Jump to ContentJump to Main Navigation
Communities and Markets in Economic Development$

Masahiko Aoki and Yujiro Hayami

Print publication date: 2001

Print ISBN-13: 9780199241019

Published to Oxford Scholarship Online: August 2004

DOI: 10.1093/0199241015.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 13 August 2020

Evolution and Consequences of Community Forest Management in the Hill Region of Nepal

Evolution and Consequences of Community Forest Management in the Hill Region of Nepal

Chapter:
(p.295) 11 Evolution and Consequences of Community Forest Management in the Hill Region of Nepal
Source:
Communities and Markets in Economic Development
Author(s):

Keijiro Otsuka

Towa Tachibana

Publisher:
Oxford University Press
DOI:10.1093/0199241015.003.0011

Abstract and Keywords

This chapter analyses the feasibility and effectiveness of forest resource management under the community management system, based on a case study of community forests in Nepal. The factors affecting the formation of community management systems are identified. The impact of community forest management on forest conditions is examined. It is argued that the benefit of community management exceeds the cost of establishing such as system only when forest resources have become sufficiently scarce and valuable.

Keywords:   evolution, consequences, community forest management, Nepal

Community management of local natural resources, such as forest, water, and rangeland, has been receiving increasing attention in the development literature as a potentially effective institutional arrangement to regulate the use of these resources (Baland and Platteau 1996; Bardhan 1993; Hayami 1997). Having perceived the failures of central government's ownership and control, international donor agencies and the governments of many developing countries began to adopt the participatory community approach, in which local resource users are entrusted to manage natural resources (FAO 1989; World Bank 1996). The promotion of this new policy has been supported by new developments in economic theory, particularly in the game‐theoretic fields, which suggests the possibility of avoiding the tragedy of the commons under a variety of conditions (Pearce 1992). Further, several authors argue that community management is potentially an effective institutional arrangement to attain efficient use of local natural resources, particularly if it is granted formal and assured land rights (Runge 1981; Ostrom 1990; McGranahan 1991; Bromley 1992; Ostrom and Gardner 1993; Baland and Platteau 1996).

Under the private ownership every owner must exert effort to protect his/her property. In the case of forest resource management, we expect that the cost of protection will be particularly high if the resources are minor forest products, such as firewood, fodder, and grasses.1 This is because it is highly costly to identify the property from which these minor products are extracted.2 The cost of protecting forests, however, may be economized by (p.296) organizing collective action such as patrolling and guarding the community resources under the common property regime.3 It is also argued that community management will be efficient in traditional communities where people know each other and are tied by kinship and other enduring personal relationships (e.g. Hayami 1997). In such circumstances, the transaction costs of formulating management rules of community resources and enforcing them may not be too high and, hence, the use of resources may be effectively regulated. A recent theoretical study based on evolutionary game theory demonstrates that whether forest resources are depleted under a common‐property regime depends on the cost of sanction that maintains the cooperative behaviors of its members (Sethi and Somanathan 1996). Further, Aoki (2000), in this volume, argues that to the extent that the threat of social ostracism is credible, cooperation in the commons game can become the norm of behaviors in the village community.

In contrast to the development in theoretical studies, there have been relatively few empirical studies exploring the efficiency and effectiveness of local community management of natural resources.4 Given the various theoretical predictions, it is clearly an empirical question whether the community management of natural resources is effective in conserving and protecting common property resources. This chapter attempts to assess the feasibility and the effectiveness of forest resource management under the community management system based on a case study of community forests in the hill region of Nepal. First, we attempt to identify factors facilitating or impeding the formation of community management systems. Second, we explore the consequences of the community forest management on the forest conditions.

Specifically we postulate the U‐shaped changes in forest resources over time, in which deforestation is followed by reforestation. The deforestation phase can be explained by open access to forest resources coupled with increased population pressure, whereas the reforestation phase can be explained by the emergence of effective community management. This hypothesis maintains that organizing community members, introducing management rules, and enforcing them are costly, and hence community management is introduced only when the benefit of initiating such management exceeds the cost. In other words, we argue that the benefit of initiating community management exceeds the cost of setting up such a system only when forest resources have become sufficiently scarce and valuable.5

(p.297) 11.1 Geographical and Historical Background

In Nepal there are five topographical regions: to range them from low to high altitudes, they are Tarai, Siwalik, Middle Mountain, High Mountain, and High Himal. The hill region corresponds to Middle Mountain, which lies between the altitudes of 700 and 2000 meters. This region has been the center of the country, although the Tarai plain has emerged as an industrial center since the eradication of malaria in the 1950s.

Because of its rugged geographical conditions, the majority of inhabitants in the hill region of Nepal have unfavorable access to urban markets. Even now it is not exceptional to walk over hilly trails for a few days to reach one's village from the nearest market town. As a result, the main economic activity has been subsistence farming with minimum dependence on purchased inputs, such as chemical fertilizer, and the main source of household energy has been almost exclusively firewood. Furthermore, farmers depend on such minor forest products as leaf fodder and grasses to feed livestock for milking, plowing, and making composts. Compost application is indispensable to maintain the soil fertility of terraced fields located on steep slopes. Even leaf litter collected from the forest is used for bedding materials for livestock to collect nutrients contained in urine.

Historically under the feudal Rana regime from 1846 to 1950, local government officials controlled the harvesting of timber trees in the forests. Such control was particularly strict in areas near Tarai and main roads, which had favorable access to timber markets in India. Even in remote areas where market access was unfavorable, permission was required to fell timber trees for own use, such as the construction of houses.6 In contrast, the collection of minor forest products was usually unregulated, which would reflect the abundance of such resources at that time.

The feudal system, however, collapsed in 1950, and was followed by the political upheaval and the accelerated population growth. It is alleged that massive deforestation began in the 1950s with the gradual collapse of the feudal forest management system, which resulted in rampant felling of timber trees (Gilmour and Fisher 1991). In order to arrest deforestation, forest areas were nationalized in 1957. With the limited number of forestry officers, however, the government could not implement the nationalization policy effectively except in Tarai and the areas near the hill district headquarters (Graner 1997). According to our interviews with village elders, forest users did not usually perceive that their forests were nationalized (p.298) until cadastral surveys were conducted one or two decades later. Other factors, such as population pressure, would have been also responsible for deforestation (Gilmour and Fisher 1991).

