questions would have been
resolved; yet these issues remain.

Thus despite promising results in resource improvement, it is not clear that community-
based management benefits all segments of the community.

2.6.4 General Equilibrium Effects

Malla (2000) and Malla et al. (2003) carefully describe in detail recent experiences in Nepal.
The observed results of the implementation of community forestry contrast markedly with
the expected results.

The most common policy implementation is exclusionary, or restriction of access to the
resource or of resource extraction. Malla (2000) finds that of the cases he studies, most forest
user groups restrict the use of forest products that have a cash value or that are in short
supply, including timber, firewood and tradable non-wood forest products.

Rules of access to forest products vary among forest user groups: some allow free collection
of dead and dry materials such as fallen twigs or branches and leaf litter. In a nation where
more than 90 percent of the population lives in rural areas, reliance on forests for livelihoods
is significant: trees are collected for fodder and bedding materials for livestock, leaf litter is
collected to be used as a fertilizer for crops, and timber and poles are used to build houses

27
and for agricultural tools. Firewood, sourced from the forests, is often the most important
source of energy for cooking and heating for rural households. Non-timber forest products
are collected by households for cash generation.

On its face, exclusion from resource use would improve the condition of the resource. If
limited access is allowed, then the equal distribution of whatever products are harvested can
presumably leave the entire community equally improved.

It is clear, however, that communities are not homogeneous, but each is characterized by
strata of income and wealth, often with underpinnings in historical or caste relationships,
perpetuated by unequal land holdings and employment opportunities,

Malla (2000) describes the unintended effects from the putative equal distribution of benefits
of the forest resource: He describes how in general little greenwood is permitted to be
harvested from community forests and what is harvested is typically distributed equally
among the households in the user group. However, since wealthier households are able to
harvest firewood from their own private trees, the distribution from the community forests
provides them with more firewood than they would previously have taken, and thus the
poorer households receive less.

Similarly, whatever timber is permitted to be harvested is usually made available equitably
within the forest user group, but the distribution mechanism used is auction. Here, prices are
bid up to such levels that only the wealthier households can afford to participate, locking out
the poorer households.

28
Non-wood forest products without trading value are normally freely collected, which would
imply equal access across income groups, but wealthier households have more livestock and
more labor, often hired labor, and therefore are able to collect more product.

Non-timber forest products which have a cash value are especially important for poorer
households as income supplements, but community forestry has focused on timber, firewood
and fodder, and use policies are incompatible with organized and systematic management of
the forest resource needed for this type of product and therefore the loss of these potentially
valuable products hurts the poor more than wealthier households.

Lastly, income generated from joint forest management or community management, which
accrues to the community, is used in a variety of purposes, including the payment of salaries
for management staff (nursery workers, planters, weeding labor, guards, etc.) and in village
development, such as schools, roads, drinking water, irrigation schemes, electricity systems,
etc. However, even these expenditures are not necessarily equitably distributed. For
example, the building of new schools often fails to include individual direct support for
students, such as for uniforms or books, which disadvantages the poorer households who
consequently cannot send their children to the new schools.

Similarly, irrigation schemes benefit wealthier families in the fertile valleys who are already
favorably possessed of better land, while most poorer households own terraced lands on the
upper slopes which are beyond the reach of irrigation systems. Other development schemes
can require additional local contributions in cash, or labor, or both; the non-poor simply give
cash, but the poor contribute labor. This disadvantages the poor since the opportunity cost of

29
this labor in the market centers is much higher than the local labor contribution value, thus in
effect costing them relatively more than the cash contributed by the non-poor.

The timber market has also seen unintended effects of the implementation of community
forestry. Exclusionary policies of community forestry prevent user group households from
harvesting timber to sell in nearby markets, a practice that was common and rather more
important for poorer households before the introduction of community forestry.
Nevertheless, population pressures and maturing tastes for consumer goods have led to
improved demand for timber and forest products, leading to the development of forest
industries and increasing the pressure on local resources. In response, farmers with land have
begun silviculture on their farmland with the aim of selling timber and firewood in the
marketplace.

The decision to not use community forests to fill local market demand has had two negative
distributional effects: (1) Only private tree growers will benefit from the larger market for
forest products, since it is the wealthier households that have more land; and (2) the
increased demand for timber and firewood will lead to higher local market prices, making
firewood no longer affordable for poorer households who, in addition to being restricted
from collecting from community forests, can no longer afford to purchase from the market
and thus are made worse off.

30
2.7 Policy Interventions

Clearly, simply prohibiting or limiting resource extraction can improve the resource, and
careful sales of excess timber or cultivated products can lead to higher incomes for the
community as a whole. Yet the distributional effects and price responses have differential
impacts on the non-poor and the poor.

Researchers have offered several policy recommendations. Among these are: (i) increasing
capacity in field staff available to deliver extension services (Acharya, 2002); (ii) increasing
training for DoF staff and for user groups (Malla et al., 2003); (iii) redistribution
arrangements (Neupane, 2003); (iv) organizing FUG meetings at a higher administrative
level to allow the inclusion of women and disadvantaged members (Neupane, 2003); (v)
increasing inputs from external sources of technical information (Neupane, 2003); (vi)
encouraging production of non-timber forest products for commercial purposes (Dev et al.,
2003); and (vii) the promotion of active forest management procedures rather than
exclusively resource protection (Yadav et al., 2003).

Recognizing the negative effects of resource restriction especially on the landless with no
access to substitutes, in 1995 Nepal began a leasehold forestry program in which parcels of
very degraded forest land are leased on a long-term basis (forty years, renewable) to a
leasehold group of poor landless families. Leasehold groups are authorized to extract forest
products and distribute them among group members but may not remove any exiting trees.
The land cannot be sold or mortgaged. Although there is little empirical evidence of this
program, Nagendra et al. (2005) studies three leasehold groups and finds inter-group

31
conflicts over use rights and social issues and conflicts with neighboring forest user groups
over membership claims and benefit sharing. Similarly, Karmacharya et al. (2003) finds
neighboring forest groups unwilling to acknowledge exclusive rights of the leasehold group
to its given parcel. By mid-2006, only 11,100 hectares of forest land had been leased to
leasehold groups, representing less than one percent of the forest area turned over to
community forestry, and involving 18,500 households, making it a relatively small portion
of the community forestry program.

2.8 Summary

The large-scale ideological shift towards decentralization in the management of natural
resources over the past twenty years has led to increasing implementation of community
forestry and joint forest management, particularly in South Asia. The design of the
institutional regimes has been varied and much studied in relation to theoretical literature
predicting success given a combination of governing conditions and community
characteristics. Still, while resource protection appears to have been successful, community
welfare effects, including distributional equity, are only now beginning to attract attention.
Case studies reveal that exclusionary policies and nominally equitable distribution of
common-pool resource products unfavorably impact the relatively poor within communities.
General equilibrium effects of policies implemented with the best of intentions have led to
unexpected results as price changes and income effects ripple through local communities.
Thus more careful design of policies to adequately account for unequal impacts is now being
suggested.

32
Chapter 3
Forestry Co-Management Model and Simulation

This chapter will present an analytic framework to uncover the underlying relationships
between community characteristics, such as preferences, group heterogeneity and income
opportunities, and the incentive regime. The framework will build upon the model presented
in Charkravarti et al. (2005).

This is a static general equilibrium model of a forestry-based economy in which a co-
management program is to be implemented. The incentive scheme includes the distribution
of a share of the harvest revenues and allowances for unauthorized timber removal.

The model captures heterogeneity of the forest community along four axes: dependence on
the forest resource; ability to capture the benefits of the incentive scheme; ability to evade
costs of maintenance; and availability of outside income opportunities.

By calibrating the parameters of the model to stylized facts from co-management field
experiences, the model will be used to simulate the effect of the introduction of co-

33
management on resource preservation and on community welfare, and how heterogeneity
impacts these effects. Thereafter, through policy variable manipulation, a combination of
strategies to achieve welfare and resource improvements simultaneously will be identified.

3.1 General Specification

We begin with a forest-based community, which is presented diagrammatically in Figure
3.1. The forests are owned by the central government, but there is a community that lives
within the forest and depends upon it as a resource for livelihood. The government plays
only a redistributive role in this scenario.

There is a forest department, however, which harvests the forest asset using local labor and
whose main goal is profit maximization. The harvested timber is sold in two markets: (i) to
an export market, with the quantity determined by the central government and the price set
on the global market, and (ii) domestically at an endogenously determined price. Wages for
harvesting labor are set by the forest department.

The forest department also hires protection or forest management labor. This serves to
improve the condition of the forest resource. Labor is supplied by the forest community. The
wage paid for this service is fixed by the forest department. In general, this condition
assumes a local economy with surplus labor in the tradition of Ranis’ development theory
(Fei and Ranis, 1964). If the forest department is not interested in promoting reforestation
using community labor, it may choose to set the wage to zero.

34
Figure 3.1 Forest-Based Economy Model

35
In addition to legal harvested timber, some forest products are collected informally by the
forest community, using its own labor and not passing through a market.

There is also a market good, which is manufactured by the market sector using labor hired
from the forest community. The market good is consumed by both the forest community and
the market sector itself and its price is endogenously determined. Income taxes are levied on
the gross profits of the market sector. The market sector consumes harvested timber in
addition to the market good, but does not otherwise remove or collect forest products from
the forest.

Thus there are three paid labor opportunities: harvesting labor, protection labor and market
labor. Harvesting labor is demand-determined. The forest community can choose to spend
unpaid time on collection activities or at leisure. If forest protective services are unpaid, the
forest community may still choose to supply labor to this activity because it values the forest
resource.

Utility for the forest community is a function of its consumption of forest products (either
harvested timber or collected forest products), of the market good, of whatever government
transfers it receives, and of leisure. Additionally, the value that forest community places on
the forest asset is represented by an externality function that is itself a function of the state of
the forest. The externality function is based on the state of the forest, which is diminished by
harvesting and by collection and is enhanced by protective and management services. The
forest asset regenerates according to a logistic biometric function.

36
Utility of the market sector is a function of its consumption of harvested timber, of the
market good, of the government transfer it receives, and of the externality value of the forest.

The forest community is segmented into two: the forest elite, designated “A,” and the non-
elite, designated “B.” It is assumed that the relative size of these two sub-communities is
10:90; that is, ten percent of the forest community is elite.

Heterogeneity is initially introduced into this model in the preference weights given to forest
products and the market good. The elite are assumed to relatively prefer the market good
over forest products. All other aspects of these two communities are identical. Shares of
labor opportunities (
H
α ), transfers ( ϕ) and fines ( γ) are proportional to population.

The model can be specified therefore as follows. Lower case superscripts refer to quantities
demanded (d) or supplied (s), while subscripts refer to the forest community (C,i where
i=A,B), manufacturing (R), export (X), protective (P), de facto (df), and externality (E). In
general, N is labor, F refers to a forest product and X refers to the manufactured good.

Forest Community A:
The problem for the elite of the forest community is:
( ) ( )
A C E
d
A C C
d
A C A df A C
L X F N N N
L F X G F F U Max
A C
d
A C
d
A C
s
A R
s
A P A df
, , , , ,
, , , , ,
, , , ,
, , , , , ,
⋅ + ϕ
subject to its budget constraint and labor time endowment, and where the externality
function is
( )
B P A P
d
R
d
B C
d
A C B df A df E E
F F F F F F F F F F
, , , , , ,
, , , + + + + =

37
where
) (
, , A df df A df
N F F = de facto collection of forest products; and
) (
, , A P P A P
N F F = increase in forest asset due to protective services of forest
community A.

As noted above, the forest community earns income from wages paid for harvesting labor
hired by the forest department and wages paid by the manufacturing sector for market labor
in producing the market good. The forest community also receives a government transfer.
For simplicity, this transfer is taken to be in-kind and thus does not affect the budget
constraint. This can be taken to represent government spending within the forest community,
such as the building of a local school or medical center. If a revenue-sharing co-management
regime is implemented, the forest community also earns a share, ν , of the gross revenues
from timber sales, of which forest community A receives ϕ (and forest community B
receives ( ϕ − 1 )).

The forest community spends its income on marketed timber products and on the market
good. Additionally, rules under co-management specify a limit to the amount of forest
products extracted informally by the forest community. Any extraction above this allowable
cut is subject to a penalty. The penalty function, ( ) ⋅ H , is based on the amount extracted
above the allowable limit,
L
F . The probability of detection, ( ) ⋅ η , is a function of the excess
amount extracted, being higher for higher levels of extraction and lower for less extraction.
The share of the penalty paid by forest community A is γ.

38
The budget constraint of forest community A is therefore:
() ( ) ( )
d
X X F
s
F
d
H H H
s
A P P
s
A R R
d
A C F
d
A C X
F P F P N w N w N w H F P X P
, , , , ,
+ + + + = ⋅ ⋅ + + ϕν α γη
where
() ) (
, , L B df A df
F F F − + = ⋅ η η is the probability of detection; and
() ) (
, , L B df A df
F F F H H − + = ⋅ is the penalty for excess extraction.

The community’s labor time endowment is fixed, and is allocated among the four types of
labor activities:

A TOT A C
d
H H
s
A R A df
s
A P
N L N N N N
, , , , ,
= + + + + α

Forest Community B:
The problem for the non-elite of the forest community is:
() ( )
B C E
d
B C C
d
B C B df B C
L X F N N N
L F X G F F U Max
B C
d
B C
d
B C
s
B R
s
B P B df
, , , , ,
, , , , ,
, , , ) 1 ( ,
, , , , , ,
⋅ − + ϕ
subject to its budget constraint and labor time endowment, and where again the externality
function is
( )
B P A P
d
R
d
B C
d
A C B df A df E E
F F F F F F F F F F
, , , , , ,
, , , + + + + =
where
) (
, , B df df B df
N F F = de facto collection of forest products; and
) (
, , B P P B P
N F F = increase in forest asset due to protective services of forest
community B.

39
The budget constraint of forest community B is:
( ) ( )
()
,, , ,
,
(1 ) (1 )
(1 )
dd s s d
XCB F CB R RB P PB H H H
sd
FFXX
PX PF H w N w N w N
PF P F
γη α
ϕν
++− ⋅ ⋅= + +−
+− +

The community’s labor time endowment is fixed, and is allocated among the four types of
labor activities:

B TOT B C
d
H H
s
B R B df
s
B P
N L N N N N
, , , , ,
) 1 ( = + − + + + α

Market Sector:
The problem for the manufacturing sector is:
( ) ( )
P
d
R
d
C df E
d
R R
d
R R R
X F N
F F F F F F X G F U U Max
d
R
d
R
d
R
, , , , , ,
, ,
+ =
subject to its budget constraint. The manufacturing sector gains income through its profits
from selling the market good, which it manufactures using labor hired from the forest
community. The profits are taxed by the government at rate t. The sector also receives a
transfer from the government. For simplicity, again, this transfer is taken to be in-kind and
thus does not affect the budget constraint; this can similarly be thought of as government
investment in the manufacturing sector.

The manufacturing sector spends its net income on marketed timber products and on the
market good. The budget constraint of the manufacturing sector is therefore:
) )( 1 (
d
R R
s
R X
d
R F
d
R X
N w X P t F P X P − − = +

40
Forest Department:
The problem for the forest department is to choose the amount of timber to sell locally to
maximize profits:
( ) ( ) ( ) C N w N w H G F P F P Max
d
P P
d
H H F
d
X X F
s
F
N
d
H
− − − ⋅ ⋅ + + + − − η ν ε
,
) 1 )( 1 (
where
) , (
d
H H
d
X
s
H
N F F F F F = + =
and where C is a fixed harvesting cost, and where ε is the share of timber revenue given to
the government. If co-management is implemented, the forest department gives a share ( ν )
of its timber revenue to the forest community, and receives the penalties levied on the forest
community for excess extraction of forest products.

Government and Market Clearing:
The government collects tax revenues and redistributes this to the three other sectors. It has
no objective function, and therefore the government’s budget constraint is simply one of the
market clearing conditions:
( )
d
X X F
s
F
d
R R
s
R X F R C
F P F P N w X P t G G G
,
) 1 ( ) ( + − + − ⋅ = + + ν ε
()
d
R
d
B C
d
A C
s
R
X X X X + + = ⋅
, ,
to clear the goods market; and
d
R
d
B C
d
A C
s
F F F F + + =
, ,
to clear the domestic timber market.

41
3.2 Functional Forms and Calibration

In order to yield interior solutions, it is assumed that the utility and production functions are
convex and continuous. Production functions throughout take the Cobb-Douglas form, and
the utility functions are CES. Functional forms are shown in Table A.1 in the Appendix; the
list of variables is in Table A.2.

It is assumed that marginal labor productivity is highest in the manufacturing sector and least
productive in protection. Labor exhibits diminishing marginal productivity in all activities.

The choice of parameter values is a difficult one, given a dearth of microeconomic data.
Discussions with researchers in the field and comparisons to results of case studies provide
approximations which are used in calibration. Thus the model is calibrated so that it yields
the following results consistent with Kumar’s study of Indian villages (Kumar, 2002):

• The introduction of a co-management regime unambiguously results in improved
forest cover
• Protection by the forest communities occurs even in the absence of a co-management
regime, but the provision of protective labor comes from poor households rather than
the non-poor
• Forest collection declines by up to fifty percent under a co-management regime
• Forest-related income as a proportion of the households’ total income is somewhat
higher for the poor than for the non-poor
• There is a shift to non-forest livelihoods

42
• There is a decline in welfare for the forest community, larger for the poor than for the
non-poor
• The village wage rate is set at one half to one-third the official wage paid for
government work.

3.3 Baseline Model

Simulation yields a baseline set of variable values as presented in Table 3.1; parameter
values used are shown in Table A.3 in the Appendix.

Because the results depend on the initial values of the variables, it is not useful to discuss the
results in absolute terms; however, the economy can be described in relative terms as
follows.

The forest community spends about fifty-eight percent of its time endowment in informal
collection activities for its livelihood. Twenty-one percent is engaged as hired labor for the
market sector; almost fifteen percent is engaged as harvesting labor for the forest
department, and over six percent spent at leisure. Little time is spent in the protection or
management of the forest resource.

To satisfy its preference for forest products, the forest community purchases harvested
timber and collects products in a ratio of approximately 37:63. The remainder of its income
is spent on the market good. It spends more of its income on the market good than on timber
(84.81 versus 25.44). Of the community segments, the elite consume relatively more of the

43
market good and of harvested timber than the non-elite and relatively less of collected forest
products.

The equilibrium price paid for harvested timber domestically is 0.37, compared to the price
of the market good at 2.49.The equilibrium market wage rate is 2.84, compared to the
exogenously-set harvesting wage of 3.50.

Although the elite earn a higher income (proportionate to population) than the non-elite,
welfare in the community is relatively equitable.

Given these values, there is a net decrease in the forest resource. These initial values seem
to be consistent with case study literature.

If the assumption of surplus labor is relaxed and the harvesting wage endogenously
determined, we would expect that competition for this wage opportunity would drive the
wage down and increase harvesting employment, leading to an increase in harvested timber,
a fall in the local timber price and increased consumption of harvested timber versus
collected products. Because an increase in wage earning opportunities relatively benefit the
poorer non-elite, this would tend to lead to a more equitable distribution of welfare.

44
Table 3.1 Variable values in baseline model with no co-management

Variable
Values

% of total (FC)
Labor supplied:
Protective
Of which: – A
– B
0.02
0.00
0.02

0.00
100.00
Harvesting
Of which: – A
– B
14.52
1.45
13.06

10.00
90.00
Collecting
Of which: – A
– B
57.97
1.36
56.61

2.34
97.66
Market
Of which: – A
– B
21.02
6.22
14.80

29.57
70.43
Leisure
Of which: – A
– B
6.50
1.00
5.51

15.32
84.68

Income:
Forest community
Of which: – A
– B
110.41
22.71
87.70

20.57
79.43
Market sector 48.97
Forest department 189.58
Government 98.87

Consumption:
Forest Community A:
Timber 9.62 13.99
Market good 7.68 22.55
Collected 3.34 2.81
Forest Community B:
Timber 59.14 86.01
Market good 26.38 77.45
Collected 115.66 97.19
Market Sector:
Timber 21.81
Market good 16.40
Fines paid 0.00

Welfare:
Forest community – A 17.49 10.08
– B 156.03 89.92
Market sector 29.57
∆ Biomass (47.54)

45
Table 3.1 (Continued)

Variable
Values

% of total (FC)

Prices:
Timber (domestic) 0.37
Export price of timber 1.00
Market good 2.49
Harvesting wage 3.50
Market labor wage 2.84

3.4 The Introduction of Co-management

Into this model, a regime of co-management is introduced. Co-management is a system
whereby the forest community is given some level of responsibility for the management and
protection of the forest asset in return for a share of the harvesting revenues and perhaps
wage payments for its management labor. Once co-management is introduced, several
resource use rules are implemented; for example, informal collection of forest products is
restricted, and violation of use rules is enforced with penalties. In addition to exclusion and
rules enforcement, the forest department may undertake a training program to educate the
local community on the value of its resource (environmental training) or on productive
protection and management methods (silvicultural training).

The share of revenues given to the forest community is an exogenous parameter and must be
chosen. It is set at sixty percent, which is the sharing percentage legislated in the Nepalese
community forestry program. From the first order conditions of the optimization problem,
one would expect that revenue sharing will impact the forest community’s budget constraint;
as the forest community receives an increasing share of the timber revenues, it is more likely

46
to increase its purchases of both harvested timber and of the market good. For the forest
department, however, sharing more of its revenue with the forest community reduces its
budget and it will be more likely to hire less harvesting labor.

The equilibrium values of the choice variables resulting from this implementation of co-
management are shown in Table 3.2.

The results are described as follows:

With the additional income from the share of the timber revenues, incomes of the forest
communities increase. This additional infusion generates increased demand for both timber
and the market good, leading to price increases. Increased production of the market good in
response to higher demand fuels the market sector, and which also sees increasing incomes.
The competition for labor bids up the market wage, however, and the price of the market
good rises. (Imports of the market good are triggered when the domestic price rises above a
given level, moderating the demand for local production.)

The forest department meets increased demand for timber by raising its price. The increase
in the price of timber induces a substitution towards the market good. The imposition of
hefty fines discourages informal collection of forest products. The combination of decreased
consumption of forest products (harvested and informal) is not offset by the increased
consumption of the market good and thus welfare falls for both forest communities, more so
for the non-elite than for the elite.

47
Table 3.2 Effect of introducing co-management, with proportional sharing of benefits

% of timber sales revenue shared
0% 60%

% ∆
Labor supplied:
Protective
Of which: – A
– B
0.02
0.00
0.02
19.96
0.00
19.96

1.17
Harvesting
Of which: – A
– B
14.52
1.45
13.06
13.55
1.36
12.19
(6.68)
(6.68)
(6.68)
Collecting
Of which: – A
– B
57.97
1.36
56.61
33.63
1.18
32.44
(42.00)
(12.91)
(42.70)
Market
Of which: – A
– B
21.02
6.22
14.80
22.25
6.29
15.97
5.88
1.10
7.88
Leisure
Of which: – A
– B
6.50
1.00
5.51
10.63
1.19
9.44
63.41
19.16
71.42

Income:
Forest community
Of which: – A
– B
110.41
22.71
87.70
252.63
39.84
212.79
128.81
75.44
142.63
Market sector 48.97 76.89 57.00
Forest department 189.58 171.28 (9.65)
Government 98.87 48.32 (51.13)

Consumption:
Forest Community A:
Timber 9.62 9.83 2.17
Market good 7.68 7.85 2.19
Collected 3.34 2.93 (12.30)
Forest Community B:
Timber 59.14 45.25 (23.49)
Market good 26.38 36.53 38.48
Collected 115.66 68.15 (41.08)
Market Sector:
Timber 21.81 19.85 (8.97)
Market good 16.40 18.53 12.99
Fines paid 0.00 63.59

Welfare:
Forest Community 173.52 128.86 (25.74)
Of which: – A 17.49 15.73 (10.10)
– B 156.03 113.13 (27.49)
Market sector 29.57 26.25 (11.22)
∆ Biomass (47.54) 32.27

48
Table 3.2 (Continued)

% of timber sales revenue shared
0% 60%

% ∆

Prices:
Timber (domestic) 0.37 0.65 74.98
Export price of timber 1.00 1.00 -
Market good 2.49 3.46 38.56
Harvesting wage 3.50 3.50 -
Market labor wage 2.84 3.59 26.65

At the same time, there is a decline in deforestation to such an extent that reforestation
occurs. Thus, consistent with Kumar’s study, the improvement in forest cover and increased
incomes are not accompanied by welfare improvement in the forest community, and the non-
elite or poor suffer more than the elite or non-poor. This is a case of “saving the forest at the
expense of the poorest.”