While the loss of forest area has been the major form of deforestation in most developing countries (see, e.g., Deacon 1994), degradation of forest conditions has been the major concern in the hill region of Nepal. Indeed, based on aerial photographs taken in 1964–65 and 1978–79, it is reported that the loss of forest area had been negligible (Metz 1991; Soussan et al. 1995). This is because most of the cultivable land had already been converted to farmland in the hill region by the first half of the twentieth century. According to our own estimates based on aerial photographs taken in 1978 and 1992–96, the average proportion of forest areas in village development council (VDC),7 a local administrative unit, changed only from 35.2 percent to 35.7 percent. In general, no visible changes in the area of our sample forests were identified by the analysis of aerial photographs.

As will be seen, proportions of crown cover declined in many sample forests for the same period, indicating that the quality of forests has generally deteriorated. In the forests near main roads, timber extraction for market sale by villagers as well as outsiders has been a serious problem. In the majority of hill forests which have unfavorable access to markets, however, timber trees were cut primarily for the construction of houses and funerals in the villages. Other direct causes for declining crown cover were grazing, and collection of firewood and leaf fodder for own consumption. Grazing is a major threat to forests, as cattle eat and step on seedlings and coppices. While the collection of dead and dry branches does not deteriorate the forest condition, cutting live branches for firewood and fodder results in major damage. As the population increased, the intensity of grazing and the extraction of firewood and fodder increased, resulting in degradation of forest conditions.

The shortage of forest resources led to a reduction in the size of livestock and increased the time allocated to collecting forest resources from distant forests, which had traditionally been used by different communities, such as hamlets (Kumar and Hotchkiss 1988). Such invasion on other forests has often created conflict among people from different communities. Responding to continued deforestation and increasing conflicts, some communities spontaneously began to form informal forest user groups to manage forests, even though the government possessed legal ownership titles. It should be noted that such informal user groups were formed based on customary access rights to forest, but not on residence within the political boundaries. Thus, it is not uncommon to find cases in (p.299) which several hamlets from different VDCs form an informal user group.8 These informal user‐groups began to exclude outsiders from the use of their own forests and to regulate the extraction of timber and other forest products by the user group members.9 Messerschmidt (1990) and Gilmour and Fisher (1991), among others, indicate the burgeoning success of such indigenous forest management systems based on the results of several case studies in the early stages of their evolution.

Having observed favorable effects of forest‐user group management on forest conditions, the government of Nepal officially declared the promotion of community management of forests in 1987 (Nepal 1988). Since then, the government has handed over use rights of forests to forest user groups, if they demonstrate their willingness and ability to manage forests effectively to the district forest officer (DFO). The submission of a ‘constitution’ and a ‘management plan’ are required, which specify the rules of forest management agreed upon by all community members. Once handed‐over, formal forest user groups are allowed to sell forest products to the markets, even though we seldom observed significant sale of those products. Furthermore, the Forest Act in 1993 granted the formal user groups the legal right to punish violators of the constitution, which seems to have further strengthened the community management.

11.2 Basic Characteristics of Sample Forests

Nepal consists of five development regions: eastern, central, western, mid‐western, and far‐western. Initially we have randomly sampled 100 forests from the whole hill region in terms of the relative importance of forest area in each region compared with the total hill area.10 The minimum forest size of 10 hectares was imposed in order to apply aerial photo analysis with sufficient accuracy. Fifty forests were chosen from the non‐remote areas and another fifty from remote areas; remoteness being defined by a distance of more than 10km from a main road and 15km from district capitals. Roughly speaking, non‐remote forests can be reached within a (p.300)

Table 11.1 Basic Characteristics of Sample Forests

Non‐remote forest

Remote forests

No. of sample forests

38

36

Traveling time to (hours):

Nearest local market

1.04

6.24

Nearest ranger office

1.29

5.90

Size of forest (ha)

86.1

117.8

No. of user households

187.9

132.8

Average forest area per household (ha)

0.46

0.89

Proportion of user group management (%)

92.1

50.0

Proportion of formal user groups (%)

74.3

33.3

half‐day's walk from the main road. Because of the inconsistency and the unreliability of some data, this study focuses on 74 sample forests.11

As is shown in Table 11.1, non‐remote forests are located substantially closer to local markets and offices of forest rangers, who are in charge of local forest management. The average size of community forests in non‐remote areas is about 70 percent of those in remote areas. In contrast, the number of user households is much larger in non‐remote areas. As a result, the average forest area per user household in non‐remote areas is nearly one‐half of that in remote areas. Therefore, population pressure on forest resources tend to be higher in non‐remote areas. Such areas, however, have better access to alternative sources of energy for firewood, such as kerosene and biogas, and to chemical fertilizer, which is a major substitute for compost. While population pressure will stimulate the emergence of spontaneous community forest management, favorable access to markets may hinder it.

According to Table 11.1, the proportion of sample forests managed by user groups, either formally or informally, is much higher in non‐remote forests.12 This may be explained by the deteriorating forest conditions in non‐remote areas due to severe population pressure that led to the formation of forest user groups. As a matter of fact, the average crown cover in 1978 was 39 percent and 49 percent in non‐remote and remote sample forests, respectively. According to the result of our group interviews, leaders of about 70 percent of forest user groups stated that the shortage of forest resources was the single most important reason for initiating user group management.

(p.301) Another important factor affecting the formation of user group management may be the direct intervention by the DFO. An informal forest user group is a voluntary organization, which manages forests officially owned by the government. Since no formal use rights are granted, the informal group may face difficulties in excluding both outsiders and inside violators. In order to register forest user groups or to acquire the official use rights of forests from the government, an informal user group has to submit a management plan to the DFO in accordance with the guidelines prepared by the Ministry of Forestry. Once approved, the forest is handed over to the user group and secure use rights are granted, unless it grossly violates government regulations, such as felling mature timber trees without permission. The proportion of registered forests managed by formal user groups is definitely higher in non‐remote areas (Table 11.1). This seems to reflect the fact that because of budget and manpower constraints, DFOs have primarily assisted those user groups which are easily accessible.

Local DFO staff also provide technical support for the forest management, including the provision of seedlings for rehabilitation of degraded forests. DFOs occasionally visit the formally managed forests to check forest conditions and user group activities. If they detect violations of management rules, it is expected that they assist user group committees to implement punishment. In fact, users of the registered forests generally perceive that the enforcement of the rules of community forest management has become more effective since the forest was registered. Another major role of DFOs is to examine whether there are any serious conflicts of management rules among forest users proposed by informal forest user groups. The DFOs refuse to hand‐over forests to user groups if conflicts are found. As is pointed out by Aoki (2000) and Baland and Platteau (1996), it is clear that the introduction of the third party, i.e. the DFO, has changed the nature of community management games.

As shown in Table 11.2, the relatively small number of informal user groups were formed before the mid‐1980s. Since the change in government

Table 11.2 The Number of User Groups Initiating Informal and Formal Management by Period

Informal

Formal

1966–70

4

1971–75

1

1976–80

4

1981–85

3

1986–90

20

1

1991–97

21

31

Total

53

32

(p.302) policy in favor of community user group management in 1987, not only were a large number of informal user group management schemes initiated, but also a sizable number of formal user groups were set up. The initiation of some user group management may have been inspired by the opportunity to acquire forest use rights. In fact, several user groups were set up as formal user groups from the beginning with the direct assistance of the DFO, even though the majority of formal user groups had been converted from informal groups.

11.3 Forest Conditions and Management Rules

In the standard economic model of common property management (e.g. Dasgupta and Heal 1979), the crux of the management issue is how to prevent socially excessive exploitation of resources arising from the failure of coordination among resource users. In the reality of Nepal, over‐exploitation has already taken place and forests have been seriously degraded. At present, a more important management issue is how to regenerate forest resources rather than simply protect them. Thus, the extraction of scarce forest resources is not only regulated but also often prohibited by user groups in the short run.

One of the critical questions is whether and to what extent the policy of handing‐over forest use rights to the community is effective in enhancing the management efficiency of forest user groups. In practice, informal user groups often perceive that the local forests, which they have used for many years, belong to them. Based on such perception, they voluntarily undertook forest management. Thus, some researchers argue that informal user groups have been successful in protecting forests in a sustainable manner, even though they do not possess official use rights (e.g. Fisher 1989; Fox 1993). Others, however, disagree (e.g. Gilmour et al. 1989). According to our field interviews, management rules are more effectively enforced under formal user group management than its informal counterparts, primarily because the enforcement of management rules is supported, implicitly or explicitly, by the authority of the DFO.

We conducted an analysis of aerial photos taken in 1978 and 1992–96, in order to obtain information on general forest conditions and their changes over time. The most important indicator of the forest conditions generated by the aerial photo analysis is the proportion of crown cover. The results, however, were not wholly satisfactory, primarily because it is difficult to distinguish between the canopies of big trees and short bushy trees generated by coppices of felled trees. As a result, relatively small differences and changes were found in crown cover across forests under no management, informal and formal user group management and over time (see Table 11.3). A closer examination suggests, however, that forest conditions (p.303)

Table 11.3 Forest Conditions in 1978, 1992–96, and 1997 by Management Regime in 1997

No user group

Informal

Formal

No. of forests

21

21

32

Crown cover (%):

1978

42.4

31.4

31.9*

1992–96

38.6

33.3

28.1*

Biomass per ha in 1997 (ton/ha)

143

135

131

No. of regenerated trees per 100 m2 in 1997

31.0

33.0

47.4*

Note: (*) indicates that there is significant difference in means between no user group management and informal or formal management at the 5 percent level.

under no user group management deteriorated more rapidly, even though they were more favorable than those under informal and formal user group management in both 1978 and recent years. These observations are consistent with our hypothesis that deforestation is followed by reforestation due to the formation of user group management. It appears that, though inaccurate, crown cover broadly captures the forest conditions. On the other hand, there is no clear indication that forest conditions were improved under formal user group management. Recall, however, that formal user group management was initiated mostly in the 1990s, so that the impact of formal user group management might not have been revealed in terms of changes in crown cover from 1978 to 1992–96. This is another limitation of the use of aerial photos for our analysis.

In the hill region of Nepal, the most important forest resources for local users are not timber but minor forest products, such as firewood, leaf fodder, and leaf litter. It is therefore appropriate to use the biomass of all trees, rather than timber volume, for the analysis of management of forest resources in this region. Thus, we conducted a forest inventory assessment in which we directly measured the density and distribution of various tree species of different sizes in sample forests. We also measured the rate of regeneration of young trees, which are shorter than breast height. Given the facts that user group management was initiated relatively recently and that the major means of restoring the forest conditions is to generate coppices by strict protection, the rate of regeneration seems to be a more appropriate indicator of the effects of user group management than biomass.

It must be pointed out that the measurement of the density and distribution of various trees in community forests in the hill region of Nepal is not an easy task, so we applied a sampling formula for the selection of sample plots. We anticipate that some measurement errors are inevitable, particularly (p.304) when forests are large and their conditions are not uniform (see Rayamajhi and Pokharel (1998) for details of forest measurement).

It is interesting to find that the average biomass per hectare was greatest under no user group management in 1997 (Table 11.3). In terms of the regeneration of trees, however, formal management seems to perform best. It is likely that the regeneration rate was low under informal management essentially because forests were so degraded that recovery requires both time and strict protection. On the other hand, regeneration was also inactive under no user group management despite the large stock of biomass, which is likely to reflect the absence of management. In contrast, the regeneration rate is high under formal management despite a meager endowment of biomass. Such differences can be explained by the effect of formal user group management on the regeneration of young trees. Overall, the above observations seem consistent with the hypothesis of U‐shaped changes in forest conditions over time, in which forest conditions deteriorate when forest resources are abundant and improve after user group management, particularly that of a formal one, is initiated.

The restrictions on cutting timber trees by user groups have been effectively enforced, as they are bulky resources and, hence, the cost of their protection is relatively low. Major management rules on the extraction of minor forest products are either a restriction on the number of days on which specific members of user groups are allowed to collect forest resources or to totally prohibit use for the time being to facilitate regeneration. Slightly less than one‐half of user groups employ watchers or adopt rotational patrolling, and mutual supervision is attempted everywhere under any user group management. While a restriction on the collection of dead and dry branches is important to prevent the excessive collection of firewood, more important regulations for the regeneration and recovery of forest conditions are restrictions on cutting green branches and grazing livestock. Cutting green branches and young trees except for deformed ones, very low branches, and those trees grown too densely, must be strictly controlled for the successful regeneration of trees.

Table 11.4 summarizes the regulations on the extraction of minor forest products implemented by informal and formal user groups. The numbers simply indicate whether the user groups adopt any regulations or not, as it is difficult to assess the severity or intensity of regulations. For example, we cannot always assume that a total prohibition of resource extraction is stricter than regulated extraction of the resources as a management rule, because it is often desirable to remove some trees, which are densely grown, and low‐lying branches to improve forest conditions.