3.5 Policy Experiments and Comparative Statics

From this baseline result, one can manipulate policy variables and parameters as is realistic
in real world experience to achieve pareto-superior results. Recall that the combination of
three effects leads to declines in welfare for both forest communities in spite of increasing
incomes: (1) the timber price increase leads to substitution away from timber; (2) the
imposition of fines discourages informal collection; and (3) the general price increase in the
market good. While both the elite and the non-elite experience welfare declines, the decline
is sharper for the non-elite because of their relative dependence on forest products from
which they are being excluded.

49
Because co-management leads to improvement in the forest resource at the expense of
welfare falls for all segments in the economy, the search for results under which at least
some improvements are made is motivated.

Results obtained from the experiments are shown in Tables 3.3 and 3.4, with for each change
in parameter the percentage change in the variable values over the baseline case which is
shown in column (2).

3.5.1 Paying wages for protective or management labor

In the baseline case, labor spent in protective or management services has been voluntarily
provided. In some implementations of community forestry, however, such labor is paid a
fixed wage by the forest department. Thus terms are added to the forest department and
forest community budget constraints, and the wage is set (i) equal to a level somewhat below
the market wage; and (ii) equal to the harvesting wage. Protective labor becomes a choice
variable for the forest communities, with the labor supplied paid at the exogenously-set rate.

The equilibrium results are shown in the third and fourth columns of Table 3.3. The variable
values from Table 3.2 are reproduced in the first two columns. The remaining columns in the
table show the percentage change in the variable values over the baseline case under co-
management (column (2)).

The results point to the importance of relative wages. The payment of protective services,
which had been offered voluntarily, improves incomes for the forest community and

50
provides an additional income-generating activity for both forest communities. Interestingly,
paying for protective labor does not increase its supply; rather, forest communities supply
approximately the same amount of labor to the forest department as if the labor was
voluntary but now enjoy higher incomes.

51
Table 3.3 Effect of co-management with proportional sharing of benefits, experiments (% change over base)

60% of timber sales revenue shared
0%
Base

P
w =2

P
w =3.5

X
β ↑

H
β ↑

df
β ↑

P
β ↑
H
β ↑ ,
P
β ↑

H
w ↓
H
w ↓ ,
P
w ↑

0
d
X
F →

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Income:
Forest comm.:
Of which: A
B
110.41
22.71
87.70
252.63
39.84
212.79
21.50
11.62
23.35
51.29
48.52
51.81
(3.12)
(11.87)
(1.48)
(2.00)
(0.93)
(2.19)
8.48
2.12
9.67
4.66
(13.96)
8.15
(0.95)
(9.15)
(1.39)
1.42
(16.60)
4.79
8.52
(7.36)
11.49
(51.13)
(63.18)
(48.87)
Market sector 48.97 76.89 24.52 (12.13) 33.05 (0.30) 0.21 37.72 (3.33) 42.94 30.27 (88.54)
Forest dept 189.58 171.28 9.52 14.91 0.88 (0.16) 4.97 2.18 (0.03) 2.24 3.15 (73.57)
Government 98.87 48.32 7.47 2.28 9.55 (0.09) 14.59 11.96 (0.93) 12.21 8.10 (76.12)

Consumption:
Forest Comm A:
Timber 9.62 9.83 67.96 123.91 (6.80) 6.71 (1.82) (7.63) 7.12 0.91 3.47 (35.32)
Market good 7.68 7.85 (10.51) 16.94 (14.39) 1.33 4.50 (16.82) 4.46 (19.88) (8.98) (40.42)
Collected 3.34 2.93 (65.19) (81.23) 23.40 (6.05) 14.59 34.81 (7.51) 15.10 4.61 65.67
Forest Comm B:
Timber 59.14 45.25 15.65 10.94 39.91 15.66 26.63 45.70 11.80 52.38 32.32 103.97
Market good 26.38 36.53 17.33 48.45 (8.87) (3.06) 9.55 6.52 (1.12) 3.22 15.07 (51.39)
Collected 115.66 68.15 (0.82) 3.24 (11.63) (8.02) (7.01) (12.12) (7.89) (10.90) (8.34) (8.75)
Market Sector:
Timber 21.81 19.85 0.20 11.69 (7.26) 5.59 7.70 15.47 5.64 21.23 21.96 (82.88)
Market good 16.40 18.53 19.00 (27.63) 41.09 2.53 1.30 42.80 (1.30) 52.33 35.46 (81.09)
Fines paid 0.00 63.59 17.82 32.24 (7.17) 0.55 7.28 (7.12) (0.80) (5.99) 1.05 (72.76)

52
Table 3.3 (Continued)

60% of timber sales revenue shared
0%
Base

P
w =2

P
w =3.5

X
β ↑

H
β ↑

df
β ↑

P
β ↑
H
β ↑ ,
P
β ↑

H
w ↓
H
w ↓ ,
P
w ↑

0
d
X
F →

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Welfare:
Forest comm 173.52 128.86 7.36 12.46 4.97 1.04 5.90 8.13 0.28 9.91 7.55 (19.85)
Of which: A 17.49 15.73 8.46 31.79 (5.37) 2.31 2.66 (5.79) 3.50 (5.60) (1.35) (42.91)
B 156.03 113.13 7.21 9.77 6.41 0.86 6.35 10.07 (0.17) 12.06 8.79 (16.65)
Market sector 29.57 26.25 11.01 (11.77) 19.16 2.86 3.47 28.00 0.72 34.28 25.63 (63.27)
∆ Biomass (47.54) 32.27 (36.66) (62.94) (20.20) (2.08) (24.98) (50.95) 10.07 (76.23) (54.57) (63.28)

53
The increased incomes fuel demand for harvested timber and for the market good. At the
lower protective wage level, the price increase in timber is relatively less than the increase in
the market good and there is some substitution into timber for the elite community. The non-
elite are able to support an increase in both timber and market good consumption.

The market sector expands, improving incomes for the market sector at the lower protective
wage. However when the protective wage is set equal to the higher harvesting wage, the
competition for labor becomes a significant factor, driving the market wage up and turning
market sector profits into losses. This loss of income inevitably leads to lower consumption
in the market sector and a fall in welfare.

At both levels of protective wages, the forest community as a whole sees improved welfare,
although the improvement for the elite is larger than the improvement for the non-elite,
worsening equity in the community.

Additionally, reforestation occurs but at a lower level than when protective labor is
voluntary because the higher level of timber production more than compensates for the fall
in collection activities.

Thus one can conclude that co-management schemes under which protective labor is
voluntary lead to welfare declines for both forest community sub-segments, although the fall
in welfare for the non-elite is more than twice as large (proportionally) than for the elite, and
are thus not likely to be sustainable in spite of the improvement in forest cover.

54
As an alternative, payment for protective services can be made, and does lead to higher
incomes and improved welfare for the forest community than when such labor is voluntary;
however, at some level market wage increases reduces welfare for the market sector and the
overall improvements are to some extent at the cost of resource improvement and
distributional equity.

3.5.2 Labor productivity improvements

The introduction of co-management has in Nepal been effected through the efforts of multi-
or bi-lateral aid organizations. Such intervention has included forestry and agricultural
education and other development efforts in conjunction with community forestry. The effect
of such efforts is to improve labor productivity in that area in which intervention is
concentrated. Columns 5 through 9 of Table 3.3 show the percent change in variables over
the baseline case given an increase of ten percent in the productivity parameter of labor in
various activities. In general, an increase in labor productivity will cause a reduction of labor
demand in that activity and a reallocation of labor into other activities. The general
equilibrium results depend on whether the labor is paid or unpaid, and if the wage paid is
fixed or variable. The results can be summarized as follows:

(i) Manufacturing productivity:

The percent changes over the variables in the baseline case of an improvement in
manufacturing productivity under co-management are shown in column 5 of Table 3.3. An
improvement in manufacturing productivity reduces the demand for manufacturing labor and

55
a fall in the market wage rate. This combination of decline in labor demand and fall in wages
leads to a decline in the income of the forest community but an increase in income for the
market sector as the more productive labor produces more market goods which are sold into
the forest community.

The decline in income for the forest community elite reduces its consumption of both timber
and market good and substitutes into collected forest products, which it does not relatively
prefer. The non-elite, meanwhile shift their consumption away from the market good and
into timber products, which it relatively prefers. These shifts in consumption lead to a
welfare decline for the elite and a welfare improvement for the non-elite.

The market sector is able to enjoy an improvement in welfare due to its increase in market
good consumption, which compensates it for the decline in consumption of the market good
by the forest community.

Whatever of the released labor not hired by the forest department is shifted into protective
activities. However, the increase domestic timber sales to the non-elite outweigh the benefit
derived from this labor and the improvement to the forest resource is not quite as large as in
the baseline case.

In summary, in comparison to the baseline case, an increase in manufacturing labor
productivity leads to improvement in welfare for the non-elite forest community and for the
market sector, a fall in welfare for the elite, and an increase in income for both the forest
department (due to higher timber sales) and for the government (due to higher taxes received

56
from market good sales). The forest resource improvement is smaller; nevertheless,
reforestation is still occurring. There is an increase in the both market good production and
timber harvesting. Equity in the forest community improves, but at the expense of the elite.

(ii) Harvesting productivity:

The percent changes over the variables in the baseline case of an improvement in harvesting
productivity under co-management are shown in column 6 of Table 3.3. An improvement in
harvesting productivity reduces the demand for harvesting labor. Because the harvesting
wage is fixed, this leads to a decline in the price of harvested timber. Consumption of
domestic timber consequently increases in all sectors (production of which is accommodated
by the improved productivity), but incomes of the forest community decline as a result of the
loss of this paid labor opportunity.

The labor released is spent on protective services, and the higher consumption of timber
allows a decline in collected products. In balance, there is a slight reduction in the level of
reforestation over the baseline case.

This substitution from collected products into harvested timber allows a small improvement
for welfare for both forest communities. The fall in the price of timber leads to an income
effect for the market sector allowing it to increase its consumption of both timber and the
market good and increasing its welfare.

57
While the forest department sells more timber, it is at a lower price, therefore forest
department income is unchanged. Government income is also unchanged.

In summary, in comparison to the baseline case, an increase in harvesting labor productivity
leads to improvement in welfare for all sectors, somewhat more for the elite among the
forest community than the non-elite. The forest resource improvement is slightly smaller;
nevertheless, reforestation is still occurring. There is a transition towards formal timber
harvesting and away from unauthorized collection. Equity between the forest communities
worsens.

(iii) Collecting productivity:

The percent changes over the variables in the baseline case of an improvement in collecting
productivity under co-management are shown in column 7 of Table 3.3. An improvement in
collecting productivity reduces the demand for collecting labor. This is an unpaid activity.
The released labor is spent in all other activities. The increase in market labor supplied
would cause a decline in the market wage; however, any additional labor spent in market
labor is paid, as is the additional harvesting labor, therefore incomes for the forest
communities increase.

This increase in incomes fuels demand for the market good and for harvesting timber,
encouraging production in both sectors and leading to increase in the market wage.

58
As incomes increase, consumption of domestic timber and the market good consequently
increases in all sectors. Although the market wage rises and the price of the market good
falls, the market sector’s income improves and it is also able to fund increased consumption.
The labor released is thus spent in market labor, harvesting labor and on increased protective
services.

The increased productivity allows for a rise in consumption of harvested timber and the
market good, raising welfare levels for the forest community, although somewhat more for
the non-elite. The market sector also experience a rise in welfare as its consumption
increases.

The forest department sells more timber at a slightly higher price, therefore forest
department income increases. Government income is slightly higher due to increased tax
revenue from the market sector.

In summary, in comparison to the baseline case, an increase in collecting labor productivity
leads to improvement in welfare for all sectors, somewhat more for the non-elite among the
forest community than the elite. The forest resource improvement is slightly smaller;
nevertheless, reforestation is still occurring. There is a transition towards formal timber
harvesting and market activities and away from unauthorized collection. Equity between the
forest communities improves.

59
(iv) Protecting productivity:

The percent changes over the variables in the baseline case of an improvement in protecting
productivity under co-management are shown in column 8 of Table 3.3. An improvement in
protecting productivity reduces the demand for protecting labor. This is an unpaid activity,
and its product is not a direct consumption good as is collected forest products.

The released labor is spent in market labor and harvesting labor. The increase in supply of
market labor leads to a fall in the market wage. Nevertheless, market labor is a paid activity
and therefore the forest community that shifts its labor away form protection sees an increase
in income. This is the non-elite in the forest community. The elite have already been
engaged in market labor and have not been engaged in protection, therefore the increase in
protection productivity does not affect their labor allocation choice. The fall in the market
wage, however, leads to a decline in the income of the elite.

The increased labor supply in the market sector causes a rise in market good output. This is
consumed by both the non-elite and the market sector; the elite having reduced its
consumption due to its fall in income.

Improved protection productivity makes more harvested timber available, as the resource is
improved. The non-elite shift their consumption of forest products from collected to
harvested timber, and the market sector increases its consumption of timber.

60
In spite of improved protective productivity, the rise in timber harvesting outweighs this
benefits and the forest resource improves at a much lower rate than the baseline case.

In summary, in comparison to the baseline case, an increase in protection labor productivity
leads to improvement in welfare for the non-elite and the market sector but a decline in the
welfare of the elite. The forest resource improvement is much smaller, although reforestation
is still occurring. There is a transition towards formal timber harvesting and market activities
and away from unauthorized collection. Equity between the forest communities improves,
but at the cost of elite welfare.

(v) Harvesting and protecting productivity:

The percent changes over the variables in the baseline case of an improvement in both
harvesting and protection productivity under co-management are shown in column 9 of
Table 3.3. The results above show that an increase in harvesting productivity reduces
demand for harvesting labor and improves welfare for all sectors as the community
transitions towards formal timber harvesting and away from informal collection, while an
increase in protection productivity worsens welfare for the forest elite because the release of
labor to the market sector induces a fall in wages for the elite and lower consumption
possibilities.

By increasing productivity in both harvesting and protection, labor demand in harvesting is
again reduced while still accommodating an increase in timber production and prices, and
therefore consumption. Incomes remain relatively flat, however, and the transition towards

61
timber harvesting and away from collection here allows an improvement in welfare for the
elite at the expense of a decline in welfare for the non-elite as their consumption of collected
products is not compensated by sufficient consumption of timber. Reforestation, however,
increases over the baseline case.

In summary, in comparison to the baseline case, an increase in harvesting labor productivity
accompanied by an increase in protection labor leads to improvement in welfare for the
forest community as a whole, for the market sector and in the forest resource, but masks a
fall in welfare for the non-elite as the community transitions towards formal timber
harvesting. Equity between the forest communities worsens. This is simply a worsening of
the case of “saving the forest at the expense of the poorest.”

3.5.3 Wage changes

There are two wages that are fixed in the community by the forest department: the wage for
harvesting labor and the wage paid for protective labor. In the baseline model, this last wage
is set to zero. The experiments above showed that if wages are paid for protective labor, both
forest communities see an increase in income and a consequent welfare improvement, but
dependent on the level at which the wage is set, the market sector can suffer from having to
compete for labor which all of a sudden has more higher paid alternatives. The end result in
both cases was a declining forest resource and widening community inequality.

Two additional experiments with wage setting are described below:

62
(i) Reduction in the harvesting wage:

The percent changes over the variables in the baseline case of a reduction in the wage paid
for harvesting labor under co-management are shown in column 10 of Table 3.3. The decline
in harvesting wage allows the forest department to hire additional harvesting labor,
increasing its domestic timber production. Consumption of timber therefore increases as the
price falls. This allows the non-elite to substitute away from collecting activities, offering its
released labor into the market. The market wage falls as a response to this increased supply
as production increases.

The reduction in both paid labor opportunities leads to a fall in income for the elite, who had
not had any unpaid labor in protective services and therefore are paid less for their paid
work. The non-elite, however, replaced some unpaid protective labor with paid market labor
and experience an increase in income. In the forest community as a whole, there is an
increase in timber consumption and a decline in collections. The market sector enjoys a
lower cost of production and therefore an improvement in welfare. The forest resource
suffers as a consequence from the increased timber production and decline in protection.

In summary, in comparison to the baseline case, a decrease in harvesting wage leads to
improvement in welfare for the forest community as a whole, and for the market sector, but
masks a fall in welfare for the elite as the community transitions towards formal timber
harvesting. Equity between the forest communities improves but at the expense of the elite
and at the expense of the forest resource.

63
(ii) Reduction in harvesting wage and concurrent increase in protective wage:

The percent changes over the variables in the baseline case of a reduction in the wage paid
for harvesting labor and a concurrent increase in the payment for protection labor under co-
management are shown in column 11 of Table 3.3. In this case, the shift of payment from
harvesting to protection is engineered so that the total payment for labor made by the forest
department is approximately unchanged.

Again because of the decline in the harvesting wage the forest department increases its
demand for harvesting labor, but less than without the payment of protection because this is
an added constraint on its budget. The release of labor into the market sector reduces the
market wage, but the opportunity to be paid for protection offers the non-elite another paid
labor choice and therefore additional income. This again is not a choice that the elite makes,
and the fall in the market wage and harvesting earnings combine to reduce its income. The
increase in the non-elite’s income is significantly larger than when protection was not paid.
As above, the increased income for the non-elite allows increased consumption and
improved welfare, but the elite experience a loss of welfare.

In summary, in comparison to the baseline case, an decrease in the harvesting wage
accompanied by payments for protective labor leads to improvement in welfare for the forest
community as a whole, for the market sector and in the forest resource, but masks a fall in
welfare for the elite as the community transitions towards the market sector. Equity between
the forest communities improves at the expense of the elite.

64
3.5.4 Preference changes

Productivity improvements are one result of intervention, effected through the education of
the workforce or introduction of new technology. An alternate type of intervention can be
the education of consumers, or the exposure of consumers to varying tastes resulting in
preference shifts. Three main examples of this are experimented with here. The first is an
increase in the relative preference for the market good by the non-elite (over forest
products); the second is an increase in the preference for the externality services of the forest
(by the elite and no-elite in turn); and lastly is an increase in the preference for leisure (again
by the elite and non-elite separately). The results are as follows:

(i) Increase in relative preference for the market good over forest products by the
non-elite:

The percent changes over the variables in the baseline case of an improvement in the relative
preference for the market good over forest products are shown in column 3 of Table 3.4. An
increase in the taste preference for the market good by the non-elite increases the demand for
the market good, inducing an increase in the demand for market labor; at the same time there
is a fall in the demand for forest products, which induces a fall in the demand for harvesting
labor and a fall in the supply of collecting labor. Because the demand for market labor is met
by an increase in the supply of market labor, the market wage falls. This leads to a reduction
in income for the elite, all of whose labor had been paid, but an increase in the income of the
non-elite as they shift from non-paying collection and protection activity toward paid market
labor.

65
The fall in the market wage increases profits for the market sector and this sector therefore
experiences an increase in welfare as its consumption of both timber and the market good
increases. Similarly, the increase in income for the non-elite forest community allows an
increase in consumption of the newly-preferred market good and timber, and a higher
welfare. The fall in income for the elite, however, reduces their consumption possibilities in
spite of the fall in price of the market good and they experience a marginal decline in
welfare.

In summary, in comparison to the baseline case, an increase in the relative preference for the
market good by the non-elite leads to improvement in welfare for both the non-elite and for
the market sector but a small decline in welfare for the elite. Equity for the forest community
therefore improves, but at the expense of the elite. There is a shift into harvested timber and
away from collection activities, and an expansion in the market sector, but the improvement
in the forest resource falls as a result of the increased harvesting and decline in protection.

(ii) Increase in preference for the externality services of the forest:

The percent changes over the variables in the baseline case of an increase in the relative
preference for externality services under co-management are shown in columns 4 and 5 of
Table 3.4. When a community increases its preference for the externality, it concurrently
decreases its preference for consumption.

66
Table 3.4 Effect of co-management with proportional sharing of benefits, experiments (% change over base)

60% of timber sales revenue shared
0%
Base
1
CB
β ↓ ,
3
CB
β ↑

ZA
β ↓

ZB
β ↓

4
CA
β ↑

4
CB
β ↑

0 ≠
Ext
P

Ext
P ↑

0 ≠
fr
β

fr
β ↑

S
df
β ↑
X Pref ↑ Externality Pref ↑ Leisure Pref ↑ Pay for Externality NTFP Trade Allowed
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Income:
Forest comm
Of which: A
B
110.41
22.71
87.70
252.63
39.84
212.79
7.53
(11.23)
11.04
9.32
(11.10)
13.14
6.86
(9.57)
9.93
7.06
(11.12)
10.47
7.54
(9.94)
10.81
6.58
(20.24)
11.60
7.43
(16.23)
11.87
25.77
86.14
14.47
32.13
185.03
3.51
40.69
138.89
22.30
Market sector 48.97 76.89 40.08 35.41 30.62 31.95 31.79 45.07 43.85 29.93 101.87 97.17
Forest dept 189.58 171.28 2.25 2.98 3.26 4.36 3.89 (5.75) (5.56) (13.42) (12.94) (12.87)
Government 98.87 48.32 13.31 14.52 10.75 11.85 11.48 33.91 36.76 5.88 13.67 18.19

Consumption:
Forest Comm A:
Timber 9.62 9.83 (1.32) (6.02) (4.88) (4.62) (5.60) 5.68 50.42 32.31 (8.64) (19.89)
Market good 7.68 7.85 (14.18) (12.73) (11.25) (13.46) (11.51) (28.25) (32.39) 111.05 262.00 197.94
Collected 3.34 2.93 18.82 36.39 18.86 15.82 20.50 152.03 130.13 (8.37) 30.47 47.80
Forest Comm B:
Timber 59.14 45.25 49.44 72.06 46.18 53.66 50.56 31.75 27.74 32.98 1.76 38.36
Market good 26.38 36.53 10.96 9.45 8.55 7.65 9.05 19.12 20.72 24.40 14.91 37.70
Collected 115.66 68.15 (11.91) (13.31) (10.39) (9.91) (10.29) (27.15) (26.71) (47.53) (48.57) (51.75)
Market Sector:
Timber 21.81 19.85 22.34 15.62 12.44 13.42 14.04 2.59 0.15 (76.15) (87.24) (66.30)
Market good 16.40 18.53 43.84 39.23 35.69 37.85 37.22 43.60 42.40 (45.72) (46.14) (9.48)
NTFP * * *
Fines paid 0.00 63.59 (8.91) (10.64) (3.81) (2.64) (3.23) (33.96) (35.07) (39.30) (33.11) (43.18)

67
Table 3.4 (Continued)

60% of timber sales revenue shared
0%
Base
1
CB
β ↓ ,
3
CB
β ↑

ZA
β ↓

ZB
β ↓

4
CA
β ↑

4
CB
β ↑

0 ≠
Ext
P

Ext
P ↑

0 ≠
fr
β

fr
β ↑

S
df
β ↑
X Pref ↑ Externality Pref ↑ Leisure Pref ↑ Pay for Externality NTFP Trade Allowed
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Welfare:
Forest comm 173.52 128.86 9.70 14.48 9.32 11.14 10.32 4.11 4.83 0.45 (3.98) 3.36
Of which: A 17.49 15.73 (3.27) (2.86) (3.43) (4.52) (3.55) (0.36) 7.56 52.02 81.07 61.35
B 156.03 113.13 11.50 16.89 11.09 13.31 12.25 4.73 4.46 (6.72) (15.81) (4.70)
Market sector 29.57 26.25 30.93 26.96 23.47 25.04 24.85 28.79 27.94 (44.91) (48.02) 19.06)
∆ Biomass (47.54) 32.27 (69.66) (92.31) (58.32) (73.07) (70.56) 2.97 (2.01) 57.02 103.32 41.31

68
In the case of the elite, it reduces its consumption of the market good and of harvested
timber and its supply of labor to the market. The reduction in supply of labor induces a fall
in the market wage, but encourages labor supply from the non-elite, which substitutes away
from unpaid labor activities (protection and collection) into market labor.

The non-elite therefore enjoy an increase in income and are able to support higher levels of
consumption, leading to increased welfare. The increased consumption levels
overcompensate for the elite’s reduction and as a result the forest resource suffers. The
externality value of the forest declines, and the elite end up with lower consumption and less
of an externality in spite of it being relatively more preferred.