Several important observations can be made from Table 11.4. First, as is expected, no management rule is adopted in the forests without a user group. This indicates that forest is open access, at least for community members, unless user group management is introduced. Second, it is clear (p.305)

Table 11.4 Frequency of Sample Forests Adopting Regulations of Forest Resource Extraction/Use by Management Regime in 1997

No user group

Informal

Formal

No. of sample forests

21

21

32

Regulations on (%):

Green firewood collection

0 (0)

19 (91)

32 (100)

Dead firewood collection

0 (0)

3 (14)

14 (44)

Grazing

0 (0)

8 (38)

12 (38)

that all the formal groups, as well as almost all the informal groups, adopt some regulations on the extraction of green firewood, which include prohibitions and regulated extractions.13 Since such regulations are of critical importance for the protection of forests, not only formal but also informal user groups implement some regulations. Thus, we cannot distinguish between the two management regimes in terms of the frequency of regulations for green firewood extraction. Third, the regulation of the collecting of dead branches is less strict than that of green firewood, as it does not affect the forest conditions of the future. Fourth, the collection of dead branches is seldom regulated in informally managed forests, whereas it is regulated in nearly one‐half of the formally managed forests. Thus, it seems reasonable to hypothesize that formal user groups manage forests more strictly than informal user groups by adopting stricter management rules, even though both of them manage forests to some extent, unlike the case of no user group management.14

(p.306) Grazing is sometimes prohibited or allowed only in areas where trees are mature under user group management. Even if grazing is not explicitly prohibited, grazing is not feasible if the forests are ‘closed,’ as will be explained shortly. While regulations of grazing have become stricter over time, regulations of firewood collection have remained largely the same.

There are several ways to regulate forest resource extraction. Generally, the prohibition implies that nobody is allowed to enter the forest except for management purposes organized by the user group committee, rather than simply prohibiting cutting of trees and green branches. Similarly, the rules of regulated use specify ‘open days’ during which specific groups of user households are allowed to collect resources under the supervision of forest user group committee members. Potentially these regulations can be effectively enforced, as those who enter the forest on ‘closed days’ are in principle immediately caught (see, e.g., Chhetri and Randy 1992).15 The extraction of forest resources by a group of people on the same day facilitates mutual supervision. Furthermore, the extracted resources are usually piled up in one location at the end of the open day and are shared equally among the participating households in most cases. This practice reduces incentives to extract forest resources excessively.16 The same regulations were widely employed in community forest management in pre‐war Japan (see, e.g. McKean 1992).

Punishment rules differ from community to community. If the illegal extraction of firewood is found by the forest watchers and if this is a first offence, a fine may be imposed which is equivalent to the value of the extracted firewood. Second and third offences are subject to escalated punishments, which are stated in the management rules in the case of the formal management. In practice, repeated offences seldom take place or even if they do take place, the escalated clauses with severe punishments may not be applied to fellow community members. According to the field interviews with the committee members of several user groups, illegal activities have been reduced after handing‐over, because such activities are more likely to be punished strictly due to the involvement of local DFO staff in forest management. In general, the use of watchers is less common and punishment is less strict in the case of informal forest management.

While concrete evidence is hard to come by, we believe that community norms play a significant role in the enforcement of community management rules. Since villages are geographically isolated in the hill region of Nepal, social interactions among villagers are intense. Farmers collectively participate not only in the management of forests but also in the management (p.307) of local schools, irrigation and drinking water systems, and festivals, funerals, and religious ceremonies. As Aoki argues (2000), linking management ‘games’ with social activities generating social surplus may lead to a situation where cooperative behaviors become the community norm. Yet, in practice, cooperation in forest management is seldom overlapped with cooperation in irrigation and drinking water management or other spheres, such as the management of schools. This is because the relevant groups for the management of forest and water resources are generally different; while forest management tends to be carried out by a certain number of wards, other social activities are often carried out by a single ward or even hamlets within a ward. It seems that although such linkages are likely to play a role in reducing non‐cooperative behaviors in forest management, it is difficult to identify the specific social activities that are closely linked with forest management.

11.4 Determinants of the Emergence of User Group Management

In order to identify the determinants of the emergence of user group management, we employed two methods of empirical estimation. First, we estimated a probit function in which the dependent variable assumes unity when any types of forest user groups were set up. We did not distinguish between informal and formal user group management in this estimation, because we suspect that the initiation of formal user group management was affected critically by the intention of DFOs, which may be considered exogenous for user groups. Second, we estimated an ordered‐probit function in which the dependent variable is a rank order dummy: 1 for informal and 2 for formal user group management. This method attempts to identify factors affecting not only the formation of informal user groups but also their conversion to formal user groups. The same set of explanatory variables are used and include the following: the number of user households; the size of forest per user household; the number of local administration units involved in a single user group; the ratio of households participating in other social activities; homogeneity of social groups represented by ethnic composition; traveling time to the nearest ranger office; crown cover in 1978; and the ratio of forest area in VDCs in 1978 in which our sample forests are located.17

The number of user households, for which we took a logarithm in the actual estimation, is supposed to capture the transaction costs of organizing collective forest management, which is related to the cost of reaching agreement. If the fixed cost component of setting up user group management is (p.308) high, the large forest area may have a positive effect on the initiation of user group management, partly because the total benefit of forest management will be larger and partly because scale advantages may exist in protection activities. In order to take into account the effect of forest area in addition to the effect of the size of user households, we added a logarithm of forest area per user household. If the transaction costs of community forest management increase with the size of user group and forest area, we expect to have negative effects of these two variables. We observed, however, that large community forests were sometimes divided into smaller units and were managed separately by different groups of people.18 To the extent that such division of a forest area is effective in reducing the transaction costs, both the number of households and the forest area per household may not exert a negative and significant influence on the formation of user groups. It must be also pointed out that the data on the number of user households would suffer from measurement errors, particularly in forests under no user group management or open access, because anyone can freely use the resources of open access forests.19 Also note that due to data limitation, we had to use current‐period variables for the size of user groups and a few other variables to explain changes in forest management institutions in the past.

The transaction costs of user group management would be affected not only by the size of the group and forest area but also by its ethnic homogeneity as well as homogeneity of administrative units. We used the ratio of the dominant ethnic group to represent the effect of ethnic homogeneity among the users. It is usually presumed that collective action can be more easily organized among homogeneous members. In fact, leaders of 85 percent of user groups stated that the user groups were formed by the decisions of users at large rather than by the initiative of selected leaders. Group homogeneity, however, may imply the lack of leadership. We also considered the number of wards (the lowest local administrative unit in Nepal), from which the users come to the forest, as an indicator of group heterogeneity. Since the ward is an administrative unit, which roughly corresponds to a village in other societies, the transaction cost within a ward tends to be low. Thus, we expect that the fewer the number of wards, the lower would be the cost of collective management.

(p.309) We considered the two variables to represent the forest conditions and access to forest resources in the earlier period. One is crown cover in 1978 obtained from aerial photo analysis. Although this indicator is not very sensitive to changes and differences in forest conditions, it seems to capture major changes and differences. We hypothesize that the lower crown cover in 1978 would lead to a greater probability of inducing the initiation of user group management. Another variable is the ratio of forest area in 1978 in the VDC in which the sample forest is located. The larger the ratio of forest area, the lower would have been the population pressure on forest resources in the sample forests. This is because, even if sample forest was degraded, users might have easily gone to other forests in nearby locations, if the ratio of forest area in the VDC was high. Thus, we expect that a higher forest ratio leads to a lower incidence of user group management.

We used the ratio of households participating in other social activities (i.e. management of irrigation, drinking water systems, and local schools) to capture the effects of possible linkages of forest management with other social activities. Finally, we included traveling time to a ranger office as a variable representing the effects of the DFO. If the formation of user group management is truly voluntary, this variable will be insignificant. To the extent that user groups were set up with the expectation of receiving official user rights or with the direct assistance from the DFO, this variable would have a negative effect on the formation of user groups.

The estimation results of probit and ordered‐probit regressions are shown in the first two columns and the last column of Table 11.5, respectively. It should be pointed out that the estimated coefficients in the probit estimation are converted into marginal effects computed at the means of regressors, while those in the ordered‐probit estimation are not. Thus, simple comparison of the estimation results by probit and ordered‐probit regressions does not make any sense. The marginal effects of regressors on predicted probabilities of ordered‐probit estimation are summarized in Table 11.6. It is found that the number of user households has a weakly significant effect, if we apply two‐tailed t‐test, whereas forest area per user household is insignificant. It appears that the size of user group per se, measured by the number of households and the forest area, does not decisively affect the cost of organizing the collective action and enforcing management rules. It may well be that both scale advantages and disadvantages coexist in forest management.20

The number of wards has negative and significant coefficients, suggesting that the larger the administrative heterogeneity, the larger will be the (p.310)

Table 11.5 Determinants of the Emergence of Forest User Group Management: Probit and Ordered‐Probit Regressionsa

Probitb

Probitb

Ordered‐probitc

Intercept

0.235 (0.482)

0.209 (0.428)

−0.178 (0.132)

Ln no. of households

0.154* (1.708)

0.146 (1.636)

0.453 (1.650)

Ln forest area per household

0.050 (0.689)

0.045 (.643)

−0.011 (0.043)

No. of wards

−0.100** (2.181)

−.100** (2.166)

−0.244* (1.748)

Ratio of households in other community activities

−0.096 (0.522)

Proportion of largest ethnic group

0.041 (0.135)

0.019 (.063)

0.337 (.391)

Traveling hours to ranger office

−0.030** (2.152)

−0.029** (2.100)

−0.123*** (3.637)

Crown cover in 1978

−0.129** (2.204)

−0.123** (2.146)

−.193 (1.166)

Ratio of forest land in VDC in 1978

−0.209 (0.550)

−0.231 (0.609)

−0.048 (0.040)

Log likelihood

−34.95

−35.08

−68.71

Percentage of correct prediction

73.0

73.0

54.0

Notes: (a) The numbers in parentheses are t‐statistics. ***indicates significance at the 1 percent level, **at the 5 percent level, and *at the 10 percent level.

(b) Dependent variable is unity where either informal or formal management has been introduced.

(c) Ranked choices of formal (=2), informal (=1), and no community management (=0) are considered as dependent variables.

Table 11.6 Marginal Effects for Ordered Probit Regression

No management

Informal user management

Formal user management

Intercept

0.055

0.014

−0.069

Ln no. of households

−0.141

−0.036

0.177

Ln forest area per household

0.003

0.001

−0.004

No. of wards

0.076

0.019

−0.096

Proportion of largest ethnic group

−0.105

−0.027

0.132

Traveling hours to ranger office

0.038

0.010

−0.048

Crown cover in 1978

0.060

0.015

−0.075

Ratio of forest land in VDC in 1978

0.015

0.004

−0.019

(p.311) cost of organizing collective action.21 On the other hand, the ratio of households participating in other social activities and the proportion of the largest ethnic group are insignificant, indicating that the presence of other community activities and ethnic homogeneity in terms of dominant caste do not directly affect the initiation of forest management. The former result was expected because, as was noted above, the sphere of community cooperation for such social activities as irrigation and school management is usually different from that for forest management.

Traveling hours to a ranger office is significant in the probit regression and highly significant in the ordered‐probit regression. The former finding indicates that the nearer the ranger office to the forest, the more likely it was that user group management was initiated. Furthermore, the higher significance in the ordered‐probit regression suggests that proximity to a ranger office was particularly conducive to the initiation of formal user group management. In fact, Table 11.6 shows that the marginal effect of traveling hours to a ranger office is small for the formation of an informal group, but negative and comparably larger for the conversion from an informal to a formal user group. It seems clear that the influence of the DFO on the formation of formal user groups has been pervasive.

Crown cover in 1978 has negative and significant coefficients in the probit regression, implying that forest user group management was more likely to be initiated in forests which were more degraded in 1978. This supports our hypothesis that deforestation leads to the formation of forest user groups. This is also consistent with a finding of Fujita et al. (1999) that collective irrigation management in the Philippines is more likely to be organized effectively where water supply is scarce. This variable, however, is insignificant in the ordered‐probit regression. This insignificance and the negative sign of the marginal effect for formal user group management seem to suggest that crown cover in 1978 did not significantly affect the transformation of informal user groups into formal groups.

The ratio of forest area in a VDC in 1978 is not significant. It appears that it was not merely the ratio of the forest area but the availability of forest resources that affected the formation of user group management.