If the preference shift were to occur in the non-elite community instead, the results are
somewhat similar. The non-elite shift their labor away from collection and into the market.
Although the wage rate falls, this is a shift into a paid activity and the non-elite enjoy an
increase in income, which allows higher levels of consumption of both timber and the
market good. Increased consumption outweighs the benefit of improved appreciation for ht
forest externality and the resource suffers. The elite community again is negatively impacted
by the fall in the market wage and experiences a welfare decline.

In summary, an increase in the preference for the externality services of the forest leads
unexpectedly to higher incomes for the non-elite and the market sector, an increase in timber
harvesting and in the market sector, and a slower rate of regeneration in the forest. Equity in
the forest community improves as the non-elite make welfare advances but at the expense of
the elite.

69
(iii) Increase in preference for leisure:

A community that increases its preference for leisure is expected to increase the amount of
time it spends at leisure. The percent changes over the variables in the baseline case of an
increase in the relative preference for leisure under co-management are shown in columns 6
and 7 of Table 3.4.

For the elite, the increase in leisure time is taken from market labor. This leads to a decline
in income and a reduction in consumption. The consumption fall is not compensated by the
increase in leisure and the elite experience a welfare decline.

For the non-elite, the increase in leisure time is taken from protective labor, which is unpaid.
Its increase in leisure preference is at the expense of forest products and the market good,
which therefore allows the community to release labor from unpaid collection activity also.
The decrease in demand for the market good induces a fall in its price, ultimately increasing
the quantity demanded. The market sector expands and income for the non-elite improves as
it is hired in the market. The increase in income allows higher levels of consumption in
addition to the increase in leisure time, leading to welfare improvement.

In summary, an increase in the preference for leisure time leads to an increase in income for
the non-elite and the market sector, an increase in timber harvesting and in the market sector,
and a slower rate of regeneration in the forest. Equity in the forest community improves as
welfare for the non-elite improves against a decline in welfare for the elite.

70
3.5.5 Payment for externality services

One innovation has been to offer payment to the user community for improvements in the
externality of the natural resource [citations needed]. Two experiments are attempted here:
the first is the introduction of such payment, and the second is an increase in the payment
rate. The results can be summarized as follows:

(i) Payment for externality improvements allowed:

The impact of introducing such a payment will be dependent on the level of payment. In this
experiment it is exogenously set at a relatively low rate, approximately ten percent of the
price of export price of timber. The percent changes over the variables in the baseline case of
a payment for externality services are shown in column 8 of Table 3.4.

Once payment is made possible, the provision of protective labor is made more attractive
thus there is an increase in the supply of labor to protection. Similarly, there is a disincentive
to collection, thus labor supply to collection activities decreases. The shift from collection to
protection leaves some surplus labor, which is offered to the market, inducing a fall in the
market wage rate. At the same time, the reduction in collection forces a substitution towards
harvested timber, increasing harvesting employment and timber production.

The combination of these effects leaves the elite with a lower income, a fall in consumption
and a slight decline in welfare. The non-elite experience an increase in income, higher levels
of consumption and welfare improvement. The market sector is able to enjoy higher profits

71
and therefore higher welfare levels. The increase in harvesting is more than compensated for
by the decline in collection and the increase in protection so the forest resource continues its
improvement at a slightly higher rate.

In summary, allowing the payment for externality serves to shift labor choices such that the
market wage falls, reducing the income of the elite and increasing incomes for the non-elite.
Equity therefore improved between the forest communities but at the expense of the elite.

(ii) Increase in payment rate:

The percent changes over the variables in the baseline case of a ten percent increase in the
payment fro externality services are shown in column 9 of Table 3.4. An increase in the
price paid for the externality has positive effects on both forest communities: the non-elite
shift some labor from the market to protection, inducing an increase in the market wage,
from which both communities benefit. The shift to protection and reduction in collection
results in a substitution towards domestic harvested timber.

The consumption effects of these income changes result in improved welfare for the elite
and a slightly lower welfare for the elite as compared to the initial payment scheme. The
forest resource regeneration continues at a marginally lower rate. Equity in the forest
community degrades over the baseline case but both forest communities benefit and the
forest resources regenerates; this can therefore be seen as a positive scenario.

72
3.5.6 Local trade in NTFPs

Another oft-cited intervention is the promotion of local trade in non-timber forest products,
such as medicinal herbs, mushrooms, etc. as a way of earning additional livelihood options
without widespread destruction of the resource. The domestic price for these products is
determined endogenously. Three experiments are shown here: first, allowing trade to occur;
second, increasing the local preference for NTFPs; and lastly increasing the productivity of
production of NTFPs. The results can be summarized as follows:

(i) Local trade in NTFPs allowed:

Forest communities now have an additional livelihood option, and the market sector is given
an additional consumption product. The percent changes over the variables in the baseline
case resulting from the introduction of local trade in NTFPs under co-management are
shown in column 10 of Table 3.4. The forest communities shift their labor supply from
collection for self-consumption to forest protection, to the market sector and to collection for
sale purposes. The share of labor in these activities is largely altered: the elite shift their
labor from the market to collection of NTFPs, while the non-elite shift from (unpaid)
collection to the (paid) market. Both forest communities see a significant increase in income,
as does the market sector as this increase in demand funnels into additional demand for the
market good.

The increase in demand and competition for labor leads to price increases for the market
good. The elite are able to support an increase in consumption of harvested timber and of the

73
market good (as are the non-elite) and enjoy a significant welfare improvement. However,
the non-elite suffer a significant reduction in the amount of colleted forest products which is
not compensated by the increase in other consumption; therefore, on balance the non-elite
see a welfare decline. The reduction in informal collection combined with increased
protection produce improved regeneration of the forest resource.

In summary, this is another example of saving the forest at the expense of the poorest.

(ii) Increase in preference for the NTFPs:

The percent changes over the variables in the baseline case of an increase in the relative
preference for NTFPs under co-management are shown in column 11 of Table 3.4. An
increase in the preference for NTFPs will raise demand for these products, inducing a price
response and a supply response. More labor is shifted to collecting NTFPs for sale by the
non-elite, and labor is shifted from protection to the market sector. The market wage falls
such that income for the non-elite suffer compared to the original preference structure, and
they experience a decline in welfare.

The increase in price of NTFPs leads to a significant increase in the income of the elite, who
substitute into consumption of the market good generating a boom in the market sector,
additional hiring and a fall in the price of the market good. Nevertheless, the increased price
of NTFPs erodes the welfare of the market sector.

74
In summary, an increase in the preference for NTFPs leads to lower market wages, higher
prices of NTFPs, a decline in the welfare of the non-elite (who are providing the market
labor), and increase in the welfare of the elite (who are supplying the NTFPs) and a decline
in welfare for the market sector. The forest resource regenerates at a rate higher than ever as
collection activities and harvesting decline and the collection of NTFPs, although higher, is
done in a sustainable manner.

(iii) Increase in productivity of production of NFPs:

Rather than increasing the preference for NTFPs, imagine that the forest communities
receive an intervention that educates them about the sustainable production of NTFPs and
improves their yield. The percent changes over the variables in the baseline case of an
increase in the labor productivity of NTFP production under co-management are shown in
column 12 of Table 3.4.

Now more productive, the forest communities are able to spend less time at collection of
NTFPs and still produce the same or more income from this activity. The released labor is
sent to the market, which rewards the increased supply with a lower wage rate.

Nevertheless, both forest communities enjoy higher levels of income than with the original
production technology and also in comparison to the baseline case.

There is a shift to market good production and away from harvested timber as competition
for harvesting labor leads the forest department to raise its domestic price of timber. While

75
the elite enjoy this substitution and experience welfare improvements, the non-elite still
relatively prefer forest products and suffer a welfare decline as a result of the shift in
consumption. The forest resource regenerates at a rate better than in the baseline case but not
quite as high as with the original production technology.

3.5.7 Loss of External Funding

So far, we have been examining potential policy interventions which are directed towards
improving livelihood support and distributional equity while maintaining an improving
resource. A cautionary note sounded in the review literature has been the sustainability of
any community forestry regime in the absence of significant continuing external support
(FAO, 2006; FAO, 1999). This is of particular concern in implementations where the forest
resource is not commercially viable and therefore the potential of revenue-sharing to impact
local incomes is limited or non-existent.

To examine how important the contribution of external support is to community forestry, we
experiment with a withdrawal of external demand for timber. Timber exports provide the
forest department with substantial financial resources and a derived demand for harvesting
labor. Thus not only do community households benefit from the availability of wage income,
but the sharing of timber revenues with the forest communities under co-management
provide significant income injections which cycle through the local economy. Removal of
this injection will have profound effects. The percent changes over the variables in the
baseline case of a removal of timber export demand under co-management are shown in
column 12 of Table 3.3.

76
The loss of income is deep to both the elite and the non-elite. This loss of income robs them
of the ability to purchase both timber and the market good, reducing the income of the
market sector, the forest department and the government. The elite, who were relatively
more employed in harvesting labor than the non-elite, experience a greater fall in income and
curtail their purchases of all products, leaving them only the option of increasing their
consumption of collected products. Price declines in timber allow the non-elite to increase
their purchases. Together, all sectors of the community contract and face severe welfare
declines. The forest resource regenerates, but at a slower rate than the base case given the
increase in collections in spite of the loss of exported timber.

The deterioration of economic conditions in this scenario highlights the importance of
external support. In prior scenarios, the external funding came in the form of demand for
locally harvested timber, producing labor opportunities and providing revenues to be shared
with local communities. The withdrawal of such external support leads to economic
contraction and significant welfare declines for all.

3.6 Conclusions

While co-management of forest resources has not been found to be consistently successful in
improving distributional equity, this paper explores through simulation exercises possible
improvements in co-management policies that may lead to more sustainable economic and
ecological outcomes. General equilibrium effects through wage and price responses can
yield results that are unanticipated.

77
A baseline equilibrium is established which is fairly consistent with results reported from a
variety of case studies. These studies show that once co-management is introduced into a
forest community, welfare declines are experienced by both elite and non-elite classes as a
result of three effects: (1) an increase in the domestic price of timber leads to substitution
away from timber; (2) the imposition of fines and exclusionary policies discourages informal
collection of forest products, upon which the non-elite are more dependent; and (3) the
general price increase in the market good reduces consumption possibilities by both the elite
and the non-elite. While both the elite and non-elite experience welfare declines, the decline
is sharper for the non-elite because of their relative dependence on forest products from
which they are being excluded.

Simulation exercises are run for three types of interventions:
1. Wage policies
2. Productivity enhancements
3. Preference shifts

Additionally, payment for externality services was introduced as a policy innovation, and the
production and trade of NFTPs was considered.

Over the baseline equilibrium case of co-management, paying wages for protection or
management services can lead to the improvement of welfare for all sectors of the forest
community and for the market sector. The wage rate paid in comparison to competitive wage
rates in the economy is a critical factor in determining whether the market sector thrives or
withers: the forest department can squeeze the market sector out of the labor market by

78
offering high enough wages. In either case, however, increased timber production outpaces
improved management and protection and improvement in the forest resource falters.
Additionally, increasing the wage earning opportunities benefits the elite more than the non-
elite and equity within the forest community suffers.

Similarly, a decline in the wage paid by the forest department for harvesting labor leads to
increased timber production and a fall in the domestic price of timber. The non-elite benefit
both from the additional hiring opportunities and from the fall in the timber price, while the
elite suffer from a loss of income. Equity within the community improves at the expense of
the elite. If this wage decline is accompanied by a payment for protection or management
services, these effects are moderated. Again, in both cases the improvement of the forest
resource suffers.

Intervention in the form of productivity enhancements can lead to a mixed combination of
equity improvement, welfare improvement across sectors and resource preservation. The
most promising result is achieved if the productivity improvement is concentrated in
collection activity, which then yields a surplus of labor that can be released to the market and
formal timber industry. While equity improvement and overall welfare improvement come at
some expense of the forest resource to some extent, regeneration still occurs, thus this can
generate a win-win-win scenario. Alternatively, a combination of productivity improvements
in harvesting and in protection can yield minimal impact on the forest department and the
non-elite sector while providing welfare improvement for the elite and the forest.

79
Intervention in the form of preference shifts in general cause a variety of welfare
improvements in the forest community, commonly with a decline in welfare for the elite and
always with a decline in forest regeneration.

The most promising institutional innovations come in the final two sets of experiments: the
introduction of payment for externality services and the promotion of NTFP production and
trade. It appears that the externality payment can be set so that welfare improvements are
seen in all sectors accompanied by continued forest regeneration, and possibly equity
improvement. Further experimentation with in this area is needed. The shift of domestic
production from collection for consumption to collection for the production of NTFPs for
sale also has the potential to realize large improvements in the forest, but negative welfare
effects for the non-elite can be significant. Nevertheless, experimentation with this
innovation appears promising.

The clearest experiment in terms of local improvement is the continuation of external
support. In this model, such support is provided by external demand for locally harvested
timber. In practice, however, funding is provided through the participation of international
aid organizations such as the World Bank and the Food and Agricultural Organization, or
through bilateral commitments such as have been made by DANIDA (Denmark), NUKCFP
and DFID (UK), GTZ (Germany) and NSCFP (Switzerland). It is clear that the sustainability
of community forestry hinges on the continued participation by external partners.

80
Chapter 4
Behavioral Responses to Forestry Management Regimes

Community forestry or joint forest management has been promoted as an institutional
innovation to advance both ecological and economic sustainability of forest conditions and
improve livelihoods of forest users, but consistent positive results have not been widely
achieved. Among policies that commonly characterize joint forest management,
exclusionary rules put into place to achieve ecological sustainability limit access of forest
users to the natural resource base on which their economic livelihoods depend. The most
marginal of households are especially subject to negative welfare effects.

A household’s response to the implementation of use restrictions depends on its cost-benefit
calculations. Faced with having to adhere to new rules, the household responds by weighing
the relative costs of its options to obey or cheat. The cost of loss of resource use is weighed
against the potential fine from rule-breaking, which will depend on the likelihood of being
caught, itself a function of the effectiveness of the monitoring system. The fine might be
supplemented by potential social effects if the infraction is serious enough, also perhaps the

81
limitation of future access to a presumably regenerated resource, given that the household is
a member of a community charged with the responsibility of total resource protection.
The cost of the loss of resource use itself depends on how dependent the household is. Are
there substitutes available? Are these substitutes costly?

On the other side of the equation is the future benefit to be received from a regenerated
resource. The community decision to forgo current resource use in favor of future resource
use implies a negative discount (time preference) factor, which is presumed to be necessary
to protect the resource but occurs at the expense of current consumption. With few
substitutable options available, one is left to wonder how households cope and to what extent
are the negative effects dependent on household wealth.

Results obtained through simulation exercises of a heterogeneous forest community under a
co-management regime (see Chapter 3) suggest that there are combinations of conditions
under which ecological sustainability and pareto improvements can be obtained, but that
distributional equality suffers under most conditions. The dependence of the poorest on the
forest resource from which they are excluded under management regimes, and price shifts
which disadvantage their consumption and production choices, conspire to reduce the
welfare of the poorest while simultaneously improving the welfare of the non-poor.

This chapter investigates these results empirically on a unique set of data collected from the
middle hills region in Nepal, which has been extensively experimenting with community
forestry. The data surveyed forest conditions, forest community user groups and households
within these communities for their behavioral responses under three separate types of

82
regimes: formal forest user groups (“FFUG”), which are defined as user groups having been
given formal use and access rights from the Department of Forests and which have thus
established a set of use and access rules; informal forest user groups (“IFUG”), which are
organized groups of community members establishing rules but without de jure enforcement
mechanisms; and open access which occurs in the absence of any forest user group
management and under which access to forests and forest products is not limited either de
jure or de facto (“No FUG”).

4.1 Data

4.1.1 Nationwide Forest Data

Tachibana, et al. (2001) conducted surveys in a random sample of forests across the middle
hills region of Nepal in 1995 to examine the effect of user group management on forest
regeneration. Communities using the sampled forests were interviewed regarding their
organizational structure, management plans, restriction rules, enforcement procedures, and
so on. Thereafter a sample of households in each of the forests was surveyed for basic
demographic information, agricultural and livestock activities, and labor allocation to forest
resource usage. In addition, an evaluation of resource conditions was also conducted,
measuring the regeneration of trees within the sampled forests by examining saplings and
seedlings. This nationwide data set differs from the case study literature in which a single
village or a small set of forest communities in a few districts is examined in detail. The
household survey does not address consumption or expenditure, however, so the potential
for welfare analysis is limited.

83
Just less than half of the sampled forests in the Tachibana study are governed by formal
forest user groups, having been identified by the Department of Forests and assisted in user
group formation, and most importantly granted official property rights to their community
forest. An additional third of their sample is governed by informal user groups, which may
formulate their own management rules and access restrictions but which have limited de jure
rights to enforce these rules. The remaining forests in the sample are subject to open access.

Their findings confirm theoretical postulation that user groups are more likely to emerge
when the resource is degraded to such an extent that its marginal value exceeds the cost of
organizing collective action. Organizational and transactions costs include the number of
user households, the number of administrative wards with jurisdiction over the forest and the
availability of the local forest office (which is responsible for implementing the community
forestry plan).

They also find that management of the forest by formal forest user groups is more effective
at regenerating forest resources than management by informal user groups or than without
management at all. The support of the Department of Forests in granting rights to the forest
to the user group, thereby allowing de jure as well as de facto enforcement of use
restrictions, appears critical in the effectiveness of the management regime.

Beyond the determinants of user group formation and the effectiveness of community
forestry at resource protection, Tachibana, et al. examine behavioral responses of
community members to the rules established by the prevailing user group through the
amount of labor allocated by households to firewood collection from the sample forests.

84
They estimate a household’s utility maximization function measured by the consumption of
firewood (which is a function of the household’s allocation of labor to the collection of
firewood), conditioned on forest characteristics, household characteristics and the
institutional framework governing the forest.

In this study, they find that exclusionary rules implemented and enforced by formal forest
user groups are the most effective means of reducing over-extraction of firewood from
community forests, which leads to resource preservation.

Clearly, simply prohibiting or limiting resource extraction can improve the resource, and
careful sales of excess timber or cultivated products can lead to higher incomes for the
community was a whole. Yet the distributional effects and price responses have differential
impacts on the non-poor and the poor. The Tachibana study is silent on distributional
concerns, which are significant given the reliance of the poorest forest community
households on the availability of common-pool resources for their livelihoods. These
multiple and interactive effects motivate the following analysis.

To begin to examine differential effects of community forestry on households within varying
wealth classes of forest communities, this section builds upon the Tachibana work. The
Tachibana study did not find that land ownership or caste had a significant effect on the
labor allocation decision (and thus on the amount of firewood collected from the sample
forest) under any type of management regime. This finding implies that wealth is not a
determining factor in the allocation of benefits from the common-pool resource, in spite of
findings from case studies as described in an earlier chapter.

85
These inconsistent results may be due to the pooling of the data and use of simple dummy
variables to test for differences among management regimes. It is plausible that household
behavior varies across regime types. For example, land ownership may well matter for
unmanaged forests, where land ownership confers wealth status and political power, but it
may not affect resource use rights under formal management if the distribution of forest
products is made equitably. Therefore, we divide the data by management regime and
estimate labor allocation decisions of each separately.

Table 4.1 displays a selection of household characteristics by management regime with the
pooled data shown in the first column (displayed more fully in Table 4.2). Although average
land ownership does not differ among groups, land ownership per active household member
is significantly higher in households without management implying a higher density living.
Household labor allocation to firewood collection is also not significantly different among
groups but time spent collecting of fodder is significantly higher for households without
management. This is hypothesized to be because households under formal management have
substitutable resources in their own private inventory of trees, which we do see, and because
households under no management have more livestock. The average livestock holdings are
shown as an indication of demand for fodder. There has been a movement towards stalling
of livestock as grazing restrictions are enforced in community forests under management.
The average value of assets owned is significantly higher in households under formal
management, where total assets include the market value of all livestock and land owned.

86
Table 4.1 Sampled Forest Dataset Descriptive Statistics – Household Characteristics

Variable Pooled FFUG IFUG No FUG

Average land owned (ropani) 20.97
(18.58)
20.55
(20..73)
20.37
(15.64)
22.10
(17.71)
Average land owned per active HH
member
8.39
(9.06)
7.74
(8.56)
7.71
(6.84)
9.91
(11.10)
Average labor days collecting firewood
from community forest
32.56
(36.51)
31.77
(43.91)
33.20
(30.10)
33.15
(29.38)
Average labor days collecting fodder from
community forest
39.71
(72.99)
23.30
(49.15)
31.75
(58.98)
70.13
(99.02)
Average # of private trees owned 163.38
(479.49)
206.90
(634.31)
124.52
(243.65)
134.68
(357.94)
Average head of livestock 6.01
(3.35)
5.33
(2.77)
6.23
(3.37)
6.79
(3.87)
Average head of oxen 1.75
(1.04)
1.65
(1.04)
1.82
(0.90)
1.85
(2.13)
Average head of cattle 1.65
(1.85)
1.36
(1.40)
1.59
(1.74)
2.13
(2.37)
Average head of grazing livestock 2.45
(3.55)
1.46
(2.60)
2.90
(3.55)
3.47
(4.31)
Average head of stall-fed livestock 3.56
(2.83)
3.87
(2.57)
3.33
(3.03)
3.33
(2.94)
Average total assets (1000 Rs.) 579.91
(1071.14)
661.53
(967.10)
592.22
(1344.98)
452.28
(931.84)
Average % of high ethnicity 61.41
(48.73)
75.81
(42.92)
50.77
(50.19)
50.00
(50.17)

87
Table 4.1 (Continued)

Numbers in parentheses are standard deviations.
Source: Tachibana, et al. (2001)

Variable Pooled FFUG IFUG No FUG

Average daily wage (Rs.) 39.62
(15.48)
41.55
(17.42)
38.38
(12.24)
37.93
(14.83)
Average grain production (kg) 744.60
(1229.41)
980.64
(1530.76)
687.12
(1121.75)
456.08
(623.80)

Obs 495 215 130 150
Forests 99 43 26 30

88
Table 4.2 Sampled Forest Dataset Descriptive Statistics – Household Characteristics
(Pooled)

495 household observations in 99 forests sampled
Source: Tachibana, et al. (2001)

Table 4.3 shows the calculated average resource extraction in the dataset, and Table 4.4
shows the data by management regime. Average firewood collections are not much different
by regime, but fodder, leaf litter and grasses are: all are higher without management than
under management, but leaf litter and cut grass collections are higher under formal
management than under informal management.
Variable Mean SD Min Max

Average land owned (ropani) 20.97 18.58 1 150
Average land owned per active
member
8.39 9.06 .33 75
Average labor days collecting
firewood from community
forest
32.56 36.51 0 336
Average labor days collecting
fodder from community forest
39.71 72.99 0 365
Average # of private trees
owned
163.38 479.49 .001 5600
Average head of livestock 6.01 3.35 0 19
Average head of oxen 1.75 1.04 0 6
Average head of cattle 1.65 1.85 0 15
Average head of grazing
livestock
2.45 3.55 0 19
Average head of stall-fed
livestock
3.56 2.82 0 17
Average total assets (1000 Rs.) 579.91 1071.13 0 14100
Average % of high ethnicity 61.41 48.73 0 100
Average daily wage (Rs.) 39.62 15.48 15 120
Average grain production (kg) 744.60 1229.41 0 15600

89
Table 4.3 Descriptive Statistics – Forest Benefits (Pooled)
495 household observations in 99 forests sampled
Source: Tachibana, et al. (2001)

Calculating the value of all products extracted from the community forest shows that
households without management extract the most average forest resources, those under
formal management approximately twenty percent less, and those under informal
management extract the least amount. Alternatively, total forest products from all sources,
including private resources and government forests, are highest for households under formal
management as they substitute resources from private land. Households without
management received a higher percent of their forest products from the community forest
than those under management, and under informal management higher than under formal
management.
Variable Mean SD Min Max

Average firewood collected
from sample forest (#
backloads/year)
24.42 33.16 0 335.08
Average fodder collected
from sample forest
34.22 75.86 0 478.68
Average leaf litter collected
from sample forest
29.87 59.99 0 364
Average cut grass collected
from sample forest
30.17 69.58 0 628.27

Average forest benefits from
sample forest (N.Rupees)
4429.35 6326.19 0 37800
Average total forest
resources from all sources
55101.66 269,788.90 0 3799710
% of forest benefits from
sample forest
41.08 39.41 0 100

90
Table 4.4 Sampled Forest Dataset Descriptive Statistics – Household / Forest Benefits

Variable Pooled FFUG IFUG No FUG

Average firewood collected from sample
forest (# backloads/year)
24.42
(33.16)
22.72
(37.84)
26.41
(28.70)
25.12
(29.46)
Average fodder collected from sample
forest
34.22
(75.86)
18.35
(45.49)
25.48
(56.54)
64.54
(109.48)
Average leaf litter collected from sample
forest
29.87
(59.99)
31.31
(71.04)
22.73
(39.95)
33.98
(56.90)
Average cut grass collected from sample
forest
30.17
(60.58)
27.20
(69.94)
21.95
(45.57)
41.57
(83.81)

Average forest benefits from sample forest
(N.Rupees)
4,429.35
(6,326.19)
4,235.64
(6,461.51)
3,363.93
(4,400.73)
5,630.35
(7,311.49)
Average total forest resources from all
sources
55,101.66
(269,788.90)
82,756.24
(362,635.60)
32,046.21
(165,676.00)
35,444.82
(163,477.00)
% of forest benefits from sample forest 41.08
(39.41)
38.82
(40.58)
39.87
(38.22)
45.29
(38.63)

Obs 495 215 130 150
Forests 99 43 26 30

Numbers in parentheses are standard deviations.
Source: Tachibana, et al. (2001)

91
To examine the data more carefully, we use a number of techniques. We expect the asset
data in general to be skewed to the right, indicating that the relatively wealthy would
command much higher levels of assets than the relatively poor. To confirm this supposition,
we look at asset data in three categories: land ownership, livestock ownership and total forest
products from all sources. Symmetry plots are constructed and shown in Figure 4.1. The
construction of symmetry plots is as follows: Each point represents a consecutive high and
low pair of variable values which is compared to the median. For example, the median land
owned is 15.5 hectares. The highest value is 150; the lowest is 1. This pair forms the point
{14.5, 134.4} at the upper right extreme of the graph. If the data was symmetrically
distributed, each observation pair would lie along the reference line. In each plot, however,
points lie above the reference line, indicating that the distribution of the variables is skewed
to the right as suspected.

Figure 4.1 Symmetry Plots

0 50 100 150
Distance above median
0 5 10 15
Distance below median
Symmetry Plot for Land Ownership

92
Figure 4.1 (Continued)

0 5 10 15
Distance above median
0 2 4 6
Distance below median
Symmetry Plot for Livestock Ownership

0 100000 200000 300000 400000 500000
Distance above median
0 2000 4000 6000 8000 10000
Distance below median
Symmetry Plot for Total Forest Benefits

How bad is the skewness? We compare the distributions of these variables to normal
distribution in Figure 4.2. If the observations were normally distributed, they would lie along
the reference line. All plots again show the skewed distributions.

Identifying and removing outliers may help to normalize the data somewhat. Outliers here
are defined as observations lying outside the range defined as three times the interquartile
range above and below the midquartile value. For land ownership, five outliers at the top end

93
of the distribution are identified, but the other variables are more problematic, having large
numbers of outliers (or very fat tails): for livestock ownership there are twelve; for total
forest benefits there are forty-one; for tree ownership there are twenty-nine – all in the upper
tail. In fact, the nature of these communities would lead us to expect very unequal
distribution, especially at the upper end, and the presence of outliers. Nevertheless, extreme
outliers should be considered; therefore the five outlier observations in land ownership are
removed from the data and from further analysis.

Figure 4.2 Deviation from Normal Distribution

3 15.5 57
-50 0 50 100 150
ownland
20.9741 51.53792 -9.589722
-50 0 50 100
Inverse Normal
Grid lines are 5, 10, 25, 50, 75, 90, and 95 percentiles
Deviation from Normal Distribution - Land Ownership

2 6 13
-5 0 5 10 15 20
livestock_tot
6.008081 11.51339 .5027668
-5 0 5 10 15
Inverse Normal
Grid lines are 5, 10, 25, 50, 75, 90, and 95 percentiles
Deviation from Normal Distribution - Livestock Ownership

94
Figure 4.2 (Continued)

0 9700 169441
-1000000 0 1000000 2000000 3000000 4000000
tot_benefits_t
55101.66 498865 -388661.6
-1000000 -500000 0 500000 1000000
Inverse Normal
Grid lines are 5, 10, 25, 50, 75, 90, and 95 percentiles
Deviation from Normal Distribution - Total Forest Benefits

4.1.2 Distribution of Forest Benefits by Wealth Class

The data allows some examination of the flow of forest products received by households.
Regardless of the labor allocation, Malla (2000) indicates that the product flows from the
community forest are distributed inequitably.

Using the data set, benefits are calculated as follows: Five different products are extracted
from the forests: green wood, dead and dried firewood, fodder, leaf litter and cut grasses.
The first two of these are valued at the recorded local market price per quantity collected.
The other three products are valued at the opportunity cost of labor, which in this case is the
daily wage rate for female labor for fertilizer application work. Households are divided into
land-wealth classes by landholdings into “land-poor,” land-middle,” and land-rich.” The

95
determination of these classes is made by defining categories of one standard deviation from
the mean.

Table 4.5 displays the average benefits extracted from forests by wealth class. Each column
pair shows the actual calculated average in Nepalese rupees in the first column, with the
percent of total benefits in the second column. For the entire sample, households obtained
8.18% of their total forest benefits from the community forest, and 88.56% from their own
private land. Of course, the land-poor will not be able to harvest such a large proportion of
their needs from their own land; they obtain 62.46% of their forest products from the
community forests and 28.13% from their own land. Clearly the land-poor are the most
dependent on community forests for their needs. The land-rich, on the other hand, obtain
only 2.22% of their needs from the community forest.

Nevertheless, the average benefits extracted by value from the community forest are
approximately the same across land-wealth categories. But not only are the land-poor land
poor, but they also receive less forest products from all sources than the land-middle or the
land-rich.

96
Table 4.5 Calculated average benefits from forests by land-owning group (ropani)

All Land-Poor
land ≤ 4 ropani
Land-Middle
4 < land < 36
Land-Rich
land ≥ 36 ropani
% % % %

From community forest 4,428.01 8.18 3,732.48 62.46 4,609.48 13.62 4,012.80 2.22
From private land 47,921.18 88.56 1,681.05 28.13 27,526.64 81.34 173,757.30 96.22
Other sources 1,765.37 3.26 562.24 9.41 1,704.70 5.04 2,815.97 1.56

Total average benefits 54,114.56 100.00 5,597.78 100.00 33,841.21 100.00 180586.07 100.00

Obs 490 49 364 77
Source: Author’s calculations

97
These wealth classes are further broken down by user group management regime. Table 4.6
shows the average benefits for each land wealth category by regime. Several things are
noted. First, with the exception of the land-poor, households under no regime obtain higher
levels of forest products from both the community forest and from all sources than do
households under any type of management. This can be interpreted to be the restrictive
nature of even informal management reducing at least some extractive behavior.

Second, households under formal regimes obtain higher levels of forest products from the
community forest than do households under informal management. This may reflect the
improvement of the resource due to formal management of some duration that allows higher
levels of extraction due to resource regeneration. Third, the land-poor are more dependent on
resources from the community forest than the other wealth classes under all types of
management regimes. Fourth, without management, the land-poor receive the least resources
from the community forest, on average, and the land-rich the most. This ordering is reversed
under formal management. This appears to be evidence of a pro-poor feature of formal under
group management.

98
Table 4.6 Calculated average benefits from forests by land-owning group, by management regime

Forest benefits from
community forest
Total forest benefits from
all sources
% of total forest benefits
obtained from the
community forest

Land-poor:
No FUG 1,561.00 2,941.00 63.16
Informal FUG 3,527.50 5,240.83 65.35
Formal FUG 4,627.83 7,426.41 52.72

Land-middle:
No FUG 5,791.63 33,773.62 48.20
Informal FUG 3,480.90 33,818.71 42.39
Formal FUG 4,456.92 33,906.97 41.54

Land-rich:
No FUG 6,747.88 50,707.16 26.70
Informal FUG 2,651.16 39,740.21 16.22
Formal FUG 2,724.73 360,072.17 18.03

Source: Author’s calculations

99
4.2 Estimation

4.2.1 Empirical Specification

Following Tachibana, et al., the simple household utility maximization model is based on its
consumption of firewood, which is produced in the aggregate from a combination of inputs
of combined community household labor in collection activities (L) and the natural tree
resource (R). The individual household’s production is dependent on the number of
households in the community (n), assuming that the labor contributed by each household is a
perfect substitute for labor of other households in the same community. The production
function for the i-th household is therefore:
), , / , ( M n R l F y
i i
=
where l
i
is equal to L/n and M references the management regime insofar as it affects
behavior governing forest resource use. In actual estimation, the management regime is
measured by a dummy variable for the type of regime.

Each household maximizes utility by choosing its labor spent on extraction of firewood from
the community forest, given household characteristics, community characteristics, forest
conditions and the forest management regime.

We begin with the following empirical specification on the entire data set:
() ( ) ( )
ij i j j j ij
ln l X' Y' FFUG IFUG α βγη γ ε =+ + + + +

100
where labor is measured by the number of days per year dedicated to the collection activity
by all members of household i in community j, X is a vector of household characteristics,
and Y is a vector of community characteristics, and FFUG and IFUG are dummy variables
indicating whether the community forest is managed by a formal forest user group or an
informal forest user group. (These dummies are omitted when the estimation is done on the
segmented data.)

The variables used in the estimation are as follows:
(i) Biomass per household is the availability of the common-pool resource to the
household, and is expected to have a positive effect on the amount of labor
allocated to collection;
(ii) Private trees is the availability of a substitute, expected to have a negative effect;
(iii) Owned land to represent the wealth of the household, a variable of interest. If
wealth confers political status, this is expected to have a positive impact; if it
works as a substitute, it will have a negative impact;
(iv) Household size, to represent the demand for forest products;
(v) Percent of households that is active (aged 15-59), to represent the relative supply
of labor;
(vi) Percent of household that is literate, to represent the opportunity cost of
collection activities;
(vii) Daily wage, to represent the opportunity cost of collection activities;
(viii) Distance to market, to represent the opportunity cost of collection;
(ix) Distance to the forest, to represent the availability of forest resources and the
time taken to reach the resource; and

101
(x) High ethnicity dummy, which takes the value one if the head of the household is
a Brahmin or Chhetri, to represent the wealth of the household.

The estimation results for labor allocated to firewood collection are shown in Table 4.7. The
first block of columns shows the results from pooled data; the second shows estimation
results on the data segmented by user group regime.

The first column (Model I) shows results using OLS estimation. Most of the coefficients in
the pooled data are of the expected signs but many are not significant. The amount of
resources available per household has a positive effect on labor allocated to firewood
collection. Private tree holdings by households do work as substitutes for firewood
collection. Higher land holdings lead to higher collection labor, which supports the
hypothesis that land ownership is a wealth variable and the wealthier have a greater ability to
field more resources into collection during open days.

Also noteworthy are the coefficients on the user group management dummies, neither of
which is significant. In fact, the coefficient on informal management is positive, implying a
positive effect on labor allocation. This can be explained as above by the competitive
collection strategy employed under management rules which are inadequately enforced.

We might be concerned about the endogeneity of the user group formation, however.
Although a single household can only have limited influence on the collective decision-
making that produces a user group regime, the decision to form a group is nevertheless based

102
on household and community characteristics and the conditions facing each community and
therefore should be corrected for potential self-selectivity bias.

Table 4.7 Estimation results for household labor allocation function for firewood from
community forest

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors
robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level

Pooled Segmented
OLS TreatmentTreatment FFUG IFUG No FUG

Intercept 2.055 **
(2.03)
2.388 **
(2.43)
1.54
(1.56)
.799
(.63)
3.828 *
(1.87)
2.839
(1.24)
Biomass
per HH
.139 **
(2.54)
.142 ***
(2.62)
.178 ***
(3.39)
.304 ***
(2.78)
.058
(1.66)
.092
(1.08)
Private
trees (#)
-.089 ***
(-4.45)
-.084 ***
(-4.07)
-.085 ***
(-3.72)
-.045
(-1.35)
-.026
(-1.08)
-.070 **
(-2.52)
Owned
land
.175 ***
(2.69)
.130 **
(1.98)
.169 ***
(2.60)
.125
(1.27)
-.047
(-.40)
.335 *
(1.89)
HH size .136
(1.14)
.201 *
(1.88)
.136
(1.21)
.308
(1.28)
.175
(.92)
-.000
(-.00)
% HH
active
.354 *
(1.72)
.253
(1.26)
.310
(1.49)
.141
(.47)
.106
(.24)
.440
(1.15)
% HH
literate
-.179
(-.91)
-.128
(-.61)
-.145
(-.80)
-.586 ***
(-2.72)
.128
(.31)
-.322
(-.74)
Daily
wage
.069
(.25)
.128
(.45)
.090
(.33)
.276
(.71)
-.400
(-.83)
-.249
(-.55)
Distance
to market
.012
(.29)
-.030
(-.75)
.004
(.09)
.101
(1.19)
-.003
(-.11)
.031
(.29)
Distance
to forest
.049
(1.19)
.011
(.28)
.041
(.98)
.019
(.36)
.059
(.37)
.180
(1.18)
Ethnicity
– High
-.045
(-.38)
.009
(.08)
-.009
(-.07)
-.078
(-.51)
.160
(.92)
-.121
(-.48)
FFUG -.045
(-.22)
-1.139 ***
(-2.72)

IFUG .249
(1.36)
1.217 ***
(2.72)

R² .3351 .5300 .3003 .2666
Obs 343 343 343 148 89 106
Clusters 80 80 80 35 22 23
Wald χ
2
74.31 82.01

103
To account for such endogeneity, a Heckman treatment effects model is employed. In these
models, the “treatment” is taken to be governance by a forest user group. Two models are
estimated: one where the dichotomous treatment variable is governance by a formal forest
users group (FFUG), the second where the treatment variable is governance by an informal
users group (IFUG). The instruments used for the user group dummies are those identified
by Tachibana, et al. as determinants of user group formation: the number of households
using the community forest, representing the transactions cost of collective action; the
number of wards (or communities) involved in accessing the forest, representing a cost of
collective action; a remote dummy representing the effect of the central government effort at
transferring control of the forests to the community; and the condition of the forest at the
beginning of the study, representing the potential benefit to user households of forming user
groups to manage their common-pool resource. The first stage probit estimation results are
shown in Table 4.8.

Table 4.8 First stage probit results for FUG formation

Numbers in parentheses are t-statistics. Standard errors robust to clustering.
Significant at * 0.10 level ** 0.05 level *** 0.01 level
FFUG IFUG

Intercept -1.055
(-1.13)
.763
(.56)
Ln(number of
households)
.344 **
(1.99)
.305
(1.23)
Ln(number of wards) -.070
(-1.25)
-.210 *
(-1.70)
Remote dummy -.809 ***
(-2.61)
-.131 *
(-1.73)
Crown cover in 1978 -.010 **
(-1.83)
-.027 **
(-2.11)

104
Once the endogeneity of user group formation is accounted for, the results in Table 4.7
confirm the importance of the household’s stock of trees on its own land as a substitute for
community forest resources: the coefficient on private trees is negative and very significant.
The coefficient on biomass per household (of the common-pool resource) is positive and
very significant, showing that households will allocate more labor when the resources
available to them are more abundant.

The coefficient on the percent of active members in the household is positive as expected,
indicating that more active households will have a greater supply of labor available for
firewood collection. Finally, the coefficient on land owned is positive and significant,
showing that households with more land are likely to allocate more labor to firewood
collection in spite of having potentially more substitutable resources. This reflects a wealth
effect such that wealthier households have greater opportunities to hire labor and therefore
can field more labor into the community forest.

Note that the coefficient on the formal user group dummy is negative, meaning a restriction
on labor allocation, while the coefficient on the informal user group dummy is positive,
implying a strategic response to the formation of an informal user group wherein the
resource has been identified as valuable and as its availability is potentially restricted
households compete for the resource by allocating more labor.

4.2.2 Differential Behavioral Responses

Are labor allocation responses different under various user group regimes?

105
Looking at the estimation results for the data segmented by user group regime in Table 4.7,
we see that the coefficient on private trees is only significant under the open access regime,
implying that the substitution effect operates as expected without a management regime and
when resources are not restricted by rules; however, once rules are imposed, whether
enforcement is adequate or not, households may substitute their own resources for the
common-pool resource but they will nevertheless allocate the same amount of labor to
collection as a strategic response. Interestingly, land wealth confers no benefit to collection
under management but is significant under no management, again supporting our
supposition of a wealth effect.

In addition to firewood, fodder is another important forest resource extracted by households.
Table 4.9 shows OLS estimation results when the dependent variable is labor allocated to
fodder collection. We employ OLS because there it is unlikely that user group formation is
endogenous to the restrictions on fodder collection. The variable representing demand for
this forest product is the household’s livestock holdings; this variable replaces household
size from the previous estimation.

The results are very similar to firewood collection, but here in the pooled estimation, both
formal and informal management dummies are negative and significant, supporting the
theory that community management reduces resource extraction. The coefficient is larger on
the formal user group dummy, meaning the reduction in extraction is larger due to the
formality of the management.

106
Table 4.9 Estimation results for labor allocation function for fodder from community
forest

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors
robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level

In the segmented estimations, we see that private trees are again not substituted under formal
or informal management for fodder supply; equal labor allocation to collection of fodder
occurs under community management in spite of households having substitutes available.
Land ownership does appear to increase labor allocation to fodder collection under either
Pooled FFUG IFUG No FUG

Intercept 3.53 **
(2.58)
4.538 **
(2.25)
.021
(.01)
2.787
(1.49)
Biomass per HH .089
(1.11)
.398 ***
(3.57)
-.132 *
(-2.08)
.098
(1.38)
Private trees (#) -.081 ***
(-2.81)
.013
(.22)
-.038
(-.50)
-.101 ***
(-3.79)
Owned land .266 **
(2.46)
.228 *
(1.95)
.281
(1.24)
.269 **
(2.08)
Livestock .257 **
(2.26)
-.121
(-.69)
.388
(1.31)
.423 *
(1.98)
% HH active -.398
(-1.20)
-.470
(-.73)
-.222
(-.31)
.144
(.32)
% HH literate .119
(.44)
.087
(.17)
-.85 **
(-2.30)
.469
(1.43)
Daily wage -.116
(-.38)
-.844 *
(-1.91)
.994 **
(2.40)
.014
(.04)
Distance to market -.041
(-.77)
-.177 **
(-2.17)
.045
(1.32)
.021
(.31)
Distance to forest -.005
(-.06)
-.020
(-.21)
-.066
(-.40)
-.100
(-.80)
Ethnicity – High -.161
(-.75)
.334
(1.18)
-.473 *
(-1.88)
-.395
(-1.16)
FFUG -.544 **
(-2.18)

IFUG -.298 **
(-1.68)

R² .2233 .4432 .4291 .4488
Obs 206 68 53 85
Clusters 66 27 16 23

107
open access or under formal management. (The coefficient is positive under informal
management but not significant.) This is weak support for the hypothesis that the relatively
more wealthy receive higher benefits from community forests. Finally, the coefficient on
livestock ownership under open access is positive and significant, as expected. Households
with more livestock will need more fodder and thus will allocate higher levels of labor to
fodder collection. This coefficient is also positive under informal management but not
significant. However, it is negative but not significant under formal management:
households with more livestock will tend to allocate less labor to the collection of fodder.

Table 4.10 shows the estimation results when the dependent variable is the total of all forest
resources extracted from the community forest. Again we see in the pooled data that forests
with higher levels of biomass per household will allow more labor allocated to forest product
collection, and that user group management has a negative impact on collection (significant
under formal management).

Looking at the segmented data, we see again that private trees are a substitute only for
households without user group management. Land ownership is not a significant determinant
of resource collection under any type of management regime. Livestock ownership under
formal management has a negative impact on labor allocation. We can view this as a
substitution effect within forest products: under formal management more livestock is
stalled, less timber is collected, which is relatively more valuable than other forest resources,
thus a negative impact on total labor allocation.

108
Table 4.10 Estimation results for labor allocation function for all resources from
community forest

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors
robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level
Pooled FFUG IFUG No FUG

Intercept 3.073 ***
(2.66)
4.673 **
(2.25)
3.321 ***
(2.85)
1.794
(.95)
Biomass per HH .169 ***
(2.90)
.286 ***
(2.71)
.108
(1.64)
.115
(1.05)
Private trees (#) -.050 *
(-1.90)
.040
(.66)
.007
(.29)
-.120 ***
(-2.97)
Owned land -.002
(-.02)
.021
(.13)
-.103
(-.68)
-.054
(-.46)
Livestock .017
(.13)
-.340 *
(-1.87)
.136
(.82)
.330
(1.36)
HH size .269 **
(1.99)
.282
(1.43)
.266
(1.29)
.098
(.37)
% HH active -.317
(-1.11)
-.475
(-.93)
-.672
(-1.27)
.146
(.49)
% HH literate .032
(.13)
-.556
(-1.50)
.324
(.74)
-.120
(-.28)
Daily wage .324
(.95)
-.108
(-.17)
.209
(.72)
.444
(1.12)
Distance to market .030
(.57)
.055
(.48)
-.007
(-.23)
.061
(.51)
Distance to forest -.021
(-.27)
-.053
(-.47)
-.033
(-.33)
.241
(1.29)
Ethnicity – High -.052
(-.31)
-.191
(-.81)
-.051
(-.25)
.231
(.86)
FFUG -.503 **
(-1.98)

IFUG -.251
(-1.21)

R² .3153 .4589 .2743 .2717
Obs 375 156 106 113
Clusters 87 38 24 25

109
Taken together, these results show that in the pooled data, once a treatment model is
employed to correct for the endogeneity of user group formation, the impact of wealth in the
form of land ownership becomes apparent. The estimations show that in spite of formal user
group management, the land-rich will allocate more of their household labor to the collection
of firewood and fodder in the community forest. This response occurs under open access,
where restrictive rules are absent, and under formal management, where enforcement is
more effective.

Further, we see that a household’s inventory of private trees is a substitute for common-pool
resources in open access situations but not when the forest is governed by a community
management regime, whether formal or informal. This implies a strategic competitiveness in
collection behavior as the existence of a management regime itself has imputed a value to
the common-pool resource and households do not reduce their extraction to consume their
private holdings. That is, the establishment of management rules highlights the importance
of the resource and thus encourages collection relative to a forest under no management; in
effect, these households are taking their “fair share” even though they have alternatives.

4.2.3 Inequality

Do the data support the findings of case studies implying a worsening of inequality within
the community once user group management is introduced? To examine this outcome, Gini
indices for a variety of assets are calculated for all communities. It would have been
preferable to calculate income inequality but the data contains no consumption or

110
expenditure data. Table 4.11 summarizes the gini indices under the three types of
management regimes.

Table 4.11 Gini Indices

Variable FFUG IFUG No FUG

Land holdings 0.311
(0.106)
0.249
(0.088)
0.249
(0.113)
Land value 0.367
(0.203)
0.326
(0.215)
0.304
(0.281)
Private tree
ownership
0.471
(0.166)
0.391
(0.167)
0.381
(0.138)
Livestock value 1996 0.189
(0.102)
0.179
(0.069)
0.194
(0.090)

Obs 43 26 30

Numbers in parentheses are standard deviations.

We note that inequality in land holdings, in land value and in private tree ownership are all
higher in communities under management than under open access, and higher under formal
management than under informal management. Livestock inequality is less systematic.

We would like to estimate the effect of user group management in the determination of
community inequality, using the ginis as calculated above. Given the nature of the data,
however, and the small sample size within each community, the results cannot be expected
to be rigorous and thus the analysis is left for future research.

111
4.3 Conclusion

All results seem to support the case studies that wealth (as measured by landownership) does
not discourage the household from supplying labor to collection of firewood from the
community forest, although there may be alternatives available. Indeed, households with
private trees do not significantly reduce their collection from community forests which are
governed by formally recognized forest user groups, indicating that there is no substitution
effect and that these households are therefore taking their “fair share” from the community
forests even though they have alternatives. This is supportive of concerns expressed by case
studies showing how putative equitable distribution of benefits disadvantages the poorest in
the community. Given this result, inequality within communities may worsen but the
evidence in this nationwide data set offers only weak evidence in support of this hypothesis.

While valuable in that it covers forest conditions and institutional information, the data set
employed in this chapter suffers from two shortcomings. First, it is cross sectional data,
sampling a moment in time. We can make no inference about the impact of the introduction
of user group management on labor allocation decisions. Second, the data set contains no
expenditure or consumption data thus we can make no conclusions about the impact of user
group management or formation on welfare. Our attempts in this chapter are rudimentary
given the nature of the data.