11.5 Consequences of User Group Management

There are several difficulties in properly assessing the effects of forest user group management on forest conditions. First of all, the representative (p.312) indicators of forest conditions (timber volume, biomass, etc.) are stock variables, which change slowly over time. Considering that user group management was initiated relatively recently, we believe that the rate of regeneration of young trees is a more appropriate indicator of the improvement of forest conditions. Second, management rules are highly complex: there are a variety of management rules from no regulation to complete prohibition of resource extraction and they vary from one resource to another. It is therefore very difficult to construct a single indicator of regulation intensity. Third, management rules have often changed over time. Ideally we have to assess the effect of each management rule with due consideration of the periods during which such rule was implemented. Finally, there are potential estimation problems associated with the endogen‐ous nature of management rules and institutions. As we have seen in the previous section, the initiation of user group management was likely to be induced by a scarcity of tree resources associated with deforestation. To the extent that degraded forest conditions deter the regeneration of trees and foster the formation of user groups, the impact of user group management on the regeneration of trees may appear negative in the regression analysis, even if it actually contributed to the rehabilitation of forests. In order to avoid this, we have to control for the effects of the forest conditions properly. The forest conditions, however, are also endogenous. In actual estimation, we used the ratio of forest area in a VDC in 1992–96 as a variable representing the pressure on forest conditions in the neighborhood of our sample community forest. The other estimation problem caused by the problem of endogeneity is that there is a possibility that the more effective informal user groups tend to become formal user groups. In that case, even if formal user group management is found to be more effective than informal user group management, it may be due to a self‐selection effect rather than the effect of receiving legal use rights from the government with the support from the DFO for community management. Since DFOs do not accept the applications for handing over if the users have any conflicts, this self‐selection problem is likely to be pervasive.

In this study, as a first step toward coping with the complexities of management rules, we used dummy variables for informal and formal management in regression analyses. As was indicated in the previous section, the current management system reflects the effects of the intervention of the DFO, forest conditions, and other factors. In the estimation of the effects of management systems on forest conditions, we include the proportion of forest area in the VDC in 1992–96, the logarithm of average altitude, and pine forest dummy, in addition to dummy variables for the management system. The proportion of forest area in the VDC in 1992–96 is supposed to capture the external pressure on the sample forests. If there are only a few forests in the neighborhood, many outsiders are likely to invade the forest under investigation. The latter two variables are included to control for the effects (p.313) of the ecological conditions; the regeneration of trees becomes less active in the higher altitude, and the presence of pine trees deters regeneration.

In the first two columns of Table 11.7, we report the results of ordinary least squares (OLS) and weighted least squares (WLS) estimation, in which the dependent variable is the logarithm of the number of regenerated trees per hectare. The number of regenerated trees is a weighted average with weights in proportion to the height of regenerated trees. In the weighted least squares regression, we used the number of sample plots in the forest inventory as weights. Since R 2 is not a proper indicator of the goodness of fit in weighted regressions, we report the correlation coefficients between the actual and estimated dependent variables in the table.

The fits of these regressions are not very high, which is likely to reflect the measurement errors of the regeneration rate of young trees. Except for the proportion of forest area, however, all the independent variables have the expected signs. The two ecological variables have generally significant effects on regeneration. The most important finding is that the coefficients of the formal forest‐user group (FFU) dummy are positive and highly significant, while those of the informal forest‐user group (IFU) dummy are positive but only marginally significant according to a one‐tailed t‐test. Furthermore, the former coefficients are numerically larger than the latter

Table 11.7 Determinants of Regeneration Per Hectare in 1997a

Model I

Model IIb

OLS

WLSc

OLS

WLSc

Constant

12.809 (7.156)

10.772 (6.249)

11.909 (5.590)

9.675 (4.549)

IFU dummy

0.264 (1.191)

0.360* (1.657)

0.428 (1.357)

0.545** (1.784)

FFU dummy

0.571*** (2.746)

0.651*** (3.289)

0.839** (2.003)

0.936*** (2.395)

Proportion of forest area in VDC, 1992/96

−0.484 (0.916)

−0.339 (0.691)

−0.471 (0.918)

−0.409 (0.845)

Ln average altitude

−0.640*** (2.534)

−0.374* (1.534)

−0.538** (1.905)

−0.241 (0.844)

Pine forest dummy

−0.364** (2.131)

−0.356** (2.166)

−0.357** (2.149)

−0.347** (2.162)

R 2

0.323

0.301

0.310

0.280

Correlation coefficients

0.571

0.559

0.562

0.534

Notes: (a) Dependent variable is Ln (regeneration per hectare). The numbers in parentheses are t‐statistics. ***indicates significance at the 1 percent level, ** at the 5 percent level, and * at the 10 percent level.

(b) Predicted values are used for formal forest user (FFU) dummy.

(c) Weighted least squares regression with weight being the number of sample plots in forest measurement.

(p.314) coefficients; the estimated coefficients of around 0.6 for formal user group management and around 0.3 for informal management indicate that the formal and informal management systems increases the rate of regeneration of young trees by approximately 80 percent and 35 percent, respectively.22 These results support the hypothesis that formal user group management is more efficient than informal management in improving forest conditions. These findings are consistent with the findings of Upadhyaya and Otsuka (1998): using the data of firewood collection from five user households in each of our sample forest, they find that the regulation of cutting green firewood by formal user groups was significantly effective in reducing the firewood collection by user households.

As was pointed out above, one may question that the significant effect of a formal forest‐user group is due to self‐selection of effective user groups. In order to check this self‐selection bias of formal user groups, we have tried the two‐step estimation known as the treatment effects model (Greene 1997: 981). Based on the assumption that self‐selection occurs only for the formation of formal user groups, we used the probit estimation for formal management as a first step estimation. The explanatory variables are the same as those which appeared in Table 11.5. The predicted dummy variable for formal user group management is used as an instrument in the second‐stage estimation for the rate of regeneration. The results are reported in the third and fourth columns of Table 11.7. The two‐step estimation does not result in a much greater difference in the qualitative results compared with those from the simple OLS and WLS estimation. That is, the effect of the formal forest‐user group dummy is positive and significant even after removing the possible self‐selection bias.

11.6 Concluding Remarks

This study attempted to explore the causes and the consequences of forest user group management based on case studies in the hill region of Nepal. Despite the difficulty in measuring the conditions of forests, their changes over time, and the size of user group members, we obtained some useful empirical evidence. First, we found that the formation of user groups was induced by a shortage of forest resources. This finding suggests that when forest resources are abundant, user group management is less likely to be practiced, which leads to deforestation and shortages of forest resources. The degraded forest conditions, in turn, stimulate the formation of user (p.315) groups, because it is costly to sustain livelihoods and subsistence farming without a sufficient supply of forest resources in the hill region of Nepal.

Second, there is no indication that the size of forest and the size of user group members hindered the initiation and the effectiveness of user group management. Unlike the usual presumption, the size of management unit does not seem to increase the transaction cost of user group management significantly. It may well be that not only scale disadvantages but also scale advantages exist in the protection, as a small number of people can enforce protection of large forest areas.