112

Chapter 5
Welfare Effects of Community Forestry

There are many studies confirming the success of community forestry in Nepal in
regenerating the forest resource (Edmonds (2002), Pokharel (2002), Yadav et al. (2003) Dev
et al. (2003)). These same studies find that benefits of the resource improvement are
distributed inequitably, and that the poorest in these communities are negatively impacted by
resource use restrictions.

Simulation exercises in the general equilibrium model of a forest-based community with two
wealth classes (see Chapter 3) show that under reasonable assumptions, welfare worsens for
all members of the community but more so for the relatively more poor, worsening
inequality within the community. As substantiated by case study literature, this is caused by
three effects: (1) exclusion for a necessary resource in the absence of available alternatives;
(2) increase in the price of resource products due to supply pressures in the face of increased
market demand; and (3) general inflation due to substitution away from the restricted
(unpriced) good. Together these results imply that we are “saving the forest at the expense of
the poorest.”

113
Analysis of behavioral responses to FUG implementation in Nepal shows that despite the
apparent effectiveness of FUGs at reducing labor allocated to firewood collection from
community forests, it is only households in communities without FUGs that substitute
private resources for community resources, that households in communities with FUGs
allocate equal time to resource collection in spite of available alternatives (see Chapter 4).
This study highlights increases in inequality once FUG management is implemented but
finds evidence that welfare is improved under community forestry.

Thus localized case studies show that FUG management is effective at reducing firewood
collection and therefore halting resource degradation, and provide anecdotal evidence that
management rules and use restrictions are negatively impacting the poorest segments of the
forest communities, exacerbating inequality within forest communities and perhaps
negatively impacting welfare in general. This evidence, however, remains in large part
anecdotal and may fail to report positive effects in a larger sample. This chapter tests the
findings on a nationwide living standards survey in Nepal to see if the results are local or
general in nature.

5.1 Data

The Nepal Living Standards Survey is a multi-topic survey following the Living Standards
Measurement Survey (LSMS) methodology developed by the World Bank, collecting data
on various aspects of household welfare, including consumption, income, housing, labor
markets, education and health, accompanied by a community-level survey on facilities,
prices, demographics, etc.

114
The first round of the NLSS (NLSS I) was administered between June 1995 and May 1996,
interviewing 3,388 households across Nepal. Our interest is in the rural hills region of Nepal,
which comprises approximately 40% of the land mass of Nepal. Of the households surveyed
in the NLSS I, 1,136 were in the rural hills region in 92 communities. The second round
(NLSS II) was administered from April 2003 to April 2004, interviewing 4,008 households
of which 1,244 were in the rural hills in 96 communities. In addition, a panel sample is
drawn from 1,232 households from the NLSS I that were re-interviewed in 2003, of which
326 were in the rural hills in 31 communities.

The data are examined for features relating to the hypotheses that both household welfare
and distributional equity within communities were negatively impacted as a result of the
institutional shift towards community forestry. The dependence of communities on forest
resources is shown in Table 5.1. The percentage of communities reporting wood as the
primary source of cooking fuel in 1995 was 97.59%; this number fell slightly by 2003 to
96.84% but clearly the dependence on forest resources is demonstrated in both periods.

The handover of national forests to communities can be detected in the data shown. In the
1995, 51.76% of the communities surveyed had a community forest; by 2003, this grew to
72.92%. There is a corresponding increase in the percentage of communities reporting the
community forest as their primary source of firewood, from 30.12% to 49.47%. This is
consequently matched by a decrease in dependence on government forests as these are being
handed over.

115
Similarly, there is a large increase in the number of communities with FUGs, from 29.35%
to 57.29% in 2003.
That there were already a sizeable number of communities with user group management
prior to their formal implementation raises a concern about self-selection which will be
addressed in the econometric specification. The average age of these user groups in the
earlier survey is 144 months, or twelve years. This is reflective of the history of community-
level attempts at forest management back to the control by panchayats, a tradition that
persevered beyond their legal dissolution.

Table 5.1 Dependence on Forests

1995 2003

% of communities reporting wood as the primary
source of cooking fuel

97.59

96.84
% of communities with a community forest 51.76 72.92
% of communities reporting community forest as
primary source of firewood

30.12

49.47
% of communities reporting government forest as
primary source of firewood

44.58

23.16

% of communities with FUGs 29.35 57.29
Average age of FUG (months) 144.07 79.49

% reporting forested area improved 36.14 53.68
% with FUGs reporting forested area improved 46.15 68.52
% w/o FUGs reporting forested area improved 31.58 34.15

Observations 92 96
Source: NLSS1 and NLSS2

Consistent with case studies, it appears that the effect of community forestry and forest user
group implementation has indeed been to improve the forest. In 1995, 36.14% of the
communities reported the forested area improving; by 2003 that number was increased to

116
53.68%. This improvement, however, was most significantly seen in communities with user
group management: by 2003, 68.52% of the communities with FUGs reported an
improvement in forested areas, an increase of 12 percentage points over 1995, compared to
only 34.15% of the communities without FUGs, an increase of less than three percentage
points.

5.1.1 Welfare Indications:

Household income and asset characteristics are shown in Table 5.2. The first two columns
show the average real values of each variable; the third column shows the percent change in
the averages between the two survey periods. The fourth column shows the percentage
change in household holdings in the panel sample.

Table 5.2 Household Income and Asset Holdings

Cross Section Panel
1995 2003 % ∆ % ∆

Land owned (hectares) 0.54 0.63 16.67 27.59
Livestock owned (head) 6.09 6.80 11.66 3.49
Ave per capita real expenditure 7,293.50 8,615.51 18.13 40.64
Ave per capita real food
expenditure
4,380.80 4,856.30 10.85 49.04
Ave value of owned house 62,669.29 65,851.45 5.08 21.11
Ave value of owned HH assets 12,290.99 12,545.41 2.07 5.67
Ave value of owned farm assets 658.18 2,211.89 236.07 202.19
Ave value of businesses owned 7,213.18 6,247.91 (13.38) (23.84)
Ave value of all assets owned 82,831.65 86,856.66 4.86 15.16
Ave value of non-house assets 20,162.35 21,005.21 4.18 2.79

Observations 1136 1152 326
Real values in 1995 Nepalese rupees.
Source: NLSS1 and NLSS2

117
We first compare the averages of the variables in the two cross sections. Average land
ownership in 1995 was 0.54 hectares per household, increasing to 0.63 hectares by 2003.
Livestock ownership increased from 6.09 head in 1995 to 6.80 head in 2003. Average
incomes as measured by per capita expenditures increased substantially over the period, by
over 18 percent. While house values and households assets showed only small increases,
farm assets (which include such items as tractors, ploughs, carts, generators, threshers, etc.)
increased more than two-fold, while the value of businesses owned declined. The category
called “all assets” includes the prior four items, while “non-house assets” includes “all
assets” less the value of the household’s owned house.

To directly compare how individual households’ incomes and asset holdings increased over
this time (as opposed to sample averages) the changes in income and asset holdings of the
panel data are shown in the fourth column of Table 5.2. Incomes increased by an average of
40.64%, consistent with the report of poverty trends prepared by the Central Bureau of
Statistics (HMGN, 2005). Furthermore, with the exception of business value, all categories
of asset ownership also increased over the period. Again, there was an especially large
growth in farm assets as households intensified agricultural work, although these values
were only a small proportion of household asset holdings.

Housing characteristics are shown in Table 5.3. House construction is a very important
component of demand for forest products. By separating the data in both survey periods by
management regime, we note a significant decline in the use of wood products in the
construction of walls, roofs and floors in those communities with user group management in
2003 compared to those without. On average, there is an increase in the use of wood in

118
construction of walls (from 4.67 percent of households in 1995 to 6.86 percent in 2003) and
floors (from 2.28 percent of households in 1995 to 4.51 percent in 2003), but these increases
were entirely within communities without user group management, and, in fact, user group
management appears to lead to a decline in the reliance on forest products for house
construction.

Table 5.3 Housing Characteristics by Management Regime

1995 2003
No FUG FUG No FUG FUG

Average number of rooms 2.84 2.78 3.39 3.74 †
Average size of interior (sq ft) 530.38 494.22 495.20 573.77 †
Average size of house plot (sq ft) 788.61 799.03 2030.20 1889.09

Percent of households reporting:
Main construction material of
walls:

Mud-bonded bricks/stones 86.63 79.88 † 72.56 86.06 †
Cement-bonded bricks /
stones
7.05 15.24 † 7.32 7.42
Wood / branches 4.83 4.27 12.80 2.42 †

Main construction material of
roof:

Straw / thatch 50.62 63.11 † 46.34 39.24 †
Tiles / slate 29.58 17.68 † 25.20 31.67 †
Galvanized iron 13.24 16.77 † 19.92 22.58 †
Concrete / cement 2.60 .30 † 4.67 3.03 †
Wood / planks 2.48 - . † 2.44 1.67 †
Earth / mud .99 1.83 † .81 1.36 †

Main material of floor:
Earth 92.95 95.73 † 88.62 90.91
Cement / tile 2.72 1.52 † 5.08 5.45
Wood / planks 3.09 2.13 † 6.30 3.18 †

Observations 808 328 492 660
Source: NLSS1 and NLSS2
† Significantly different from “No FUG” value of that year at 5% level of significance.

119
To be convinced that communities with or without FUGs are different, we must examine
income and asset holdings under the different management regimes. The first two columns
of Table 5.4 show the average income and asset holdings of households in communities
without FUGs in 1995 and 2003, while the fourth and fifth columns show these variables for
households in communities with FUGs in both periods. As above, the percentage changes in
the averages over time are also shown. For example, average land holdings increased
12.73% for households in communities without FUGS compared to increase of 25.49% for
households in communities with FUGs. The average increase in income is larger for
households in communities with FUGs than without, increasing by 23.82% versus 10.71%.
Average household assets fell dramatically in communities with FUGs, while average total
assets increased by 52%, versus a decline of 3.73% in communities without FUGs. These
data would appear to indicate that user group management is beneficial to households in
terms of income growth and total asset growth.

Again, sample composition has changed over time in these tables because these are cross
sectional data. We therefore examine the changes in these variables in the panel set in Table
5.5. Again we see a similar pattern. The percentage change in land ownership is 12.73
percent for households in communities without user group management as opposed to 37.88
percent in communities with user group management. There is a significant increase in
average livestock ownership, in average household assets, in farm assets, in housing values
and in total asset holdings for households under user group management. Again these data
indicate that households in communities with user group management have seen significant
improvement in income and asset holdings during the survey periods, and significantly
greater improvement than households in communities without user groups.

120
Table 5.4 Household Assets by Management Regime

Real values in 1995 Nepalese rupees.
† Significantly different from “No FUG” of same year value at 5% level of significance.
Source: NLSS1 and NLSS2

No FUG With FUG
1995 2003 % ∆ 1995 2003 % ∆

Mean land owned
(hectares)
0.55 0.62 12.73 0.51 0.64 25.49
Mean livestock owned
(head)
6.28 6.95 10.67 5.62 6.70 19.22
Mean per capita real
expenditure
7,299.00 8,080.80 10.71 7,279.96 9,014.11 † 23.82
Mean pc real food
expenditure
4,342.70 4,319.27 (0.54) 4,474.66 5,204.46 † 16.31
Mean HH assets 13,505.29 12,224.64 (9.48) 34,388.15 † 12,784.52 (62.82)
Mean farm assets 681.24 3,018.09 343.03 602.37 1,610.91† 167.43
Mean business value 5,681.66 5,853.60 3.03 10,985.96 6,541.85 (40.45)
Mean total assets 94,107.94 90,598.59 (3.73) 55,275.16 † 84,067.22 52.09
Mean housing asset 74,149.75 69,502.25 (6.27) 34,388.15 † 63,129.94 (83.58)
Mean NH assets 19,868.19 21,096.34 6.18 20,887.00 20,937.28 0.24

Observations 808 492 328 660

121
Table 5.5 Household Assets by Management Regime - Panel

Real values in 1995 Nepalese rupees.
† Significantly different from “No FUG” of same year value at 5% level of significance.
Source: NLSS1 and NLSS2
No FUG With FUG
1995 2003 % ∆ 1995 2003 % ∆

Mean land owned
(hectares)
.55 .62 12.73 .66 .91 † 37.88
Mean livestock owned
(head)
6.60 6.18 (6.36) 5.65 7.18 † 27.08
Mean per capita real
expenditure
6933.85 8011.36 15.54 6897.47 8160.25 18.31
Mean pc real food
expenditure
3636.82 4474.49 23.03 4100.71 † 5041.15 † 22.94
Mean HH assets 13802.51 12154.62 (11.94) 8249.50 13970.79 69.35
Mean farm assets 604.16 1776.41 194.03 469.89 1602.73 241.09
Mean business value 5813.76 4675.06 (19.59) 4900.40 3541.13 (27.74)
Mean total assets 80090.42 86764.18 8.33 44272.62 † 70265.56 58.71
Mean housing asset 61194.55 70682.46 15.50 31652.38 † 53213.44 † 68.12
Mean NH assets 20220.44 18606.09 (7.98) 13619.79 19114.66 40.34

Observations 231 196 95 130

122
However, as noted above, there may be self-selection issues involved. Those communities
without user groups may be characteristically different from communities with user group
management that also lead to slower growth in household income and asset holdings. To
explore this thought, we examine the differences in average percent change in the same
variables for households under three different types of communities: “No FUG” includes
those community that did not have user group management in 1995 nor in 2003; “Old FUG”
includes communities that had user group management in both periods, and therefore may
have self-selected into user group formation; and “New FUG” includes those communities
that established user groups after 1995, presumably in response to the institutional shock.

The percent changes in income and asset holdings for households under these different types
of regimes are shown in Table 5.6.

Table 5.6 Household Asset Growth by Management Regime - Panel
Real values in 1995 Nepalese rupees.
Source: NLSS1 and NLSS2

No FUG Old FUG New FUG

Mean land owned (hectares) 12.00 13.11 73.53
Mean livestock owned (head) 8.84 13.67 (4.42)
Mean per capita real expenditure 14.11 36.63 8.04
Mean pc real food expenditure 25.43 49.26 12.62
Mean HH assets (22.28) 55.75 40.30
Mean farm assets 198.86 159.84 152.10
Mean business value (19.26) (66.97) 29.28
Mean total assets (7.40) 83.32 45.81
Mean housing asset (3.14) 145.01 43.23
Mean NH assets (15.45) .87 42.37

Observations 164 63 67

123
Here the differences become more striking. For example, the growth in land ownership in
households without user group management is not much different from households under
“Old FUG” management, but households in communities new to user group management
saw growth in land ownership of 73.53 percent.

On the other hand, growth in livestock ownership was significantly higher for households in
old user groups than without user group management, but negative in new user groups.
Together these would indicate a shift towards agricultural activity and away from livestock
holdings in new user group households.

More striking is the difference among the groups in income and total asset growth. While
average income growth for households in old user groups was 36.63 percent, which is much
more than without user group management at 14.11 percent, both of these were in turn much
higher than households in new user groups at 8.04 percent. Looking at growth in total asset
holdings, there was an average of 83.32 percent growth for households in old user groups,
which is much better than the fall in asset growth of 7.40 percent in households without user
groups, but still almost twice as high as the 45.81 percent growth rate for households in new
user groups.

These data together show that the improvement in livestock ownership, income, household
assets, farm assets, housing assets and total assets were significantly lower for households in
communities new to user group management than those with user groups of longer
durations.

124
To some extent this would appear to support the alarm expressed by local case studies
showing a worsening of welfare in user group communities, particularly those new to the
institutional regime and suggest that the benefits of FUG management are felt in asset
accumulation and income improvement over the long run but that incomes may be
negatively affected in the short run as forest user groups are first implemented.

5.1.2 Inequality Indications:

Turning to indications of inequality within communities, the literature indicates that the
impact of user group management varies according to wealth status and that the poor are
negatively affected. If so, this should be evident in the distributional changes over time and
management regime. To look for this evidence, Gini indices are constructed for each income
and asset category in each community. To simplify the exposition, I reduce the discussion of
asset categories by dropping the individual house, farm and business assets.

Table 5.7 displays the average values of the Ginis in the cross-sectional data: the first
column is of the 1995 survey period and the second column is the 2003 survey period; the
third column shows the percent change in the average Gini over the time period. From this
table, it would appear that with the exception of inequality of per capita real expenditures
(which remains virtually unchanged) inequality in the rural hills communities has fallen
between 1995 and 2003. In particular, significant declines in inequality are shown in land
ownership, total asset ownership and non-house asset ownership.

125
Table 5.7 Inequality (Ginis) in Rural Hill Communities

Variable 1995 2003 % ∆

Land ownership (hectares) 0.5399 0.4273 (20.86)
Livestock ownership (head) 0.3928 0.3676 (6.42)
Per capita real expenditures 0.2489 0.2491 0.08
Per capita real food expenditures 0.2285 0.2102 (8.01)
Real total assets 0.4172 0.3577 (14.26)
Real non-house assets 0.5011 0.4373 (12.73)

Observations 92 96
Source: Author’s calculations from NLSS1 and NLSS2

Examining the Ginis by management regime, shown in Table 5.8, it is clear that inequality in
the communities without FUGs has fallen during the time period in all asset and income
categories. The results are more mixed in communities with FUGs, with inequality in asset
ownership improving while inequality in incomes widening.

The changes in the average Ginis over time do not wholly support the supposition expressed
in case studies that inequality is worsening in these communities. However, the panel sample
suggests a different result. Table 5.9 shows the Ginis for those communities that have
recently implemented community forestry, compared to those without forest user groups.
Although the sample size is small, for all asset and income categories except land ownership,
inequality increases in the former sample while in the latter again we see inequality has
declined over the period.

From the descriptive statistics, therefore, it seems plausible that there is a negative (short
term) impact of user group management on income and that equity worsens in those
communities that are new at organizing forest user groups.

126
Table 5.8 Inequality (Ginis) in Rural Hill Communities

Source: Author’s calculations from NLSS1 and NLSS2

Variable No FUG FUG
1995 2003 % ∆ 1995 2003 % ∆

Land ownership (hectares) 0.5340 0.4496 (15.81) 0.5542 0.4108 (25.88)
Livestock ownership (head) 0.3865 0.3506 ( 9.29) 0.4078 0.3803 ( 6.74)
Per capita real expenditures 0.2506 0.2380 ( 5.03) 0.2446 0.2574 5.23
Per capita real food expend 0.2343 0.2009 (14.26) 0.2145 0.2170 1.17
Real total assets 0.4192 0.3618 (13.69) 0.4125 0.3547 (14.01)
Real non-house assets 0.5055 0.4359 (13.77) 0.4905 0.4383 (10.64)

Observations 6527 4155

127
Table 5.9 Inequality (Ginis) in Panel Rural Hill Communities

† Significantly different from “No FUG” of same year value at 5% level of significance.
Source: Author’s calculations from NLSS1 and NLSS2

New FUG No FUG
Variable 1995 2003 % ∆ 1995 2003 % ∆

Land ownership (hectares) 0.4521 0.4021 (11.06) 0.5066 0.4173 (17.63)
Livestock ownership (head) 0.3601 0.3704 2.86 0.3562 0.3361 (5.64) †
Per capita real expenditures 0.1890 0.2509 32.75 0.2813 † 0.2692 (4.30) †
Per capita real food expenditures 0.2044 0.2478 21.47 0.2241 0.2201 (1.83)
Real total assets 0.3187 0.3261 2.32 0.3943 0.3157 (19.93) †
Real non-house assets 0.4357 0.4803 10.24 0.4920 0.4072 (17.24) †

Observations 6 6 16 16

128
5.2 Estimation

5.2.1 Welfare Effects

(i) Cross section data

The estimation strategy employed is straightforward. To determine if there is support in the
data for a negative effect of forest user group management on welfare, the following
reduced-form welfare function is estimated individually on data from both periods (t = 1995,
2003):
() ( )
ijt it jt jt it
ln welfare X' Y' FUG α βγη ε =+ + + +
where welfare is measured by per capita real expenditure and by asset ownership of
household i in community j, X is a vector of household characteristics, and Y is a vector of
community characteristics.

The dependent variables used to represent welfare in the estimation are: per capita real
expenditures, value of housing, net household earnings, and total asset holdings. Explanatory
household variables include the following:
(i) The percent of the household that is active is taken to represent the
household’s supply of labor. It includes any household member between the
age of 15 and 59, and is expected to have a positive effect on welfare;
(ii) The percent of the household that is literate represents an income advantage
and is expected to have a positive effect on welfare;

129
(iii) High ethnicity, a dummy variable taking the value of one if the household is
Brahmin or Chhetri, also representing an income advantage and is expected
to have a positive effect on welfare;
(iv) The amount of land owned by the household, representing the wealth of
household;
(v) Net remittances received by the household, representing income received
from outside the household and expecting to have a positive effect on
welfare.

Remittances from abroad are an important component of income for households in the rural
hills, comprising between 16.0% and 25.9% of total household income in a sample of seven
hill districts [LFP (2003)]. It is possible, however, that remittances are not independent of
household welfare, that poorer households are more likely to send a household member
outside the community for wage opportunities. This variable is therefore dropped in one
specification; the results are not significantly different.

Community variables include:
(i) The distance to the nearest market center, which is indicative of how remote
the community is and is expected to have negative effect on welfare;
(ii) An irrigation dummy which takes the value of one if there are irrigation
facilities in the community such as canals or borings, to capture the general
wealth of the community;

130
(iii) The percent of households in the community with electricity, also to capture
the general wealth of the community and expected to have a positive effect
on household welfare.

The variable of interest is forest user group management. For each period, a dummy variable
representing the existence of a community forest user group is included. Recall that in 1995,
less than a third of communities have forest user groups, this number almost doubling by
2003. An additional specification includes the age (in months) of the user group, under the
expectation that user groups that have been in existence for longer periods of time have been
able to overcome transitional shocks and ameliorate welfare declines.

To test whether user group management has different impacts depending on household land
ownership, an interaction term is included in a third specification. Private land ownership
may provide a valuable alternative source of firewood and fodder if restriction from the
community forest is enforced. On the other hand, land ownership may confer status to the
household and therefore can proxy for political power. Under both conditions, the expected
effect of land ownership under user group management on welfare is positive.

Table 5.10 shows the cross section OLS estimation results for per capita expenditures. In
both survey periods, model I includes the forest user group dummy, while model II adds the
age of the user group in months; model III includes an interaction term between the user
group dummy and the amount of land owned. In the 1995 survey, all the household variables
have coefficients of the expected signs and are very significant; similarly, in 2003 all
household variables with the exception of the high ethnicity dummy are positive and

131
significant. Looking at the community variables, the distance to market is negative as
expected but only marginally significant in 2003 and not significant at all in 1995. The
community irrigation dummy and percent of households with electricity are also significant
and positive in 1995 as expected, but the irrigation dummy is negative in 2003.

Our variables of interest show that in 1995, there is no evidence of any effect of forest user
group management on household welfare. By 2003, however, there is a positive effect,
although once the age of the user group is introduced, the effect is positive for older user
groups only. This validates our supposition that a newly formed user group has an immediate
negative impact on welfare. Land ownership under user group management has no effect on
welfare in either survey period but interestingly, the coefficient is negative in 1995 and
positive in 2003.

It is likely that there is endogeneity among the regressors, particularly in the earlier period
when forest user group formation occurred as a result of internal processes and therefore
community characteristics that affect per capita expenditure may also have been those that
encouraged collective action to protect the forest. For example, communities in forests that
were very degraded and had no outside livelihood option may have been motivated to form
protective user groups but would then also have been those that experienced lower incomes
(and hence lower per capita expenditures). Our estimated coefficient for the FUG dummy
variable would in this case be biased downwards.

132
Table 5.10 Estimation Results (OLS) for Per Capita Real Expenditure

1995 2003
I II III I II III

Intercept 8.205 ***
(54.23)
8.225 ***
(51.64)
8.217 ***
(50.98)
8.346 ***
(66.24)
8.340 ***
(67.60)
8.349 ***
(63.00)
% of household active .369 ***
(4.15)
.369 ***
(4.14)
.366 ***
(4.14)
.356 ***
(4.59)
.356 ***
(4.70)
.359 ***
(4.69)
% of household literate .477 ***
(7.33)
.480 ***
(7.45)
.481 ***
(7.43)
.533 ***
(8.26)
.533 ***
(8.19)
.533 ***
(8.18)
High ethnicity dummy .109 **
(2.55)
.112 **
(2.60)
.112 ***
(2.63)
.062
(1.38)
.056
(1.34)
.056
(1.33)
Land owned (hectares) .106 ***
(3.80)
.107 ***
(3.82)
.119 ***
(3.36)
.106 **
(2.56)
.106 **
(2.56)
.095
(1.36)
Net remittances received (1995 Rs) .002 ***
(5.02)
.002 ***
(4.90)
.002 ***
(4.93)
.003 **
(2.21)
.003 **
(2.37)
.003 **
(2.50)
Ln(distance to market) -.011
(-.69)
-.013
(-.80)
-.013
(-.79)
-.031 *
(-1.66)
-.030 *
(-1.69)
-.030 *
(-1.71)
Community irrigation dummy .156 ***
(2.93)
.149 ***
(2.77)
.148 ***
(2.76)
-.102 *
(-1.92)
-.098 *
(-1.83)
-.097 *
(-1.84)
% of households with electricity .004 ***
(2.93)
.004 ***
(2.81)
.004 ***
(2.81)
.005 ***
(5.78)
.005 ***
(5.61)
.005 ***
(5.60)

133
Table 5.10 (Continued)

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors robust to clustering.
Significant at * 0.10 level ** 0.05 level *** 0.01 level

1995 2003
I II III I II III

FUG dummy -.015
(-.23)
.014
(.19)
.036
(.44)
.128 **
(2.45)
-.004
(-.05)
-.018
(-.19)
FUG age -.000
(-.84)
-.000
(-.73)
.002 **
(2.00)
.002 **
(2.00)
FUG * Land owned -.048
(-.83)
.022
(.30)

R
2
.3354 .3364 .3370 .3551 .3601 .3603
F 37.48 35.46 33.02 23.10 22.02 20.82
Observations 1058 1058 1058 1128 1128 1128
Clusters 86 86 86 94 94 94

134
Ideally, estimation with a two-stage model to accommodate such endogeneity should be
employed, with the existence of a FUG estimated using relevant community variables. The
predicted value of the FUG dummy would then be used in the second-stage estimation for
per capita expenditure. This strategy is attempted, with instruments based on hypotheses of
collective action as in Chapter 4. For example, user groups are assumed to be dependent on
the availability of a community forest, therefore one of the instruments is the community
forest dummy variable. Second, a more degraded forest has a higher marginal benefit of
protection and is expected to be more likely to encourage user group formation; hence a
second variable is a “forest decreased” dummy. This variable takes the value of 1 if the
respondent household feels that the forest area has decreased over the past five years.

Other community variables used as instruments include: the number of households in the
community to represent the transaction cost of collective action; the percentage of
households in the community owning land to represent dependence on the common-pool
resource; the distance to the forest to represent the availability of the resource; the distance
to market or the distance to the nearest paved road representing remoteness and
unavailability of alternatives; and the existence of another user group in the community to
represent the community’s general ability to take collective action.

These instruments are first tested in probit estimations, the results of which are shown in
Table 5.11. From the OLS estimates, we know that the distance to the nearest market is not
significant in 1995 but it is in 2003. Therefore, the instrument chosen to represent
remoteness in 2003 is the distance to the nearest paved road.

135
Not all of the variables assumed to affect user group formation turn out to be significant and
therefore will not be used as instruments in the estimations that follow. For the user group
dummy in 1995, we will use as instruments the proportion of households with land, the
community forest dummy and the user group dummy. For the user group dummy in 2003,
we will use the distance to the nearest paved road, the community forest dummy and a
dummy variable taking the value of one if households felt hat the forest area has decreased
within the past five years.

The results of the estimation of per capita real expenditure using a two-stage strategy are
shown in Table 5.12. All first-stage results are shown in Table 5.21.

Once we correct for the endogeneity of user group formation, the forest user group dummy
in 1995 has a positive and significant coefficient, indicating that forest user group
management has a positive impact on household welfare. This is true whether we account for
the age of the user group or not. By 2003, however, the positive coefficient on the forest user
group is made insignificant once the age of the user group is accounted for. This would
suggest that only user groups of longer duration have a positive effect on welfare, although
neither of the coefficients in 2003 is significant. The effect of land ownership under user
group management is negative in 1995. This is puzzling and opposite to the effect we expect.
It indicates that land owners suffered an excess loss of welfare under user group
management in 1995. The effect disappears in 2003.

136
Table 5.11 Probit Results for FUG formation

Numbers in parentheses are t-statistics. Standard errors robust to clustering.
Significant at * 0.10 level ** 0.05 level *** 0.01 level
1995 2003
I II I II

Intercept 1.775
(.92)
-.003
(-.00)
-3.066 *
(-1.94)
-4.324 ***
(-2.68)
Ln(number of households) -.315
(-1.07)
-.142
(-.49)
.147
(.60)
.276
(1.09)
% of HH owning land -1.786 **
(-2.20)
-1.968 **
(-2.39)
.860
(1.28)
1.093 *
(1.73)
Ln(distance to forest) -.069
(-.33)
-.136
(-.59)
-.129
(-.64)
-.176
(-.86)
Ln(distance to market) -.087
(-.89)
.144
(1.50)

Ln(distance to paved road) .144 *
(1.66)
.250 ***
(3.04)
Community forest dummy 1.651 ***
(3.44)
1.738 ***
(3.48)
2.184 ***
(4.15)
2.105 ***
(3.94)
Other UG dummy 1.031 ***
(2.68)
1.208 ***
(3.14)
.157
(.42)
.268
(.69)
Forest decreased dummy .163
(.47)
-.062
(-.15)
-1.138 **
(-2.45)
-1.261 **
(-2.50)

Pseudo R
2
.3730 .3911 .3983 .4426
Wald χ
2
19.76 21.71 33.22 36.25
Observations 1004 1004 1116 1116
Clusters 81 81 93 93

137
Table 5.12 Estimation Results (2SLS) for Per Capita Real Expenditure

1995 2003
I II III I II III

Intercept 8.171 ***
(89.30)
8.158 ***
(87.71)
8.125 ***
(82.84)
8.347 ***
(64.25)
8.338 ***
(66.25)
8.353 ***
(57.19)
% of household active .351 ***
(3.54)
.347 ***
(2.94)
.327 ***
(3.28)
.359 ***
(4.65)
.363 ***
(4.76)
.361 ***
(4.73)
% of household literate .518 ***
(6.88)
.538 ***
(7.11)
.555 ***
(6.91)
.536 ***
(8.22)
.535 ***
(8.11)
.536 ***
(8.10)
High ethnicity dummy .083 **
(1.96)
.101 **
(2.17)
.109 **
(2.31)
.057
(1.29)
.054
(1.27)
.052
(1.22)
Land owned (hectares) .110 ***
(3.84)
.114 ***
(3.95)
.166 ***
(3.42)
.105 ***
(2.57)
.106 ***
(2.58)
.091
(1.16)
Net remittances received (1995
Rs)
.002 ***
(3.97)
.002 ***
(3.18)
.002 ***
(3.45)
.003 **
(2.25)
.003 **
(2.36)
.003 **
(2.54)
Ln (distance to market) -.030 *
(-1.69)
-.030 *
(-1.72)
-.030 *
(-1.74)
Community irrigation dummy .106 *
(1.68)
.078
(1.16)
.074
(1.12)
-.103 *
(-1.93)
-.097 *
(-1.78)
-.094 *
(-1.75)
% of households with
electricity
.004 ***
(2.61)
.004 **
(2.38)
.004 **
(2.42)
.005 ***
(5.61)
.005 ***
(5.43)
.005 ***
(5.42)

138
Table 5.12 (Continued)

Numbers in parentheses are z-statistics. Regional dummies included but not reported. Standard errors robust to clustering. Predicted value of FUG
in 1995 based on (%_households_with_land_95, community_forest_dummy_95, user_group_dummy_95). F-statistic on first stage is 20.27;
13.80; 12.17. Predicted value of FUG in 2003 based on (community_forest_dummy_03, ln_distance_pvrd_03, forest_decreased_dummy_03). F-
statistic on first stage is 49.08; 6.88;7.43. Significant at * 0.10 level ** 0.05 level *** 0.01 level
1995 2003
I II III I II III

FUG dummy .265 ***
(2.69)
.404 ***
(2.94)
.502 ***
(3.02)
.121
(1.54)
-.000
(-.00)
-.048
(-.20)
FUG age -.001 ***
(-2.66)
-.001 **
(-2.42)
.002
(.99)
.002
(1.18)
FUG * Land owned -.253 **
(-2.32)
.031
(.30)

R
2
.3033 .2911 .2818 .3525 .3576 .3577
F 34.50 28.91 28.05 21.30 21.40 20.45
Observations 986 986 986 1116 1116 1116
Clusters 80 80 80 93 93 93

139
In fact, using a two-stage strategy for the data in 2003 renders our coefficients on the
variables of interest insignificant. Recall that endogeneity is plausible for the earlier time
period (1995) because some user groups had been formed in communities prior to the
institutional shock. By 2003, however, the implementation of user groups was widespread
and initiated by Department of Forest staff, rather than entirely by the communities
themselves. Of course, communities with a propensity to form user groups may have been
those selected early on by the DoF and therefore some self-selection issues remain.
Nevertheless, the endogeneity is likely diluted due to the large enrolment between the two
survey periods.

A Hausman test comparing the two strategies on the 2003 data finds no support for a
systematic difference in the coefficients; thus we return to the original results of Table 5.10
for this later survey period and conclude that user group management has a positive effect on
welfare, but only for those groups of some duration.

Is the two-stage estimation strategy defensible for the 1995 cross-sectional data? A Hausman
test finds a systematic difference in the coefficients under IV and OLS estimations
significant at the 0.001 level. Under-identification is rejected at the 0.001 level of
significance; weak identification is rejected with a Kleibergen-Paap Wald F- statistic of
20.27 compared to a Stock-Yogo critical value of 13.91 at 5% maximal relative bias; and
Hansen J-statistic of over-identification is 2.647 with a p-value of 0.2662. Given these
results, the use of these instruments in a two-stage estimation strategy appears valid for the
1995 cross-sectional data.

140
Table 5.13 shows the estimation results when the dependent variable is the household’s
housing value. In both survey periods, Model I is the OLS estimation; Model II shows the
results using a two-stage strategy with instruments as above. Since houses are manufactured
from timber or other forest products, the restrictions from extraction will affect the
household’s ability to construct its house. Here we see that in the earlier survey period, using
an OLS strategy, the management of a forest user group had a significantly negative impact
on the reported value of the household’s house. While this appears straightforward, again
there may be causation in the opposite direction – communities with such degraded forests
that user groups formed organically in the earlier years are also those without the timber
resources with which to build substantial homes.

The two-stage strategy in Model II in 1995 leaves the coefficient on our variable of interest
negative but no longer significant, substantiating the endogeneity. Still, the coefficient is
negative, providing some evidence that housing values were negatively impacted by user
group management in 1995. By 2003, however, households in communities with forest user
groups have significantly higher housing values, under both the OLS and the 2SLS
strategies. A Hausman test again fails to find evidence of systematic differences in the
coefficients between the two models.

Finally, the estimation results when the dependent variable is total asset ownership are
shown in Table 5.14. Total assets include household assets, farm assets, value of all
businesses owned and housing values. A two-stage IV strategy is tested but a Hausman test
fails to find evidence of a systematic difference in the coefficients; thus a simple OLS
strategy is employed. These results show that user group formation had a negative and

141
significant impact on total asset holdings in the earlier survey period, although duration has a
positive impact. The magnitudes of the coefficients, however, imply an extremely lengthy
catch-up period. By 2003, user group management has a positive effect on asset holdings,
although this effect is muted over time.

Taken together, for the two survey periods, user group management has a positive effect on
per capita expenditure, housing values and total asset holdings in the second survey period.
Earlier results are ambiguous but once endogenous user group formation is accounted for,
forest user groups have a positive effect in per capita expenditure but negative on asset
holdings and housing values. Because housing values are a substantial portion of total assets,
the negative impact on housing values, as a result of restrictions from forest products,
weighs down any potential improvement in all other assets.

Nevertheless, cross section data is problematic because the sample compositions have
changed over time. We turn, therefore, to an examination of the panel data.

142

Table 5.13 Estimation Results for Household Housing Values

1995 2003
I II I II

Intercept 10.416 ***
(30.86)
10.311 ***
(28.24)
9.944 ***
(39.86)
9.902 ***
(36.52)
% of household active -.028
(-.17)
.058
(.18)
.110
(.89)
.112
(.90)
% of household literate 1.019 ***
(7.39)
1.036 ***
(7.22)
.873 ***
(7.26)
.841 ***
(6.49)
High ethnicity dummy .179 **
(2.35)
.172 **
(2.15)
.043
(.59)
.061
(.80)
Land owned (hectares) .217 ***
(5.64)
.228 ***
(5.60)
.251 ***
(4.04)
.254 ***
(4.11)
Net remittances received (1995
Rs)
.004 ***
(3.22)
.004 ***
(3.17)
.003
(1.41)
.003
(1.35)
Ln(distance to market) -.161 ***
(-4.50)
-.154 ***
(-4.15)
-.136 ***
(-4.52)
-.138 ***
(-4.62)
Community irrigation dummy .173
(1.58)
.131
(1.16)
.050
(.44)
.055
(.50)
% of households with electricity .012 ***
(3.57)
.012 ***
(3.21)
.007 ***
(4.77)
.007 ***
(4.30)

143
Table 5.13 (Continued)

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors robust to clustering. Model II 1995:
Predicted value of FUG in 1995 based on (%_households_with_land_95, community_forest_dummy_95, user_group_dummy_95). F-statistic
on first stage is 15.20. Model II 2003: Predicted value of FUG in 2003 based on (community_forest_dummy_03, n_distance_paved_road_03,
forest_decreased_dummy). F-statistic on first stage is 46.07. Significant at * 0.10 level ** 0.05 level *** 0.01 level

1995 2003
I II I II

FUG dummy -.373 ***
(-2.77)
-.270
(-1.32)
.219 **
(2.05)
.321 *
(1.84)

R
2
.3577 .3436 .3818 .3748
F 21.24 16.34 19.72 20.75
Observations 1012 942 1077 1066
Clusters 86 80 94 93

144
Table 5.14 Estimation Results (OLS) for Household Total Asset Holdings

1995 2003
I II I II

Intercept 10.522 ***
(34.56)
10.425 ***
(33.57)
9.691 ***
(49.82)
9.691 ***
(49.67)
% of household active .144
(.91)
.148
(.93)
.166
(1.28)
.166
(1.28)
% of household literate 1.084 ***
(7.97)
1.072 ***
(8.00)
1.016 ***
(8.43)
1.016 ***
(8.43)
High ethnicity dummy .226 ***
(3.27)
.211 ***
(3.02)
.122 *
(1.59)
.122
(1.60)
Land owned (hectares) .294 ***
(5.75)
.291 ***
(5.76)
.392 ***
(6.43)
.392 ***
(6.43)
Net remittances received (1995 Rs) .005 ***
(3.50)
.005 ***
(3.39)
.003
(1.38)
.003
(1.38)
Ln(distance to market) -.143 ***
(-4.63)
-.130 ***
(-4.10)
-.070 ***
(-2.84)
-.070 ***
(-2.84)
Community irrigation dummy .074
(.77)
.107
(1.12)
.033
(.36)
.033
(.36)
% of households with electricity .009 ***
(3.21)
.010 ***
(3.41)
.007 ***
(5.59)
.007 ***
(5.59)

145
Table 5.14 (Continued)

Numbers in parentheses are t-statistics. Regional dummies included but not reported. Standard errors robust to clustering.
Significant at * 0.10 level ** 0.05 level *** 0.01 level

1995 2003
I II I II

FUG dummy -.360 ***
(-3.20)
-.497 ***
(-3.96)
.217 **
(2.47)
.220 *
(1.66)
FUG age .001 ***
(2.78)
-.000
(-.04)

R
2
.3402 .3451 .3470 .3470
F 27.92 24.79 22.61 20.99
Observations 1058 1058 1128 1128
Clusters 86 86 94 94

146
(ii) Panel data

The estimation of the panel data uses the following empirical specification:
()( ) ( )( ) ( )
ijt ijt-1 it it-1 jt jt-1 jt jt-1 i
ln welfare ln welfare X X ' Y Y ' FUG FUG α βγη ε −=+− +−+ − +

where the same welfare measures and household and community characteristics as above are
used. This is essentially a difference in differences estimation. By using this strategy, all
time-invariant community characteristics are differenced out and we are relieved of
endogeneity concerns as presented in the earlier section. Moreover, the remaining time trend
is captured by the intercept term, α. In addition to the change in user group status, a user
group duration variables is added in a second specification, and an interaction term of user
group management in 2003 with 1995 land ownership is added in a third specification to
capture the potential wealth effect of land ownership under new user group management.

Because the dependent variable is the difference in logs, the coefficients on the independent
variables will represent the effect on the growth of welfare. Recall that on average there has
been an improvement of approximately 43% in per capita expenditure over this period in
Nepal.

Table 5.15 shows the results when the dependent variable is the change in per capita real
expenditure. The coefficient on the change in user group dummy is negative across models
with increasing significance and magnitude as modifying variables are included. The
magnitude of the coefficient implies a loss of income growth of 12.19% due to the
implementation of user group management.

147
Table 5.15 Estimation Results for ∆ ln Per Capita Real Expenditure (1995-2003)

Numbers in parentheses are t-statistics. Regional dummies included but not reported.
Standard errors robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level

Adding a duration variable in Model II makes the coefficient on the change in user group
dummy larger, implying a loss of income growth of 17%, and significant at the 5% level,
although the coefficient on the duration variable is not significant. The weak implication is
again that persistent user group management has an ameliorating effect on the negative
impact of user group management on welfare growth, with a catch-up of approximately
0.1% per month of user group age. This implies that income growth is recouped after
approximately 14 years. Model III introduces an interaction term with land ownership,
which has a positive but not significant coefficient. It appears that there is no evidence to
support the hypothesis that land ownership conveys a benefit in terms of improved income
I II III

Intercept -.037
(-.33)
-.041
(-.36)
-.041
(-.36
∆ % of household active .423 ***
(3.23)
.428 ***
(3.29)
.429 ***
(3.38)
∆ % of household literate .259 *
(2.03)
.257 *
(2.00)
.257 *
(2.00)
High ethnicity dummy .093
(1.30)
.097
(1.38)
.097
(1.38)
∆ Land owned (hectares) .040
(1.05)
.042
(1.12)
.043
(1.22)

∆ FUG -.115 *
(-1.91)
-.157 **
(-2.21)
-.158 **
(-2.17)
FUG age in 2003 (months) .001
(1.31)
.001
(.99)
FUG 2003 * Land in 1995 .004
.04)

R
2
.3341 .3383 .3381
F 15.10 14.75 14.47
Observations 325 325 325
Clusters 31 31 31

148
growth to households under user group management, a result not consistent with case study
literature showing that the relatively more wealthy in the community capture more of the
benefits.

The results shown in Table 5.15 therefore provide evidence for the conclusion that the
introduction of user group management may be beneficial to household welfare in the longer
term, and possibly more so for land owners than not, but that there are short term costs
measured by a loss of income growth.

Is there a shift towards non-forest livelihoods? Table 5.16 shows the estimation results for
net earnings received (revenues less expenditures) for the household from two different
activities (agricultural activity and livestock-related activity) and from all sources, which
includes the first two categories and: business activities and agricultural and non-agricultural
wages. Because net earnings can be less than zero, these variables are not logarithmed and
therefore the dependent variables are changes in earnings over time. The coefficients on the
change in user group dummy are negative for agricultural and total earnings but not quite
significant. As such they provide only weak evidence to suggest that the implementation of
user group management has a negative impact on earnings growth. The coefficient on the
duration variable is negative but not significant, while the coefficient on the land interaction
variable is positive, large and significant. Taken together, these results imply that the
introduction of user group management may lead to a fall in earnings growth which is not
compensated over time but that the wealthy are able to improve their earnings growth under
user group management. In this case, land ownership is clearly an important indicator of
livelihood success.

149
Table 5.16 Estimation Results for ∆ Net Earnings (1995-2003)

Numbers in parentheses are t-statistics. Regional dummies included but not reported.
Standard errors robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level

Finally, the results for the growth in assets are shown in Table 5.17. Three asset categories
are shown. The first is total assets, which includes household asset holdings, farm assets,
business assets (if any) and housing value. Because this latter item is a significant portion of
a household’s asset holdings, and because it may be prone to measurement error, it is shown
separately as a dependent variable in the second column. Lastly, the change in assets net of
housing values is estimated. The results are similar across categories. The coefficients on the
change in user group dummy are all negative, but not significant in Model I. The
introduction of user group management has a negative and significant impact on the growth
Variable Agricultural Livestock Total

Intercept -1060.07
(-1.40)
-1107.57
(-1.48)
-1155.66
(-.60)
∆ % of household active -20.08
(-.02)
355.81
(1.13)
-793.26
(-.34)
∆ % of household literate 1222.32
(.95)
578.50
(1.20)
5146.71 **
(2.38)
High ethnicity dummy -1546.64 **
(-2.46)
-568.39
(-1.40)
494.40
(.29)
∆ Land owned (hectares) 1155.01 *
(1.76)
191.09
(1.33)
1158.18
(1.53)

∆ FUG -305.69
(-1.24)
26.02
(.06)
-288.57
(-1.15)
FUG age in 2003 (months) -9.32
(-.95)
1.43
(.45)
-7.84
(-.47)
FUG 2003 * Land in 1995 1168.00 *
(1.74)
111.24
(.37)
2349.86 *
(1.84)

R
2
.2302 .0612 .2732
F 13.79 3.99 14.73
Observations 325 325 325
Clusters 31 31 31

150
of housing assets and on the growth of all non-house assets. The effect is not modified by
user group duration for housing assets but is for non-housing assets (although not
significant). Land wealth confers benefits under user group management as evidenced by the
positive coefficients on the interaction term, which is significant at the 5% level for non-
housing assets.

Table 5.17 Estimation Results for ∆ ln Total Asset Holdings (1995-2003)

Numbers in parentheses are t-statistics. Regional dummies included but not reported.
Standard errors robust to clustering. Significant at * 0.10 level ** 0.05 level *** 0.01 level

Total Assets House Asset Non-House
Assets

Intercept 5.602 ***
(11.51)
6.222 ***
(7.03)
4.520 ***
(6.24)
∆ % of household active .160
(.83)
.045
(.25)
.501 **
(2.15)
∆ % of household literate .257
(1.60)
.117
(.71)
.288
(1.29)
High ethnicity dummy .080
(.85)
.063
(.45)
.210
(1.30)
∆ Land owned (hectares) .039
(.92)
.003
(.06)
.163 ***
(4.10)
Asset holdings 1995 (ln) -.510 ***
(-12.55)
-.554 ***
(-7.32)
-.404 ***
(-5.77)

∆ FUG -.121
(-1.25)
-.115 *
(-1.72)
-.391 **
(-2.04)
FUG age in 2003 (months) -.001
(-1.11)
-.001
(-.46)
.002
(1.21)
FUG 2003 * Land in 1995 .115
(1.56)
.089
(.78)
.178 **
(1.99)

R
2
.4175 .2921 .3909
F 30.30 19.40 27.33
Observations 325 307 325
Clusters 31 31 31

151
Taken together, these results show that the implementation of user group management has a
negative impact on asset growth which is improved over time for total assets excluding
housing, and land ownership under user group management confers additional benefits
translating into asset growth.

5.2.2 Inequality

(i) Cross section data

A similar estimation strategy is employed as above to examine the impact of community
forestry on distributional equity. The following function is estimated individually on data
from both periods (t = 1995, 2003):
() ( )
ijt jt jt ijt
ln gini X' FUG α βη ε =+ + +
where the Ginis are those calculated and discussed above, estimated conditional on X, a
vector of community characteristics, and the management regime.

Explanatory community variables include:
(i) The population of the community, in the expectation that a large community
would be more urban and therefore somewhat more unequal;
(ii) The distance to the nearest paved road, to capture the remoteness of the
community, in the expectation that more remote communities would be less
unequal;
(iii) The inequality in land distribution, measured by the Gini index for land,
which is expected to have a positive impact on inequality;

152
(iv) The average household income in the community, in the expectation of a
positive relationship between average income and inequality;
(v) The number of local schools per household to capture the ability of the
community to find alternate livelihoods, in the expectation that more schools
will reduce inequality; and
(vi) Percentage of the households head by a member of the Chhetri or Brahmin
castes, as a measure of ethnic heterogeneity.

Again, the variable of interest is forest user group management. The models include a
dummy variable that takes the value of one if there is a forest user group in the community.
Also included is the age of the forest user group, if any.