Third and most importantly, we found that management of community forests by formal user groups seems to be more effective in regenerating trees than management by informal user groups, not to mention the case of no management. It seems clear that the formation of formal user groups, which are granted official use rights of forests and are supported by the local forestry department, is critically important for the effective management of community forests. In fact, our field surveys strongly indicate that handing‐over can be an effective policy instrument, only if it is supplemented by the supporting activities of the local forestry department. In other words, community initiatives and government supporting activities are complements but not substitutes.

Thus, community management will not function well without governmental support. Even with such support it will not function well either, if the forest resources are not scarce, unless truly strong support measures are provided. Herein lies the dilemma of forest management policy: in order to prevent deforestation as soon as it begins to appear, the government needs to allocate large efforts and resources to community forest management.

References

Bibliography references:

Aoki, M. (2000) ‘Community Norms and Embeddedness: A Game‐Theoretic Approach,’ in this volume, Chapter 4.

Arnold, J. E. M. (2000) ‘Devolution of Control of Common Pool Resources to Local Communities: Experiences in Forestry,’ in de Janvry, A., Platteau, J.‐P., and Sadoulet, E. (eds.), Access to Land, Rural Poverty, and Public Action, Oxford University Press, Oxford.

Baland, J.‐M. and Platteau, J.‐P. (1996) Halting Degradation of Natural Resources: Is There a Role of Rural Communities? Clarendon Press, Oxford.

Bardhan, P. K. (1993) ‘Symposium on Management of Local Commons,’ Journal of Economic Perspectives, 7: 87–92.

—— (1999) ‘Water Community: An Empirical Analysis of Cooperation on Irrigation in South India,’ in this volume, Chapter 9.

Bromley, D. W. (1992) ‘The Commons, Property, and Common‐Property Regimes,’ in Bromley, D. W. (ed.), Making the Commons Work, ICS Press, San Francisco.

(p.316) Chhetri, R. B. and Randy, T. R. (1992) User Group Forestry in the Far‐Western Region of Nepal, International Centre for Integrated Mountain Development, Kathmandu.

Dasgupta, P. S. and Heal, G. M. (1979) Economic Theory and Exhaustible Resources, Cambridge University Press, Cambridge.

Deacon, R. T. (1994) ‘Deforestation and the Rule of Law in a Cross‐Section of Countries,’ Land Economics, 70: 414–30.

Fisher, R. J. (1989) ‘Indigenous Systems of Common Property Forest Management in Nepal,’ EAPI Working Paper No. 18, Environment and Policy Institute, East–West Center, Hawaii.

FAO (Food and Agriculture Organization of the United Nations) (1989) Community Forestry, Rome.

Fox, J. (1993) ‘Forest Resources in a Nepali Village in 1980 and 1990: The Positive Influence of Population Growth,’ Mountain Research and Development, 18: 89–98.

Fujita, M., Hayami, Y., and Kikuchi, M. (1999) ‘The Conditions of Farmer Participation in Irrigation Management: A Cross‐Section Analysis for the Philippines,’ mimeo, International Rice Research Institute, Los Banos, Philippines.

Gilmour, D. A. and Fisher, R. J. (1991) Villagers, Forests, and Foresters, Sahayogi Press, Kathmandu.

—— King, G. C., and Hobley, M. (1989) ‘Management of Forests for Local Use in the Hills of Nepal: Part 1, Changing Forest Management Paradigm,’ Journal of World Forest Resource Management, 4: 93–110.

Graner, E. (1997) The Political Ecology of Community Forestry in Nepal, Verlag fur Entwicklungspolitik, Saarbrucken.

Greene, W. H. (1997) Econometric Analysis, 3rd edn., Prentice Hall, New Jersey.

Hayami, Y. (1997) Development Economics: From the Poverty to the Wealth of Nations, Clarendon Press, Oxford.

Kijima, Y., Sakurai, T., and Otsuka, K. (2000) ‘Iriaichi: Collective vs. Individualized Management of Community Forests in Post‐War Japan,’ Economic Development and Cultural Change.

Kumar, S. K., and Hotchkiss, D. (1988) ‘Consequences of Deforestation for Women's Time Allocation, Agricultural Production, and Nutrition in Hill Areas of Nepal,’ IFPRI Research Report No. 69, International Food Policy Research Institute, Washington, DC.

McGranahan, G. (1991) ‘Fuelwood, Subsistence Foraging, and the Decline of Common Property,’ World Development, 19: 1275–87.

McKean, M. A. (1992) ‘Management of Traditional Common Lands (Iriaichi) in Japan,’ in Bromley, D. W. (ed.), Making the Commons Work: Theory, Practice, and Policy, ICS Press, San Francisco.

Messerschmidt, D. A. (1990) ‘Indigenous Environmental Management and Adaptation: An Introduction to Four Case Studies from Nepal,’ Mountain Research and Development, 10: 3–4.

Metz, J. J. (1991) ‘A Reassessment of the Causes and Severity of Nepal's Environmental Crisis,’ World Development, 19: 805–20.

Nepal, Ministry of Forests and Soil Conservation (1988) Master Plan for the Forestry Sector, Nepal, Kathmandu.

(p.317) Ostrom, E. (1990) Governing the Commons: The Evolution of Institutions for Collective Action, Cambridge University Press, Cambridge.

—— and Gardner, R. (1993) ‘Coping with Asymmetries in the Commons: Self‐Governing Irrigation System Can Work,’ Journal of Economic Perspectives, 7: 93–112.

Pearce, D. G. (1992) ‘Repeated Games: Cooperation and Rationality,’ in Laffont, J.‐J. (ed.), Advances in Economic Theory, Cambridge University Press, Cambridge.

Rayamajhi, S. and Pokharel, R. (1998) ‘From Deforestation to Reforestation: Common‐Property Forest Management in the Hill Region of Nepal,’ a report submitted to the Environment and Production Technology Division, International Food Policy Research Institute, Washington, DC.

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.

Sakurai, T., Pokharel, R., Rayamajhi, S., and Otsuka, K. (1998) ‘Private, Collective, and Centralized Community Management: A Comparative Study of Timber Forest and Plantation Management in Inner Tarai of Nepal,’ a paper presented at the workshop on Land Tenure and the Management of Land and Trees, Tokyo Metropolitan University, Tokyo.

Sethi, R. and Somanathan, E. (1996) ‘The Evolution of Social Norms in Common Property Resource Use,’ American Economic Review, 86: 766–88.

Soussan, J., Shrestha, B. K., and Uprety, L. A. (1995) The Social Dynamics of Deforestation: A Case Study from Nepal, Parthenon Publishing Group, New York.