Table 5.18 displays the OLS estimation results for 1995. Each column pair represents the
estimation results for a different Gini as dependent variable. Within each pair, one model
includes a dummy variable for the presence of a forest user group; the second includes a
variable indicating the duration of the forest user group. The explanatory power of these
models to account for inequality is very low, with F-statistics significant at the 5% level (or
less).

Nonetheless, some variables are significant: the land Gini is positive across all models and
significant. Average household income is also positively related to inequality. Larger
villages appear to have less inequality in income but not in assets, as do remote villages. The
number of schools per household reduces inequality in this time period but the coefficients
are not significant.

153
Table 5.18 Estimation Results for Inequality (1995)

Numbers in parentheses are t-statistics. Significant at * 0.10 level ** 0.05 level *** 0.01 level
Predicted value of FUG instrumented by: cf_dummy_95, %_households_with_land, UG_dummy_95.
F-statistic on first stage: 16.54.

Lastly, forest user group management did not have a significant impact on inequality in any
of the measures used in this earlier time period, although across all models the coefficients
are negative meaning there isles inequality under user group management.

The results for 2003 appear in Table 5.19. The explanatory power of these models is
somewhat higher, with F-statistics significant at the 1% level or better, as the community
characteristics better explain inequality in the communities. For example, larger
Variable Gini (PCE) 2SLS Gini (Total Assets) OLS

Intercept .060
(.25)
-.022
(-.09)
-.825 **
(-2.29)
-.822 **
(-2.24)
Ln (population) -.042 ***
(-2.65)
-.049 ***
(-2.79)
-.012
(-.47)
-.011
(-.46)
Ln (distance to paved road) -.024 ***
(-3.61)
-.007 *
(-1.81)
-.009
(-.96)
-.009
(-.96)
Gini (Land) .130 **
(2.50)
.149 ***
(2.70)
.152 **
(1.96)
.152 **
(1.96)
Ln (average household
income)
.044 **
(1.99)
.047 **
(1.97)
.119 ***
(3.47)
.119 ***
(3.39)
Ln (number of local schools
per household)
-.023
(-1.22)
-.025
(-1.21)
-.027
(-.93)
-.027
(-.92)
% of HH of high ethnicity -.012
(-.41)
-.016
(-.52)
-.041
(-.96)
-.041
(-.95)

FUG dummy -.048 *
(-1.66)
-.009
(-1.25)
-.030
(-1.10)
-.030
(-.96)
FUG * FUG age (years) -.000
(-.94)
-.000
(-.07)

R
2
.2187 .1809 .1841 .1841
F 3.48 1.99 2.55 2.20
Observations 88 81 87 87

154
communities are in general more unequal, as expected, as are richer communities, and
communities with more unequal land distribution. Remoteness is significant for income
inequality but not for asset inequality; the coefficients are consistently negative suggesting
that remote communities are more equal.

Table 5.19 Estimation Results for Inequality (2003)

Numbers in parentheses are t-statistics. Significant at * 0.10 level ** 0.05 level *** 0.01 level

However, the coefficients on the forest user group dummy are again not significant. After
accounting for differences in community characteristics such as size, remoteness, average
income level and land distribution, there is no evidence to support the hypothesis that forest
Variable Gini (PCE) Gini (Total Assets)

Intercept -.924 **
(-4.25)
-.919 **
(-4.19)
-1.465 ***
(-5.45)
-1.461 ***
(-5.39)
Ln (population) .004
(.29)
.004
(.28)
.068 ***
(3.79)
.068 ***
(3.77)
Ln (distance to paved road) -.023 ***
(-2.66)
-.023 ***
(-2.63)
-.008
(-.79)
-.008
(-.78)
Gini (Land) .079
(1.17)
.080
(1.18)
.337 ***
(4.04)
.337 ***
(4.02)
Ln (average household
income)
.116 ***
(5.15)
.115 ***
(5.07)
.133 ***
(4.78)
.132 ***
(4.71)
Ln (number of local schools
per household)
-.011
(-.66)
-.012
(-.68)
.035
(1.63)
.035
(1.60)
% of HH of high ethnicity -.094 ***
(-3.75)
-.095 ***
(-3.74)
-.030
(-.97)
-.031
(-.98)

FUG dummy .007
(.40)
.010
(.36)
-.010
(-.50)
-.016
(-.47)
FUG * FUG age (years) -.000
(-.34)
-.000
(-.22)

R
2
.4133 .4141 .5612 .5614
F 8.76 7.60 15.90 13.76
Observations 95 95 95 95

155
user group implementation has a negative impact on distributional equity in per capita
expenditures or total asset holdings.

This is not inconsistent with the descriptive data seen above, but there appeared to be
significant worsening in distribution for communities with newly-implemented user groups.
To find evidence for this again we need to look at the panel data.

(ii) Panel data

The estimation of the panel data uses the second empirical specification, in the following
variations:
() ( ) ( ) ln( )
( )
ijt ijt-1 it it-1 ijt
jtjt-1 ij
ln gini ln gini X X ' gini
FUG FUG
αβρ
η ε
−=+− +
+ −+

() ( ) ( ) ln()
( ) ( )
ijt ijt-1 it it-1 ijt
jtjt-1 jt ij
ln gini ln gini X X ' gini
FUG FUG FUG age
αβρ
η ϕε
−=+− +
+ −+ +

where the same Gini measures and community characteristics are used as above. The first
variation introduces again the change in user group management while the second variation
introduces the age of the user group to capture duration effects. The dependent variables are
the difference in the Gini indices for per capita real expenditure and total household assets;
the independent variables are the changes in the variables used above. As above, this is a
difference in differences estimation strategy which allows time-invariant community
variables to be differenced out.

156
The estimation results are shown in Table 5.20. These models are again not very good fits;
the sample size is small. However, we do see consistently positive coefficients on the change
in user group dummy, implying a worsening inequality for those communities introducing
user group management, although the coefficients are not significant. Further, the
coefficients on user group duration are also positive, lending only weak support to the
hypothesis that inequality within the community worsens as user group management
endures.

Table 5.20 Estimation Results for Change in Inequality (1995-2003)

Dependent variable ∆ Gini (·). Numbers in parentheses are t-statistics.
Significant at * 0.10 level ** 0.05 level *** 0.01 level

Variable Gini (PCE) Gini (Total Assets)
I II I II

Intercept .592 **
(2.80)
.537 **
(2.30)
.178 *
(1.98)
.177 *
(1.92)
∆ population -.000
(-.20)
-.000
(-.31)
.000
(1.20)
.000
(1.15)
∆ distance to paved road -.057
(-1.54)
-.058
(-1.54)
.018
(1.47)
.018
(1.44)
∆ average household income -.000
(-1.57)
-.000
(-1.56)
.000
(.59)
.000
(.58)
∆ number of local schools
per household
-.072
(-.45)
-.033
(-.19)
.039
(.74)
.040
(.70)
Gini(·) 1995 -2.076 **
(-2.89)
-2.005 **
(-2.72)
-.579 ***
(-2.92)
-.581 ***
(-2.84)

∆ FUG .072
(.60)
.033
(.24)
.029
(.71)
.028
(.57)
FUG age in 2003 (months) .001
(.62)
.000
(.08)

R
2
.4792 .4890 .3964 .3966
F 3.22 2.73 2.30 1.88
Observations 28 28 28 28

157
5.3 Conclusions

The large-scale institutional shift from centralized to local-level forest management in the
middle hills region of Nepal in 1993 was intended to halt what had been alarmingly high
rates of deforestation. Early evaluation of the success of this shift was therefore focused on
resource protection and regeneration, much of which has been encouraging. Subsequent
investigation however noted that the very institutional details enabling resource
improvement, such as use restrictions and limited exploitation, were negatively impacting
especially those households with few livelihood alternatives, in effect “saving the forest at
the expense of the poorest.” This not only left some households in the cold but resulted in
increasing inequality within these communities that were necessarily now involved in
promoting collective activity to enable effective resource management.

Employing a nationwide living standards data set, we test the hypotheses that community
forestry is detrimental to welfare and that distributional equity worsens. We find that on the
contrary, after correcting for possible endogeneity of user group formation, by 2003 per
capita expenditure and housing values have been positively impacted in those communities
with FUGs. However, panel data provides strong evidence that newly-formed forest user
groups have a significantly negative effect on growth of per capita expenditures, and on
growth of total assets excluding housing.

On the question of community inequality, there is only very weak evidence to support the
hypothesis that forest user group management has an effect on distributional equality in

158
general; in particular, inequality in per capita real expenditures and asset holdings increases
in communities with newly-formed forest user groups.

These results imply that community managed forests can improve household welfare in the
longer term but that the early transition period is welfare-reducing. These negative effects
can be especially critical given that the implementation of community-level resource
management is occurring in developing countries where income levels are low and
livelihoods are especially dependent on local resource exploitation. Exclusionary rules and
mandatory participatory requirements have uneven effects on households with limited means
for livelihood diversification. To prevent the poor from slipping further away, policymakers
will need to consider transition plans that might include compensation for negative
distribution effects and thereby encourage participation of all segments of those communities
involved in and affected by this institutional shift. Such policy decisions have wider
implications for the management of other resources as decentralization of authority becomes
widespread.

159
Chapter 6
Conclusions

The design of public policy often occurs through long debate among interested (and often
disinterested) parties, some of whom expect the payoffs from this debate to affect the
success of policy positions on other issues. This can result in unintentional incentives given
to a variety of stakeholders. The design of natural resource management regimes accurately
addressing incentives of involved participants is especially important given the enormous
cachet that participatory processes and decentralization now carry.

Within the communities themselves, the effects of heterogeneity are not well understood,
particularly when underlying relationships and super-imposed incentive mechanisms
combine to alter behavioral responses; thus the development of analytical models to depict
the impact of co-management regimes must include considerations of community
heterogeneity.

Because there is as yet an absence of definitive causal relationships between economic
behavioral choices and resource degradation when constrained within an existing

160
institutional framework, this research advances current knowledge by producing a
parameterized model of natural resource co-management as a candidate for application to
other sectors or nations.

In simulation results, we have shown how policy innovations can improve welfare and
distributional equity while ensuring resource improvement. A baseline equilibrium is first
established which is fairly consistent with results reported from a variety of case studies.
These studies show that once co-management is introduced into a forest community, welfare
declines are experienced by both elite and non-elite classes as a result of three effects: (1) an
increase in the domestic price of timber leads to substitution away from timber; (2) the
imposition of fines and exclusionary policies discourages informal collection of forest
products, upon which the non-elite are more dependent; and (3) the general price increase in
the market good reduces consumption possibilities by both the elite and the non-elite. The
welfare decline is sharper for the non-elite because of their relative dependence on forest
products from which they are being excluded.

Simulation exercises are run for three types of interventions: (i) wage policies; (ii)
productivity enhancements; and (iii) preference shifts. Additionally, payment for externality
services is introduced as a policy innovation, and the production and trade of NFTPs is
considered.

Over the baseline equilibrium case of co-management, paying wages for protection or
management services can lead to the improvement of welfare for all sectors of the forest
community and for the market sector. The wage rate paid in comparison to other competitive

161
wage rates in the economy is a critical factor in determining whether the market sector
thrives or withers: the forest department can squeeze the market sector out of the labor
market by offering high enough wages. In either case, however, increased timber production
outpaces improved management and protection and improvement in the forest resource
falters, although regeneration can still be accommodated. Additionally, increasing the wage
earning opportunities benefits the elite more than the non-elite and equity within the forest
community suffers.

Similarly, a decline in the wage paid by the forest department for harvesting labor leads to
increased timber production and a fall in the domestic price of timber. The non-elite benefit
both from the additional hiring opportunities and from the fall in the timber price, while the
elite suffer from a loss of income. Equity within the community improves at the expense of
the elite. If this wage decline is accompanied by a payment for protection or management
services, these effects are moderated. Again, in both cases the improvement of the forest
resource suffers.

Intervention in the form of productivity enhancements can lead to a mixed combination of
equity improvement, welfare improvement across sectors and resource preservation. The
most promising result is achieved if the productivity improvement is concentrated in
collection activity, which then yields a surplus of labor that can be released to the market and
formal timber industry. While equity improvement and overall welfare improvement come at
some expense of the forest resource to some extent, regeneration still occurs, thus this can
generate a win-win-win scenario. Alternatively, a combination of productivity improvements

162
in harvesting and in protection can yield minimal impact on the forest department and the
non-elite sector while providing welfare improvement for the elite and the forest.

The most promising institutional innovations come with the introduction of payment for
externality services and the promotion of NTFP production and trade. It appears possible to
set the externality payment so that welfare improvements are seen in all sectors accompanied
by continued forest regeneration, and possibly equity improvement. Further work with in
this area is needed to calibrate payments, but the experiments add weight to arguments in
favor of continued external support. Without sufficient funding from external sources to add
liquidity to the local economy, and technical assistance to aid the implementation of
improved silvicultural techniques, community forestry will cause immiseration throughout
the local economy.

The shift of domestic production from collection for consumption to collection for the
production of NTFPs for sale also has the potential to realize large improvements in the
forest, but negative welfare effects for the non-elite can be significant. Nevertheless,
experimentation with this innovation appears promising.

In another chapter, we examine household behavioral responses to varying degrees of user
management and find that having substitutable resources does not necessarily lead to
households reducing their labor allocation from resource extraction under a formal
management regime. This is in opposition to hypothesized behavioral responses and gives
evidence to strategic competitive behavior.

163
All results appear to support the case studies that wealth (as measured by landownership)
does not discourage the household from supplying labor to collection of firewood from the
community forest, even though there are alternatives available. Indeed, households with
private trees do not significantly reduce their collection from community forests which are
governed by formally recognized forest user groups, indicating that there is no substitution
effect and that these households are therefore taking their “fair share” from the community
forests even though they have alternatives. This is supportive of concerns expressed by case
studies showing how the putative equitable distribution of benefits disadvantages the poorest
in the community. Given this result, inequality within communities may worsen but the
evidence in this nationwide data set offers only weak evidence in support of this hypothesis.

In a final chapter, we test the hypothesis advanced in case studies that user group
management is welfare-reducing for the poorest households using two waves of a national
living standards survey which effectively bookend the large-scale institutional shock
implemented in Nepal in the mid-1990s. We find that although the shock has a negative
impact on growth of income in newly-formed user groups, the effect dissipates as user group
management persists and leads to improved incomes and asset holdings over the long term.
The data gives only weak evidence to support widening inequality within communities in the
short-run as a result of user group management.

The large-scale institutional shift from centralized to local-level forest management in the
middle hills region of Nepal in 1993 was intended to halt what had been alarmingly high
rates of deforestation. Early evaluation of the success of this shift was therefore focused on
resource protection and regeneration, much of which has been encouraging. Subsequent

164
investigation however noted that the very institutional details enabling resource
improvement, such as use restrictions and limited exploitation, were negatively impacting
especially those households with few livelihood alternatives, in effect “saving the forest at
the expense of the poorest.” This not only left some households in the cold but resulted in
increasing inequality within these communities that were necessarily now involved in
promoting collective activity to enable effective resource management.

Employing a nationwide living standards data set, we test the hypotheses that community
forestry is detrimental to welfare and that distributional equity worsens. We find that on the
contrary, after correcting for possible endogeneity of user group formation, by 2003 per
capita expenditure and housing values have been positively impacted in those communities
with FUGs. However, panel data provides strong evidence that newly-formed forest user
groups have a significantly negative effect on growth of per capita expenditures, and on
growth of total assets excluding housing.

On the question of community inequality, there is only very weak evidence to support the
hypothesis that forest user group management has an effect on distributional equality in
general; in particular, inequality in per capita real expenditures and asset holdings increases
in communities with newly-formed forest user groups.

These results imply that community managed forests can improve household welfare in the
longer term but that the early transition period is welfare-reducing. These negative effects
can be especially critical given that the implementation of community-level resource
management is occurring in developing countries where income levels are low and

165
livelihoods are especially dependent on local resource exploitation. Exclusionary rules and
mandatory participatory requirements have uneven effects on households with limited means
for livelihood diversification. To prevent the poor from slipping further away, policymakers
will need to consider transition plans that might include compensation for negative
distribution effects and thereby encourage participation of all segments of those communities
involved in and affected by this institutional shift. Such policy decisions have wider
implications for the management of other resources as decentralization of authority becomes
widespread.

A policy prescription that emerges from the findings includes increasing the availability of
land to marginal households. Although the leasehold program in Nepal has been aimed at the
poor, the implementation has suffered from procedural limitations and has encouraged intra-
community conflicts. A redesign of the program to grant individual land ownership rather
than group leasing to targeted marginal households similar to the program found in Vietnam
(Nguyen, 2006) may allow such conflicts to be avoided. Additionally, direct support
payments in compensation for resource exclusion are clearly justifiable, given that
substantial funding requirements already exist. Such payments can be marketed as payment-
for-externality-services to avoid negative connotations of conditional aid and can be
graduated over time as the resource regenerates.

The implications of correct policy prescriptions are significant. As global population
increases and the demands placed on natural resources rise as a consequence of economic
growth and industrialization, institutional arrangements for the protection of global wealth
are of critical importance. Local-level management regimes appear to be more efficient but

166
may result in poor immediate outcomes for those with few alternatives and little leeway in
livelihood support. The design of appropriate institutional details is therefore critical so that
the welfare of the poorest is not ignored in exchange for the protection of our natural
resources.

167
Bibliography

Acharya, K.P. 2002. “Twenty-Four Years of Community Forestry in Nepal.” International
Forestry Review 4(2): 149-156.

Agrawal, Arun. 2002. “Common Resources and Institutional Sustainability.” In Drama of
the Commons. Elinor Ostrom, Thomas Dietz, Nives Dolsak, Paul Stern, Susan Stonach, Elke
Weber, eds. Pp. 41-85. Washington, D.C.: National Academy Press.

Agrawal, Arun and Elinor Ostrom. 2001. “Collective Action, Property Rights and
Decentralization in Resource Use in India and Nepal.” Politics and Society 29(4): 485-514.

Akida, A. and R. Bromley, 2007. “Trends in Forest Ownership, Forest Resources Tenure and
Institutional Arrangements: A Case Study from Tanzania.” Rome: FAO.

Axelrod, R. 1984. The Evolution of Cooperation. Princeton, NJ: Princeton University Press.

Baland, Jean-Marie, Pranab Bardhan, Sanghamitra Das, Dilip Mookherjee and Rinki Sarkar,
2002. “Inequality, Collective Action and the Environment: Evidence from Firewood
Collection in Nepal.” Working paper.

Baland, Jean-Marie and Jean-Phillippe Platteau. 1996. Halting Degradation of Natural
Resources: Is There a Role for Rural Communities? New York: Oxford University Press.

Becker, C. and R. Leon. 2000. “Indigenous Forest Management in the Bolivian Amazon:
Lessons from the Yucarare People.” In People and Forests: Communities, Institutions and
Governance. Clark Gibson, Margaret McKean and Elinor Ostrom, eds. Pp. 135-161.
Cambridge, Massachusetts: MIT Press.

Berkes, Fikret. 1992. “Success and Failure in Marine Coastal Fisheries of Turkey.” In
Making the Commons Work: Theory, Practice, and Policy. Daniel Bromley, ed. Pp. 161-182.
San Francisco: Institute for Contemporary Studies Press.

Bhattarai, Basundhara and Hemant Ojhi, 2000. “Distributional Impact of Community
Forestry: Who is Benefiting from Nepal’s Community Forests?” Forest Action Research
Series: 00/01.

Bray, David, Leticia Merino-Pérez and Deborah Barry. 2005. The Community Forests of
Mexico: Managing for Sustainable Landscapes. University of Texas Press.

168
Bromley, Daniel. 1989. Economic Interests and Institutions – The Conceptual Foundations
of Public Policy. Oxford: Basil Blackwell.

Bromley, Daniel. 1992. “The Commons, Property, and Common Property Regimes.” In
Making the Commons Work. D. Bromley, ed. Pp. 3-16. San Francisco: Institute for
Contemporary Studies.

Bulte, Erwin and Richard Horan. 2003. “Habitat Conservation, Wildlife Extraction and
Agricultural Expansion.” Journal of Environmental Economics 45: 109-127.

Cardenas, J. 2003. “Real Wealth and Experimental Cooperation: Experiments in the Field
Lab.” Journal of Development Economics 70: 263-289.

Charkravarti, Milindo, Samar Datta, E. Lance Howe and Jeffrey Nugent. 2005. “Joint Forest
Management: Experience and Modeling.” In Economics, Sustainability, and Natural
Resources. Shashi Kant and Albert Berry, eds.

Clarke, J. 2007. “Trends in Forest Ownership, Forest Resources Tenure and Institutional
Arrangements: A Case Study from South Africa.” Rome: FAO.

Cornes, Richard, and Todd Sandler. 1983. “On Commons and Tragedies.” American
Economic Review 73: 787-792.

Cooper, Christine, Juliette Finzi, Jay Hasbrouck and Vincent Todd. 2004. A Case Study of
Sustainability: Jalcomulco, Veracruz. Los Angeles: Center for Sustainable Cities, University
of Southern California.

Crutchfield, J.A. “The Marine Fisheries: The Problem in International Cooperation.”
American Economic Review 54:207-218.

Dasgupta, Partha, and Geoffrey Heal. 1979. Economic Theory and Exhaustible Resources.
Cambridge, Massachusetts: Cambridge University Press.

Davis, L. and Douglas North. 1971. Institutional Change and American Economic Growth.
Cambridge, Massachusetts: Cambridge University Press.

Demsetz, H. 1967. “Toward a Theory of Property Rights.” American Economic Review
62:347-359.

Dev, O.P., N.P. Yadav, O. Springate-Baginski and J. Soussan. 2003. “Impacts of
Community Forestry on Livelihoods in the Middle Hills of Nepal.” Journal of Forest and
Livelihood 3(1): 64-77.

Edmonds, Eric. 2002. “Government-Initiated Community Resources Management and Local
Resource Extraction from Nepal’s Forests.” Journal of Development Economics 68: 89-115.

169
Ellickson, Robert. 1991. Order Without Law: How Neighbors Settle Disputes. Cambridge,
MA: Harvard University Press.

Fei, John C. and Gustav Ranis. 1964. Development of the Labor Surplus Economy.
Homewood, IL: Irwin.

Food and Agriculture Organization (FAO). 2006. Understanding Forest Tenure in South and
Southeast Asia. Rome: United Nations Publications.

Food and Agriculture Organization (FAO). 2003. State of the World’s Forests 2003. Rome:
United Nations Publications.

Food and Agriculture Organization (FAO), 1999. FRA2000: Forest Resources of Nepal:
Counntry Report. Rome: United Nations Publications.

Food and Agriculture Organization (FAO). 1999. Status and Progress in the Implementation
of National Forest Programmes: Outcomes of an FAO Worldwide Survey. Rome: United
Nations Publications.

Feder, G. and D. Feeny. 1991. “Land Tenure and Property Rights: Theory and Implications
for Development Policy.” World Bank Economic Review 5(1): 135-153.

Gautam, A.P., G.P. Shivakoti and E.L. Webb. 2004. “A Review of Forest Policies,
Institutions, and Changes in the Resource Condition in Nepal.” International Forestry
Review 6(2): 136-148.

Gebremedhin, B., J. Pender and G. Tesfay. 2000. “Community Natural Resource
Management: The Case of Woodlots in Northern Ethiopia.” EPTD Discussion Paper.
Washington D.C.: International Food Policy Research Institute.

Gibson, Clark and C. Dustin Becker. 2000. “A Lack of Institutional Demand: Why a Strong
Local Community in Western Ecuador Fails to Protect Its Forest.” In People and Forests:
Communities, Institutions and Governance. Clark Gibson, Margaret McKean and Elinor
Ostrom, eds. Pp. 135-161. Cambridge, Massachusetts: MIT Press.

Gibson, Clark, Margaret McKean and Elinor Ostrom. 2000. “Explaining Deforestation: The
Role of Local Institutions.” In People and Forests: Communities, Institutions and
Governance. Clark Gibson, Margaret McKean and Elinor Ostrom, eds. Pp. 1-26. Cambridge,
Massachusetts: MIT Press.

Gibson, Clark, Elinor Ostrom and Margaret McKean. 2000. “Forests, People and
Governance: Some Initial Theoretical Lessons.” In People and Forests: Communities,
Institutions and Governance. Clark Gibson, Margaret McKean and Elinor Ostrom, eds. Pp.
227-242. Cambridge, Massachusetts: MIT Press.

Gordon, H.S.1954. “The Economic Thery of a Common-Property Resource: The Fishery.”
Journal of Political Economy 62(April): 124-142.