Stevenson, G. G. (1991) Common Property Economics: A General Theory and Land Use Applications, Cambridge University Press, Cambridge.

Upadhyaya, H. K. and Otsuka, K. (1998) ‘Community Forest Management in Hill Region of Nepal,’ a paper presented at the workshop on Land Tenure and the Management of Land and Trees, Tokyo Metropolitan University, Tokyo.

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

World Bank (1996) World Bank Participation Source Book, World Bank, Washington, DC.

Notes:

We would like to thank Bob Allen, Yujiro Hayami, and other participants of the World Bank Conference in Stanford (1999). We are also grateful to our collaborators in Nepal, Ridish Pokharel, Santosh Rayamajhi, and Hari Upadhyaya. This research was partly supported by a grant from the Japan Economic Research Foundation.

(1) Tiny private forests are developed near farmers' residence in some places and fodder trees are often planted on edges of private fields in the hill region of Nepal, because it is easy to oversee such forests and plantations.

(2) It is likely that the cost of protection will be lower if major forest products are timber, because felling and hauling timber trees illegally may well be detected by someone in the village community. Moreover, the production of high‐quality timber trees requires silvicultural operations such as pruning, singling, thinning, and weeding. Kijima et al. (2000) demonstrate that timber trees were more actively planted and thinned under individualized management than traditional collective management in post‐war Japan. Similar findings are made in timber forests and plantations in the inner Tarai region of Nepal by Sakurai et al. (1998).

(3) Community ownership or a common property system is defined as a system whereby resources are subtractable like private goods, and are owned and used jointly by a group of people. It is generally presumed that the exclusion of users is costly to a varying extent. See Ostrom (1990), who proposes to call resources with imperfect excludability common‐pool resources.

(4) See Wade (1988), Stevenson (1991), Sakurai et al. (1998), and Kijima et al. (2000), Bardhan (Chapter 9 in this volume), and Fujita et al. (1999), among others. See also Arnold (2000) for a recent survey of the literature on community forest management.

(5) Our hypothesis is consistent with the induced institutional innovation hypothesis (Hayami 1997) which argues that the increasing scarcity of a resource induces an innovation to save that resource relative to other resources.

(6) It is not clear how strictly the regulation on timber harvesting was applied to the remote hill regions. Graner (1997, ch. 2) suggests that the Rana regime attached great importance to the forests even in the hill region. Gilmour and Fisher (1991, ch. 1) argue, however, that the central government of Rana promoted forest clearance in most areas of the hill region and protected limited forest areas as a source of fuel for manufacturing arms. In our field interviews, many older respondents recalled that, with required payment, it was not too difficult for them to obtain permission for timber harvesting during the Rana regime.

(7) The local administrative hierarchy in Nepal consists, from larger to smaller units, of development region, district, VDC, and ward. Usually there are several hamlets in each ward.

(8) Hamlet (tol in Nepali language) is not the geographical minimum unit for the membership of a forest user group. Though not so common, we encountered cases in which some households are members of a forest user group and others are those of another user group even within the same hamlet.

(9) While the organizational structures of the informal user group are diverse, it is common to select committee members in general meeting to entrust the management activities of forest, such as the selection of trees which are allowed to be felled. There are also cases in which the leaders of local administration are chosen to be the leaders of informal community forest management. In contrast, the formal user groups must select committee members by election.

(10) By chance, we happened to select a few forests from the lower hills called the Siwalik mountains at an altitude of 300–700 meters.

(11) In certain areas, our survey teams had difficulty in locating sample forests chosen by the use of aerial photos. In this study, we have omitted those 26 sample forests, which might not have been investigated by our survey teams by mistake with positive probabilities. We hope that this procedure will not result in a serious bias in the estimation results of regression analyses.

(12) Actually there are several informal user groups which are inactive as evidenced by the absence of management rules and activities. Based on the results of resurvey, we treated these cases as no user group management.

(13) We gathered that young trees are not allowed to be cut for firewood even in those two informal user groups which do not have explicit rules on the extraction of green firewood. In these user groups, the users can enter their forests only with their neighbor households, usually four to five, for mutual supervision. Furthermore, their forests can be easily seen from their village residential areas.

(14) In order to corroborate this hypothesis statistically, we attempted to estimate the simultaneous equation systems in which the selected dummy variable for the presence of regulation (i.e. for the collection of green firewood, and dead and dry branches) is regressed on management regimes, among others, in the first‐stage probit regression and the regeneration rate of young trees is regressed on the predicted management‐rule dummy in the second‐stage regression. This formulation, however, is not workable, because the management regime dummies used as regressors in the first‐stage probit regression almost fully ‘explain’ the dependent variable. In other words, the difficulty in estimation arises from the facts that no management rule is adopted without user groups and that either almost all formal and informal user groups adopt some regulations, as in the case of green firewood extraction, or only a very few informal user groups adopt some regulations, as in the case of dead firewood extraction. See Greene (1997: 892) on this point.

(15) We found an interesting regulation in which one tree is allocated to each household each year without specifying open and closed days. A household can cut its tree any time but only under the supervision of the committee members.

(16) However, other sharing rules seem to have been increasingly adopted, such as the sharing based simply on one's collection of firewood on open days.

(17) When we calculated the ratio of forest area in a VDC, we included the shrub area because this area also produces minor forest products.

(18) In our observation, the number of community forests implementing the division of user groups is increasing. It will be an important issue to what extent such division will continue towards the direction of management by a few or even private management.

(19) In the survey of the open access forests, we asked villagers residing near the forest the names of hamlets from which residents come to the forest under investigation. Then, we visit‐ed each hamlet and asked the number of households. This method may lead to underestimation because of the possible omission of hamlets and to overestimation because some households in hamlets are not users of the forest under investigation. According to Graner (1997), there are also cases in which user group management excluded socially and economically disadvantaged users.

(20) Suspecting that scale economies dominate in the range of small‐scale forest management operations, whereas scale diseconomies dominate in the range of large operations, we estimated regression functions which added square terms of either the number of forest users or forest areas. The estimated coefficients, however, are not significant.

(21) It must be pointed out that the number of wards may partly capture the effects of forest size and geographical conditions. If the forest is large or lies on the top of a hill and ridges, usually multiple numbers of wards have access to it. For the effective protection and management of such forest, it is necessary to organize all the wards around it, since it is difficult to protect the forest from one or a few sides.

(22) Since a logarithm is taken for the dependent variable, exponential of estimated dummy coefficients must be taken to assess the quantitative impacts of the formal and informal management institutions.