170
Greif, A. 1992. “Institutions and International Trade: Lessons for the Commercial
Revolution.” American Economic Review.

Greif, A. 1997. “Microtheory and Recent Developments in the Study of Economic
Institutions through Economic History.” In Advances in Economic Theory, Volume II. D.M.
Kreps and K.F. Wallis, eds. Cambridge, Massachusetts: Cambridge University Press.

Hardin, Garrett. 1968. “The Tragedy of the Commons.” Science 162: 1243-1248.

HMGN, 2005. Poverty Trends in Nepal (1995-96 and 2003-4). Kathmandu: Central Bureau
of Statistics.

Johnson, O. 1972. “Economic Analysis, the Legal Framework and Land Tenure Systems.”
Journal of Law and Economics 15(1): 259-276.

Kaimowitz, David and Arild Angelson. 1998. Economic Models of Tropical Deforestation:
A Review. Center for International Forestry Research.

Kant, Shashi. 2000. “A Dynamic Approach to Forest Regimes in Developing Economics.”
Ecological Economics 23: 287-300.

Kant, Shashi and R. Albert Berry. 2001. “A Theoretical Model of Optimal Forest Resource
Regimes in Developing Economies.” Journal of Institutional and Theoretical Economics
157(2): 331-355.

Kenneth, R. 1989. “Solving the Common-Property Dilemma: Village Fisheries Rights in
Japanese Coastal Waters.” In Common Property Resources: Ecology and Community-Based
Sustainable Development. F. Berkes, ed. London: Belhaven Press.

Kigenyi, F.W. 2007. “Trends in Forest Ownership, Forest Resources Tenure and
Institutional Arrangements: A Case Study from Uganda.” Rome: FAO.

Klooster, Dan. 2000. “Institutional Choice, Community, and Struggle: A Case Study of
Forest Co-Management in Mexico.” World Development 28(1): 1-20.

Kopelman, Shirli, J. Mark Weber and David Messick. 2002. “Factors Influencing
Cooperation in Commons Dilemmas: A Review of Experimental Psychological Research.”
In Drama of the Commons. Elinor Ostrom, Thomas Dietz, Nives Dolsak, Paul Stern, Susan
Stonach, Elke Weber, eds. Pp. 113-156. Washington, D.C.: National Academy Press.

Krishna, A. 1999. “Creating and Harnessing Social Capital.” In Social Capital: A
Multifaceted Perspective. P. Dasgupta and I. Serageldin, eds. Washington, D.C.: World
Bank.

Kumar, Nalini. 2002. “The Challenges of Community Participation in Forest Development
in Nepal.” IEG Working Paper. Washington, D.C.: World Bank.

171
Kumar, Sanjay. 2002. “Does Participation in Common Pool Resource Management Help the
Poor? A Social Cost-Benefit Analysis of Joint Forest Management in Jharkhand, India.”
World Development 30(5): 763-782.

Livelihoods and Forestry Programme (LFP), 2003. Hill Livelihoods Baseline Study.
Kathmandu.

Lloyd, William. 1833. “On the Checks to Population.” In Managing the Commons. G.
Hardin and J. Baden, eds. (1977) W.H. San Francisco: Freeman & Co.

Malla, Yam Bahadur. 2000. “Impact of Community Forestry Policy on Rural Livelihoods
and Food Security in Nepal. Unasylva: International Journal of Forestry and Forest
Industries Vol 51 No.202: 35-42.

Malla, Yam B., Hari Neupane and Peter Branney, 2003. “Why aren’t Poor People Benefiting
More from Community Forestry?” Journal of Forest and Livelihood 3(1): 78-93

Maskey, Vishakha, Tesfa Gebremedhin and Timothy Dalton, 2003. “ Survey Analysis of
Participation in a Community Forest Management in Nepal.” Working paper.

Mason, Charles, Todd Sandler and Richard Cornes. 1988. “Expectations, the Commons and
Optimal Group Size.” Journal of Environmental Economics and Management 15: 99-110.

McCay, Bonnie. 2002. “Emergence of Institutions for the Commons: Contexts, Situations,
and Events.” In Drama of the Commons. Elinor Ostrom, Thomas Dietz, Nives Dolsak, Paul
Stern, Susan Stonach, Elke Weber, eds. Pp. 361-402. Washington, D.C.: National Academy
Press.

McKean, M. 1992. “Success on the Commons: A Comparative Examination of Institutions
for Common Property Management.” Journal of Theoretical Politics 4(3): 247-281.

Messerschmidt, Donald. 1987. “Conservation and Society in Nepal: Traditional Forest
Management and Innovative Development.” In Lands At Risk in the Third World: Local-
Level Perspectives. Peter D. Little and Michael M. Horowitz, eds. Pp 373-397. Boulder,
Colorado: Westview Press.

Muñoz-Piña, Carlos, Alain de Janvry and Elisabeth Sadoulet. 2002. “Recrafting Rights Over
Common Property Resources in Mexico: Divide, Incorporate and Equalize.” Working paper.
University of California, Berkeley.

Nagendra, Harini, Birendra Karna and Mukunda Karmacharya. 2005. “Examining
Institutional Change: Social Conflict in Nepal’s Leasehold Forestry Programme.”
Conservation and Society 3(1): 72-91.

Nasir, S.M., 2006. “Trends in Forest Ownership, Forest Resources Tenure and Institutional
Arrangements: A Case Study from Pakistan.” Rome: FAO.

172
Neupane, Hari, 2003. “Contested Impact of Community Forestry on Equity: Some
Evidences from Nepal.” Journal of Forest and Livelihood 2(2): 55-62

Nguyen, Q.T. 2006. “Trends in Forest Ownership, Forest Resources Tenure and Institutional
Arrangements: A Case Study from Vietnam.” Rome: FAO.

North, Douglas. 1981. Structure and Change in Economic History. New York, New York:
Norton.

North, Douglas. 1990. Institutions, Institutional Change and Economic Performance.
Cambridge, Massachusetts: Cambridge University Press.

Nugent, J. and N. Sanchez. 1993. “Tribes, Chiefs, and Transhumance: A Comparative
Institutional Analysis.” Economic Development and Cultural Change 42(1): 87-113.

Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of
Groups. Cambridge, Massachusetts: Harvard University Press.

Ostrom, Elinor. 1990. Governing the Commons: The Evolution of Institutions for Collective
Action. Cambridge, Massachusetts: Cambridge University Press.

Ostrom, Elinor. 1992. Crafting Institutions for Self-Governing Irrigation Systems. San
Francisco: Institute for Contemporary Studies.

Ostrom, Elinor. 1998. “A Behavioral Approach to the Rational Choice Theory of Collective
Action.” American Political Science Review 92(1).

Ostrom, Elinor. 1999. “Coping with Tragedies of the Commons.” Annual Review of Political
Science 2: 493-535.

Ostrom, Elinor. 1999. “Institutional Rational Choice: An Assessment of the IAD
Framework.” In Theories of the Policy Process. P. Sabatier, ed. Boulder, Colorado:
Westview Press.

Ostrom, Elinor. 2000. “Collective Action and the Evolution of Social Norms.” Journal of
Economic Perspectives 14(3).

Ostrom, Elinor, R. Gardner, and J. Walker. 1994. Rules, Games and Common-Pool
Resources, Ann Arbor, Michigan: University of Michigan Press.

Ostrom, Elinor, J. Walker and R. Gardner. 1992. “Covenants with and without a Sword:
Self-Governance is Possible.” American Political Science Review 86(2): 404-417.

Ostrom, Elinor and Mary Beth Wertime. 2000. “International Forestry Resources and
Institutions Research Strategy.” In People and Forests: Communities, Institutions and
Governance. Clark Gibson, Margaret McKean and Elinor Ostrom, eds. Pp. 243-268.
Cambridge, Massachusetts: MIT Press.

173
Pearce, David and Katrina Brown. 1994. “Saving the World’s Tropical Forests.” In The
Causes of Tropical Deforestation. David Pearce and Katrina Brown, eds. Pp. 2-26. London:
UCL Press.

Poole, Peter. 1989. Developing a Partnership of Indigenous Peoples, Conservationists and
Land Use Planners in Latin America. Unpublished report. Washington, D.C.: World Bank.

Richards, Michael. 2000. “Can Sustainable Tropical Forestry Be Made Profitable? The
Potential Limitations of Innovative Incentive Mechanisms.” World Development 28(6):
1001-1016.

Runge, C.F. 1981. “Common Property Externalities: Isolation, Assurance and Resource
Depletion in a Traditional Grazing Context.” American Journal of Agricultural Economics
63: 595-606.

Runge, C.F. 1984. “Strategic Interdependence in Models of Property Rights.” American
Journal of Agricultural Economics 66(5): 807-813.

Runge, C.F. 1986. “Common Property and Collective Action in Economic Development.”
World Development 14(5): 623-635.

Runge, C.F. 1992. “Common Property and Collective Action in Economic Development.” In
Making the Commons Work: Theory, Practice and Policy. Daniel Bromley, ed. Pp. 17-40.
San Francisco: Institute for Contemporary Studies Press.

Rusnack, Gerett. 1997. “Co-management of Natural Resources in Canada: A Review of
Concepts and Case Studies.” Ottawa: International Development Research Centre.

Sandler, T. 1992. Collective Action: Theory and Applications. Ann Arbor, MI: University of
Michigan Press.

Schaefer, M.B. 1957. “Some Considerations of Population Dynamics and Economics in
Relation to the Management of Commercial Marine Fisheries.” Journal of the Fisheries
Research Board of Canada 14:669-681.

Simorangkir, D. and Sardjono, M.A. 2006. “Implications of Forest Utilization, Conversion
Policy and Tenure Dynamics on Resource Management and Poverty Reduction: A Case
Study form Pasir District, East Kalimantan, Indonesia.” Rome: FAO.

Singh, K.D., J.P. Singh and B. Sinha, 2006. “Trends in Forest Ownership, Forest Resources
Tenure and Institutional Arrangements: A Case Study form Orissa, India. Rome: FAO.

Sitoe, A. and Tchaúque, F.J. 2007. “Trends in Forest Ownership, Forest Resources Tenure
and Institutional Arrangements: A Case Study from Mozambique.” Rome: FAO.

174
Springate-Baginski, O., O.P. Dev, N.P. Yadav and J. Soussan. 2003. “Community Forest
Management in the Middle Hills of Nepal: The Changing Context.” Journal of Forest and
Livelihood 3(1): 5-20.

Tachibana, Towa, Hari Upadhyaya, Ridish Pokharel, Santosh Rayamajhi and Keijiro Otsuka,
2001. “Common Property Forest Management in the Hill Region of Nepal.” In Land Tenure
and Natural Resource Management: A Comparative Study of Agrarian Communities in Asia
and Africa. Keijiro Otsuka and Frank Place, eds.

Taylor, M. 1992. “The Economics and Politics of Property Rights and Common Pool
Resources.” Natural Resources Journal 32(3): 633-648.

Varughese, G. and Elinor Ostrom. 2001. “The Contested Role of Heterogeneity in Collective
Action: Some Evidence from Community Forestry in Nepal.” World Development 29(3):
747-765.

Wade, R. 1988. Village Republics: Economic Conditions for Collective Action in South
India. Cambridge, Massachusetts: Cambridge University Press.

WRI, World Resources Institute. 1995. Balancing Acts: Community-Based Forest
Management and National Law in Asia and the Pacific. Washington, D.C.

Yadav, N.P., O.P. Dev, O. Springate-Baginski and J. Soussan. 2003. “Forest Management
and Utilization Under Community Forestry.” Journal of Forest and Livelihood 3(1): 37-50.

Zheng, B., 2006. “Changes and Trends in Forest Tenure and Institutional Arrangements for
Collective Forest Resources in Yunnan Province, China.” Rome: FAO

175
Appendix A
Model Specification

A.1 Functional Forms

Production:

Protection: Collection: Market Good: Harvested Timber:
()
P
e d
P P P
N F ) ( β = ⋅ ( )
df
e
df df df
N F ) ( β = ⋅ ( )
X
e d
R X
s
R
N X ) ( β = ⋅ ( )
H H
e d
H
e
H H
N F F ) (
) 1 ( −
= ⋅ β

Forest Community A:
Utility:
()
,
,
,, ,
,
,,
1
1
1 2
,, , , , , ,
,
,,
34
,, , ,
() (1 )( ) ( )
( ) ( )
CA
CA
FA FA C A
FA
CA CA
d
CA F A df A F A C A C A C A
ZA
CA C A
d
CA CA CA C A
FF G
UA
XL
ρ
ρ
ρρ ρ
ρ
ρρ
ββ β β ϕ
β
ββ




+− +
 



=



++
 

,
,
,
1
,
(1 )( )
ZA
Z A
ZA
ZA E
F
ρ
ρ
ρ
β
 
 

 

 

+−
 

 

 
 
 

where ()
f
e
d
X
d
R
d
B C
d
A C f B df A df f B P A P f f E
F F F F F F F F F F ) ( ) ( ) (
, ,
3
, ,
2
, ,
1 0
+ + + − + − + + = α α α α

176

Budget Constraint:

()
1
,
,, ,
,,
,, 1 , ,
(1 ) ( )
exp( )
s
PA ss d s d
RRA P PA H H H F FE X ss
PA P B
b
dd
XCA F CA dfA dfB L
N
wN wN w N P F P F
NN
PX PF a F F F
αϕφ φ ν
γ

++ + +− +


+

=+ + + −

Time Constraint:

A TOT A C
d
H H
s
A R
s
A P A df
N L N N N N
, , , , ,
= + + + + α

Forest Community B:
Utility:
()
,
,
,, ,
,
,,
1
1
1 2
,, , , , , ,
,
,,
34
,, , ,
( ) (1 )( ) ((1 ) )
( ) ( )
CB
CB
FB FB C B
FB
CB CB
d
CB F B df B F B C B C B C B
ZB
CB C B
d
CB CB C B CB
FF G
UA
XL
ρ
ρ
ρρ ρ
ρ
ρρ
ββ β β ϕ
β
ββ




+− + −
 



=


++


,
,
,
1
,
(1 )( )
ZB
Z B
ZB
ZB E
F
ρ
ρ
ρ
β
 

 

 

 

+−
 

 

  
 
 

Budget Constraint:

()
1
,
,, ,
,,
,, 1 , ,
(1 ) ( 1 ) ( )
(1 ) exp( )
s
PB ss d s d
RRB P PB H H H F FE X ss
PA P B
b
dd
XCB F CB dfA dfB L
N
wN wN w N P F P F
NN
PX PF a F F F
αϕφ φ ν
γ

++ +− +− +


+

=+ +− + −

177
Time Constraint:

B TOT B C
d
H H
s
B R
s
B P B df
N L N N N N
, , , , ,
= + + + + α

Residual Sector:
Utility:
()
R Z
R Z
R Z
R
R R R
E R Z
d
R R R R
d
R R R Z R R
F X G F A U
,
,
,
1
,
1
3 2 1
,
) )( 1 ( ) ( ) ( ) (
ρ
ρ
ρ
ρ
ρ ρ ρ
β β β β β








− +








+ + =

Budget Constraint:

d
R F
d
R X
d
R R
s
R X
F P X P N w X P t + = − − ) )( 1 (

Forest Department:
Objective Function:

C N w N w F F a G F P F P
d
H H
d
P P
b
df FD
d
X E F
s
F
− − − − + + + − −
1
) (exp( ) )( 1 )( 1 (
1 ,
ν ε

Government:

FD R C
d
X E F
s
F
d
R R
s
R X
G G G F P F P N w X P t + + = + − + − ) )( 1 ( ) (
,
ν ε

178
Market Clearing:

()
d
R
d
B C
d
A C
s
R
X X X X + + = ⋅
, ,

d
X
d
R
d
B C
d
A C H
F F F F F + + + =
, ,

179
Table A.2 Variables

VARIABLE DESCRIPTION
A df
N
,

Labor supplied to collection activities by forest community “A” (FC-A)
B df
N
,

Labor supplied to collection activities by forest community “B” (FC-B)
s
A P
N
,

Labor supplied to protective services by FC-A
s
B P
N
,

Labor supplied to protective services by FC-B
s
A R
N
,

Labor supplied to the market by FC-A
s
B R
N
,

Labor supplied to the market by FC-B

d
H
N
Harvesting labor demanded by the forest department
d
R
N
Market labor demanded by the market sector
d
A C
F
,

Demand for harvested timber by FC-A
d
B C
F
,

Demand for harvested timber by FC-B
d
R
F
Demand for harvested timber by the residual sector
d
A C
X
,

Demand for the market good by FC-A
d
B C
X
,

Demand for the market good by FC-B
d
R
X
Demand for the market good by the residual sector
A C
L
,

Demand for leisure by FC-A
A C
L
,

Demand for leisure by FC-B

R
w
Wage paid for market labor
F
P
Local price of harvested timber
X
P
Local price of the market good

180
Table A.3 Parameter Values

PARAMETER DESCRIPTION VALUE
A C
A
,

Multiplier on utility of forest community “A” (FC-A) 3.0
B C
A
,

Multiplier on utility of forest community “B” (FC-B) 3.0
R
A
Multiplier on utility of the market sector (R) 3.0
A C,
σ
Elasticity of substitution between consumption items for
FC-A
1.2
B C,
σ
Elasticity of substitution between consumption items for
FC-B
1.2
R
σ
Elasticity of substitution between consumption items for R 1.2
A F ,
σ
Elasticity of substitution between harvested timber and
collected forest products for FC-A
3.0
B F ,
σ
Elasticity of substitution between harvested timber and
collected forest products for FC-B
3.0
R F ,
σ
Elasticity of substitution between harvested timber and
collected forest products for R
3.0
A Z ,
σ
Elasticity of substitution between current consumption and
the forest externality for FC-A
0.5
B Z ,
σ Elasticity of substitution between current consumption and
the forest externality for FC-B
0.5
R Z ,
σ
Elasticity of substitution between current consumption and
the forest externality for R
0.5
1
,A C
β
Coefficient on first argument (forest products) in utility
function of FC-A
0.35
2
,A C
β
Coefficient on second argument (government transfer) in
utility function of FC-A
0.2
3
,A C
β
Coefficient on third argument function (market good) in
utility of FC-A
0.4
4
,A C
β
Coefficient on fourth argument (leisure) in utility function
of FC-A
0.05
A F ,
β
Coefficient on collected forest products in inner nested CES
of the first argument of utility function of FC-A
0.3
A Z ,
β
Coefficient on current consumption in outer nested CES of
utility function of FC-A
0.999
1
,B C
β
Coefficient on first argument (forest products) in utility
function of FC-B
0.6
2
,B C
β
Coefficient on second argument (government transfer) in
utility function of FC-B
0.2
3
,B C
β
Coefficient on third argument (market good) in utility
function of FC-B
0.15
4
,B C
β
Coefficient on fourth argument (leisure) in utility function
of FC-B
0.05

181
B F ,
β
Coefficient on collected forest products in inner nested CES
of the first argument of utility function of FC-B
0.5

Table A.3 Parameter Values (Continued)

PARAMETER DESCRIPTION VALUE
B Z ,
β
Coefficient on current consumption in outer nested CES of
utility function of FC-B
0.99
1
R
β
Coefficient on first argument function (forest products) in
utility of R
0.25
2
R
β
Coefficient on second argument (government transfer) in
utility function of R
0.2
3
R
β
Coefficient on third argument (market good) in utility
function of R
0.45
R F ,
β
Coefficient on collected forest products in inner nested CES
of the first argument of utility function of R
0.0
R Z ,
β
Coefficient on current consumption in outer nested CES of
utility function of R
0.999
df
β
Coefficient for labor in the production of collected forest
products
2.50
df
e
Labor share in the Cobb-Douglas production function for
forest products
0.95
P
β
Coefficient for labor in the production of protected forest 1.00
P
e
Labor share in the Cobb-Douglas production function for
protection
0.90
X
β
Coefficient for labor in the production of the market good 8.00
X
e
Labor share in the Cobb-Douglas production function for
the market good
0.80
H
β
Coefficient for labor in the harvesting of timber 6.00
H
e
Labor share in the Cobb-Douglas production function for
harvesting
0.80
C Fixed costs of harvesting 1.00
0
f
α
Coefficient on forest stock in externality function 0.005
1
f
α
Coefficient on forest production through protection in
externality function
0.05
2
f
α
Coefficient on forest product collection in externality
function
0.06
3
f
α
Coefficient on harvested timber in externality function 0.02
f
e
Elasticity on externality function 1.5
1
a
Multiplier on the penalty function 4.0
1
b
Exponent on the penalty function 0.05
2
b
Exponent on the detection probability function 1.0

182
F
Initial forest stock 6000
L
F
Allowable level of informal forest product extraction 10
Table A.3 Parameter Values (Continued)

PARAMETER DESCRIPTION VALUE
TOT
N
Total labor supply of the forest community 100
A TOT
N
,

Labor supply of forest community “A” 10
P
w
Protecting wage 0.0
H
w
Harvesting wage 3.50
E F
P
,

Export price of harvested timber 1.00
d
X
F
Export demand for harvested timber 200
0
R
α
Constant term in import function of market good 3.0
1
R
α
Coefficient on first-order term in market import function 1.0
2
R
α
Coefficient on second-order term in market import function 5.0
X
P
Domestic price ceiling for market good, above which
imports are triggered
2.0
t Tax rate on profits in the market sector 0.1
ε Share of timber sales revenues (net of transfer to forest
community) given to government
0.40
C
ω
Share of government revenues transferred to the forest
community
0.3
R
ω
Share of government revenues transferred to the market
sector
0.2
H
α
Share of harvest labor captured by FC-A 0.1
γ
Share of excess collection penalties paid by FC-A 0.1
φ Share of timber sales revenues which is distributed equitably
to the forest community
1.0
ϕ
Share of transfers to the forest community (both government
and share of timber sales revenues) which are captured by
FC-A
0.1
ν Share of timber sales revenues which are transferred from
the forest department to the forest community
0.6

Abstract (if available)

Abstract Globalization, population pressures and consumptive choices coincident with higher incomes have induced an urgent search for optimal management techniques as indigenous resources have become increasingly accessible and hence more vulnerable to over-exploitation. Devastating losses of forest cover and persistently high rates of deforestation in south Asia have led governments to devolve authority over forest management to the community level.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Against the wind: labor force participation of women in Iran

PDF

Armed conflict, education and the marriage market: evidence from Tajikistan

PDF

The Catholic church in Latin America: an evaluation of the institutional and political impacts of progressive church reforms

PDF

Chinese government’s role in crisis management: case studies of three major crises in recent years

PDF

One size fits none: theorizing weight management in the everyday lives of adults with serious mental illness

PDF

The financial crisis: signaling a new order of communications

PDF

The role of race in configuring park use: a political ecology perspective

PDF

Communities of innovation: cyborganic and the birth of networked social media

PDF

A call for a new code of responsibility: the value of establishing a CSR reporting standard

PDF

Regulation of bone formation in vitro and in vivo

PDF

Achieving equity in educational outcomes through organizational learning: enhancing the institutional effectiveness of community colleges

PDF

DIGGing for votes: an analysis of 2008 presidential candidates' use of new media

PDF

Plato's Magnesia and philosophical polities in Magna Graecia

PDF

Plasma levels of paclitaxel in the elderly. Effect of liver and renal function.

PDF

Communication and home-based interventions in Alzheimer’s disease -- a review

PDF

The effects of online courses for student success in basic skills mathematics classes at California community colleges

PDF

Measuring the alignment of high school and community college math assessments

PDF

Legal origin and bilateral and multilateral dispute settlement mechanisms: the case of patent protection legislation

PDF

Creating community online: the effects of online social networking communities on college students' experiences

PDF

Wonder boys: tales of the extraordinarily queer adolescent

Asset Metadata

Creator Cooper, Christine (author)

Core Title Community forest management in Nepal: saving the forest at the expense of the poorest

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Economics

Degree Conferral Date 2008-08

Publication Date 08/01/2008

Defense Date 05/13/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag community management,Forest management,Inequality,natural resources,Nepal,OAI-PMH Harvest,Welfare

Place Name Nepal (countries)

Language English

Advisor Nugent, Jeffrey B. (committee chair), Strauss, John (committee member), Wolch, Jennifer (committee member)

Creator Email ccooper@reconcorp.com,ccooper@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1467

Unique identifier UC1182595

Identifier etd-Cooper-2206 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-100828 (legacy record id),usctheses-m1467 (legacy record id)

Legacy Identifier etd-Cooper-2206.pdf

Dmrecord 100828

Document Type Dissertation

Rights Cooper, Christine

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu